diff options
Diffstat (limited to 'src/grp-machine')
88 files changed, 19873 insertions, 0 deletions
diff --git a/src/grp-machine/GNUmakefile b/src/grp-machine/GNUmakefile new file mode 120000 index 0000000000..54fdd42278 --- /dev/null +++ b/src/grp-machine/GNUmakefile @@ -0,0 +1 @@ +../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/Makefile b/src/grp-machine/Makefile new file mode 100644 index 0000000000..98e33243b7 --- /dev/null +++ b/src/grp-machine/Makefile @@ -0,0 +1,32 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +nested.subdirs += grp-import +nested.subdirs += libmachine-core +nested.subdirs += machinectl +nested.subdirs += nss-mymachines +nested.subdirs += systemd-machined + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/GNUmakefile b/src/grp-machine/grp-import/GNUmakefile new file mode 120000 index 0000000000..95e5924740 --- /dev/null +++ b/src/grp-machine/grp-import/GNUmakefile @@ -0,0 +1 @@ +../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/Makefile b/src/grp-machine/grp-import/Makefile new file mode 100644 index 0000000000..dfa189b317 --- /dev/null +++ b/src/grp-machine/grp-import/Makefile @@ -0,0 +1,32 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +nested.subdirs += libimport +nested.subdirs += systemd-export +nested.subdirs += systemd-import +nested.subdirs += systemd-importd +nested.subdirs += systemd-pull + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/libimport/GNUmakefile b/src/grp-machine/grp-import/libimport/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/grp-import/libimport/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/libimport/Makefile b/src/grp-machine/grp-import/libimport/Makefile new file mode 100644 index 0000000000..585001b2fa --- /dev/null +++ b/src/grp-machine/grp-import/libimport/Makefile @@ -0,0 +1,41 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +manual_tests += \ + test-qcow2 + +test_qcow2_SOURCES = \ + src/import/test-qcow2.c \ + src/import/qcow2-util.c \ + src/import/qcow2-util.h + +test_qcow2_CFLAGS = \ + $(ZLIB_CFLAGS) + +test_qcow2_LDADD = \ + libsystemd-shared.la \ + $(ZLIB_LIBS) + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/libimport/import-common.c b/src/grp-machine/grp-import/libimport/import-common.c new file mode 100644 index 0000000000..b0a34d5745 --- /dev/null +++ b/src/grp-machine/grp-import/libimport/import-common.c @@ -0,0 +1,222 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sched.h> +#include <sys/prctl.h> +#include <sys/stat.h> +#include <unistd.h> + +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/capability-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/util.h" + +#include "import-common.h" + +int import_make_read_only_fd(int fd) { + int r; + + assert(fd >= 0); + + /* First, let's make this a read-only subvolume if it refers + * to a subvolume */ + r = btrfs_subvol_set_read_only_fd(fd, true); + if (r == -ENOTTY || r == -ENOTDIR || r == -EINVAL) { + struct stat st; + + /* This doesn't refer to a subvolume, or the file + * system isn't even btrfs. In that, case fall back to + * chmod()ing */ + + r = fstat(fd, &st); + if (r < 0) + return log_error_errno(errno, "Failed to stat temporary image: %m"); + + /* Drop "w" flag */ + if (fchmod(fd, st.st_mode & 07555) < 0) + return log_error_errno(errno, "Failed to chmod() final image: %m"); + + return 0; + + } else if (r < 0) + return log_error_errno(r, "Failed to make subvolume read-only: %m"); + + return 0; +} + +int import_make_read_only(const char *path) { + _cleanup_close_ int fd = 1; + + fd = open(path, O_RDONLY|O_NOCTTY|O_CLOEXEC); + if (fd < 0) + return log_error_errno(errno, "Failed to open %s: %m", path); + + return import_make_read_only_fd(fd); +} + +int import_fork_tar_x(const char *path, pid_t *ret) { + _cleanup_close_pair_ int pipefd[2] = { -1, -1 }; + pid_t pid; + int r; + + assert(path); + assert(ret); + + if (pipe2(pipefd, O_CLOEXEC) < 0) + return log_error_errno(errno, "Failed to create pipe for tar: %m"); + + pid = fork(); + if (pid < 0) + return log_error_errno(errno, "Failed to fork off tar: %m"); + + if (pid == 0) { + int null_fd; + uint64_t retain = + (1ULL << CAP_CHOWN) | + (1ULL << CAP_FOWNER) | + (1ULL << CAP_FSETID) | + (1ULL << CAP_MKNOD) | + (1ULL << CAP_SETFCAP) | + (1ULL << CAP_DAC_OVERRIDE); + + /* Child */ + + (void) reset_all_signal_handlers(); + (void) reset_signal_mask(); + assert_se(prctl(PR_SET_PDEATHSIG, SIGTERM) == 0); + + pipefd[1] = safe_close(pipefd[1]); + + if (dup2(pipefd[0], STDIN_FILENO) != STDIN_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (pipefd[0] != STDIN_FILENO) + pipefd[0] = safe_close(pipefd[0]); + + null_fd = open("/dev/null", O_WRONLY|O_NOCTTY); + if (null_fd < 0) { + log_error_errno(errno, "Failed to open /dev/null: %m"); + _exit(EXIT_FAILURE); + } + + if (dup2(null_fd, STDOUT_FILENO) != STDOUT_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (null_fd != STDOUT_FILENO) + null_fd = safe_close(null_fd); + + stdio_unset_cloexec(); + + if (unshare(CLONE_NEWNET) < 0) + log_error_errno(errno, "Failed to lock tar into network namespace, ignoring: %m"); + + r = capability_bounding_set_drop(retain, true); + if (r < 0) + log_error_errno(r, "Failed to drop capabilities, ignoring: %m"); + + execlp("tar", "tar", "--numeric-owner", "-C", path, "-px", "--xattrs", "--xattrs-include=*", NULL); + log_error_errno(errno, "Failed to execute tar: %m"); + _exit(EXIT_FAILURE); + } + + pipefd[0] = safe_close(pipefd[0]); + r = pipefd[1]; + pipefd[1] = -1; + + *ret = pid; + + return r; +} + +int import_fork_tar_c(const char *path, pid_t *ret) { + _cleanup_close_pair_ int pipefd[2] = { -1, -1 }; + pid_t pid; + int r; + + assert(path); + assert(ret); + + if (pipe2(pipefd, O_CLOEXEC) < 0) + return log_error_errno(errno, "Failed to create pipe for tar: %m"); + + pid = fork(); + if (pid < 0) + return log_error_errno(errno, "Failed to fork off tar: %m"); + + if (pid == 0) { + int null_fd; + uint64_t retain = (1ULL << CAP_DAC_OVERRIDE); + + /* Child */ + + (void) reset_all_signal_handlers(); + (void) reset_signal_mask(); + assert_se(prctl(PR_SET_PDEATHSIG, SIGTERM) == 0); + + pipefd[0] = safe_close(pipefd[0]); + + if (dup2(pipefd[1], STDOUT_FILENO) != STDOUT_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (pipefd[1] != STDOUT_FILENO) + pipefd[1] = safe_close(pipefd[1]); + + null_fd = open("/dev/null", O_RDONLY|O_NOCTTY); + if (null_fd < 0) { + log_error_errno(errno, "Failed to open /dev/null: %m"); + _exit(EXIT_FAILURE); + } + + if (dup2(null_fd, STDIN_FILENO) != STDIN_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (null_fd != STDIN_FILENO) + null_fd = safe_close(null_fd); + + stdio_unset_cloexec(); + + if (unshare(CLONE_NEWNET) < 0) + log_error_errno(errno, "Failed to lock tar into network namespace, ignoring: %m"); + + r = capability_bounding_set_drop(retain, true); + if (r < 0) + log_error_errno(r, "Failed to drop capabilities, ignoring: %m"); + + execlp("tar", "tar", "-C", path, "-c", "--xattrs", "--xattrs-include=*", ".", NULL); + log_error_errno(errno, "Failed to execute tar: %m"); + _exit(EXIT_FAILURE); + } + + pipefd[1] = safe_close(pipefd[1]); + r = pipefd[0]; + pipefd[0] = -1; + + *ret = pid; + + return r; +} diff --git a/src/grp-machine/grp-import/libimport/import-common.h b/src/grp-machine/grp-import/libimport/import-common.h new file mode 100644 index 0000000000..07d3250e71 --- /dev/null +++ b/src/grp-machine/grp-import/libimport/import-common.h @@ -0,0 +1,26 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +int import_make_read_only_fd(int fd); +int import_make_read_only(const char *path); + +int import_fork_tar_c(const char *path, pid_t *ret); +int import_fork_tar_x(const char *path, pid_t *ret); diff --git a/src/grp-machine/grp-import/libimport/import-compress.c b/src/grp-machine/grp-import/libimport/import-compress.c new file mode 100644 index 0000000000..2e116df26c --- /dev/null +++ b/src/grp-machine/grp-import/libimport/import-compress.c @@ -0,0 +1,470 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "systemd-basic/string-table.h" +#include "systemd-basic/util.h" + +#include "import-compress.h" + +void import_compress_free(ImportCompress *c) { + assert(c); + + if (c->type == IMPORT_COMPRESS_XZ) + lzma_end(&c->xz); + else if (c->type == IMPORT_COMPRESS_GZIP) { + if (c->encoding) + deflateEnd(&c->gzip); + else + inflateEnd(&c->gzip); + } else if (c->type == IMPORT_COMPRESS_BZIP2) { + if (c->encoding) + BZ2_bzCompressEnd(&c->bzip2); + else + BZ2_bzDecompressEnd(&c->bzip2); + } + + c->type = IMPORT_COMPRESS_UNKNOWN; +} + +int import_uncompress_detect(ImportCompress *c, const void *data, size_t size) { + static const uint8_t xz_signature[] = { + 0xfd, '7', 'z', 'X', 'Z', 0x00 + }; + static const uint8_t gzip_signature[] = { + 0x1f, 0x8b + }; + static const uint8_t bzip2_signature[] = { + 'B', 'Z', 'h' + }; + + int r; + + assert(c); + + if (c->type != IMPORT_COMPRESS_UNKNOWN) + return 1; + + if (size < MAX3(sizeof(xz_signature), + sizeof(gzip_signature), + sizeof(bzip2_signature))) + return 0; + + assert(data); + + if (memcmp(data, xz_signature, sizeof(xz_signature)) == 0) { + lzma_ret xzr; + + xzr = lzma_stream_decoder(&c->xz, UINT64_MAX, LZMA_TELL_UNSUPPORTED_CHECK); + if (xzr != LZMA_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_XZ; + + } else if (memcmp(data, gzip_signature, sizeof(gzip_signature)) == 0) { + r = inflateInit2(&c->gzip, 15+16); + if (r != Z_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_GZIP; + + } else if (memcmp(data, bzip2_signature, sizeof(bzip2_signature)) == 0) { + r = BZ2_bzDecompressInit(&c->bzip2, 0, 0); + if (r != BZ_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_BZIP2; + } else + c->type = IMPORT_COMPRESS_UNCOMPRESSED; + + c->encoding = false; + + return 1; +} + +int import_uncompress(ImportCompress *c, const void *data, size_t size, ImportCompressCallback callback, void *userdata) { + int r; + + assert(c); + assert(callback); + + r = import_uncompress_detect(c, data, size); + if (r <= 0) + return r; + + if (c->encoding) + return -EINVAL; + + if (size <= 0) + return 1; + + assert(data); + + switch (c->type) { + + case IMPORT_COMPRESS_UNCOMPRESSED: + r = callback(data, size, userdata); + if (r < 0) + return r; + + break; + + case IMPORT_COMPRESS_XZ: + c->xz.next_in = data; + c->xz.avail_in = size; + + while (c->xz.avail_in > 0) { + uint8_t buffer[16 * 1024]; + lzma_ret lzr; + + c->xz.next_out = buffer; + c->xz.avail_out = sizeof(buffer); + + lzr = lzma_code(&c->xz, LZMA_RUN); + if (lzr != LZMA_OK && lzr != LZMA_STREAM_END) + return -EIO; + + r = callback(buffer, sizeof(buffer) - c->xz.avail_out, userdata); + if (r < 0) + return r; + } + + break; + + case IMPORT_COMPRESS_GZIP: + c->gzip.next_in = (void*) data; + c->gzip.avail_in = size; + + while (c->gzip.avail_in > 0) { + uint8_t buffer[16 * 1024]; + + c->gzip.next_out = buffer; + c->gzip.avail_out = sizeof(buffer); + + r = inflate(&c->gzip, Z_NO_FLUSH); + if (r != Z_OK && r != Z_STREAM_END) + return -EIO; + + r = callback(buffer, sizeof(buffer) - c->gzip.avail_out, userdata); + if (r < 0) + return r; + } + + break; + + case IMPORT_COMPRESS_BZIP2: + c->bzip2.next_in = (void*) data; + c->bzip2.avail_in = size; + + while (c->bzip2.avail_in > 0) { + uint8_t buffer[16 * 1024]; + + c->bzip2.next_out = (char*) buffer; + c->bzip2.avail_out = sizeof(buffer); + + r = BZ2_bzDecompress(&c->bzip2); + if (r != BZ_OK && r != BZ_STREAM_END) + return -EIO; + + r = callback(buffer, sizeof(buffer) - c->bzip2.avail_out, userdata); + if (r < 0) + return r; + } + + break; + + default: + assert_not_reached("Unknown compression"); + } + + return 1; +} + +int import_compress_init(ImportCompress *c, ImportCompressType t) { + int r; + + assert(c); + + switch (t) { + + case IMPORT_COMPRESS_XZ: { + lzma_ret xzr; + + xzr = lzma_easy_encoder(&c->xz, LZMA_PRESET_DEFAULT, LZMA_CHECK_CRC64); + if (xzr != LZMA_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_XZ; + break; + } + + case IMPORT_COMPRESS_GZIP: + r = deflateInit2(&c->gzip, Z_DEFAULT_COMPRESSION, Z_DEFLATED, 15 + 16, 8, Z_DEFAULT_STRATEGY); + if (r != Z_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_GZIP; + break; + + case IMPORT_COMPRESS_BZIP2: + r = BZ2_bzCompressInit(&c->bzip2, 9, 0, 0); + if (r != BZ_OK) + return -EIO; + + c->type = IMPORT_COMPRESS_BZIP2; + break; + + case IMPORT_COMPRESS_UNCOMPRESSED: + c->type = IMPORT_COMPRESS_UNCOMPRESSED; + break; + + default: + return -EOPNOTSUPP; + } + + c->encoding = true; + return 0; +} + +static int enlarge_buffer(void **buffer, size_t *buffer_size, size_t *buffer_allocated) { + size_t l; + void *p; + + if (*buffer_allocated > *buffer_size) + return 0; + + l = MAX(16*1024U, (*buffer_size * 2)); + p = realloc(*buffer, l); + if (!p) + return -ENOMEM; + + *buffer = p; + *buffer_allocated = l; + + return 1; +} + +int import_compress(ImportCompress *c, const void *data, size_t size, void **buffer, size_t *buffer_size, size_t *buffer_allocated) { + int r; + + assert(c); + assert(buffer); + assert(buffer_size); + assert(buffer_allocated); + + if (!c->encoding) + return -EINVAL; + + if (size <= 0) + return 0; + + assert(data); + + *buffer_size = 0; + + switch (c->type) { + + case IMPORT_COMPRESS_XZ: + + c->xz.next_in = data; + c->xz.avail_in = size; + + while (c->xz.avail_in > 0) { + lzma_ret lzr; + + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->xz.next_out = (uint8_t*) *buffer + *buffer_size; + c->xz.avail_out = *buffer_allocated - *buffer_size; + + lzr = lzma_code(&c->xz, LZMA_RUN); + if (lzr != LZMA_OK) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->xz.avail_out; + } + + break; + + case IMPORT_COMPRESS_GZIP: + + c->gzip.next_in = (void*) data; + c->gzip.avail_in = size; + + while (c->gzip.avail_in > 0) { + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->gzip.next_out = (uint8_t*) *buffer + *buffer_size; + c->gzip.avail_out = *buffer_allocated - *buffer_size; + + r = deflate(&c->gzip, Z_NO_FLUSH); + if (r != Z_OK) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->gzip.avail_out; + } + + break; + + case IMPORT_COMPRESS_BZIP2: + + c->bzip2.next_in = (void*) data; + c->bzip2.avail_in = size; + + while (c->bzip2.avail_in > 0) { + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->bzip2.next_out = (void*) ((uint8_t*) *buffer + *buffer_size); + c->bzip2.avail_out = *buffer_allocated - *buffer_size; + + r = BZ2_bzCompress(&c->bzip2, BZ_RUN); + if (r != BZ_RUN_OK) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->bzip2.avail_out; + } + + break; + + case IMPORT_COMPRESS_UNCOMPRESSED: + + if (*buffer_allocated < size) { + void *p; + + p = realloc(*buffer, size); + if (!p) + return -ENOMEM; + + *buffer = p; + *buffer_allocated = size; + } + + memcpy(*buffer, data, size); + *buffer_size = size; + break; + + default: + return -EOPNOTSUPP; + } + + return 0; +} + +int import_compress_finish(ImportCompress *c, void **buffer, size_t *buffer_size, size_t *buffer_allocated) { + int r; + + assert(c); + assert(buffer); + assert(buffer_size); + assert(buffer_allocated); + + if (!c->encoding) + return -EINVAL; + + *buffer_size = 0; + + switch (c->type) { + + case IMPORT_COMPRESS_XZ: { + lzma_ret lzr; + + c->xz.avail_in = 0; + + do { + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->xz.next_out = (uint8_t*) *buffer + *buffer_size; + c->xz.avail_out = *buffer_allocated - *buffer_size; + + lzr = lzma_code(&c->xz, LZMA_FINISH); + if (lzr != LZMA_OK && lzr != LZMA_STREAM_END) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->xz.avail_out; + } while (lzr != LZMA_STREAM_END); + + break; + } + + case IMPORT_COMPRESS_GZIP: + c->gzip.avail_in = 0; + + do { + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->gzip.next_out = (uint8_t*) *buffer + *buffer_size; + c->gzip.avail_out = *buffer_allocated - *buffer_size; + + r = deflate(&c->gzip, Z_FINISH); + if (r != Z_OK && r != Z_STREAM_END) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->gzip.avail_out; + } while (r != Z_STREAM_END); + + break; + + case IMPORT_COMPRESS_BZIP2: + c->bzip2.avail_in = 0; + + do { + r = enlarge_buffer(buffer, buffer_size, buffer_allocated); + if (r < 0) + return r; + + c->bzip2.next_out = (void*) ((uint8_t*) *buffer + *buffer_size); + c->bzip2.avail_out = *buffer_allocated - *buffer_size; + + r = BZ2_bzCompress(&c->bzip2, BZ_FINISH); + if (r != BZ_FINISH_OK && r != BZ_STREAM_END) + return -EIO; + + *buffer_size += (*buffer_allocated - *buffer_size) - c->bzip2.avail_out; + } while (r != BZ_STREAM_END); + + break; + + case IMPORT_COMPRESS_UNCOMPRESSED: + break; + + default: + return -EOPNOTSUPP; + } + + return 0; +} + +static const char* const import_compress_type_table[_IMPORT_COMPRESS_TYPE_MAX] = { + [IMPORT_COMPRESS_UNKNOWN] = "unknown", + [IMPORT_COMPRESS_UNCOMPRESSED] = "uncompressed", + [IMPORT_COMPRESS_XZ] = "xz", + [IMPORT_COMPRESS_GZIP] = "gzip", + [IMPORT_COMPRESS_BZIP2] = "bzip2", +}; + +DEFINE_STRING_TABLE_LOOKUP(import_compress_type, ImportCompressType); diff --git a/src/grp-machine/grp-import/libimport/import-compress.h b/src/grp-machine/grp-import/libimport/import-compress.h new file mode 100644 index 0000000000..025dd030be --- /dev/null +++ b/src/grp-machine/grp-import/libimport/import-compress.h @@ -0,0 +1,61 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <bzlib.h> +#include <lzma.h> +#include <sys/types.h> +#include <zlib.h> + +#include "systemd-basic/macro.h" + +typedef enum ImportCompressType { + IMPORT_COMPRESS_UNKNOWN, + IMPORT_COMPRESS_UNCOMPRESSED, + IMPORT_COMPRESS_XZ, + IMPORT_COMPRESS_GZIP, + IMPORT_COMPRESS_BZIP2, + _IMPORT_COMPRESS_TYPE_MAX, + _IMPORT_COMPRESS_TYPE_INVALID = -1, +} ImportCompressType; + +typedef struct ImportCompress { + ImportCompressType type; + bool encoding; + union { + lzma_stream xz; + z_stream gzip; + bz_stream bzip2; + }; +} ImportCompress; + +typedef int (*ImportCompressCallback)(const void *data, size_t size, void *userdata); + +void import_compress_free(ImportCompress *c); + +int import_uncompress_detect(ImportCompress *c, const void *data, size_t size); +int import_uncompress(ImportCompress *c, const void *data, size_t size, ImportCompressCallback callback, void *userdata); + +int import_compress_init(ImportCompress *c, ImportCompressType t); +int import_compress(ImportCompress *c, const void *data, size_t size, void **buffer, size_t *buffer_size, size_t *buffer_allocated); +int import_compress_finish(ImportCompress *c, void **buffer, size_t *buffer_size, size_t *buffer_allocated); + +const char* import_compress_type_to_string(ImportCompressType t) _const_; +ImportCompressType import_compress_type_from_string(const char *s) _pure_; diff --git a/src/grp-machine/grp-import/libimport/qcow2-util.c b/src/grp-machine/grp-import/libimport/qcow2-util.c new file mode 100644 index 0000000000..4c81fbebc7 --- /dev/null +++ b/src/grp-machine/grp-import/libimport/qcow2-util.c @@ -0,0 +1,353 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <zlib.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/sparse-endian.h" +#include "systemd-basic/util.h" + +#include "qcow2-util.h" + +#define QCOW2_MAGIC 0x514649fb + +#define QCOW2_COPIED (1ULL << 63) +#define QCOW2_COMPRESSED (1ULL << 62) +#define QCOW2_ZERO (1ULL << 0) + +typedef struct _packed_ Header { + be32_t magic; + be32_t version; + + be64_t backing_file_offset; + be32_t backing_file_size; + + be32_t cluster_bits; + be64_t size; + be32_t crypt_method; + + be32_t l1_size; + be64_t l1_table_offset; + + be64_t refcount_table_offset; + be32_t refcount_table_clusters; + + be32_t nb_snapshots; + be64_t snapshots_offset; + + /* The remainder is only present on QCOW3 */ + be64_t incompatible_features; + be64_t compatible_features; + be64_t autoclear_features; + + be32_t refcount_order; + be32_t header_length; +} Header; + +#define HEADER_MAGIC(header) be32toh((header)->magic) +#define HEADER_VERSION(header) be32toh((header)->version) +#define HEADER_CLUSTER_BITS(header) be32toh((header)->cluster_bits) +#define HEADER_CLUSTER_SIZE(header) (1ULL << HEADER_CLUSTER_BITS(header)) +#define HEADER_L2_BITS(header) (HEADER_CLUSTER_BITS(header) - 3) +#define HEADER_SIZE(header) be64toh((header)->size) +#define HEADER_CRYPT_METHOD(header) be32toh((header)->crypt_method) +#define HEADER_L1_SIZE(header) be32toh((header)->l1_size) +#define HEADER_L2_SIZE(header) (HEADER_CLUSTER_SIZE(header)/sizeof(uint64_t)) +#define HEADER_L1_TABLE_OFFSET(header) be64toh((header)->l1_table_offset) + +static uint32_t HEADER_HEADER_LENGTH(const Header *h) { + if (HEADER_VERSION(h) < 3) + return offsetof(Header, incompatible_features); + + return be32toh(h->header_length); +} + +static int copy_cluster( + int sfd, uint64_t soffset, + int dfd, uint64_t doffset, + uint64_t cluster_size, + void *buffer) { + + ssize_t l; + int r; + + r = btrfs_clone_range(sfd, soffset, dfd, doffset, cluster_size); + if (r >= 0) + return r; + + l = pread(sfd, buffer, cluster_size, soffset); + if (l < 0) + return -errno; + if ((uint64_t) l != cluster_size) + return -EIO; + + l = pwrite(dfd, buffer, cluster_size, doffset); + if (l < 0) + return -errno; + if ((uint64_t) l != cluster_size) + return -EIO; + + return 0; +} + +static int decompress_cluster( + int sfd, uint64_t soffset, + int dfd, uint64_t doffset, + uint64_t compressed_size, + uint64_t cluster_size, + void *buffer1, + void *buffer2) { + + _cleanup_free_ void *large_buffer = NULL; + z_stream s = {}; + uint64_t sz; + ssize_t l; + int r; + + if (compressed_size > cluster_size) { + /* The usual cluster buffer doesn't suffice, let's + * allocate a larger one, temporarily */ + + large_buffer = malloc(compressed_size); + if (!large_buffer) + return -ENOMEM; + + buffer1 = large_buffer; + } + + l = pread(sfd, buffer1, compressed_size, soffset); + if (l < 0) + return -errno; + if ((uint64_t) l != compressed_size) + return -EIO; + + s.next_in = buffer1; + s.avail_in = compressed_size; + s.next_out = buffer2; + s.avail_out = cluster_size; + + r = inflateInit2(&s, -12); + if (r != Z_OK) + return -EIO; + + r = inflate(&s, Z_FINISH); + sz = (uint8_t*) s.next_out - (uint8_t*) buffer2; + inflateEnd(&s); + if (r != Z_STREAM_END || sz != cluster_size) + return -EIO; + + l = pwrite(dfd, buffer2, cluster_size, doffset); + if (l < 0) + return -errno; + if ((uint64_t) l != cluster_size) + return -EIO; + + return 0; +} + +static int normalize_offset( + const Header *header, + uint64_t p, + uint64_t *ret, + bool *compressed, + uint64_t *compressed_size) { + + uint64_t q; + + q = be64toh(p); + + if (q & QCOW2_COMPRESSED) { + uint64_t sz, csize_shift, csize_mask; + + if (!compressed) + return -EOPNOTSUPP; + + csize_shift = 64 - 2 - (HEADER_CLUSTER_BITS(header) - 8); + csize_mask = (1ULL << (HEADER_CLUSTER_BITS(header) - 8)) - 1; + sz = (((q >> csize_shift) & csize_mask) + 1) * 512 - (q & 511); + q &= ((1ULL << csize_shift) - 1); + + if (compressed_size) + *compressed_size = sz; + + *compressed = true; + + } else { + if (compressed) { + *compressed = false; + *compressed_size = 0; + } + + if (q & QCOW2_ZERO) { + /* We make no distinction between zero blocks and holes */ + *ret = 0; + return 0; + } + + q &= ~QCOW2_COPIED; + } + + *ret = q; + return q > 0; /* returns positive if not a hole */ +} + +static int verify_header(const Header *header) { + assert(header); + + if (HEADER_MAGIC(header) != QCOW2_MAGIC) + return -EBADMSG; + + if (HEADER_VERSION(header) != 2 && + HEADER_VERSION(header) != 3) + return -EOPNOTSUPP; + + if (HEADER_CRYPT_METHOD(header) != 0) + return -EOPNOTSUPP; + + if (HEADER_CLUSTER_BITS(header) < 9) /* 512K */ + return -EBADMSG; + + if (HEADER_CLUSTER_BITS(header) > 21) /* 2MB */ + return -EBADMSG; + + if (HEADER_SIZE(header) % HEADER_CLUSTER_SIZE(header) != 0) + return -EBADMSG; + + if (HEADER_L1_SIZE(header) > 32*1024*1024) /* 32MB */ + return -EBADMSG; + + if (HEADER_VERSION(header) == 3) { + + if (header->incompatible_features != 0) + return -EOPNOTSUPP; + + if (HEADER_HEADER_LENGTH(header) < sizeof(Header)) + return -EBADMSG; + } + + return 0; +} + +int qcow2_convert(int qcow2_fd, int raw_fd) { + _cleanup_free_ void *buffer1 = NULL, *buffer2 = NULL; + _cleanup_free_ be64_t *l1_table = NULL, *l2_table = NULL; + uint64_t sz, i; + Header header; + ssize_t l; + int r; + + l = pread(qcow2_fd, &header, sizeof(header), 0); + if (l < 0) + return -errno; + if (l != sizeof(header)) + return -EIO; + + r = verify_header(&header); + if (r < 0) + return r; + + l1_table = new(be64_t, HEADER_L1_SIZE(&header)); + if (!l1_table) + return -ENOMEM; + + l2_table = malloc(HEADER_CLUSTER_SIZE(&header)); + if (!l2_table) + return -ENOMEM; + + buffer1 = malloc(HEADER_CLUSTER_SIZE(&header)); + if (!buffer1) + return -ENOMEM; + + buffer2 = malloc(HEADER_CLUSTER_SIZE(&header)); + if (!buffer2) + return -ENOMEM; + + /* Empty the file if it exists, we rely on zero bits */ + if (ftruncate(raw_fd, 0) < 0) + return -errno; + + if (ftruncate(raw_fd, HEADER_SIZE(&header)) < 0) + return -errno; + + sz = sizeof(uint64_t) * HEADER_L1_SIZE(&header); + l = pread(qcow2_fd, l1_table, sz, HEADER_L1_TABLE_OFFSET(&header)); + if (l < 0) + return -errno; + if ((uint64_t) l != sz) + return -EIO; + + for (i = 0; i < HEADER_L1_SIZE(&header); i ++) { + uint64_t l2_begin, j; + + r = normalize_offset(&header, l1_table[i], &l2_begin, NULL, NULL); + if (r < 0) + return r; + if (r == 0) + continue; + + l = pread(qcow2_fd, l2_table, HEADER_CLUSTER_SIZE(&header), l2_begin); + if (l < 0) + return -errno; + if ((uint64_t) l != HEADER_CLUSTER_SIZE(&header)) + return -EIO; + + for (j = 0; j < HEADER_L2_SIZE(&header); j++) { + uint64_t data_begin, p, compressed_size; + bool compressed; + + p = ((i << HEADER_L2_BITS(&header)) + j) << HEADER_CLUSTER_BITS(&header); + + r = normalize_offset(&header, l2_table[j], &data_begin, &compressed, &compressed_size); + if (r < 0) + return r; + if (r == 0) + continue; + + if (compressed) + r = decompress_cluster( + qcow2_fd, data_begin, + raw_fd, p, + compressed_size, HEADER_CLUSTER_SIZE(&header), + buffer1, buffer2); + else + r = copy_cluster( + qcow2_fd, data_begin, + raw_fd, p, + HEADER_CLUSTER_SIZE(&header), buffer1); + if (r < 0) + return r; + } + } + + return 0; +} + +int qcow2_detect(int fd) { + be32_t id; + ssize_t l; + + l = pread(fd, &id, sizeof(id), 0); + if (l < 0) + return -errno; + if (l != sizeof(id)) + return -EIO; + + return htobe32(QCOW2_MAGIC) == id; +} diff --git a/src/grp-machine/grp-import/libimport/qcow2-util.h b/src/grp-machine/grp-import/libimport/qcow2-util.h new file mode 100644 index 0000000000..6dddac8cdf --- /dev/null +++ b/src/grp-machine/grp-import/libimport/qcow2-util.h @@ -0,0 +1,23 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +int qcow2_detect(int fd); +int qcow2_convert(int qcow2_fd, int raw_fd); diff --git a/src/grp-machine/grp-import/libimport/test-qcow2.c b/src/grp-machine/grp-import/libimport/test-qcow2.c new file mode 100644 index 0000000000..7c973970ba --- /dev/null +++ b/src/grp-machine/grp-import/libimport/test-qcow2.c @@ -0,0 +1,54 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "systemd-basic/fd-util.h" +#include "systemd-basic/log.h" +#include "systemd-basic/util.h" + +#include "qcow2-util.h" + +int main(int argc, char *argv[]) { + _cleanup_close_ int sfd = -1, dfd = -1; + int r; + + if (argc != 3) { + log_error("Needs two arguments."); + return EXIT_FAILURE; + } + + sfd = open(argv[1], O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (sfd < 0) { + log_error_errno(errno, "Can't open source file: %m"); + return EXIT_FAILURE; + } + + dfd = open(argv[2], O_WRONLY|O_CREAT|O_CLOEXEC|O_NOCTTY, 0666); + if (dfd < 0) { + log_error_errno(errno, "Can't open destination file: %m"); + return EXIT_FAILURE; + } + + r = qcow2_convert(sfd, dfd); + if (r < 0) { + log_error_errno(r, "Failed to unpack: %m"); + return EXIT_FAILURE; + } + + return EXIT_SUCCESS; +} diff --git a/src/grp-machine/grp-import/systemd-export/GNUmakefile b/src/grp-machine/grp-import/systemd-export/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/systemd-export/Makefile b/src/grp-machine/grp-import/systemd-export/Makefile new file mode 100644 index 0000000000..fa6306226f --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/Makefile @@ -0,0 +1,50 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +rootlibexec_PROGRAMS += systemd-export + +systemd_export_SOURCES = \ + src/import/export.c \ + src/import/export-tar.c \ + src/import/export-tar.h \ + src/import/export-raw.c \ + src/import/export-raw.h \ + src/import/import-common.c \ + src/import/import-common.h \ + src/import/import-compress.c \ + src/import/import-compress.h + +systemd_export_CFLAGS = \ + $(XZ_CFLAGS) \ + $(ZLIB_CFLAGS) \ + $(BZIP2_CFLAGS) + +systemd_export_LDADD = \ + libsystemd-shared.la \ + $(XZ_LIBS) \ + $(ZLIB_LIBS) \ + $(BZIP2_LIBS) + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/systemd-export/export-raw.c b/src/grp-machine/grp-import/systemd-export/export-raw.c new file mode 100644 index 0000000000..64d0aa6197 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/export-raw.c @@ -0,0 +1,352 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sys/sendfile.h> + +/* When we include libgen.h because we need dirname() we immediately + * undefine basename() since libgen.h defines it as a macro to the POSIX + * version which is really broken. We prefer GNU basename(). */ +#include <libgen.h> +#undef basename + +#include <systemd/sd-daemon.h> + +#include "import-common.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/missing.h" +#include "systemd-basic/ratelimit.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/util.h" + +#include "export-raw.h" + +#define COPY_BUFFER_SIZE (16*1024) + +struct RawExport { + sd_event *event; + + RawExportFinished on_finished; + void *userdata; + + char *path; + + int input_fd; + int output_fd; + + ImportCompress compress; + + sd_event_source *output_event_source; + + void *buffer; + size_t buffer_size; + size_t buffer_allocated; + + uint64_t written_compressed; + uint64_t written_uncompressed; + + unsigned last_percent; + RateLimit progress_rate_limit; + + struct stat st; + + bool eof; + bool tried_reflink; + bool tried_sendfile; +}; + +RawExport *raw_export_unref(RawExport *e) { + if (!e) + return NULL; + + sd_event_source_unref(e->output_event_source); + + import_compress_free(&e->compress); + + sd_event_unref(e->event); + + safe_close(e->input_fd); + + free(e->buffer); + free(e->path); + return mfree(e); +} + +int raw_export_new( + RawExport **ret, + sd_event *event, + RawExportFinished on_finished, + void *userdata) { + + _cleanup_(raw_export_unrefp) RawExport *e = NULL; + int r; + + assert(ret); + + e = new0(RawExport, 1); + if (!e) + return -ENOMEM; + + e->output_fd = e->input_fd = -1; + e->on_finished = on_finished; + e->userdata = userdata; + + RATELIMIT_INIT(e->progress_rate_limit, 100 * USEC_PER_MSEC, 1); + e->last_percent = (unsigned) -1; + + if (event) + e->event = sd_event_ref(event); + else { + r = sd_event_default(&e->event); + if (r < 0) + return r; + } + + *ret = e; + e = NULL; + + return 0; +} + +static void raw_export_report_progress(RawExport *e) { + unsigned percent; + assert(e); + + if (e->written_uncompressed >= (uint64_t) e->st.st_size) + percent = 100; + else + percent = (unsigned) ((e->written_uncompressed * UINT64_C(100)) / (uint64_t) e->st.st_size); + + if (percent == e->last_percent) + return; + + if (!ratelimit_test(&e->progress_rate_limit)) + return; + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_info("Exported %u%%.", percent); + + e->last_percent = percent; +} + +static int raw_export_process(RawExport *e) { + ssize_t l; + int r; + + assert(e); + + if (!e->tried_reflink && e->compress.type == IMPORT_COMPRESS_UNCOMPRESSED) { + + /* If we shall take an uncompressed snapshot we can + * reflink source to destination directly. Let's see + * if this works. */ + + r = btrfs_reflink(e->input_fd, e->output_fd); + if (r >= 0) { + r = 0; + goto finish; + } + + e->tried_reflink = true; + } + + if (!e->tried_sendfile && e->compress.type == IMPORT_COMPRESS_UNCOMPRESSED) { + + l = sendfile(e->output_fd, e->input_fd, NULL, COPY_BUFFER_SIZE); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + e->tried_sendfile = true; + } else if (l == 0) { + r = 0; + goto finish; + } else { + e->written_uncompressed += l; + e->written_compressed += l; + + raw_export_report_progress(e); + + return 0; + } + } + + while (e->buffer_size <= 0) { + uint8_t input[COPY_BUFFER_SIZE]; + + if (e->eof) { + r = 0; + goto finish; + } + + l = read(e->input_fd, input, sizeof(input)); + if (l < 0) { + r = log_error_errno(errno, "Failed to read raw file: %m"); + goto finish; + } + + if (l == 0) { + e->eof = true; + r = import_compress_finish(&e->compress, &e->buffer, &e->buffer_size, &e->buffer_allocated); + } else { + e->written_uncompressed += l; + r = import_compress(&e->compress, input, l, &e->buffer, &e->buffer_size, &e->buffer_allocated); + } + if (r < 0) { + r = log_error_errno(r, "Failed to encode: %m"); + goto finish; + } + } + + l = write(e->output_fd, e->buffer, e->buffer_size); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + r = log_error_errno(errno, "Failed to write output file: %m"); + goto finish; + } + + assert((size_t) l <= e->buffer_size); + memmove(e->buffer, (uint8_t*) e->buffer + l, e->buffer_size - l); + e->buffer_size -= l; + e->written_compressed += l; + + raw_export_report_progress(e); + + return 0; + +finish: + if (r >= 0) { + (void) copy_times(e->input_fd, e->output_fd); + (void) copy_xattr(e->input_fd, e->output_fd); + } + + if (e->on_finished) + e->on_finished(e, r, e->userdata); + else + sd_event_exit(e->event, r); + + return 0; +} + +static int raw_export_on_output(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + RawExport *i = userdata; + + return raw_export_process(i); +} + +static int raw_export_on_defer(sd_event_source *s, void *userdata) { + RawExport *i = userdata; + + return raw_export_process(i); +} + +static int reflink_snapshot(int fd, const char *path) { + char *p, *d; + int new_fd, r; + + p = strdupa(path); + d = dirname(p); + + new_fd = open(d, O_TMPFILE|O_CLOEXEC|O_NOCTTY|O_RDWR, 0600); + if (new_fd < 0) { + _cleanup_free_ char *t = NULL; + + r = tempfn_random(path, NULL, &t); + if (r < 0) + return r; + + new_fd = open(t, O_CLOEXEC|O_CREAT|O_NOCTTY|O_RDWR, 0600); + if (new_fd < 0) + return -errno; + + (void) unlink(t); + } + + r = btrfs_reflink(fd, new_fd); + if (r < 0) { + safe_close(new_fd); + return r; + } + + return new_fd; +} + +int raw_export_start(RawExport *e, const char *path, int fd, ImportCompressType compress) { + _cleanup_close_ int sfd = -1, tfd = -1; + int r; + + assert(e); + assert(path); + assert(fd >= 0); + assert(compress < _IMPORT_COMPRESS_TYPE_MAX); + assert(compress != IMPORT_COMPRESS_UNKNOWN); + + if (e->output_fd >= 0) + return -EBUSY; + + r = fd_nonblock(fd, true); + if (r < 0) + return r; + + r = free_and_strdup(&e->path, path); + if (r < 0) + return r; + + sfd = open(path, O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (sfd < 0) + return -errno; + + if (fstat(sfd, &e->st) < 0) + return -errno; + if (!S_ISREG(e->st.st_mode)) + return -ENOTTY; + + /* Try to take a reflink snapshot of the file, if we can t make the export atomic */ + tfd = reflink_snapshot(sfd, path); + if (tfd >= 0) { + e->input_fd = tfd; + tfd = -1; + } else { + e->input_fd = sfd; + sfd = -1; + } + + r = import_compress_init(&e->compress, compress); + if (r < 0) + return r; + + r = sd_event_add_io(e->event, &e->output_event_source, fd, EPOLLOUT, raw_export_on_output, e); + if (r == -EPERM) { + r = sd_event_add_defer(e->event, &e->output_event_source, raw_export_on_defer, e); + if (r < 0) + return r; + + r = sd_event_source_set_enabled(e->output_event_source, SD_EVENT_ON); + } + if (r < 0) + return r; + + e->output_fd = fd; + return r; +} diff --git a/src/grp-machine/grp-import/systemd-export/export-raw.h b/src/grp-machine/grp-import/systemd-export/export-raw.h new file mode 100644 index 0000000000..89aef76eed --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/export-raw.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "import-compress.h" +#include "systemd-basic/macro.h" + +typedef struct RawExport RawExport; +typedef void (*RawExportFinished)(RawExport *export, int error, void *userdata); + +int raw_export_new(RawExport **export, sd_event *event, RawExportFinished on_finished, void *userdata); +RawExport* raw_export_unref(RawExport *export); + +DEFINE_TRIVIAL_CLEANUP_FUNC(RawExport*, raw_export_unref); + +int raw_export_start(RawExport *export, const char *path, int fd, ImportCompressType compress); diff --git a/src/grp-machine/grp-import/systemd-export/export-tar.c b/src/grp-machine/grp-import/systemd-export/export-tar.c new file mode 100644 index 0000000000..541f033c3e --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/export-tar.c @@ -0,0 +1,327 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-daemon.h> + +#include "import-common.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/ratelimit.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/util.h" + +#include "export-tar.h" + +#define COPY_BUFFER_SIZE (16*1024) + +struct TarExport { + sd_event *event; + + TarExportFinished on_finished; + void *userdata; + + char *path; + char *temp_path; + + int output_fd; + int tar_fd; + + ImportCompress compress; + + sd_event_source *output_event_source; + + void *buffer; + size_t buffer_size; + size_t buffer_allocated; + + uint64_t written_compressed; + uint64_t written_uncompressed; + + pid_t tar_pid; + + struct stat st; + uint64_t quota_referenced; + + unsigned last_percent; + RateLimit progress_rate_limit; + + bool eof; + bool tried_splice; +}; + +TarExport *tar_export_unref(TarExport *e) { + if (!e) + return NULL; + + sd_event_source_unref(e->output_event_source); + + if (e->tar_pid > 1) { + (void) kill_and_sigcont(e->tar_pid, SIGKILL); + (void) wait_for_terminate(e->tar_pid, NULL); + } + + if (e->temp_path) { + (void) btrfs_subvol_remove(e->temp_path, BTRFS_REMOVE_QUOTA); + free(e->temp_path); + } + + import_compress_free(&e->compress); + + sd_event_unref(e->event); + + safe_close(e->tar_fd); + + free(e->buffer); + free(e->path); + return mfree(e); +} + +int tar_export_new( + TarExport **ret, + sd_event *event, + TarExportFinished on_finished, + void *userdata) { + + _cleanup_(tar_export_unrefp) TarExport *e = NULL; + int r; + + assert(ret); + + e = new0(TarExport, 1); + if (!e) + return -ENOMEM; + + e->output_fd = e->tar_fd = -1; + e->on_finished = on_finished; + e->userdata = userdata; + e->quota_referenced = (uint64_t) -1; + + RATELIMIT_INIT(e->progress_rate_limit, 100 * USEC_PER_MSEC, 1); + e->last_percent = (unsigned) -1; + + if (event) + e->event = sd_event_ref(event); + else { + r = sd_event_default(&e->event); + if (r < 0) + return r; + } + + *ret = e; + e = NULL; + + return 0; +} + +static void tar_export_report_progress(TarExport *e) { + unsigned percent; + assert(e); + + /* Do we have any quota info? If not, we don't know anything about the progress */ + if (e->quota_referenced == (uint64_t) -1) + return; + + if (e->written_uncompressed >= e->quota_referenced) + percent = 100; + else + percent = (unsigned) ((e->written_uncompressed * UINT64_C(100)) / e->quota_referenced); + + if (percent == e->last_percent) + return; + + if (!ratelimit_test(&e->progress_rate_limit)) + return; + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_info("Exported %u%%.", percent); + + e->last_percent = percent; +} + +static int tar_export_process(TarExport *e) { + ssize_t l; + int r; + + assert(e); + + if (!e->tried_splice && e->compress.type == IMPORT_COMPRESS_UNCOMPRESSED) { + + l = splice(e->tar_fd, NULL, e->output_fd, NULL, COPY_BUFFER_SIZE, 0); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + e->tried_splice = true; + } else if (l == 0) { + r = 0; + goto finish; + } else { + e->written_uncompressed += l; + e->written_compressed += l; + + tar_export_report_progress(e); + + return 0; + } + } + + while (e->buffer_size <= 0) { + uint8_t input[COPY_BUFFER_SIZE]; + + if (e->eof) { + r = 0; + goto finish; + } + + l = read(e->tar_fd, input, sizeof(input)); + if (l < 0) { + r = log_error_errno(errno, "Failed to read tar file: %m"); + goto finish; + } + + if (l == 0) { + e->eof = true; + r = import_compress_finish(&e->compress, &e->buffer, &e->buffer_size, &e->buffer_allocated); + } else { + e->written_uncompressed += l; + r = import_compress(&e->compress, input, l, &e->buffer, &e->buffer_size, &e->buffer_allocated); + } + if (r < 0) { + r = log_error_errno(r, "Failed to encode: %m"); + goto finish; + } + } + + l = write(e->output_fd, e->buffer, e->buffer_size); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + r = log_error_errno(errno, "Failed to write output file: %m"); + goto finish; + } + + assert((size_t) l <= e->buffer_size); + memmove(e->buffer, (uint8_t*) e->buffer + l, e->buffer_size - l); + e->buffer_size -= l; + e->written_compressed += l; + + tar_export_report_progress(e); + + return 0; + +finish: + if (e->on_finished) + e->on_finished(e, r, e->userdata); + else + sd_event_exit(e->event, r); + + return 0; +} + +static int tar_export_on_output(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + TarExport *i = userdata; + + return tar_export_process(i); +} + +static int tar_export_on_defer(sd_event_source *s, void *userdata) { + TarExport *i = userdata; + + return tar_export_process(i); +} + +int tar_export_start(TarExport *e, const char *path, int fd, ImportCompressType compress) { + _cleanup_close_ int sfd = -1; + int r; + + assert(e); + assert(path); + assert(fd >= 0); + assert(compress < _IMPORT_COMPRESS_TYPE_MAX); + assert(compress != IMPORT_COMPRESS_UNKNOWN); + + if (e->output_fd >= 0) + return -EBUSY; + + sfd = open(path, O_DIRECTORY|O_RDONLY|O_NOCTTY|O_CLOEXEC); + if (sfd < 0) + return -errno; + + if (fstat(sfd, &e->st) < 0) + return -errno; + + r = fd_nonblock(fd, true); + if (r < 0) + return r; + + r = free_and_strdup(&e->path, path); + if (r < 0) + return r; + + e->quota_referenced = (uint64_t) -1; + + if (e->st.st_ino == 256) { /* might be a btrfs subvolume? */ + BtrfsQuotaInfo q; + + r = btrfs_subvol_get_subtree_quota_fd(sfd, 0, &q); + if (r >= 0) + e->quota_referenced = q.referenced; + + e->temp_path = mfree(e->temp_path); + + r = tempfn_random(path, NULL, &e->temp_path); + if (r < 0) + return r; + + /* Let's try to make a snapshot, if we can, so that the export is atomic */ + r = btrfs_subvol_snapshot_fd(sfd, e->temp_path, BTRFS_SNAPSHOT_READ_ONLY|BTRFS_SNAPSHOT_RECURSIVE); + if (r < 0) { + log_debug_errno(r, "Couldn't create snapshot %s of %s, not exporting atomically: %m", e->temp_path, path); + e->temp_path = mfree(e->temp_path); + } + } + + r = import_compress_init(&e->compress, compress); + if (r < 0) + return r; + + r = sd_event_add_io(e->event, &e->output_event_source, fd, EPOLLOUT, tar_export_on_output, e); + if (r == -EPERM) { + r = sd_event_add_defer(e->event, &e->output_event_source, tar_export_on_defer, e); + if (r < 0) + return r; + + r = sd_event_source_set_enabled(e->output_event_source, SD_EVENT_ON); + } + if (r < 0) + return r; + + e->tar_fd = import_fork_tar_c(e->temp_path ?: e->path, &e->tar_pid); + if (e->tar_fd < 0) { + e->output_event_source = sd_event_source_unref(e->output_event_source); + return e->tar_fd; + } + + e->output_fd = fd; + return r; +} diff --git a/src/grp-machine/grp-import/systemd-export/export-tar.h b/src/grp-machine/grp-import/systemd-export/export-tar.h new file mode 100644 index 0000000000..2c50c1e5bf --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/export-tar.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "import-compress.h" +#include "systemd-basic/macro.h" + +typedef struct TarExport TarExport; +typedef void (*TarExportFinished)(TarExport *export, int error, void *userdata); + +int tar_export_new(TarExport **export, sd_event *event, TarExportFinished on_finished, void *userdata); +TarExport* tar_export_unref(TarExport *export); + +DEFINE_TRIVIAL_CLEANUP_FUNC(TarExport*, tar_export_unref); + +int tar_export_start(TarExport *export, const char *path, int fd, ImportCompressType compress); diff --git a/src/grp-machine/grp-import/systemd-export/export.c b/src/grp-machine/grp-import/systemd-export/export.c new file mode 100644 index 0000000000..a07273da7c --- /dev/null +++ b/src/grp-machine/grp-import/systemd-export/export.c @@ -0,0 +1,321 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <getopt.h> + +#include <systemd/sd-event.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/verbs.h" +#include "systemd-shared/import-util.h" +#include "systemd-shared/machine-image.h" + +#include "export-raw.h" +#include "export-tar.h" + +static ImportCompressType arg_compress = IMPORT_COMPRESS_UNKNOWN; + +static void determine_compression_from_filename(const char *p) { + + if (arg_compress != IMPORT_COMPRESS_UNKNOWN) + return; + + if (!p) { + arg_compress = IMPORT_COMPRESS_UNCOMPRESSED; + return; + } + + if (endswith(p, ".xz")) + arg_compress = IMPORT_COMPRESS_XZ; + else if (endswith(p, ".gz")) + arg_compress = IMPORT_COMPRESS_GZIP; + else if (endswith(p, ".bz2")) + arg_compress = IMPORT_COMPRESS_BZIP2; + else + arg_compress = IMPORT_COMPRESS_UNCOMPRESSED; +} + +static int interrupt_signal_handler(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) { + log_notice("Transfer aborted."); + sd_event_exit(sd_event_source_get_event(s), EINTR); + return 0; +} + +static void on_tar_finished(TarExport *export, int error, void *userdata) { + sd_event *event = userdata; + assert(export); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int export_tar(int argc, char *argv[], void *userdata) { + _cleanup_(tar_export_unrefp) TarExport *export = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + _cleanup_(image_unrefp) Image *image = NULL; + const char *path = NULL, *local = NULL; + _cleanup_close_ int open_fd = -1; + int r, fd; + + if (machine_name_is_valid(argv[1])) { + r = image_find(argv[1], &image); + if (r < 0) + return log_error_errno(r, "Failed to look for machine %s: %m", argv[1]); + if (r == 0) { + log_error("Machine image %s not found.", argv[1]); + return -ENOENT; + } + + local = image->path; + } else + local = argv[1]; + + if (argc >= 3) + path = argv[2]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + determine_compression_from_filename(path); + + if (path) { + open_fd = open(path, O_WRONLY|O_CREAT|O_TRUNC|O_CLOEXEC|O_NOCTTY, 0666); + if (open_fd < 0) + return log_error_errno(errno, "Failed to open tar image for export: %m"); + + fd = open_fd; + + log_info("Exporting '%s', saving to '%s' with compression '%s'.", local, path, import_compress_type_to_string(arg_compress)); + } else { + _cleanup_free_ char *pretty = NULL; + + fd = STDOUT_FILENO; + + (void) readlink_malloc("/proc/self/fd/1", &pretty); + log_info("Exporting '%s', saving to '%s' with compression '%s'.", local, strna(pretty), import_compress_type_to_string(arg_compress)); + } + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = tar_export_new(&export, event, on_tar_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate exporter: %m"); + + r = tar_export_start(export, local, fd, arg_compress); + if (r < 0) + return log_error_errno(r, "Failed to export image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static void on_raw_finished(RawExport *export, int error, void *userdata) { + sd_event *event = userdata; + assert(export); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int export_raw(int argc, char *argv[], void *userdata) { + _cleanup_(raw_export_unrefp) RawExport *export = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + _cleanup_(image_unrefp) Image *image = NULL; + const char *path = NULL, *local = NULL; + _cleanup_close_ int open_fd = -1; + int r, fd; + + if (machine_name_is_valid(argv[1])) { + r = image_find(argv[1], &image); + if (r < 0) + return log_error_errno(r, "Failed to look for machine %s: %m", argv[1]); + if (r == 0) { + log_error("Machine image %s not found.", argv[1]); + return -ENOENT; + } + + local = image->path; + } else + local = argv[1]; + + if (argc >= 3) + path = argv[2]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + determine_compression_from_filename(path); + + if (path) { + open_fd = open(path, O_WRONLY|O_CREAT|O_TRUNC|O_CLOEXEC|O_NOCTTY, 0666); + if (open_fd < 0) + return log_error_errno(errno, "Failed to open raw image for export: %m"); + + fd = open_fd; + + log_info("Exporting '%s', saving to '%s' with compression '%s'.", local, path, import_compress_type_to_string(arg_compress)); + } else { + _cleanup_free_ char *pretty = NULL; + + fd = STDOUT_FILENO; + + (void) readlink_malloc("/proc/self/fd/1", &pretty); + log_info("Exporting '%s', saving to '%s' with compression '%s'.", local, strna(pretty), import_compress_type_to_string(arg_compress)); + } + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = raw_export_new(&export, event, on_raw_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate exporter: %m"); + + r = raw_export_start(export, local, fd, arg_compress); + if (r < 0) + return log_error_errno(r, "Failed to export image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static int help(int argc, char *argv[], void *userdata) { + + printf("%s [OPTIONS...] {COMMAND} ...\n\n" + "Export container or virtual machine images.\n\n" + " -h --help Show this help\n" + " --version Show package version\n" + " --format=FORMAT Select format\n\n" + "Commands:\n" + " tar NAME [FILE] Export a TAR image\n" + " raw NAME [FILE] Export a RAW image\n", + program_invocation_short_name); + + return 0; +} + +static int parse_argv(int argc, char *argv[]) { + + enum { + ARG_VERSION = 0x100, + ARG_FORMAT, + }; + + static const struct option options[] = { + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "format", required_argument, NULL, ARG_FORMAT }, + {} + }; + + int c; + + assert(argc >= 0); + assert(argv); + + while ((c = getopt_long(argc, argv, "h", options, NULL)) >= 0) + + switch (c) { + + case 'h': + return help(0, NULL, NULL); + + case ARG_VERSION: + return version(); + + case ARG_FORMAT: + if (streq(optarg, "uncompressed")) + arg_compress = IMPORT_COMPRESS_UNCOMPRESSED; + else if (streq(optarg, "xz")) + arg_compress = IMPORT_COMPRESS_XZ; + else if (streq(optarg, "gzip")) + arg_compress = IMPORT_COMPRESS_GZIP; + else if (streq(optarg, "bzip2")) + arg_compress = IMPORT_COMPRESS_BZIP2; + else { + log_error("Unknown format: %s", optarg); + return -EINVAL; + } + break; + + case '?': + return -EINVAL; + + default: + assert_not_reached("Unhandled option"); + } + + return 1; +} + +static int export_main(int argc, char *argv[]) { + + static const Verb verbs[] = { + { "help", VERB_ANY, VERB_ANY, 0, help }, + { "tar", 2, 3, 0, export_tar }, + { "raw", 2, 3, 0, export_raw }, + {} + }; + + return dispatch_verb(argc, argv, verbs, NULL); +} + +int main(int argc, char *argv[]) { + int r; + + setlocale(LC_ALL, ""); + log_parse_environment(); + log_open(); + + r = parse_argv(argc, argv); + if (r <= 0) + goto finish; + + (void) ignore_signals(SIGPIPE, -1); + + r = export_main(argc, argv); + +finish: + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/grp-import/systemd-import/GNUmakefile b/src/grp-machine/grp-import/systemd-import/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/systemd-import/Makefile b/src/grp-machine/grp-import/systemd-import/Makefile new file mode 100644 index 0000000000..8d04ec18c5 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/Makefile @@ -0,0 +1,51 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +rootlibexec_PROGRAMS += systemd-import +systemd_import_SOURCES = \ + src/import/import.c \ + src/import/import-raw.c \ + src/import/import-raw.h \ + src/import/import-tar.c \ + src/import/import-tar.h \ + src/import/import-common.c \ + src/import/import-common.h \ + src/import/import-compress.c \ + src/import/import-compress.h \ + src/import/qcow2-util.c \ + src/import/qcow2-util.h + +systemd_import_CFLAGS = \ + $(XZ_CFLAGS) \ + $(ZLIB_CFLAGS) \ + $(BZIP2_CFLAGS) + +systemd_import_LDADD = \ + libsystemd-shared.la \ + $(XZ_LIBS) \ + $(ZLIB_LIBS) \ + $(BZIP2_LIBS) + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/systemd-import/import-raw.c b/src/grp-machine/grp-import/systemd-import/import-raw.c new file mode 100644 index 0000000000..fa484377f1 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/import-raw.c @@ -0,0 +1,466 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <linux/fs.h> + +#include <systemd/sd-daemon.h> +#include <systemd/sd-event.h> + +#include "import-common.h" +#include "import-compress.h" +#include "qcow2-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/chattr-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/ratelimit.h" +#include "systemd-basic/rm-rf.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/util.h" +#include "systemd-shared/machine-pool.h" + +#include "import-raw.h" + +struct RawImport { + sd_event *event; + + char *image_root; + + RawImportFinished on_finished; + void *userdata; + + char *local; + bool force_local; + bool read_only; + bool grow_machine_directory; + + char *temp_path; + char *final_path; + + int input_fd; + int output_fd; + + ImportCompress compress; + + uint64_t written_since_last_grow; + + sd_event_source *input_event_source; + + uint8_t buffer[16*1024]; + size_t buffer_size; + + uint64_t written_compressed; + uint64_t written_uncompressed; + + struct stat st; + + unsigned last_percent; + RateLimit progress_rate_limit; +}; + +RawImport* raw_import_unref(RawImport *i) { + if (!i) + return NULL; + + sd_event_unref(i->event); + + if (i->temp_path) { + (void) unlink(i->temp_path); + free(i->temp_path); + } + + import_compress_free(&i->compress); + + sd_event_source_unref(i->input_event_source); + + safe_close(i->output_fd); + + free(i->final_path); + free(i->image_root); + free(i->local); + return mfree(i); +} + +int raw_import_new( + RawImport **ret, + sd_event *event, + const char *image_root, + RawImportFinished on_finished, + void *userdata) { + + _cleanup_(raw_import_unrefp) RawImport *i = NULL; + int r; + + assert(ret); + + i = new0(RawImport, 1); + if (!i) + return -ENOMEM; + + i->input_fd = i->output_fd = -1; + i->on_finished = on_finished; + i->userdata = userdata; + + RATELIMIT_INIT(i->progress_rate_limit, 100 * USEC_PER_MSEC, 1); + i->last_percent = (unsigned) -1; + + i->image_root = strdup(image_root ?: "/var/lib/machines"); + if (!i->image_root) + return -ENOMEM; + + i->grow_machine_directory = path_startswith(i->image_root, "/var/lib/machines"); + + if (event) + i->event = sd_event_ref(event); + else { + r = sd_event_default(&i->event); + if (r < 0) + return r; + } + + *ret = i; + i = NULL; + + return 0; +} + +static void raw_import_report_progress(RawImport *i) { + unsigned percent; + assert(i); + + /* We have no size information, unless the source is a regular file */ + if (!S_ISREG(i->st.st_mode)) + return; + + if (i->written_compressed >= (uint64_t) i->st.st_size) + percent = 100; + else + percent = (unsigned) ((i->written_compressed * UINT64_C(100)) / (uint64_t) i->st.st_size); + + if (percent == i->last_percent) + return; + + if (!ratelimit_test(&i->progress_rate_limit)) + return; + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_info("Imported %u%%.", percent); + + i->last_percent = percent; +} + +static int raw_import_maybe_convert_qcow2(RawImport *i) { + _cleanup_close_ int converted_fd = -1; + _cleanup_free_ char *t = NULL; + int r; + + assert(i); + + r = qcow2_detect(i->output_fd); + if (r < 0) + return log_error_errno(r, "Failed to detect whether this is a QCOW2 image: %m"); + if (r == 0) + return 0; + + /* This is a QCOW2 image, let's convert it */ + r = tempfn_random(i->final_path, NULL, &t); + if (r < 0) + return log_oom(); + + converted_fd = open(t, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (converted_fd < 0) + return log_error_errno(errno, "Failed to create %s: %m", t); + + r = chattr_fd(converted_fd, FS_NOCOW_FL, FS_NOCOW_FL); + if (r < 0) + log_warning_errno(r, "Failed to set file attributes on %s: %m", t); + + log_info("Unpacking QCOW2 file."); + + r = qcow2_convert(i->output_fd, converted_fd); + if (r < 0) { + unlink(t); + return log_error_errno(r, "Failed to convert qcow2 image: %m"); + } + + (void) unlink(i->temp_path); + free(i->temp_path); + i->temp_path = t; + t = NULL; + + safe_close(i->output_fd); + i->output_fd = converted_fd; + converted_fd = -1; + + return 1; +} + +static int raw_import_finish(RawImport *i) { + int r; + + assert(i); + assert(i->output_fd >= 0); + assert(i->temp_path); + assert(i->final_path); + + /* In case this was a sparse file, make sure the file system is right */ + if (i->written_uncompressed > 0) { + if (ftruncate(i->output_fd, i->written_uncompressed) < 0) + return log_error_errno(errno, "Failed to truncate file: %m"); + } + + r = raw_import_maybe_convert_qcow2(i); + if (r < 0) + return r; + + if (S_ISREG(i->st.st_mode)) { + (void) copy_times(i->input_fd, i->output_fd); + (void) copy_xattr(i->input_fd, i->output_fd); + } + + if (i->read_only) { + r = import_make_read_only_fd(i->output_fd); + if (r < 0) + return r; + } + + if (i->force_local) + (void) rm_rf(i->final_path, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + + r = rename_noreplace(AT_FDCWD, i->temp_path, AT_FDCWD, i->final_path); + if (r < 0) + return log_error_errno(r, "Failed to move image into place: %m"); + + i->temp_path = mfree(i->temp_path); + + return 0; +} + +static int raw_import_open_disk(RawImport *i) { + int r; + + assert(i); + + assert(!i->final_path); + assert(!i->temp_path); + assert(i->output_fd < 0); + + i->final_path = strjoin(i->image_root, "/", i->local, ".raw", NULL); + if (!i->final_path) + return log_oom(); + + r = tempfn_random(i->final_path, NULL, &i->temp_path); + if (r < 0) + return log_oom(); + + (void) mkdir_parents_label(i->temp_path, 0700); + + i->output_fd = open(i->temp_path, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (i->output_fd < 0) + return log_error_errno(errno, "Failed to open destination %s: %m", i->temp_path); + + r = chattr_fd(i->output_fd, FS_NOCOW_FL, FS_NOCOW_FL); + if (r < 0) + log_warning_errno(r, "Failed to set file attributes on %s: %m", i->temp_path); + + return 0; +} + +static int raw_import_try_reflink(RawImport *i) { + off_t p; + int r; + + assert(i); + assert(i->input_fd >= 0); + assert(i->output_fd >= 0); + + if (i->compress.type != IMPORT_COMPRESS_UNCOMPRESSED) + return 0; + + if (!S_ISREG(i->st.st_mode)) + return 0; + + p = lseek(i->input_fd, 0, SEEK_CUR); + if (p == (off_t) -1) + return log_error_errno(errno, "Failed to read file offset of input file: %m"); + + /* Let's only try a btrfs reflink, if we are reading from the beginning of the file */ + if ((uint64_t) p != (uint64_t) i->buffer_size) + return 0; + + r = btrfs_reflink(i->input_fd, i->output_fd); + if (r >= 0) + return 1; + + return 0; +} + +static int raw_import_write(const void *p, size_t sz, void *userdata) { + RawImport *i = userdata; + ssize_t n; + + if (i->grow_machine_directory && i->written_since_last_grow >= GROW_INTERVAL_BYTES) { + i->written_since_last_grow = 0; + grow_machine_directory(); + } + + n = sparse_write(i->output_fd, p, sz, 64); + if (n < 0) + return -errno; + if ((size_t) n < sz) + return -EIO; + + i->written_uncompressed += sz; + i->written_since_last_grow += sz; + + return 0; +} + +static int raw_import_process(RawImport *i) { + ssize_t l; + int r; + + assert(i); + assert(i->buffer_size < sizeof(i->buffer)); + + l = read(i->input_fd, i->buffer + i->buffer_size, sizeof(i->buffer) - i->buffer_size); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + r = log_error_errno(errno, "Failed to read input file: %m"); + goto finish; + } + if (l == 0) { + if (i->compress.type == IMPORT_COMPRESS_UNKNOWN) { + log_error("Premature end of file: %m"); + r = -EIO; + goto finish; + } + + r = raw_import_finish(i); + goto finish; + } + + i->buffer_size += l; + + if (i->compress.type == IMPORT_COMPRESS_UNKNOWN) { + r = import_uncompress_detect(&i->compress, i->buffer, i->buffer_size); + if (r < 0) { + log_error("Failed to detect file compression: %m"); + goto finish; + } + if (r == 0) /* Need more data */ + return 0; + + r = raw_import_open_disk(i); + if (r < 0) + goto finish; + + r = raw_import_try_reflink(i); + if (r < 0) + goto finish; + if (r > 0) { + r = raw_import_finish(i); + goto finish; + } + } + + r = import_uncompress(&i->compress, i->buffer, i->buffer_size, raw_import_write, i); + if (r < 0) { + log_error_errno(r, "Failed to decode and write: %m"); + goto finish; + } + + i->written_compressed += i->buffer_size; + i->buffer_size = 0; + + raw_import_report_progress(i); + + return 0; + +finish: + if (i->on_finished) + i->on_finished(i, r, i->userdata); + else + sd_event_exit(i->event, r); + + return 0; +} + +static int raw_import_on_input(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + RawImport *i = userdata; + + return raw_import_process(i); +} + +static int raw_import_on_defer(sd_event_source *s, void *userdata) { + RawImport *i = userdata; + + return raw_import_process(i); +} + +int raw_import_start(RawImport *i, int fd, const char *local, bool force_local, bool read_only) { + int r; + + assert(i); + assert(fd >= 0); + assert(local); + + if (!machine_name_is_valid(local)) + return -EINVAL; + + if (i->input_fd >= 0) + return -EBUSY; + + r = fd_nonblock(fd, true); + if (r < 0) + return r; + + r = free_and_strdup(&i->local, local); + if (r < 0) + return r; + i->force_local = force_local; + i->read_only = read_only; + + if (fstat(fd, &i->st) < 0) + return -errno; + + r = sd_event_add_io(i->event, &i->input_event_source, fd, EPOLLIN, raw_import_on_input, i); + if (r == -EPERM) { + /* This fd does not support epoll, for example because it is a regular file. Busy read in that case */ + r = sd_event_add_defer(i->event, &i->input_event_source, raw_import_on_defer, i); + if (r < 0) + return r; + + r = sd_event_source_set_enabled(i->input_event_source, SD_EVENT_ON); + } + if (r < 0) + return r; + + i->input_fd = fd; + return r; +} diff --git a/src/grp-machine/grp-import/systemd-import/import-raw.h b/src/grp-machine/grp-import/systemd-import/import-raw.h new file mode 100644 index 0000000000..64c532242d --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/import-raw.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "systemd-basic/macro.h" +#include "systemd-shared/import-util.h" + +typedef struct RawImport RawImport; +typedef void (*RawImportFinished)(RawImport *import, int error, void *userdata); + +int raw_import_new(RawImport **import, sd_event *event, const char *image_root, RawImportFinished on_finished, void *userdata); +RawImport* raw_import_unref(RawImport *import); + +DEFINE_TRIVIAL_CLEANUP_FUNC(RawImport*, raw_import_unref); + +int raw_import_start(RawImport *i, int fd, const char *local, bool force_local, bool read_only); diff --git a/src/grp-machine/grp-import/systemd-import/import-tar.c b/src/grp-machine/grp-import/systemd-import/import-tar.c new file mode 100644 index 0000000000..458f905d80 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/import-tar.c @@ -0,0 +1,387 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <linux/fs.h> + +#include <systemd/sd-daemon.h> +#include <systemd/sd-event.h> + +#include "import-common.h" +#include "import-compress.h" +#include "qcow2-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/ratelimit.h" +#include "systemd-basic/rm-rf.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/util.h" +#include "systemd-shared/machine-pool.h" + +#include "import-tar.h" + +struct TarImport { + sd_event *event; + + char *image_root; + + TarImportFinished on_finished; + void *userdata; + + char *local; + bool force_local; + bool read_only; + bool grow_machine_directory; + + char *temp_path; + char *final_path; + + int input_fd; + int tar_fd; + + ImportCompress compress; + + uint64_t written_since_last_grow; + + sd_event_source *input_event_source; + + uint8_t buffer[16*1024]; + size_t buffer_size; + + uint64_t written_compressed; + uint64_t written_uncompressed; + + struct stat st; + + pid_t tar_pid; + + unsigned last_percent; + RateLimit progress_rate_limit; +}; + +TarImport* tar_import_unref(TarImport *i) { + if (!i) + return NULL; + + sd_event_source_unref(i->input_event_source); + + if (i->tar_pid > 1) { + (void) kill_and_sigcont(i->tar_pid, SIGKILL); + (void) wait_for_terminate(i->tar_pid, NULL); + } + + if (i->temp_path) { + (void) rm_rf(i->temp_path, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + free(i->temp_path); + } + + import_compress_free(&i->compress); + + sd_event_unref(i->event); + + safe_close(i->tar_fd); + + free(i->final_path); + free(i->image_root); + free(i->local); + return mfree(i); +} + +int tar_import_new( + TarImport **ret, + sd_event *event, + const char *image_root, + TarImportFinished on_finished, + void *userdata) { + + _cleanup_(tar_import_unrefp) TarImport *i = NULL; + int r; + + assert(ret); + + i = new0(TarImport, 1); + if (!i) + return -ENOMEM; + + i->input_fd = i->tar_fd = -1; + i->on_finished = on_finished; + i->userdata = userdata; + + RATELIMIT_INIT(i->progress_rate_limit, 100 * USEC_PER_MSEC, 1); + i->last_percent = (unsigned) -1; + + i->image_root = strdup(image_root ?: "/var/lib/machines"); + if (!i->image_root) + return -ENOMEM; + + i->grow_machine_directory = path_startswith(i->image_root, "/var/lib/machines"); + + if (event) + i->event = sd_event_ref(event); + else { + r = sd_event_default(&i->event); + if (r < 0) + return r; + } + + *ret = i; + i = NULL; + + return 0; +} + +static void tar_import_report_progress(TarImport *i) { + unsigned percent; + assert(i); + + /* We have no size information, unless the source is a regular file */ + if (!S_ISREG(i->st.st_mode)) + return; + + if (i->written_compressed >= (uint64_t) i->st.st_size) + percent = 100; + else + percent = (unsigned) ((i->written_compressed * UINT64_C(100)) / (uint64_t) i->st.st_size); + + if (percent == i->last_percent) + return; + + if (!ratelimit_test(&i->progress_rate_limit)) + return; + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_info("Imported %u%%.", percent); + + i->last_percent = percent; +} + +static int tar_import_finish(TarImport *i) { + int r; + + assert(i); + assert(i->tar_fd >= 0); + assert(i->temp_path); + assert(i->final_path); + + i->tar_fd = safe_close(i->tar_fd); + + if (i->tar_pid > 0) { + r = wait_for_terminate_and_warn("tar", i->tar_pid, true); + i->tar_pid = 0; + if (r < 0) + return r; + } + + if (i->read_only) { + r = import_make_read_only(i->temp_path); + if (r < 0) + return r; + } + + if (i->force_local) + (void) rm_rf(i->final_path, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + + r = rename_noreplace(AT_FDCWD, i->temp_path, AT_FDCWD, i->final_path); + if (r < 0) + return log_error_errno(r, "Failed to move image into place: %m"); + + i->temp_path = mfree(i->temp_path); + + return 0; +} + +static int tar_import_fork_tar(TarImport *i) { + int r; + + assert(i); + + assert(!i->final_path); + assert(!i->temp_path); + assert(i->tar_fd < 0); + + i->final_path = strjoin(i->image_root, "/", i->local, NULL); + if (!i->final_path) + return log_oom(); + + r = tempfn_random(i->final_path, NULL, &i->temp_path); + if (r < 0) + return log_oom(); + + (void) mkdir_parents_label(i->temp_path, 0700); + + r = btrfs_subvol_make(i->temp_path); + if (r == -ENOTTY) { + if (mkdir(i->temp_path, 0755) < 0) + return log_error_errno(errno, "Failed to create directory %s: %m", i->temp_path); + } else if (r < 0) + return log_error_errno(r, "Failed to create subvolume %s: %m", i->temp_path); + else + (void) import_assign_pool_quota_and_warn(i->temp_path); + + i->tar_fd = import_fork_tar_x(i->temp_path, &i->tar_pid); + if (i->tar_fd < 0) + return i->tar_fd; + + return 0; +} + +static int tar_import_write(const void *p, size_t sz, void *userdata) { + TarImport *i = userdata; + int r; + + if (i->grow_machine_directory && i->written_since_last_grow >= GROW_INTERVAL_BYTES) { + i->written_since_last_grow = 0; + grow_machine_directory(); + } + + r = loop_write(i->tar_fd, p, sz, false); + if (r < 0) + return r; + + i->written_uncompressed += sz; + i->written_since_last_grow += sz; + + return 0; +} + +static int tar_import_process(TarImport *i) { + ssize_t l; + int r; + + assert(i); + assert(i->buffer_size < sizeof(i->buffer)); + + l = read(i->input_fd, i->buffer + i->buffer_size, sizeof(i->buffer) - i->buffer_size); + if (l < 0) { + if (errno == EAGAIN) + return 0; + + r = log_error_errno(errno, "Failed to read input file: %m"); + goto finish; + } + if (l == 0) { + if (i->compress.type == IMPORT_COMPRESS_UNKNOWN) { + log_error("Premature end of file: %m"); + r = -EIO; + goto finish; + } + + r = tar_import_finish(i); + goto finish; + } + + i->buffer_size += l; + + if (i->compress.type == IMPORT_COMPRESS_UNKNOWN) { + r = import_uncompress_detect(&i->compress, i->buffer, i->buffer_size); + if (r < 0) { + log_error("Failed to detect file compression: %m"); + goto finish; + } + if (r == 0) /* Need more data */ + return 0; + + r = tar_import_fork_tar(i); + if (r < 0) + goto finish; + } + + r = import_uncompress(&i->compress, i->buffer, i->buffer_size, tar_import_write, i); + if (r < 0) { + log_error_errno(r, "Failed to decode and write: %m"); + goto finish; + } + + i->written_compressed += i->buffer_size; + i->buffer_size = 0; + + tar_import_report_progress(i); + + return 0; + +finish: + if (i->on_finished) + i->on_finished(i, r, i->userdata); + else + sd_event_exit(i->event, r); + + return 0; +} + +static int tar_import_on_input(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + TarImport *i = userdata; + + return tar_import_process(i); +} + +static int tar_import_on_defer(sd_event_source *s, void *userdata) { + TarImport *i = userdata; + + return tar_import_process(i); +} + +int tar_import_start(TarImport *i, int fd, const char *local, bool force_local, bool read_only) { + int r; + + assert(i); + assert(fd >= 0); + assert(local); + + if (!machine_name_is_valid(local)) + return -EINVAL; + + if (i->input_fd >= 0) + return -EBUSY; + + r = fd_nonblock(fd, true); + if (r < 0) + return r; + + r = free_and_strdup(&i->local, local); + if (r < 0) + return r; + i->force_local = force_local; + i->read_only = read_only; + + if (fstat(fd, &i->st) < 0) + return -errno; + + r = sd_event_add_io(i->event, &i->input_event_source, fd, EPOLLIN, tar_import_on_input, i); + if (r == -EPERM) { + /* This fd does not support epoll, for example because it is a regular file. Busy read in that case */ + r = sd_event_add_defer(i->event, &i->input_event_source, tar_import_on_defer, i); + if (r < 0) + return r; + + r = sd_event_source_set_enabled(i->input_event_source, SD_EVENT_ON); + } + if (r < 0) + return r; + + i->input_fd = fd; + return r; +} diff --git a/src/grp-machine/grp-import/systemd-import/import-tar.h b/src/grp-machine/grp-import/systemd-import/import-tar.h new file mode 100644 index 0000000000..01c3f2d542 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/import-tar.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "systemd-basic/macro.h" +#include "systemd-shared/import-util.h" + +typedef struct TarImport TarImport; +typedef void (*TarImportFinished)(TarImport *import, int error, void *userdata); + +int tar_import_new(TarImport **import, sd_event *event, const char *image_root, TarImportFinished on_finished, void *userdata); +TarImport* tar_import_unref(TarImport *import); + +DEFINE_TRIVIAL_CLEANUP_FUNC(TarImport*, tar_import_unref); + +int tar_import_start(TarImport *import, int fd, const char *local, bool force_local, bool read_only); diff --git a/src/grp-machine/grp-import/systemd-import/import.c b/src/grp-machine/grp-import/systemd-import/import.c new file mode 100644 index 0000000000..d6f772a159 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-import/import.c @@ -0,0 +1,338 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <getopt.h> + +#include <systemd/sd-event.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/verbs.h" +#include "systemd-shared/import-util.h" +#include "systemd-shared/machine-image.h" + +#include "import-raw.h" +#include "import-tar.h" + +static bool arg_force = false; +static bool arg_read_only = false; +static const char *arg_image_root = "/var/lib/machines"; + +static int interrupt_signal_handler(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) { + log_notice("Transfer aborted."); + sd_event_exit(sd_event_source_get_event(s), EINTR); + return 0; +} + +static void on_tar_finished(TarImport *import, int error, void *userdata) { + sd_event *event = userdata; + assert(import); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int import_tar(int argc, char *argv[], void *userdata) { + _cleanup_(tar_import_unrefp) TarImport *import = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + const char *path = NULL, *local = NULL; + _cleanup_free_ char *ll = NULL; + _cleanup_close_ int open_fd = -1; + int r, fd; + + if (argc >= 2) + path = argv[1]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (argc >= 3) + local = argv[2]; + else if (path) + local = basename(path); + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = tar_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local image name '%s' is not valid.", local); + return -EINVAL; + } + + if (!arg_force) { + r = image_find(local, NULL); + if (r < 0) + return log_error_errno(r, "Failed to check whether image '%s' exists: %m", local); + else if (r > 0) { + log_error("Image '%s' already exists.", local); + return -EEXIST; + } + } + } else + local = "imported"; + + if (path) { + open_fd = open(path, O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (open_fd < 0) + return log_error_errno(errno, "Failed to open tar image to import: %m"); + + fd = open_fd; + + log_info("Importing '%s', saving as '%s'.", path, local); + } else { + _cleanup_free_ char *pretty = NULL; + + fd = STDIN_FILENO; + + (void) readlink_malloc("/proc/self/fd/0", &pretty); + log_info("Importing '%s', saving as '%s'.", strna(pretty), local); + } + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = tar_import_new(&import, event, arg_image_root, on_tar_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate importer: %m"); + + r = tar_import_start(import, fd, local, arg_force, arg_read_only); + if (r < 0) + return log_error_errno(r, "Failed to import image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static void on_raw_finished(RawImport *import, int error, void *userdata) { + sd_event *event = userdata; + assert(import); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int import_raw(int argc, char *argv[], void *userdata) { + _cleanup_(raw_import_unrefp) RawImport *import = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + const char *path = NULL, *local = NULL; + _cleanup_free_ char *ll = NULL; + _cleanup_close_ int open_fd = -1; + int r, fd; + + if (argc >= 2) + path = argv[1]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (argc >= 3) + local = argv[2]; + else if (path) + local = basename(path); + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = raw_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local image name '%s' is not valid.", local); + return -EINVAL; + } + + if (!arg_force) { + r = image_find(local, NULL); + if (r < 0) + return log_error_errno(r, "Failed to check whether image '%s' exists: %m", local); + else if (r > 0) { + log_error("Image '%s' already exists.", local); + return -EEXIST; + } + } + } else + local = "imported"; + + if (path) { + open_fd = open(path, O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (open_fd < 0) + return log_error_errno(errno, "Failed to open raw image to import: %m"); + + fd = open_fd; + + log_info("Importing '%s', saving as '%s'.", path, local); + } else { + _cleanup_free_ char *pretty = NULL; + + fd = STDIN_FILENO; + + (void) readlink_malloc("/proc/self/fd/0", &pretty); + log_info("Importing '%s', saving as '%s'.", pretty, local); + } + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = raw_import_new(&import, event, arg_image_root, on_raw_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate importer: %m"); + + r = raw_import_start(import, fd, local, arg_force, arg_read_only); + if (r < 0) + return log_error_errno(r, "Failed to import image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static int help(int argc, char *argv[], void *userdata) { + + printf("%s [OPTIONS...] {COMMAND} ...\n\n" + "Import container or virtual machine images.\n\n" + " -h --help Show this help\n" + " --version Show package version\n" + " --force Force creation of image\n" + " --image-root=PATH Image root directory\n" + " --read-only Create a read-only image\n\n" + "Commands:\n" + " tar FILE [NAME] Import a TAR image\n" + " raw FILE [NAME] Import a RAW image\n", + program_invocation_short_name); + + return 0; +} + +static int parse_argv(int argc, char *argv[]) { + + enum { + ARG_VERSION = 0x100, + ARG_FORCE, + ARG_IMAGE_ROOT, + ARG_READ_ONLY, + }; + + static const struct option options[] = { + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "force", no_argument, NULL, ARG_FORCE }, + { "image-root", required_argument, NULL, ARG_IMAGE_ROOT }, + { "read-only", no_argument, NULL, ARG_READ_ONLY }, + {} + }; + + int c; + + assert(argc >= 0); + assert(argv); + + while ((c = getopt_long(argc, argv, "h", options, NULL)) >= 0) + + switch (c) { + + case 'h': + return help(0, NULL, NULL); + + case ARG_VERSION: + return version(); + + case ARG_FORCE: + arg_force = true; + break; + + case ARG_IMAGE_ROOT: + arg_image_root = optarg; + break; + + case ARG_READ_ONLY: + arg_read_only = true; + break; + + case '?': + return -EINVAL; + + default: + assert_not_reached("Unhandled option"); + } + + return 1; +} + +static int import_main(int argc, char *argv[]) { + + static const Verb verbs[] = { + { "help", VERB_ANY, VERB_ANY, 0, help }, + { "tar", 2, 3, 0, import_tar }, + { "raw", 2, 3, 0, import_raw }, + {} + }; + + return dispatch_verb(argc, argv, verbs, NULL); +} + +int main(int argc, char *argv[]) { + int r; + + setlocale(LC_ALL, ""); + log_parse_environment(); + log_open(); + + r = parse_argv(argc, argv); + if (r <= 0) + goto finish; + + (void) ignore_signals(SIGPIPE, -1); + + r = import_main(argc, argv); + +finish: + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/grp-import/systemd-importd/.gitignore b/src/grp-machine/grp-import/systemd-importd/.gitignore new file mode 100644 index 0000000000..01106e2e68 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/.gitignore @@ -0,0 +1 @@ +/org.freedesktop.import1.policy diff --git a/src/grp-machine/grp-import/systemd-importd/GNUmakefile b/src/grp-machine/grp-import/systemd-importd/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/systemd-importd/Makefile b/src/grp-machine/grp-import/systemd-importd/Makefile new file mode 100644 index 0000000000..f473e34ea3 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/Makefile @@ -0,0 +1,65 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +rootlibexec_PROGRAMS += systemd-importd +systemd_importd_SOURCES = \ + src/import/importd.c + +systemd_importd_CFLAGS = \ + -D SYSTEMD_PULL_PATH=\"$(rootlibexecdir)/systemd-pull\" \ + -D SYSTEMD_IMPORT_PATH=\"$(rootlibexecdir)/systemd-import\" \ + -D SYSTEMD_EXPORT_PATH=\"$(rootlibexecdir)/systemd-export\" + +systemd_importd_LDADD = \ + libsystemd-shared.la + +nodist_systemunit_DATA += \ + units/systemd-importd.service + +dist_systemunit_DATA_busnames += \ + units/org.freedesktop.import1.busname + +BUSNAMES_TARGET_WANTS += \ + org.freedesktop.import1.busname + +SYSTEM_UNIT_ALIASES += \ + systemd-importd.service dbus-org.freedesktop.import1.service + +dist_dbussystemservice_DATA += \ + src/import/org.freedesktop.import1.service + +dist_dbuspolicy_DATA += \ + src/import/org.freedesktop.import1.conf + +polkitpolicy_files += \ + src/import/org.freedesktop.import1.policy + +polkitpolicy_in_files += \ + src/import/org.freedesktop.import1.policy.in + +EXTRA_DIST += \ + units/systemd-importd.service.in + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/systemd-importd/importd.c b/src/grp-machine/grp-import/systemd-importd/importd.c new file mode 100644 index 0000000000..68c52e564e --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/importd.c @@ -0,0 +1,1217 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sys/prctl.h> + +#include <systemd/sd-bus.h> + +#include "sd-bus/bus-common-errors.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/def.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/missing.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/parse-util.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/socket-util.h" +#include "systemd-basic/string-table.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/syslog-util.h" +#include "systemd-basic/user-util.h" +#include "systemd-basic/util.h" +#include "systemd-basic/web-util.h" +#include "systemd-shared/import-util.h" +#include "systemd-shared/machine-pool.h" + +typedef struct Transfer Transfer; +typedef struct Manager Manager; + +typedef enum TransferType { + TRANSFER_IMPORT_TAR, + TRANSFER_IMPORT_RAW, + TRANSFER_EXPORT_TAR, + TRANSFER_EXPORT_RAW, + TRANSFER_PULL_TAR, + TRANSFER_PULL_RAW, + _TRANSFER_TYPE_MAX, + _TRANSFER_TYPE_INVALID = -1, +} TransferType; + +struct Transfer { + Manager *manager; + + uint32_t id; + char *object_path; + + TransferType type; + ImportVerify verify; + + char *remote; + char *local; + bool force_local; + bool read_only; + + char *format; + + pid_t pid; + + int log_fd; + + char log_message[LINE_MAX]; + size_t log_message_size; + + sd_event_source *pid_event_source; + sd_event_source *log_event_source; + + unsigned n_canceled; + unsigned progress_percent; + + int stdin_fd; + int stdout_fd; +}; + +struct Manager { + sd_event *event; + sd_bus *bus; + + uint32_t current_transfer_id; + Hashmap *transfers; + + Hashmap *polkit_registry; + + int notify_fd; + + sd_event_source *notify_event_source; +}; + +#define TRANSFERS_MAX 64 + +static const char* const transfer_type_table[_TRANSFER_TYPE_MAX] = { + [TRANSFER_IMPORT_TAR] = "import-tar", + [TRANSFER_IMPORT_RAW] = "import-raw", + [TRANSFER_EXPORT_TAR] = "export-tar", + [TRANSFER_EXPORT_RAW] = "export-raw", + [TRANSFER_PULL_TAR] = "pull-tar", + [TRANSFER_PULL_RAW] = "pull-raw", +}; + +DEFINE_PRIVATE_STRING_TABLE_LOOKUP_TO_STRING(transfer_type, TransferType); + +static Transfer *transfer_unref(Transfer *t) { + if (!t) + return NULL; + + if (t->manager) + hashmap_remove(t->manager->transfers, UINT32_TO_PTR(t->id)); + + sd_event_source_unref(t->pid_event_source); + sd_event_source_unref(t->log_event_source); + + free(t->remote); + free(t->local); + free(t->format); + free(t->object_path); + + if (t->pid > 0) { + (void) kill_and_sigcont(t->pid, SIGKILL); + (void) wait_for_terminate(t->pid, NULL); + } + + safe_close(t->log_fd); + safe_close(t->stdin_fd); + safe_close(t->stdout_fd); + + return mfree(t); +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(Transfer*, transfer_unref); + +static int transfer_new(Manager *m, Transfer **ret) { + _cleanup_(transfer_unrefp) Transfer *t = NULL; + uint32_t id; + int r; + + assert(m); + assert(ret); + + if (hashmap_size(m->transfers) >= TRANSFERS_MAX) + return -E2BIG; + + r = hashmap_ensure_allocated(&m->transfers, &trivial_hash_ops); + if (r < 0) + return r; + + t = new0(Transfer, 1); + if (!t) + return -ENOMEM; + + t->type = _TRANSFER_TYPE_INVALID; + t->log_fd = -1; + t->stdin_fd = -1; + t->stdout_fd = -1; + t->verify = _IMPORT_VERIFY_INVALID; + + id = m->current_transfer_id + 1; + + if (asprintf(&t->object_path, "/org/freedesktop/import1/transfer/_%" PRIu32, id) < 0) + return -ENOMEM; + + r = hashmap_put(m->transfers, UINT32_TO_PTR(id), t); + if (r < 0) + return r; + + m->current_transfer_id = id; + + t->manager = m; + t->id = id; + + *ret = t; + t = NULL; + + return 0; +} + +static void transfer_send_log_line(Transfer *t, const char *line) { + int r, priority = LOG_INFO; + + assert(t); + assert(line); + + syslog_parse_priority(&line, &priority, true); + + log_full(priority, "(transfer%" PRIu32 ") %s", t->id, line); + + r = sd_bus_emit_signal( + t->manager->bus, + t->object_path, + "org.freedesktop.import1.Transfer", + "LogMessage", + "us", + priority, + line); + if (r < 0) + log_error_errno(r, "Cannot emit message: %m"); + } + +static void transfer_send_logs(Transfer *t, bool flush) { + assert(t); + + /* Try to send out all log messages, if we can. But if we + * can't we remove the messages from the buffer, but don't + * fail */ + + while (t->log_message_size > 0) { + _cleanup_free_ char *n = NULL; + char *e; + + if (t->log_message_size >= sizeof(t->log_message)) + e = t->log_message + sizeof(t->log_message); + else { + char *a, *b; + + a = memchr(t->log_message, 0, t->log_message_size); + b = memchr(t->log_message, '\n', t->log_message_size); + + if (a && b) + e = a < b ? a : b; + else if (a) + e = a; + else + e = b; + } + + if (!e) { + if (!flush) + return; + + e = t->log_message + t->log_message_size; + } + + n = strndup(t->log_message, e - t->log_message); + + /* Skip over NUL and newlines */ + while ((e < t->log_message + t->log_message_size) && (*e == 0 || *e == '\n')) + e++; + + memmove(t->log_message, e, t->log_message + sizeof(t->log_message) - e); + t->log_message_size -= e - t->log_message; + + if (!n) { + log_oom(); + continue; + } + + if (isempty(n)) + continue; + + transfer_send_log_line(t, n); + } +} + +static int transfer_finalize(Transfer *t, bool success) { + int r; + + assert(t); + + transfer_send_logs(t, true); + + r = sd_bus_emit_signal( + t->manager->bus, + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "TransferRemoved", + "uos", + t->id, + t->object_path, + success ? "done" : + t->n_canceled > 0 ? "canceled" : "failed"); + + if (r < 0) + log_error_errno(r, "Cannot emit message: %m"); + + transfer_unref(t); + return 0; +} + +static int transfer_cancel(Transfer *t) { + int r; + + assert(t); + + r = kill_and_sigcont(t->pid, t->n_canceled < 3 ? SIGTERM : SIGKILL); + if (r < 0) + return r; + + t->n_canceled++; + return 0; +} + +static int transfer_on_pid(sd_event_source *s, const siginfo_t *si, void *userdata) { + Transfer *t = userdata; + bool success = false; + + assert(s); + assert(t); + + if (si->si_code == CLD_EXITED) { + if (si->si_status != 0) + log_error("Import process failed with exit code %i.", si->si_status); + else { + log_debug("Import process succeeded."); + success = true; + } + + } else if (si->si_code == CLD_KILLED || + si->si_code == CLD_DUMPED) + + log_error("Import process terminated by signal %s.", signal_to_string(si->si_status)); + else + log_error("Import process failed due to unknown reason."); + + t->pid = 0; + + return transfer_finalize(t, success); +} + +static int transfer_on_log(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + Transfer *t = userdata; + ssize_t l; + + assert(s); + assert(t); + + l = read(fd, t->log_message + t->log_message_size, sizeof(t->log_message) - t->log_message_size); + if (l <= 0) { + /* EOF/read error. We just close the pipe here, and + * close the watch, waiting for the SIGCHLD to arrive, + * before we do anything else. */ + + if (l < 0) + log_error_errno(errno, "Failed to read log message: %m"); + + t->log_event_source = sd_event_source_unref(t->log_event_source); + return 0; + } + + t->log_message_size += l; + + transfer_send_logs(t, false); + + return 0; +} + +static int transfer_start(Transfer *t) { + _cleanup_close_pair_ int pipefd[2] = { -1, -1 }; + int r; + + assert(t); + assert(t->pid <= 0); + + if (pipe2(pipefd, O_CLOEXEC) < 0) + return -errno; + + t->pid = fork(); + if (t->pid < 0) + return -errno; + if (t->pid == 0) { + const char *cmd[] = { + NULL, /* systemd-import, systemd-export or systemd-pull */ + NULL, /* tar, raw */ + NULL, /* --verify= */ + NULL, /* verify argument */ + NULL, /* maybe --force */ + NULL, /* maybe --read-only */ + NULL, /* if so: the actual URL */ + NULL, /* maybe --format= */ + NULL, /* if so: the actual format */ + NULL, /* remote */ + NULL, /* local */ + NULL + }; + unsigned k = 0; + + /* Child */ + + (void) reset_all_signal_handlers(); + (void) reset_signal_mask(); + assert_se(prctl(PR_SET_PDEATHSIG, SIGTERM) == 0); + + pipefd[0] = safe_close(pipefd[0]); + + if (dup2(pipefd[1], STDERR_FILENO) != STDERR_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (t->stdout_fd >= 0) { + if (dup2(t->stdout_fd, STDOUT_FILENO) != STDOUT_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (t->stdout_fd != STDOUT_FILENO) + safe_close(t->stdout_fd); + } else { + if (dup2(pipefd[1], STDOUT_FILENO) != STDOUT_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + } + + if (pipefd[1] != STDOUT_FILENO && pipefd[1] != STDERR_FILENO) + pipefd[1] = safe_close(pipefd[1]); + + if (t->stdin_fd >= 0) { + if (dup2(t->stdin_fd, STDIN_FILENO) != STDIN_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (t->stdin_fd != STDIN_FILENO) + safe_close(t->stdin_fd); + } else { + int null_fd; + + null_fd = open("/dev/null", O_RDONLY|O_NOCTTY); + if (null_fd < 0) { + log_error_errno(errno, "Failed to open /dev/null: %m"); + _exit(EXIT_FAILURE); + } + + if (dup2(null_fd, STDIN_FILENO) != STDIN_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (null_fd != STDIN_FILENO) + safe_close(null_fd); + } + + stdio_unset_cloexec(); + + setenv("SYSTEMD_LOG_TARGET", "console-prefixed", 1); + setenv("NOTIFY_SOCKET", "/run/systemd/import/notify", 1); + + if (IN_SET(t->type, TRANSFER_IMPORT_TAR, TRANSFER_IMPORT_RAW)) + cmd[k++] = SYSTEMD_IMPORT_PATH; + else if (IN_SET(t->type, TRANSFER_EXPORT_TAR, TRANSFER_EXPORT_RAW)) + cmd[k++] = SYSTEMD_EXPORT_PATH; + else + cmd[k++] = SYSTEMD_PULL_PATH; + + if (IN_SET(t->type, TRANSFER_IMPORT_TAR, TRANSFER_EXPORT_TAR, TRANSFER_PULL_TAR)) + cmd[k++] = "tar"; + else + cmd[k++] = "raw"; + + if (t->verify != _IMPORT_VERIFY_INVALID) { + cmd[k++] = "--verify"; + cmd[k++] = import_verify_to_string(t->verify); + } + + if (t->force_local) + cmd[k++] = "--force"; + if (t->read_only) + cmd[k++] = "--read-only"; + + if (t->format) { + cmd[k++] = "--format"; + cmd[k++] = t->format; + } + + if (!IN_SET(t->type, TRANSFER_EXPORT_TAR, TRANSFER_EXPORT_RAW)) { + if (t->remote) + cmd[k++] = t->remote; + else + cmd[k++] = "-"; + } + + if (t->local) + cmd[k++] = t->local; + cmd[k] = NULL; + + execv(cmd[0], (char * const *) cmd); + log_error_errno(errno, "Failed to execute %s tool: %m", cmd[0]); + _exit(EXIT_FAILURE); + } + + pipefd[1] = safe_close(pipefd[1]); + t->log_fd = pipefd[0]; + pipefd[0] = -1; + + t->stdin_fd = safe_close(t->stdin_fd); + + r = sd_event_add_child(t->manager->event, &t->pid_event_source, t->pid, WEXITED, transfer_on_pid, t); + if (r < 0) + return r; + + r = sd_event_add_io(t->manager->event, &t->log_event_source, t->log_fd, EPOLLIN, transfer_on_log, t); + if (r < 0) + return r; + + /* Make sure always process logging before SIGCHLD */ + r = sd_event_source_set_priority(t->log_event_source, SD_EVENT_PRIORITY_NORMAL -5); + if (r < 0) + return r; + + r = sd_bus_emit_signal( + t->manager->bus, + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "TransferNew", + "uo", + t->id, + t->object_path); + if (r < 0) + return r; + + return 0; +} + +static Manager *manager_unref(Manager *m) { + Transfer *t; + + if (!m) + return NULL; + + sd_event_source_unref(m->notify_event_source); + safe_close(m->notify_fd); + + while ((t = hashmap_first(m->transfers))) + transfer_unref(t); + + hashmap_free(m->transfers); + + bus_verify_polkit_async_registry_free(m->polkit_registry); + + m->bus = sd_bus_flush_close_unref(m->bus); + sd_event_unref(m->event); + + return mfree(m); +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(Manager*, manager_unref); + +static int manager_on_notify(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + + char buf[NOTIFY_BUFFER_MAX+1]; + struct iovec iovec = { + .iov_base = buf, + .iov_len = sizeof(buf)-1, + }; + union { + struct cmsghdr cmsghdr; + uint8_t buf[CMSG_SPACE(sizeof(struct ucred)) + + CMSG_SPACE(sizeof(int) * NOTIFY_FD_MAX)]; + } control = {}; + struct msghdr msghdr = { + .msg_iov = &iovec, + .msg_iovlen = 1, + .msg_control = &control, + .msg_controllen = sizeof(control), + }; + struct ucred *ucred = NULL; + Manager *m = userdata; + struct cmsghdr *cmsg; + unsigned percent; + char *p, *e; + Transfer *t; + Iterator i; + ssize_t n; + int r; + + n = recvmsg(fd, &msghdr, MSG_DONTWAIT|MSG_CMSG_CLOEXEC); + if (n < 0) { + if (errno == EAGAIN || errno == EINTR) + return 0; + + return -errno; + } + + cmsg_close_all(&msghdr); + + CMSG_FOREACH(cmsg, &msghdr) + if (cmsg->cmsg_level == SOL_SOCKET && + cmsg->cmsg_type == SCM_CREDENTIALS && + cmsg->cmsg_len == CMSG_LEN(sizeof(struct ucred))) + ucred = (struct ucred*) CMSG_DATA(cmsg); + + if (msghdr.msg_flags & MSG_TRUNC) { + log_warning("Got overly long notification datagram, ignoring."); + return 0; + } + + if (!ucred || ucred->pid <= 0) { + log_warning("Got notification datagram lacking credential information, ignoring."); + return 0; + } + + HASHMAP_FOREACH(t, m->transfers, i) + if (ucred->pid == t->pid) + break; + + if (!t) { + log_warning("Got notification datagram from unexpected peer, ignoring."); + return 0; + } + + buf[n] = 0; + + p = startswith(buf, "X_IMPORT_PROGRESS="); + if (!p) { + p = strstr(buf, "\nX_IMPORT_PROGRESS="); + if (!p) + return 0; + + p += 19; + } + + e = strchrnul(p, '\n'); + *e = 0; + + r = safe_atou(p, &percent); + if (r < 0 || percent > 100) { + log_warning("Got invalid percent value, ignoring."); + return 0; + } + + t->progress_percent = percent; + + log_debug("Got percentage from client: %u%%", percent); + return 0; +} + +static int manager_new(Manager **ret) { + _cleanup_(manager_unrefp) Manager *m = NULL; + static const union sockaddr_union sa = { + .un.sun_family = AF_UNIX, + .un.sun_path = "/run/systemd/import/notify", + }; + static const int one = 1; + int r; + + assert(ret); + + m = new0(Manager, 1); + if (!m) + return -ENOMEM; + + r = sd_event_default(&m->event); + if (r < 0) + return r; + + sd_event_set_watchdog(m->event, true); + + r = sd_bus_default_system(&m->bus); + if (r < 0) + return r; + + m->notify_fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0); + if (m->notify_fd < 0) + return -errno; + + (void) mkdir_parents_label(sa.un.sun_path, 0755); + (void) unlink(sa.un.sun_path); + + if (bind(m->notify_fd, &sa.sa, SOCKADDR_UN_LEN(sa.un)) < 0) + return -errno; + + if (setsockopt(m->notify_fd, SOL_SOCKET, SO_PASSCRED, &one, sizeof(one)) < 0) + return -errno; + + r = sd_event_add_io(m->event, &m->notify_event_source, m->notify_fd, EPOLLIN, manager_on_notify, m); + if (r < 0) + return r; + + *ret = m; + m = NULL; + + return 0; +} + +static Transfer *manager_find(Manager *m, TransferType type, const char *remote) { + Transfer *t; + Iterator i; + + assert(m); + assert(type >= 0); + assert(type < _TRANSFER_TYPE_MAX); + + HASHMAP_FOREACH(t, m->transfers, i) { + + if (t->type == type && + streq_ptr(t->remote, remote)) + return t; + } + + return NULL; +} + +static int method_import_tar_or_raw(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + _cleanup_(transfer_unrefp) Transfer *t = NULL; + int fd, force, read_only, r; + const char *local, *object; + Manager *m = userdata; + TransferType type; + uint32_t id; + + assert(msg); + assert(m); + + r = bus_verify_polkit_async( + msg, + CAP_SYS_ADMIN, + "org.freedesktop.import1.import", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = sd_bus_message_read(msg, "hsbb", &fd, &local, &force, &read_only); + if (r < 0) + return r; + + if (!machine_name_is_valid(local)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Local name %s is invalid", local); + + r = setup_machine_directory((uint64_t) -1, error); + if (r < 0) + return r; + + type = streq_ptr(sd_bus_message_get_member(msg), "ImportTar") ? TRANSFER_IMPORT_TAR : TRANSFER_IMPORT_RAW; + + r = transfer_new(m, &t); + if (r < 0) + return r; + + t->type = type; + t->force_local = force; + t->read_only = read_only; + + t->local = strdup(local); + if (!t->local) + return -ENOMEM; + + t->stdin_fd = fcntl(fd, F_DUPFD_CLOEXEC, 3); + if (t->stdin_fd < 0) + return -errno; + + r = transfer_start(t); + if (r < 0) + return r; + + object = t->object_path; + id = t->id; + t = NULL; + + return sd_bus_reply_method_return(msg, "uo", id, object); +} + +static int method_export_tar_or_raw(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + _cleanup_(transfer_unrefp) Transfer *t = NULL; + int fd, r; + const char *local, *object, *format; + Manager *m = userdata; + TransferType type; + uint32_t id; + + assert(msg); + assert(m); + + r = bus_verify_polkit_async( + msg, + CAP_SYS_ADMIN, + "org.freedesktop.import1.export", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = sd_bus_message_read(msg, "shs", &local, &fd, &format); + if (r < 0) + return r; + + if (!machine_name_is_valid(local)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Local name %s is invalid", local); + + type = streq_ptr(sd_bus_message_get_member(msg), "ExportTar") ? TRANSFER_EXPORT_TAR : TRANSFER_EXPORT_RAW; + + r = transfer_new(m, &t); + if (r < 0) + return r; + + t->type = type; + + if (!isempty(format)) { + t->format = strdup(format); + if (!t->format) + return -ENOMEM; + } + + t->local = strdup(local); + if (!t->local) + return -ENOMEM; + + t->stdout_fd = fcntl(fd, F_DUPFD_CLOEXEC, 3); + if (t->stdout_fd < 0) + return -errno; + + r = transfer_start(t); + if (r < 0) + return r; + + object = t->object_path; + id = t->id; + t = NULL; + + return sd_bus_reply_method_return(msg, "uo", id, object); +} + +static int method_pull_tar_or_raw(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + _cleanup_(transfer_unrefp) Transfer *t = NULL; + const char *remote, *local, *verify, *object; + Manager *m = userdata; + ImportVerify v; + TransferType type; + int force, r; + uint32_t id; + + assert(msg); + assert(m); + + r = bus_verify_polkit_async( + msg, + CAP_SYS_ADMIN, + "org.freedesktop.import1.pull", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = sd_bus_message_read(msg, "sssb", &remote, &local, &verify, &force); + if (r < 0) + return r; + + if (!http_url_is_valid(remote)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "URL %s is invalid", remote); + + if (isempty(local)) + local = NULL; + else if (!machine_name_is_valid(local)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Local name %s is invalid", local); + + if (isempty(verify)) + v = IMPORT_VERIFY_SIGNATURE; + else + v = import_verify_from_string(verify); + if (v < 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Unknown verification mode %s", verify); + + r = setup_machine_directory((uint64_t) -1, error); + if (r < 0) + return r; + + type = streq_ptr(sd_bus_message_get_member(msg), "PullTar") ? TRANSFER_PULL_TAR : TRANSFER_PULL_RAW; + + if (manager_find(m, type, remote)) + return sd_bus_error_setf(error, BUS_ERROR_TRANSFER_IN_PROGRESS, "Transfer for %s already in progress.", remote); + + r = transfer_new(m, &t); + if (r < 0) + return r; + + t->type = type; + t->verify = v; + t->force_local = force; + + t->remote = strdup(remote); + if (!t->remote) + return -ENOMEM; + + if (local) { + t->local = strdup(local); + if (!t->local) + return -ENOMEM; + } + + r = transfer_start(t); + if (r < 0) + return r; + + object = t->object_path; + id = t->id; + t = NULL; + + return sd_bus_reply_method_return(msg, "uo", id, object); +} + +static int method_list_transfers(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + Manager *m = userdata; + Transfer *t; + Iterator i; + int r; + + assert(msg); + assert(m); + + r = sd_bus_message_new_method_return(msg, &reply); + if (r < 0) + return r; + + r = sd_bus_message_open_container(reply, 'a', "(usssdo)"); + if (r < 0) + return r; + + HASHMAP_FOREACH(t, m->transfers, i) { + + r = sd_bus_message_append( + reply, + "(usssdo)", + t->id, + transfer_type_to_string(t->type), + t->remote, + t->local, + (double) t->progress_percent / 100.0, + t->object_path); + if (r < 0) + return r; + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +static int method_cancel(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + Transfer *t = userdata; + int r; + + assert(msg); + assert(t); + + r = bus_verify_polkit_async( + msg, + CAP_SYS_ADMIN, + "org.freedesktop.import1.pull", + NULL, + false, + UID_INVALID, + &t->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = transfer_cancel(t); + if (r < 0) + return r; + + return sd_bus_reply_method_return(msg, NULL); +} + +static int method_cancel_transfer(sd_bus_message *msg, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Transfer *t; + uint32_t id; + int r; + + assert(msg); + assert(m); + + r = bus_verify_polkit_async( + msg, + CAP_SYS_ADMIN, + "org.freedesktop.import1.pull", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = sd_bus_message_read(msg, "u", &id); + if (r < 0) + return r; + if (id <= 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid transfer id"); + + t = hashmap_get(m->transfers, UINT32_TO_PTR(id)); + if (!t) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_TRANSFER, "No transfer by id %" PRIu32, id); + + r = transfer_cancel(t); + if (r < 0) + return r; + + return sd_bus_reply_method_return(msg, NULL); +} + +static int property_get_progress( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + Transfer *t = userdata; + + assert(bus); + assert(reply); + assert(t); + + return sd_bus_message_append(reply, "d", (double) t->progress_percent / 100.0); +} + +static BUS_DEFINE_PROPERTY_GET_ENUM(property_get_type, transfer_type, TransferType); +static BUS_DEFINE_PROPERTY_GET_ENUM(property_get_verify, import_verify, ImportVerify); + +static const sd_bus_vtable transfer_vtable[] = { + SD_BUS_VTABLE_START(0), + SD_BUS_PROPERTY("Id", "u", NULL, offsetof(Transfer, id), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Local", "s", NULL, offsetof(Transfer, local), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Remote", "s", NULL, offsetof(Transfer, remote), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Type", "s", property_get_type, offsetof(Transfer, type), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Verify", "s", property_get_verify, offsetof(Transfer, verify), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Progress", "d", property_get_progress, 0, 0), + SD_BUS_METHOD("Cancel", NULL, NULL, method_cancel, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_SIGNAL("LogMessage", "us", 0), + SD_BUS_VTABLE_END, +}; + +static const sd_bus_vtable manager_vtable[] = { + SD_BUS_VTABLE_START(0), + SD_BUS_METHOD("ImportTar", "hsbb", "uo", method_import_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ImportRaw", "hsbb", "uo", method_import_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ExportTar", "shs", "uo", method_export_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ExportRaw", "shs", "uo", method_export_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("PullTar", "sssb", "uo", method_pull_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("PullRaw", "sssb", "uo", method_pull_tar_or_raw, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ListTransfers", NULL, "a(usssdo)", method_list_transfers, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CancelTransfer", "u", NULL, method_cancel_transfer, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_SIGNAL("TransferNew", "uo", 0), + SD_BUS_SIGNAL("TransferRemoved", "uos", 0), + SD_BUS_VTABLE_END, +}; + +static int transfer_object_find(sd_bus *bus, const char *path, const char *interface, void *userdata, void **found, sd_bus_error *error) { + Manager *m = userdata; + Transfer *t; + const char *p; + uint32_t id; + int r; + + assert(bus); + assert(path); + assert(interface); + assert(found); + assert(m); + + p = startswith(path, "/org/freedesktop/import1/transfer/_"); + if (!p) + return 0; + + r = safe_atou32(p, &id); + if (r < 0 || id == 0) + return 0; + + t = hashmap_get(m->transfers, UINT32_TO_PTR(id)); + if (!t) + return 0; + + *found = t; + return 1; +} + +static int transfer_node_enumerator(sd_bus *bus, const char *path, void *userdata, char ***nodes, sd_bus_error *error) { + _cleanup_strv_free_ char **l = NULL; + Manager *m = userdata; + Transfer *t; + unsigned k = 0; + Iterator i; + + l = new0(char*, hashmap_size(m->transfers) + 1); + if (!l) + return -ENOMEM; + + HASHMAP_FOREACH(t, m->transfers, i) { + + l[k] = strdup(t->object_path); + if (!l[k]) + return -ENOMEM; + + k++; + } + + *nodes = l; + l = NULL; + + return 1; +} + +static int manager_add_bus_objects(Manager *m) { + int r; + + assert(m); + + r = sd_bus_add_object_vtable(m->bus, NULL, "/org/freedesktop/import1", "org.freedesktop.import1.Manager", manager_vtable, m); + if (r < 0) + return log_error_errno(r, "Failed to register object: %m"); + + r = sd_bus_add_fallback_vtable(m->bus, NULL, "/org/freedesktop/import1/transfer", "org.freedesktop.import1.Transfer", transfer_vtable, transfer_object_find, m); + if (r < 0) + return log_error_errno(r, "Failed to register object: %m"); + + r = sd_bus_add_node_enumerator(m->bus, NULL, "/org/freedesktop/import1/transfer", transfer_node_enumerator, m); + if (r < 0) + return log_error_errno(r, "Failed to add transfer enumerator: %m"); + + r = sd_bus_request_name(m->bus, "org.freedesktop.import1", 0); + if (r < 0) + return log_error_errno(r, "Failed to register name: %m"); + + r = sd_bus_attach_event(m->bus, m->event, 0); + if (r < 0) + return log_error_errno(r, "Failed to attach bus to event loop: %m"); + + return 0; +} + +static bool manager_check_idle(void *userdata) { + Manager *m = userdata; + + return hashmap_isempty(m->transfers); +} + +static int manager_run(Manager *m) { + assert(m); + + return bus_event_loop_with_idle( + m->event, + m->bus, + "org.freedesktop.import1", + DEFAULT_EXIT_USEC, + manager_check_idle, + m); +} + +int main(int argc, char *argv[]) { + _cleanup_(manager_unrefp) Manager *m = NULL; + int r; + + log_set_target(LOG_TARGET_AUTO); + log_parse_environment(); + log_open(); + + umask(0022); + + if (argc != 1) { + log_error("This program takes no arguments."); + r = -EINVAL; + goto finish; + } + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGCHLD, -1) >= 0); + + r = manager_new(&m); + if (r < 0) { + log_error_errno(r, "Failed to allocate manager object: %m"); + goto finish; + } + + r = manager_add_bus_objects(m); + if (r < 0) + goto finish; + + r = manager_run(m); + if (r < 0) { + log_error_errno(r, "Failed to run event loop: %m"); + goto finish; + } + +finish: + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.conf b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.conf new file mode 100644 index 0000000000..ed2539a03b --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.conf @@ -0,0 +1,62 @@ +<?xml version="1.0"?> <!--*-nxml-*--> +<!DOCTYPE busconfig PUBLIC "-//freedesktop//DTD D-BUS Bus Configuration 1.0//EN" + "http://www.freedesktop.org/standards/dbus/1.0/busconfig.dtd"> + +<!-- + This file is part of systemd. + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. +--> + +<busconfig> + + <policy user="root"> + <allow own="org.freedesktop.import1"/> + <allow send_destination="org.freedesktop.import1"/> + <allow receive_sender="org.freedesktop.import1"/> + </policy> + + <policy context="default"> + <deny send_destination="org.freedesktop.import1"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.DBus.Introspectable"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.DBus.Peer"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.DBus.Properties" + send_member="Get"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.DBus.Properties" + send_member="GetAll"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.import1.Manager" + send_member="ListTransfers"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.import1.Manager" + send_member="CancelTransfer"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.import1.Manager" + send_member="PullTar"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.import1.Manager" + send_member="PullRaw"/> + + <allow send_destination="org.freedesktop.import1" + send_interface="org.freedesktop.import1.Transfer" + send_member="Cancel"/> + + <allow receive_sender="org.freedesktop.import1"/> + </policy> + +</busconfig> diff --git a/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.policy.in b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.policy.in new file mode 100644 index 0000000000..85924ed743 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.policy.in @@ -0,0 +1,49 @@ +<?xml version="1.0" encoding="UTF-8"?> <!--*-nxml-*--> +<!DOCTYPE policyconfig PUBLIC "-//freedesktop//DTD PolicyKit Policy Configuration 1.0//EN" + "http://www.freedesktop.org/standards/PolicyKit/1/policyconfig.dtd"> + +<!-- + This file is part of systemd. + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. +--> + +<policyconfig> + + <vendor>The systemd Project</vendor> + <vendor_url>http://www.freedesktop.org/wiki/Software/systemd</vendor_url> + + <action id="org.freedesktop.import1.import"> + <_description>Import a VM or container image</_description> + <_message>Authentication is required to import a VM or container image</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.import1.export"> + <_description>Export a VM or container image</_description> + <_message>Authentication is required to export a VM or container image</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.import1.pull"> + <_description>Download a VM or container image</_description> + <_message>Authentication is required to download a VM or container image</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + +</policyconfig> diff --git a/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.service b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.service new file mode 100644 index 0000000000..8fc4c47881 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/org.freedesktop.import1.service @@ -0,0 +1,12 @@ +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[D-BUS Service] +Name=org.freedesktop.import1 +Exec=/bin/false +User=root +SystemdService=dbus-org.freedesktop.import1.service diff --git a/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.in b/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.in new file mode 100644 index 0000000000..ac27c2bcba --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.in @@ -0,0 +1,23 @@ +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[Unit] +Description=Virtual Machine and Container Download Service +Documentation=man:systemd-importd.service(8) +Documentation=http://www.freedesktop.org/wiki/Software/systemd/importd + +[Service] +ExecStart=@rootlibexecdir@/systemd-importd +BusName=org.freedesktop.import1 +WatchdogSec=3min +KillMode=mixed +CapabilityBoundingSet=CAP_CHOWN CAP_FOWNER CAP_FSETID CAP_MKNOD CAP_SETFCAP CAP_SYS_ADMIN CAP_SETPCAP CAP_DAC_OVERRIDE +NoNewPrivileges=yes +MemoryDenyWriteExecute=yes +RestrictRealtime=yes +RestrictAddressFamilies=AF_UNIX AF_INET AF_INET6 +SystemCallFilter=~@clock @cpu-emulation @debug @keyring @module @obsolete @raw-io diff --git a/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.xml b/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.xml new file mode 100644 index 0000000000..8fdced475c --- /dev/null +++ b/src/grp-machine/grp-import/systemd-importd/systemd-importd.service.xml @@ -0,0 +1,82 @@ +<?xml version='1.0'?> <!--*-nxml-*--> +<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" + "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"> + +<!-- + This file is part of systemd. + + Copyright 2016 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +--> + +<refentry id="systemd-importd.service" conditional='ENABLE_IMPORTD'> + + <refentryinfo> + <title>systemd-importd.service</title> + <productname>systemd</productname> + + <authorgroup> + <author> + <contrib>Developer</contrib> + <firstname>Lennart</firstname> + <surname>Poettering</surname> + <email>lennart@poettering.net</email> + </author> + </authorgroup> + </refentryinfo> + + <refmeta> + <refentrytitle>systemd-importd.service</refentrytitle> + <manvolnum>8</manvolnum> + </refmeta> + + <refnamediv> + <refname>systemd-importd.service</refname> + <refname>systemd-importd</refname> + <refpurpose>VM and container image import and export service</refpurpose> + </refnamediv> + + <refsynopsisdiv> + <para><filename>systemd-importd.service</filename></para> + <para><filename>/usr/lib/systemd/systemd-importd</filename></para> + </refsynopsisdiv> + + <refsect1> + <title>Description</title> + + <para><command>systemd-importd</command> is a system service that allows importing, exporting and downloading of + system images suitable for running as VM or containers. It is a companion service for + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>, and provides the implementation for + <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>'s + <command>pull-raw</command>, <command>pull-tar</command>, <command>import-raw</command>, + <command>import-tar</command>, <command>export-raw</command>, and <command>export-tar</command> commands.</para> + + <para>See the + <ulink url="http://www.freedesktop.org/wiki/Software/systemd/importd"> + importd D-Bus API Documentation</ulink> for information about the + APIs <filename>systemd-importd</filename> provides.</para> + </refsect1> + + <refsect1> + <title>See Also</title> + <para> + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + </para> + </refsect1> + +</refentry> diff --git a/src/grp-machine/grp-import/systemd-pull/GNUmakefile b/src/grp-machine/grp-import/systemd-pull/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/grp-import/systemd-pull/Makefile b/src/grp-machine/grp-import/systemd-pull/Makefile new file mode 100644 index 0000000000..b89d456f37 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/Makefile @@ -0,0 +1,66 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +rootlibexec_PROGRAMS += systemd-pull +systemd_pull_SOURCES = \ + src/import/pull.c \ + src/import/pull-raw.c \ + src/import/pull-raw.h \ + src/import/pull-tar.c \ + src/import/pull-tar.h \ + src/import/pull-job.c \ + src/import/pull-job.h \ + src/import/pull-common.c \ + src/import/pull-common.h \ + src/import/import-common.c \ + src/import/import-common.h \ + src/import/import-compress.c \ + src/import/import-compress.h \ + src/import/curl-util.c \ + src/import/curl-util.h \ + src/import/qcow2-util.c \ + src/import/qcow2-util.h + +systemd_pull_CFLAGS = \ + $(LIBCURL_CFLAGS) \ + $(XZ_CFLAGS) \ + $(ZLIB_CFLAGS) \ + $(BZIP2_CFLAGS) \ + $(GCRYPT_CFLAGS) \ + -D VENDOR_KEYRING_PATH=\"$(rootlibexecdir)/import-pubring.gpg\" \ + -D USER_KEYRING_PATH=\"$(pkgsysconfdir)/import-pubring.gpg\" + +systemd_pull_LDADD = \ + libsystemd-shared.la \ + $(LIBCURL_LIBS) \ + $(XZ_LIBS) \ + $(ZLIB_LIBS) \ + $(BZIP2_LIBS) \ + $(GCRYPT_LIBS) + +dist_rootlibexec_DATA += \ + src/import/import-pubring.gpg + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/grp-import/systemd-pull/curl-util.c b/src/grp-machine/grp-import/systemd-pull/curl-util.c new file mode 100644 index 0000000000..141ce9e875 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/curl-util.c @@ -0,0 +1,447 @@ +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/string-util.h" + +#include "curl-util.h" + +static void curl_glue_check_finished(CurlGlue *g) { + CURLMsg *msg; + int k = 0; + + assert(g); + + msg = curl_multi_info_read(g->curl, &k); + if (!msg) + return; + + if (msg->msg != CURLMSG_DONE) + return; + + if (g->on_finished) + g->on_finished(g, msg->easy_handle, msg->data.result); +} + +static int curl_glue_on_io(sd_event_source *s, int fd, uint32_t revents, void *userdata) { + CurlGlue *g = userdata; + int action, k = 0, translated_fd; + + assert(s); + assert(g); + + translated_fd = PTR_TO_FD(hashmap_get(g->translate_fds, FD_TO_PTR(fd))); + + if ((revents & (EPOLLIN|EPOLLOUT)) == (EPOLLIN|EPOLLOUT)) + action = CURL_POLL_INOUT; + else if (revents & EPOLLIN) + action = CURL_POLL_IN; + else if (revents & EPOLLOUT) + action = CURL_POLL_OUT; + else + action = 0; + + if (curl_multi_socket_action(g->curl, translated_fd, action, &k) < 0) { + log_debug("Failed to propagate IO event."); + return -EINVAL; + } + + curl_glue_check_finished(g); + return 0; +} + +static int curl_glue_socket_callback(CURLM *curl, curl_socket_t s, int action, void *userdata, void *socketp) { + sd_event_source *io; + CurlGlue *g = userdata; + uint32_t events = 0; + int r; + + assert(curl); + assert(g); + + io = hashmap_get(g->ios, FD_TO_PTR(s)); + + if (action == CURL_POLL_REMOVE) { + if (io) { + int fd; + + fd = sd_event_source_get_io_fd(io); + assert(fd >= 0); + + sd_event_source_set_enabled(io, SD_EVENT_OFF); + sd_event_source_unref(io); + + hashmap_remove(g->ios, FD_TO_PTR(s)); + hashmap_remove(g->translate_fds, FD_TO_PTR(fd)); + + safe_close(fd); + } + + return 0; + } + + r = hashmap_ensure_allocated(&g->ios, &trivial_hash_ops); + if (r < 0) { + log_oom(); + return -1; + } + + r = hashmap_ensure_allocated(&g->translate_fds, &trivial_hash_ops); + if (r < 0) { + log_oom(); + return -1; + } + + if (action == CURL_POLL_IN) + events = EPOLLIN; + else if (action == CURL_POLL_OUT) + events = EPOLLOUT; + else if (action == CURL_POLL_INOUT) + events = EPOLLIN|EPOLLOUT; + + if (io) { + if (sd_event_source_set_io_events(io, events) < 0) + return -1; + + if (sd_event_source_set_enabled(io, SD_EVENT_ON) < 0) + return -1; + } else { + _cleanup_close_ int fd = -1; + + /* When curl needs to remove an fd from us it closes + * the fd first, and only then calls into us. This is + * nasty, since we cannot pass the fd on to epoll() + * anymore. Hence, duplicate the fds here, and keep a + * copy for epoll which we control after use. */ + + fd = fcntl(s, F_DUPFD_CLOEXEC, 3); + if (fd < 0) + return -1; + + if (sd_event_add_io(g->event, &io, fd, events, curl_glue_on_io, g) < 0) + return -1; + + (void) sd_event_source_set_description(io, "curl-io"); + + r = hashmap_put(g->ios, FD_TO_PTR(s), io); + if (r < 0) { + log_oom(); + sd_event_source_unref(io); + return -1; + } + + r = hashmap_put(g->translate_fds, FD_TO_PTR(fd), FD_TO_PTR(s)); + if (r < 0) { + log_oom(); + hashmap_remove(g->ios, FD_TO_PTR(s)); + sd_event_source_unref(io); + return -1; + } + + fd = -1; + } + + return 0; +} + +static int curl_glue_on_timer(sd_event_source *s, uint64_t usec, void *userdata) { + CurlGlue *g = userdata; + int k = 0; + + assert(s); + assert(g); + + if (curl_multi_socket_action(g->curl, CURL_SOCKET_TIMEOUT, 0, &k) != CURLM_OK) { + log_debug("Failed to propagate timeout."); + return -EINVAL; + } + + curl_glue_check_finished(g); + return 0; +} + +static int curl_glue_timer_callback(CURLM *curl, long timeout_ms, void *userdata) { + CurlGlue *g = userdata; + usec_t usec; + + assert(curl); + assert(g); + + if (timeout_ms < 0) { + if (g->timer) { + if (sd_event_source_set_enabled(g->timer, SD_EVENT_OFF) < 0) + return -1; + } + + return 0; + } + + usec = now(clock_boottime_or_monotonic()) + (usec_t) timeout_ms * USEC_PER_MSEC + USEC_PER_MSEC - 1; + + if (g->timer) { + if (sd_event_source_set_time(g->timer, usec) < 0) + return -1; + + if (sd_event_source_set_enabled(g->timer, SD_EVENT_ONESHOT) < 0) + return -1; + } else { + if (sd_event_add_time(g->event, &g->timer, clock_boottime_or_monotonic(), usec, 0, curl_glue_on_timer, g) < 0) + return -1; + + (void) sd_event_source_set_description(g->timer, "curl-timer"); + } + + return 0; +} + +CurlGlue *curl_glue_unref(CurlGlue *g) { + sd_event_source *io; + + if (!g) + return NULL; + + if (g->curl) + curl_multi_cleanup(g->curl); + + while ((io = hashmap_steal_first(g->ios))) { + int fd; + + fd = sd_event_source_get_io_fd(io); + assert(fd >= 0); + + hashmap_remove(g->translate_fds, FD_TO_PTR(fd)); + + safe_close(fd); + sd_event_source_unref(io); + } + + hashmap_free(g->ios); + + sd_event_source_unref(g->timer); + sd_event_unref(g->event); + return mfree(g); +} + +int curl_glue_new(CurlGlue **glue, sd_event *event) { + _cleanup_(curl_glue_unrefp) CurlGlue *g = NULL; + int r; + + g = new0(CurlGlue, 1); + if (!g) + return -ENOMEM; + + if (event) + g->event = sd_event_ref(event); + else { + r = sd_event_default(&g->event); + if (r < 0) + return r; + } + + g->curl = curl_multi_init(); + if (!g->curl) + return -ENOMEM; + + if (curl_multi_setopt(g->curl, CURLMOPT_SOCKETDATA, g) != CURLM_OK) + return -EINVAL; + + if (curl_multi_setopt(g->curl, CURLMOPT_SOCKETFUNCTION, curl_glue_socket_callback) != CURLM_OK) + return -EINVAL; + + if (curl_multi_setopt(g->curl, CURLMOPT_TIMERDATA, g) != CURLM_OK) + return -EINVAL; + + if (curl_multi_setopt(g->curl, CURLMOPT_TIMERFUNCTION, curl_glue_timer_callback) != CURLM_OK) + return -EINVAL; + + *glue = g; + g = NULL; + + return 0; +} + +int curl_glue_make(CURL **ret, const char *url, void *userdata) { + const char *useragent; + CURL *c; + int r; + + assert(ret); + assert(url); + + c = curl_easy_init(); + if (!c) + return -ENOMEM; + + /* curl_easy_setopt(c, CURLOPT_VERBOSE, 1L); */ + + if (curl_easy_setopt(c, CURLOPT_URL, url) != CURLE_OK) { + r = -EIO; + goto fail; + } + + if (curl_easy_setopt(c, CURLOPT_PRIVATE, userdata) != CURLE_OK) { + r = -EIO; + goto fail; + } + + useragent = strjoina(program_invocation_short_name, "/" PACKAGE_VERSION); + if (curl_easy_setopt(c, CURLOPT_USERAGENT, useragent) != CURLE_OK) { + r = -EIO; + goto fail; + } + + if (curl_easy_setopt(c, CURLOPT_FOLLOWLOCATION, 1L) != CURLE_OK) { + r = -EIO; + goto fail; + } + + *ret = c; + return 0; + +fail: + curl_easy_cleanup(c); + return r; +} + +int curl_glue_add(CurlGlue *g, CURL *c) { + assert(g); + assert(c); + + if (curl_multi_add_handle(g->curl, c) != CURLM_OK) + return -EIO; + + return 0; +} + +void curl_glue_remove_and_free(CurlGlue *g, CURL *c) { + assert(g); + + if (!c) + return; + + if (g->curl) + curl_multi_remove_handle(g->curl, c); + + curl_easy_cleanup(c); +} + +struct curl_slist *curl_slist_new(const char *first, ...) { + struct curl_slist *l; + va_list ap; + + if (!first) + return NULL; + + l = curl_slist_append(NULL, first); + if (!l) + return NULL; + + va_start(ap, first); + + for (;;) { + struct curl_slist *n; + const char *i; + + i = va_arg(ap, const char*); + if (!i) + break; + + n = curl_slist_append(l, i); + if (!n) { + va_end(ap); + curl_slist_free_all(l); + return NULL; + } + + l = n; + } + + va_end(ap); + return l; +} + +int curl_header_strdup(const void *contents, size_t sz, const char *field, char **value) { + const char *p = contents; + size_t l; + char *s; + + l = strlen(field); + if (sz < l) + return 0; + + if (memcmp(p, field, l) != 0) + return 0; + + p += l; + sz -= l; + + if (memchr(p, 0, sz)) + return 0; + + /* Skip over preceeding whitespace */ + while (sz > 0 && strchr(WHITESPACE, p[0])) { + p++; + sz--; + } + + /* Truncate trailing whitespace*/ + while (sz > 0 && strchr(WHITESPACE, p[sz-1])) + sz--; + + s = strndup(p, sz); + if (!s) + return -ENOMEM; + + *value = s; + return 1; +} + +int curl_parse_http_time(const char *t, usec_t *ret) { + const char *e; + locale_t loc; + struct tm tm; + time_t v; + + assert(t); + assert(ret); + + loc = newlocale(LC_TIME_MASK, "C", (locale_t) 0); + if (loc == (locale_t) 0) + return -errno; + + /* RFC822 */ + e = strptime_l(t, "%a, %d %b %Y %H:%M:%S %Z", &tm, loc); + if (!e || *e != 0) + /* RFC 850 */ + e = strptime_l(t, "%A, %d-%b-%y %H:%M:%S %Z", &tm, loc); + if (!e || *e != 0) + /* ANSI C */ + e = strptime_l(t, "%a %b %d %H:%M:%S %Y", &tm, loc); + freelocale(loc); + if (!e || *e != 0) + return -EINVAL; + + v = timegm(&tm); + if (v == (time_t) -1) + return -EINVAL; + + *ret = (usec_t) v * USEC_PER_SEC; + return 0; +} diff --git a/src/grp-machine/grp-import/systemd-pull/curl-util.h b/src/grp-machine/grp-import/systemd-pull/curl-util.h new file mode 100644 index 0000000000..1619aeb18c --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/curl-util.h @@ -0,0 +1,56 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <curl/curl.h> +#include <sys/types.h> + +#include <systemd/sd-event.h> + +#include "systemd-basic/hashmap.h" + +typedef struct CurlGlue CurlGlue; + +struct CurlGlue { + sd_event *event; + CURLM *curl; + sd_event_source *timer; + Hashmap *ios; + Hashmap *translate_fds; + + void (*on_finished)(CurlGlue *g, CURL *curl, CURLcode code); + void *userdata; +}; + +int curl_glue_new(CurlGlue **glue, sd_event *event); +CurlGlue* curl_glue_unref(CurlGlue *glue); + +DEFINE_TRIVIAL_CLEANUP_FUNC(CurlGlue*, curl_glue_unref); + +int curl_glue_make(CURL **ret, const char *url, void *userdata); +int curl_glue_add(CurlGlue *g, CURL *c); +void curl_glue_remove_and_free(CurlGlue *g, CURL *c); + +struct curl_slist *curl_slist_new(const char *first, ...) _sentinel_; +int curl_header_strdup(const void *contents, size_t sz, const char *field, char **value); +int curl_parse_http_time(const char *t, usec_t *ret); + +DEFINE_TRIVIAL_CLEANUP_FUNC(CURL*, curl_easy_cleanup); +DEFINE_TRIVIAL_CLEANUP_FUNC(struct curl_slist*, curl_slist_free_all); diff --git a/src/grp-machine/grp-import/systemd-pull/import-pubring.gpg b/src/grp-machine/grp-import/systemd-pull/import-pubring.gpg Binary files differnew file mode 100644 index 0000000000..be27776896 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/import-pubring.gpg diff --git a/src/grp-machine/grp-import/systemd-pull/pull-common.c b/src/grp-machine/grp-import/systemd-pull/pull-common.c new file mode 100644 index 0000000000..9830d8a10f --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-common.c @@ -0,0 +1,548 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sys/prctl.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/capability-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/dirent-util.h" +#include "systemd-basic/escape.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/rm-rf.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/siphash24.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/util.h" +#include "systemd-basic/web-util.h" + +#include "pull-common.h" +#include "pull-job.h" + +#define FILENAME_ESCAPE "/.#\"\'" +#define HASH_URL_THRESHOLD_LENGTH (_POSIX_PATH_MAX - 16) + +int pull_find_old_etags( + const char *url, + const char *image_root, + int dt, + const char *prefix, + const char *suffix, + char ***etags) { + + _cleanup_free_ char *escaped_url = NULL; + _cleanup_closedir_ DIR *d = NULL; + _cleanup_strv_free_ char **l = NULL; + struct dirent *de; + int r; + + assert(url); + assert(etags); + + if (!image_root) + image_root = "/var/lib/machines"; + + escaped_url = xescape(url, FILENAME_ESCAPE); + if (!escaped_url) + return -ENOMEM; + + d = opendir(image_root); + if (!d) { + if (errno == ENOENT) { + *etags = NULL; + return 0; + } + + return -errno; + } + + FOREACH_DIRENT_ALL(de, d, return -errno) { + const char *a, *b; + char *u; + + if (de->d_type != DT_UNKNOWN && + de->d_type != dt) + continue; + + if (prefix) { + a = startswith(de->d_name, prefix); + if (!a) + continue; + } else + a = de->d_name; + + a = startswith(a, escaped_url); + if (!a) + continue; + + a = startswith(a, "."); + if (!a) + continue; + + if (suffix) { + b = endswith(de->d_name, suffix); + if (!b) + continue; + } else + b = strchr(de->d_name, 0); + + if (a >= b) + continue; + + r = cunescape_length(a, b - a, 0, &u); + if (r < 0) + return r; + + if (!http_etag_is_valid(u)) { + free(u); + continue; + } + + r = strv_consume(&l, u); + if (r < 0) + return r; + } + + *etags = l; + l = NULL; + + return 0; +} + +int pull_make_local_copy(const char *final, const char *image_root, const char *local, bool force_local) { + const char *p; + int r; + + assert(final); + assert(local); + + if (!image_root) + image_root = "/var/lib/machines"; + + p = strjoina(image_root, "/", local); + + if (force_local) + (void) rm_rf(p, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + + r = btrfs_subvol_snapshot(final, p, BTRFS_SNAPSHOT_QUOTA); + if (r == -ENOTTY) { + r = copy_tree(final, p, false); + if (r < 0) + return log_error_errno(r, "Failed to copy image: %m"); + } else if (r < 0) + return log_error_errno(r, "Failed to create local image: %m"); + + log_info("Created new local image '%s'.", local); + + return 0; +} + +static int hash_url(const char *url, char **ret) { + uint64_t h; + static const sd_id128_t k = SD_ID128_ARRAY(df,89,16,87,01,cc,42,30,98,ab,4a,19,a6,a5,63,4f); + + assert(url); + + h = siphash24(url, strlen(url), k.bytes); + if (asprintf(ret, "%"PRIx64, h) < 0) + return -ENOMEM; + + return 0; +} + +int pull_make_path(const char *url, const char *etag, const char *image_root, const char *prefix, const char *suffix, char **ret) { + _cleanup_free_ char *escaped_url = NULL, *escaped_etag = NULL; + char *path; + + assert(url); + assert(ret); + + if (!image_root) + image_root = "/var/lib/machines"; + + escaped_url = xescape(url, FILENAME_ESCAPE); + if (!escaped_url) + return -ENOMEM; + + if (etag) { + escaped_etag = xescape(etag, FILENAME_ESCAPE); + if (!escaped_etag) + return -ENOMEM; + } + + path = strjoin(image_root, "/", strempty(prefix), escaped_url, escaped_etag ? "." : "", + strempty(escaped_etag), strempty(suffix), NULL); + if (!path) + return -ENOMEM; + + /* URLs might make the path longer than the maximum allowed length for a file name. + * When that happens, a URL hash is used instead. Paths returned by this function + * can be later used with tempfn_random() which adds 16 bytes to the resulting name. */ + if (strlen(path) >= HASH_URL_THRESHOLD_LENGTH) { + _cleanup_free_ char *hash = NULL; + int r; + + free(path); + + r = hash_url(url, &hash); + if (r < 0) + return r; + + path = strjoin(image_root, "/", strempty(prefix), hash, escaped_etag ? "." : "", + strempty(escaped_etag), strempty(suffix), NULL); + if (!path) + return -ENOMEM; + } + + *ret = path; + return 0; +} + +int pull_make_settings_job( + PullJob **ret, + const char *url, + CurlGlue *glue, + PullJobFinished on_finished, + void *userdata) { + + _cleanup_free_ char *last_component = NULL, *ll = NULL, *settings_url = NULL; + _cleanup_(pull_job_unrefp) PullJob *job = NULL; + const char *q; + int r; + + assert(ret); + assert(url); + assert(glue); + + r = import_url_last_component(url, &last_component); + if (r < 0) + return r; + + r = tar_strip_suffixes(last_component, &ll); + if (r < 0) + return r; + + q = strjoina(ll, ".nspawn"); + + r = import_url_change_last_component(url, q, &settings_url); + if (r < 0) + return r; + + r = pull_job_new(&job, settings_url, glue, userdata); + if (r < 0) + return r; + + job->on_finished = on_finished; + job->compressed_max = job->uncompressed_max = 1ULL * 1024ULL * 1024ULL; + + *ret = job; + job = NULL; + + return 0; +} + +int pull_make_verification_jobs( + PullJob **ret_checksum_job, + PullJob **ret_signature_job, + ImportVerify verify, + const char *url, + CurlGlue *glue, + PullJobFinished on_finished, + void *userdata) { + + _cleanup_(pull_job_unrefp) PullJob *checksum_job = NULL, *signature_job = NULL; + int r; + + assert(ret_checksum_job); + assert(ret_signature_job); + assert(verify >= 0); + assert(verify < _IMPORT_VERIFY_MAX); + assert(url); + assert(glue); + + if (verify != IMPORT_VERIFY_NO) { + _cleanup_free_ char *checksum_url = NULL; + + /* Queue job for the SHA256SUMS file for the image */ + r = import_url_change_last_component(url, "SHA256SUMS", &checksum_url); + if (r < 0) + return r; + + r = pull_job_new(&checksum_job, checksum_url, glue, userdata); + if (r < 0) + return r; + + checksum_job->on_finished = on_finished; + checksum_job->uncompressed_max = checksum_job->compressed_max = 1ULL * 1024ULL * 1024ULL; + } + + if (verify == IMPORT_VERIFY_SIGNATURE) { + _cleanup_free_ char *signature_url = NULL; + + /* Queue job for the SHA256SUMS.gpg file for the image. */ + r = import_url_change_last_component(url, "SHA256SUMS.gpg", &signature_url); + if (r < 0) + return r; + + r = pull_job_new(&signature_job, signature_url, glue, userdata); + if (r < 0) + return r; + + signature_job->on_finished = on_finished; + signature_job->uncompressed_max = signature_job->compressed_max = 1ULL * 1024ULL * 1024ULL; + } + + *ret_checksum_job = checksum_job; + *ret_signature_job = signature_job; + + checksum_job = signature_job = NULL; + + return 0; +} + +int pull_verify(PullJob *main_job, + PullJob *settings_job, + PullJob *checksum_job, + PullJob *signature_job) { + + _cleanup_close_pair_ int gpg_pipe[2] = { -1, -1 }; + _cleanup_free_ char *fn = NULL; + _cleanup_close_ int sig_file = -1; + const char *p, *line; + char sig_file_path[] = "/tmp/sigXXXXXX", gpg_home[] = "/tmp/gpghomeXXXXXX"; + _cleanup_(sigkill_waitp) pid_t pid = 0; + bool gpg_home_created = false; + int r; + + assert(main_job); + assert(main_job->state == PULL_JOB_DONE); + + if (!checksum_job) + return 0; + + assert(main_job->calc_checksum); + assert(main_job->checksum); + assert(checksum_job->state == PULL_JOB_DONE); + + if (!checksum_job->payload || checksum_job->payload_size <= 0) { + log_error("Checksum is empty, cannot verify."); + return -EBADMSG; + } + + r = import_url_last_component(main_job->url, &fn); + if (r < 0) + return log_oom(); + + if (!filename_is_valid(fn)) { + log_error("Cannot verify checksum, could not determine valid server-side file name."); + return -EBADMSG; + } + + line = strjoina(main_job->checksum, " *", fn, "\n"); + + p = memmem(checksum_job->payload, + checksum_job->payload_size, + line, + strlen(line)); + + if (!p || (p != (char*) checksum_job->payload && p[-1] != '\n')) { + log_error("DOWNLOAD INVALID: Checksum did not check out, payload has been tampered with."); + return -EBADMSG; + } + + log_info("SHA256 checksum of %s is valid.", main_job->url); + + assert(!settings_job || IN_SET(settings_job->state, PULL_JOB_DONE, PULL_JOB_FAILED)); + + if (settings_job && + settings_job->state == PULL_JOB_DONE && + settings_job->error == 0 && + !settings_job->etag_exists) { + + _cleanup_free_ char *settings_fn = NULL; + + assert(settings_job->calc_checksum); + assert(settings_job->checksum); + + r = import_url_last_component(settings_job->url, &settings_fn); + if (r < 0) + return log_oom(); + + if (!filename_is_valid(settings_fn)) { + log_error("Cannot verify checksum, could not determine server-side settings file name."); + return -EBADMSG; + } + + line = strjoina(settings_job->checksum, " *", settings_fn, "\n"); + + p = memmem(checksum_job->payload, + checksum_job->payload_size, + line, + strlen(line)); + + if (!p || (p != (char*) checksum_job->payload && p[-1] != '\n')) { + log_error("DOWNLOAD INVALID: Checksum of settings file did not checkout, settings file has been tampered with."); + return -EBADMSG; + } + + log_info("SHA256 checksum of %s is valid.", settings_job->url); + } + + if (!signature_job) + return 0; + + assert(signature_job->state == PULL_JOB_DONE); + + if (!signature_job->payload || signature_job->payload_size <= 0) { + log_error("Signature is empty, cannot verify."); + return -EBADMSG; + } + + r = pipe2(gpg_pipe, O_CLOEXEC); + if (r < 0) + return log_error_errno(errno, "Failed to create pipe for gpg: %m"); + + sig_file = mkostemp(sig_file_path, O_RDWR); + if (sig_file < 0) + return log_error_errno(errno, "Failed to create temporary file: %m"); + + r = loop_write(sig_file, signature_job->payload, signature_job->payload_size, false); + if (r < 0) { + log_error_errno(r, "Failed to write to temporary file: %m"); + goto finish; + } + + if (!mkdtemp(gpg_home)) { + r = log_error_errno(errno, "Failed to create tempory home for gpg: %m"); + goto finish; + } + + gpg_home_created = true; + + pid = fork(); + if (pid < 0) + return log_error_errno(errno, "Failed to fork off gpg: %m"); + if (pid == 0) { + const char *cmd[] = { + "gpg", + "--no-options", + "--no-default-keyring", + "--no-auto-key-locate", + "--no-auto-check-trustdb", + "--batch", + "--trust-model=always", + NULL, /* --homedir= */ + NULL, /* --keyring= */ + NULL, /* --verify */ + NULL, /* signature file */ + NULL, /* dash */ + NULL /* trailing NULL */ + }; + unsigned k = ELEMENTSOF(cmd) - 6; + int null_fd; + + /* Child */ + + (void) reset_all_signal_handlers(); + (void) reset_signal_mask(); + assert_se(prctl(PR_SET_PDEATHSIG, SIGTERM) == 0); + + gpg_pipe[1] = safe_close(gpg_pipe[1]); + + if (dup2(gpg_pipe[0], STDIN_FILENO) != STDIN_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (gpg_pipe[0] != STDIN_FILENO) + gpg_pipe[0] = safe_close(gpg_pipe[0]); + + null_fd = open("/dev/null", O_WRONLY|O_NOCTTY); + if (null_fd < 0) { + log_error_errno(errno, "Failed to open /dev/null: %m"); + _exit(EXIT_FAILURE); + } + + if (dup2(null_fd, STDOUT_FILENO) != STDOUT_FILENO) { + log_error_errno(errno, "Failed to dup2() fd: %m"); + _exit(EXIT_FAILURE); + } + + if (null_fd != STDOUT_FILENO) + null_fd = safe_close(null_fd); + + cmd[k++] = strjoina("--homedir=", gpg_home); + + /* We add the user keyring only to the command line + * arguments, if it's around since gpg fails + * otherwise. */ + if (access(USER_KEYRING_PATH, F_OK) >= 0) + cmd[k++] = "--keyring=" USER_KEYRING_PATH; + else + cmd[k++] = "--keyring=" VENDOR_KEYRING_PATH; + + cmd[k++] = "--verify"; + cmd[k++] = sig_file_path; + cmd[k++] = "-"; + cmd[k++] = NULL; + + stdio_unset_cloexec(); + + execvp("gpg2", (char * const *) cmd); + execvp("gpg", (char * const *) cmd); + log_error_errno(errno, "Failed to execute gpg: %m"); + _exit(EXIT_FAILURE); + } + + gpg_pipe[0] = safe_close(gpg_pipe[0]); + + r = loop_write(gpg_pipe[1], checksum_job->payload, checksum_job->payload_size, false); + if (r < 0) { + log_error_errno(r, "Failed to write to pipe: %m"); + goto finish; + } + + gpg_pipe[1] = safe_close(gpg_pipe[1]); + + r = wait_for_terminate_and_warn("gpg", pid, true); + pid = 0; + if (r < 0) + goto finish; + if (r > 0) { + log_error("DOWNLOAD INVALID: Signature verification failed."); + r = -EBADMSG; + } else { + log_info("Signature verification succeeded."); + r = 0; + } + +finish: + if (sig_file >= 0) + (void) unlink(sig_file_path); + + if (gpg_home_created) + (void) rm_rf(gpg_home, REMOVE_ROOT|REMOVE_PHYSICAL); + + return r; +} diff --git a/src/grp-machine/grp-import/systemd-pull/pull-common.h b/src/grp-machine/grp-import/systemd-pull/pull-common.h new file mode 100644 index 0000000000..1eb8f27fe8 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-common.h @@ -0,0 +1,37 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <stdbool.h> + +#include "systemd-shared/import-util.h" + +#include "pull-job.h" + +int pull_make_local_copy(const char *final, const char *root, const char *local, bool force_local); + +int pull_find_old_etags(const char *url, const char *root, int dt, const char *prefix, const char *suffix, char ***etags); + +int pull_make_path(const char *url, const char *etag, const char *image_root, const char *prefix, const char *suffix, char **ret); + +int pull_make_settings_job(PullJob **ret, const char *url, CurlGlue *glue, PullJobFinished on_finished, void *userdata); +int pull_make_verification_jobs(PullJob **ret_checksum_job, PullJob **ret_signature_job, ImportVerify verify, const char *url, CurlGlue *glue, PullJobFinished on_finished, void *userdata); + +int pull_verify(PullJob *main_job, PullJob *settings_job, PullJob *checksum_job, PullJob *signature_job); diff --git a/src/grp-machine/grp-import/systemd-pull/pull-job.c b/src/grp-machine/grp-import/systemd-pull/pull-job.c new file mode 100644 index 0000000000..86d974952f --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-job.c @@ -0,0 +1,617 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sys/xattr.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/hexdecoct.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/parse-util.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/xattr-util.h" +#include "systemd-shared/machine-pool.h" + +#include "pull-job.h" + +PullJob* pull_job_unref(PullJob *j) { + if (!j) + return NULL; + + curl_glue_remove_and_free(j->glue, j->curl); + curl_slist_free_all(j->request_header); + + safe_close(j->disk_fd); + + import_compress_free(&j->compress); + + if (j->checksum_context) + gcry_md_close(j->checksum_context); + + free(j->url); + free(j->etag); + strv_free(j->old_etags); + free(j->payload); + free(j->checksum); + + return mfree(j); +} + +static void pull_job_finish(PullJob *j, int ret) { + assert(j); + + if (j->state == PULL_JOB_DONE || + j->state == PULL_JOB_FAILED) + return; + + if (ret == 0) { + j->state = PULL_JOB_DONE; + j->progress_percent = 100; + log_info("Download of %s complete.", j->url); + } else { + j->state = PULL_JOB_FAILED; + j->error = ret; + } + + if (j->on_finished) + j->on_finished(j); +} + +void pull_job_curl_on_finished(CurlGlue *g, CURL *curl, CURLcode result) { + PullJob *j = NULL; + CURLcode code; + long status; + int r; + + if (curl_easy_getinfo(curl, CURLINFO_PRIVATE, (char **)&j) != CURLE_OK) + return; + + if (!j || j->state == PULL_JOB_DONE || j->state == PULL_JOB_FAILED) + return; + + if (result != CURLE_OK) { + log_error("Transfer failed: %s", curl_easy_strerror(result)); + r = -EIO; + goto finish; + } + + code = curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &status); + if (code != CURLE_OK) { + log_error("Failed to retrieve response code: %s", curl_easy_strerror(code)); + r = -EIO; + goto finish; + } else if (status == 304) { + log_info("Image already downloaded. Skipping download."); + j->etag_exists = true; + r = 0; + goto finish; + } else if (status >= 300) { + log_error("HTTP request to %s failed with code %li.", j->url, status); + r = -EIO; + goto finish; + } else if (status < 200) { + log_error("HTTP request to %s finished with unexpected code %li.", j->url, status); + r = -EIO; + goto finish; + } + + if (j->state != PULL_JOB_RUNNING) { + log_error("Premature connection termination."); + r = -EIO; + goto finish; + } + + if (j->content_length != (uint64_t) -1 && + j->content_length != j->written_compressed) { + log_error("Download truncated."); + r = -EIO; + goto finish; + } + + if (j->checksum_context) { + uint8_t *k; + + k = gcry_md_read(j->checksum_context, GCRY_MD_SHA256); + if (!k) { + log_error("Failed to get checksum."); + r = -EIO; + goto finish; + } + + j->checksum = hexmem(k, gcry_md_get_algo_dlen(GCRY_MD_SHA256)); + if (!j->checksum) { + r = log_oom(); + goto finish; + } + + log_debug("SHA256 of %s is %s.", j->url, j->checksum); + } + + if (j->disk_fd >= 0 && j->allow_sparse) { + /* Make sure the file size is right, in case the file was + * sparse and we just seeked for the last part */ + + if (ftruncate(j->disk_fd, j->written_uncompressed) < 0) { + r = log_error_errno(errno, "Failed to truncate file: %m"); + goto finish; + } + + if (j->etag) + (void) fsetxattr(j->disk_fd, "user.source_etag", j->etag, strlen(j->etag), 0); + if (j->url) + (void) fsetxattr(j->disk_fd, "user.source_url", j->url, strlen(j->url), 0); + + if (j->mtime != 0) { + struct timespec ut[2]; + + timespec_store(&ut[0], j->mtime); + ut[1] = ut[0]; + (void) futimens(j->disk_fd, ut); + + (void) fd_setcrtime(j->disk_fd, j->mtime); + } + } + + r = 0; + +finish: + pull_job_finish(j, r); +} + +static int pull_job_write_uncompressed(const void *p, size_t sz, void *userdata) { + PullJob *j = userdata; + ssize_t n; + + assert(j); + assert(p); + + if (sz <= 0) + return 0; + + if (j->written_uncompressed + sz < j->written_uncompressed) { + log_error("File too large, overflow"); + return -EOVERFLOW; + } + + if (j->written_uncompressed + sz > j->uncompressed_max) { + log_error("File overly large, refusing"); + return -EFBIG; + } + + if (j->disk_fd >= 0) { + + if (j->grow_machine_directory && j->written_since_last_grow >= GROW_INTERVAL_BYTES) { + j->written_since_last_grow = 0; + grow_machine_directory(); + } + + if (j->allow_sparse) + n = sparse_write(j->disk_fd, p, sz, 64); + else + n = write(j->disk_fd, p, sz); + if (n < 0) + return log_error_errno(errno, "Failed to write file: %m"); + if ((size_t) n < sz) { + log_error("Short write"); + return -EIO; + } + } else { + + if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz)) + return log_oom(); + + memcpy(j->payload + j->payload_size, p, sz); + j->payload_size += sz; + } + + j->written_uncompressed += sz; + j->written_since_last_grow += sz; + + return 0; +} + +static int pull_job_write_compressed(PullJob *j, void *p, size_t sz) { + int r; + + assert(j); + assert(p); + + if (sz <= 0) + return 0; + + if (j->written_compressed + sz < j->written_compressed) { + log_error("File too large, overflow"); + return -EOVERFLOW; + } + + if (j->written_compressed + sz > j->compressed_max) { + log_error("File overly large, refusing."); + return -EFBIG; + } + + if (j->content_length != (uint64_t) -1 && + j->written_compressed + sz > j->content_length) { + log_error("Content length incorrect."); + return -EFBIG; + } + + if (j->checksum_context) + gcry_md_write(j->checksum_context, p, sz); + + r = import_uncompress(&j->compress, p, sz, pull_job_write_uncompressed, j); + if (r < 0) + return r; + + j->written_compressed += sz; + + return 0; +} + +static int pull_job_open_disk(PullJob *j) { + int r; + + assert(j); + + if (j->on_open_disk) { + r = j->on_open_disk(j); + if (r < 0) + return r; + } + + if (j->disk_fd >= 0) { + /* Check if we can do sparse files */ + + if (lseek(j->disk_fd, SEEK_SET, 0) == 0) + j->allow_sparse = true; + else { + if (errno != ESPIPE) + return log_error_errno(errno, "Failed to seek on file descriptor: %m"); + + j->allow_sparse = false; + } + } + + if (j->calc_checksum) { + if (gcry_md_open(&j->checksum_context, GCRY_MD_SHA256, 0) != 0) { + log_error("Failed to initialize hash context."); + return -EIO; + } + } + + return 0; +} + +static int pull_job_detect_compression(PullJob *j) { + _cleanup_free_ uint8_t *stub = NULL; + size_t stub_size; + + int r; + + assert(j); + + r = import_uncompress_detect(&j->compress, j->payload, j->payload_size); + if (r < 0) + return log_error_errno(r, "Failed to initialize compressor: %m"); + if (r == 0) + return 0; + + log_debug("Stream is compressed: %s", import_compress_type_to_string(j->compress.type)); + + r = pull_job_open_disk(j); + if (r < 0) + return r; + + /* Now, take the payload we read so far, and decompress it */ + stub = j->payload; + stub_size = j->payload_size; + + j->payload = NULL; + j->payload_size = 0; + j->payload_allocated = 0; + + j->state = PULL_JOB_RUNNING; + + r = pull_job_write_compressed(j, stub, stub_size); + if (r < 0) + return r; + + return 0; +} + +static size_t pull_job_write_callback(void *contents, size_t size, size_t nmemb, void *userdata) { + PullJob *j = userdata; + size_t sz = size * nmemb; + int r; + + assert(contents); + assert(j); + + switch (j->state) { + + case PULL_JOB_ANALYZING: + /* Let's first check what it actually is */ + + if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz)) { + r = log_oom(); + goto fail; + } + + memcpy(j->payload + j->payload_size, contents, sz); + j->payload_size += sz; + + r = pull_job_detect_compression(j); + if (r < 0) + goto fail; + + break; + + case PULL_JOB_RUNNING: + + r = pull_job_write_compressed(j, contents, sz); + if (r < 0) + goto fail; + + break; + + case PULL_JOB_DONE: + case PULL_JOB_FAILED: + r = -ESTALE; + goto fail; + + default: + assert_not_reached("Impossible state."); + } + + return sz; + +fail: + pull_job_finish(j, r); + return 0; +} + +static size_t pull_job_header_callback(void *contents, size_t size, size_t nmemb, void *userdata) { + PullJob *j = userdata; + size_t sz = size * nmemb; + _cleanup_free_ char *length = NULL, *last_modified = NULL; + char *etag; + int r; + + assert(contents); + assert(j); + + if (j->state == PULL_JOB_DONE || j->state == PULL_JOB_FAILED) { + r = -ESTALE; + goto fail; + } + + assert(j->state == PULL_JOB_ANALYZING); + + r = curl_header_strdup(contents, sz, "ETag:", &etag); + if (r < 0) { + log_oom(); + goto fail; + } + if (r > 0) { + free(j->etag); + j->etag = etag; + + if (strv_contains(j->old_etags, j->etag)) { + log_info("Image already downloaded. Skipping download."); + j->etag_exists = true; + pull_job_finish(j, 0); + return sz; + } + + return sz; + } + + r = curl_header_strdup(contents, sz, "Content-Length:", &length); + if (r < 0) { + log_oom(); + goto fail; + } + if (r > 0) { + (void) safe_atou64(length, &j->content_length); + + if (j->content_length != (uint64_t) -1) { + char bytes[FORMAT_BYTES_MAX]; + + if (j->content_length > j->compressed_max) { + log_error("Content too large."); + r = -EFBIG; + goto fail; + } + + log_info("Downloading %s for %s.", format_bytes(bytes, sizeof(bytes), j->content_length), j->url); + } + + return sz; + } + + r = curl_header_strdup(contents, sz, "Last-Modified:", &last_modified); + if (r < 0) { + log_oom(); + goto fail; + } + if (r > 0) { + (void) curl_parse_http_time(last_modified, &j->mtime); + return sz; + } + + if (j->on_header) { + r = j->on_header(j, contents, sz); + if (r < 0) + goto fail; + } + + return sz; + +fail: + pull_job_finish(j, r); + return 0; +} + +static int pull_job_progress_callback(void *userdata, curl_off_t dltotal, curl_off_t dlnow, curl_off_t ultotal, curl_off_t ulnow) { + PullJob *j = userdata; + unsigned percent; + usec_t n; + + assert(j); + + if (dltotal <= 0) + return 0; + + percent = ((100 * dlnow) / dltotal); + n = now(CLOCK_MONOTONIC); + + if (n > j->last_status_usec + USEC_PER_SEC && + percent != j->progress_percent && + dlnow < dltotal) { + char buf[FORMAT_TIMESPAN_MAX]; + + if (n - j->start_usec > USEC_PER_SEC && dlnow > 0) { + char y[FORMAT_BYTES_MAX]; + usec_t left, done; + + done = n - j->start_usec; + left = (usec_t) (((double) done * (double) dltotal) / dlnow) - done; + + log_info("Got %u%% of %s. %s left at %s/s.", + percent, + j->url, + format_timespan(buf, sizeof(buf), left, USEC_PER_SEC), + format_bytes(y, sizeof(y), (uint64_t) ((double) dlnow / ((double) done / (double) USEC_PER_SEC)))); + } else + log_info("Got %u%% of %s.", percent, j->url); + + j->progress_percent = percent; + j->last_status_usec = n; + + if (j->on_progress) + j->on_progress(j); + } + + return 0; +} + +int pull_job_new(PullJob **ret, const char *url, CurlGlue *glue, void *userdata) { + _cleanup_(pull_job_unrefp) PullJob *j = NULL; + + assert(url); + assert(glue); + assert(ret); + + j = new0(PullJob, 1); + if (!j) + return -ENOMEM; + + j->state = PULL_JOB_INIT; + j->disk_fd = -1; + j->userdata = userdata; + j->glue = glue; + j->content_length = (uint64_t) -1; + j->start_usec = now(CLOCK_MONOTONIC); + j->compressed_max = j->uncompressed_max = 8LLU * 1024LLU * 1024LLU * 1024LLU; /* 8GB */ + + j->url = strdup(url); + if (!j->url) + return -ENOMEM; + + *ret = j; + j = NULL; + + return 0; +} + +int pull_job_begin(PullJob *j) { + int r; + + assert(j); + + if (j->state != PULL_JOB_INIT) + return -EBUSY; + + if (j->grow_machine_directory) + grow_machine_directory(); + + r = curl_glue_make(&j->curl, j->url, j); + if (r < 0) + return r; + + if (!strv_isempty(j->old_etags)) { + _cleanup_free_ char *cc = NULL, *hdr = NULL; + + cc = strv_join(j->old_etags, ", "); + if (!cc) + return -ENOMEM; + + hdr = strappend("If-None-Match: ", cc); + if (!hdr) + return -ENOMEM; + + if (!j->request_header) { + j->request_header = curl_slist_new(hdr, NULL); + if (!j->request_header) + return -ENOMEM; + } else { + struct curl_slist *l; + + l = curl_slist_append(j->request_header, hdr); + if (!l) + return -ENOMEM; + + j->request_header = l; + } + } + + if (j->request_header) { + if (curl_easy_setopt(j->curl, CURLOPT_HTTPHEADER, j->request_header) != CURLE_OK) + return -EIO; + } + + if (curl_easy_setopt(j->curl, CURLOPT_WRITEFUNCTION, pull_job_write_callback) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_WRITEDATA, j) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_HEADERFUNCTION, pull_job_header_callback) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_HEADERDATA, j) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_XFERINFOFUNCTION, pull_job_progress_callback) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_XFERINFODATA, j) != CURLE_OK) + return -EIO; + + if (curl_easy_setopt(j->curl, CURLOPT_NOPROGRESS, 0) != CURLE_OK) + return -EIO; + + r = curl_glue_add(j->glue, j->curl); + if (r < 0) + return r; + + j->state = PULL_JOB_ANALYZING; + + return 0; +} diff --git a/src/grp-machine/grp-import/systemd-pull/pull-job.h b/src/grp-machine/grp-import/systemd-pull/pull-job.h new file mode 100644 index 0000000000..8a99883794 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-job.h @@ -0,0 +1,106 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <gcrypt.h> + +#include "import-compress.h" +#include "systemd-basic/macro.h" + +#include "curl-util.h" + +typedef struct PullJob PullJob; +typedef void (*PullJobFinished)(PullJob *job); +typedef int (*PullJobOpenDisk)(PullJob *job); +typedef int (*PullJobHeader)(PullJob *job, const char *header, size_t sz); +typedef void (*PullJobProgress)(PullJob *job); + +typedef enum PullJobState { + PULL_JOB_INIT, + PULL_JOB_ANALYZING, /* Still reading into ->payload, to figure out what we have */ + PULL_JOB_RUNNING, /* Writing to destination */ + PULL_JOB_DONE, + PULL_JOB_FAILED, + _PULL_JOB_STATE_MAX, + _PULL_JOB_STATE_INVALID = -1, +} PullJobState; + +#define PULL_JOB_IS_COMPLETE(j) (IN_SET((j)->state, PULL_JOB_DONE, PULL_JOB_FAILED)) + +struct PullJob { + PullJobState state; + int error; + + char *url; + + void *userdata; + PullJobFinished on_finished; + PullJobOpenDisk on_open_disk; + PullJobHeader on_header; + PullJobProgress on_progress; + + CurlGlue *glue; + CURL *curl; + struct curl_slist *request_header; + + char *etag; + char **old_etags; + bool etag_exists; + + uint64_t content_length; + uint64_t written_compressed; + uint64_t written_uncompressed; + + uint64_t uncompressed_max; + uint64_t compressed_max; + + uint8_t *payload; + size_t payload_size; + size_t payload_allocated; + + int disk_fd; + + usec_t mtime; + + ImportCompress compress; + + unsigned progress_percent; + usec_t start_usec; + usec_t last_status_usec; + + bool allow_sparse; + + bool calc_checksum; + gcry_md_hd_t checksum_context; + + char *checksum; + + bool grow_machine_directory; + uint64_t written_since_last_grow; +}; + +int pull_job_new(PullJob **job, const char *url, CurlGlue *glue, void *userdata); +PullJob* pull_job_unref(PullJob *job); + +int pull_job_begin(PullJob *j); + +void pull_job_curl_on_finished(CurlGlue *g, CURL *curl, CURLcode result); + +DEFINE_TRIVIAL_CLEANUP_FUNC(PullJob*, pull_job_unref); diff --git a/src/grp-machine/grp-import/systemd-pull/pull-raw.c b/src/grp-machine/grp-import/systemd-pull/pull-raw.c new file mode 100644 index 0000000000..f15fb07a8e --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-raw.c @@ -0,0 +1,651 @@ +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <curl/curl.h> +#include <sys/xattr.h> + +#include <linux/fs.h> + +#include <systemd/sd-daemon.h> + +#include "import-common.h" +#include "qcow2-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/chattr-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/macro.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/rm-rf.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/utf8.h" +#include "systemd-basic/util.h" +#include "systemd-basic/web-util.h" +#include "systemd-shared/import-util.h" + +#include "curl-util.h" +#include "pull-common.h" +#include "pull-job.h" +#include "pull-raw.h" + +typedef enum RawProgress { + RAW_DOWNLOADING, + RAW_VERIFYING, + RAW_UNPACKING, + RAW_FINALIZING, + RAW_COPYING, +} RawProgress; + +struct RawPull { + sd_event *event; + CurlGlue *glue; + + char *image_root; + + PullJob *raw_job; + PullJob *settings_job; + PullJob *checksum_job; + PullJob *signature_job; + + RawPullFinished on_finished; + void *userdata; + + char *local; + bool force_local; + bool grow_machine_directory; + bool settings; + + char *final_path; + char *temp_path; + + char *settings_path; + char *settings_temp_path; + + ImportVerify verify; +}; + +RawPull* raw_pull_unref(RawPull *i) { + if (!i) + return NULL; + + pull_job_unref(i->raw_job); + pull_job_unref(i->settings_job); + pull_job_unref(i->checksum_job); + pull_job_unref(i->signature_job); + + curl_glue_unref(i->glue); + sd_event_unref(i->event); + + if (i->temp_path) { + (void) unlink(i->temp_path); + free(i->temp_path); + } + + if (i->settings_temp_path) { + (void) unlink(i->settings_temp_path); + free(i->settings_temp_path); + } + + free(i->final_path); + free(i->settings_path); + free(i->image_root); + free(i->local); + return mfree(i); +} + +int raw_pull_new( + RawPull **ret, + sd_event *event, + const char *image_root, + RawPullFinished on_finished, + void *userdata) { + + _cleanup_(raw_pull_unrefp) RawPull *i = NULL; + int r; + + assert(ret); + + i = new0(RawPull, 1); + if (!i) + return -ENOMEM; + + i->on_finished = on_finished; + i->userdata = userdata; + + i->image_root = strdup(image_root ?: "/var/lib/machines"); + if (!i->image_root) + return -ENOMEM; + + i->grow_machine_directory = path_startswith(i->image_root, "/var/lib/machines"); + + if (event) + i->event = sd_event_ref(event); + else { + r = sd_event_default(&i->event); + if (r < 0) + return r; + } + + r = curl_glue_new(&i->glue, i->event); + if (r < 0) + return r; + + i->glue->on_finished = pull_job_curl_on_finished; + i->glue->userdata = i; + + *ret = i; + i = NULL; + + return 0; +} + +static void raw_pull_report_progress(RawPull *i, RawProgress p) { + unsigned percent; + + assert(i); + + switch (p) { + + case RAW_DOWNLOADING: { + unsigned remain = 80; + + percent = 0; + + if (i->settings_job) { + percent += i->settings_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->checksum_job) { + percent += i->checksum_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->signature_job) { + percent += i->signature_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->raw_job) + percent += i->raw_job->progress_percent * remain / 100; + break; + } + + case RAW_VERIFYING: + percent = 80; + break; + + case RAW_UNPACKING: + percent = 85; + break; + + case RAW_FINALIZING: + percent = 90; + break; + + case RAW_COPYING: + percent = 95; + break; + + default: + assert_not_reached("Unknown progress state"); + } + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_debug("Combined progress %u%%", percent); +} + +static int raw_pull_maybe_convert_qcow2(RawPull *i) { + _cleanup_close_ int converted_fd = -1; + _cleanup_free_ char *t = NULL; + int r; + + assert(i); + assert(i->raw_job); + + r = qcow2_detect(i->raw_job->disk_fd); + if (r < 0) + return log_error_errno(r, "Failed to detect whether this is a QCOW2 image: %m"); + if (r == 0) + return 0; + + /* This is a QCOW2 image, let's convert it */ + r = tempfn_random(i->final_path, NULL, &t); + if (r < 0) + return log_oom(); + + converted_fd = open(t, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (converted_fd < 0) + return log_error_errno(errno, "Failed to create %s: %m", t); + + r = chattr_fd(converted_fd, FS_NOCOW_FL, FS_NOCOW_FL); + if (r < 0) + log_warning_errno(r, "Failed to set file attributes on %s: %m", t); + + log_info("Unpacking QCOW2 file."); + + r = qcow2_convert(i->raw_job->disk_fd, converted_fd); + if (r < 0) { + unlink(t); + return log_error_errno(r, "Failed to convert qcow2 image: %m"); + } + + (void) unlink(i->temp_path); + free(i->temp_path); + i->temp_path = t; + t = NULL; + + safe_close(i->raw_job->disk_fd); + i->raw_job->disk_fd = converted_fd; + converted_fd = -1; + + return 1; +} + +static int raw_pull_make_local_copy(RawPull *i) { + _cleanup_free_ char *tp = NULL; + _cleanup_close_ int dfd = -1; + const char *p; + int r; + + assert(i); + assert(i->raw_job); + + if (!i->local) + return 0; + + if (!i->final_path) { + r = pull_make_path(i->raw_job->url, i->raw_job->etag, i->image_root, ".raw-", ".raw", &i->final_path); + if (r < 0) + return log_oom(); + } + + if (i->raw_job->etag_exists) { + /* We have downloaded this one previously, reopen it */ + + assert(i->raw_job->disk_fd < 0); + + i->raw_job->disk_fd = open(i->final_path, O_RDONLY|O_NOCTTY|O_CLOEXEC); + if (i->raw_job->disk_fd < 0) + return log_error_errno(errno, "Failed to open vendor image: %m"); + } else { + /* We freshly downloaded the image, use it */ + + assert(i->raw_job->disk_fd >= 0); + + if (lseek(i->raw_job->disk_fd, SEEK_SET, 0) == (off_t) -1) + return log_error_errno(errno, "Failed to seek to beginning of vendor image: %m"); + } + + p = strjoina(i->image_root, "/", i->local, ".raw"); + + if (i->force_local) + (void) rm_rf(p, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + + r = tempfn_random(p, NULL, &tp); + if (r < 0) + return log_oom(); + + dfd = open(tp, O_WRONLY|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (dfd < 0) + return log_error_errno(errno, "Failed to create writable copy of image: %m"); + + /* Turn off COW writing. This should greatly improve + * performance on COW file systems like btrfs, since it + * reduces fragmentation caused by not allowing in-place + * writes. */ + r = chattr_fd(dfd, FS_NOCOW_FL, FS_NOCOW_FL); + if (r < 0) + log_warning_errno(r, "Failed to set file attributes on %s: %m", tp); + + r = copy_bytes(i->raw_job->disk_fd, dfd, (uint64_t) -1, true); + if (r < 0) { + unlink(tp); + return log_error_errno(r, "Failed to make writable copy of image: %m"); + } + + (void) copy_times(i->raw_job->disk_fd, dfd); + (void) copy_xattr(i->raw_job->disk_fd, dfd); + + dfd = safe_close(dfd); + + r = rename(tp, p); + if (r < 0) { + r = log_error_errno(errno, "Failed to move writable image into place: %m"); + unlink(tp); + return r; + } + + log_info("Created new local image '%s'.", i->local); + + if (i->settings) { + const char *local_settings; + assert(i->settings_job); + + if (!i->settings_path) { + r = pull_make_path(i->settings_job->url, i->settings_job->etag, i->image_root, ".settings-", NULL, &i->settings_path); + if (r < 0) + return log_oom(); + } + + local_settings = strjoina(i->image_root, "/", i->local, ".nspawn"); + + r = copy_file_atomic(i->settings_path, local_settings, 0644, i->force_local, 0); + if (r == -EEXIST) + log_warning_errno(r, "Settings file %s already exists, not replacing.", local_settings); + else if (r == -ENOENT) + log_debug_errno(r, "Skipping creation of settings file, since none was found."); + else if (r < 0) + log_warning_errno(r, "Failed to copy settings files %s, ignoring: %m", local_settings); + else + log_info("Created new settings file %s.", local_settings); + } + + return 0; +} + +static bool raw_pull_is_done(RawPull *i) { + assert(i); + assert(i->raw_job); + + if (!PULL_JOB_IS_COMPLETE(i->raw_job)) + return false; + if (i->settings_job && !PULL_JOB_IS_COMPLETE(i->settings_job)) + return false; + if (i->checksum_job && !PULL_JOB_IS_COMPLETE(i->checksum_job)) + return false; + if (i->signature_job && !PULL_JOB_IS_COMPLETE(i->signature_job)) + return false; + + return true; +} + +static void raw_pull_job_on_finished(PullJob *j) { + RawPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + if (j == i->settings_job) { + if (j->error != 0) + log_info_errno(j->error, "Settings file could not be retrieved, proceeding without."); + } else if (j->error != 0) { + if (j == i->checksum_job) + log_error_errno(j->error, "Failed to retrieve SHA256 checksum, cannot verify. (Try --verify=no?)"); + else if (j == i->signature_job) + log_error_errno(j->error, "Failed to retrieve signature file, cannot verify. (Try --verify=no?)"); + else + log_error_errno(j->error, "Failed to retrieve image file. (Wrong URL?)"); + + r = j->error; + goto finish; + } + + /* This is invoked if either the download completed + * successfully, or the download was skipped because we + * already have the etag. In this case ->etag_exists is + * true. + * + * We only do something when we got all three files */ + + if (!raw_pull_is_done(i)) + return; + + if (i->settings_job) + i->settings_job->disk_fd = safe_close(i->settings_job->disk_fd); + + if (!i->raw_job->etag_exists) { + /* This is a new download, verify it, and move it into place */ + assert(i->raw_job->disk_fd >= 0); + + raw_pull_report_progress(i, RAW_VERIFYING); + + r = pull_verify(i->raw_job, i->settings_job, i->checksum_job, i->signature_job); + if (r < 0) + goto finish; + + raw_pull_report_progress(i, RAW_UNPACKING); + + r = raw_pull_maybe_convert_qcow2(i); + if (r < 0) + goto finish; + + raw_pull_report_progress(i, RAW_FINALIZING); + + r = import_make_read_only_fd(i->raw_job->disk_fd); + if (r < 0) + goto finish; + + r = rename_noreplace(AT_FDCWD, i->temp_path, AT_FDCWD, i->final_path); + if (r < 0) { + log_error_errno(r, "Failed to move RAW file into place: %m"); + goto finish; + } + + i->temp_path = mfree(i->temp_path); + + if (i->settings_job && + i->settings_job->error == 0 && + !i->settings_job->etag_exists) { + + assert(i->settings_temp_path); + assert(i->settings_path); + + r = import_make_read_only(i->settings_temp_path); + if (r < 0) + goto finish; + + r = rename_noreplace(AT_FDCWD, i->settings_temp_path, AT_FDCWD, i->settings_path); + if (r < 0) { + log_error_errno(r, "Failed to rename settings file: %m"); + goto finish; + } + + i->settings_temp_path = mfree(i->settings_temp_path); + } + } + + raw_pull_report_progress(i, RAW_COPYING); + + r = raw_pull_make_local_copy(i); + if (r < 0) + goto finish; + + r = 0; + +finish: + if (i->on_finished) + i->on_finished(i, r, i->userdata); + else + sd_event_exit(i->event, r); +} + +static int raw_pull_job_on_open_disk_raw(PullJob *j) { + RawPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + assert(i->raw_job == j); + assert(!i->final_path); + assert(!i->temp_path); + + r = pull_make_path(j->url, j->etag, i->image_root, ".raw-", ".raw", &i->final_path); + if (r < 0) + return log_oom(); + + r = tempfn_random(i->final_path, NULL, &i->temp_path); + if (r < 0) + return log_oom(); + + (void) mkdir_parents_label(i->temp_path, 0700); + + j->disk_fd = open(i->temp_path, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (j->disk_fd < 0) + return log_error_errno(errno, "Failed to create %s: %m", i->temp_path); + + r = chattr_fd(j->disk_fd, FS_NOCOW_FL, FS_NOCOW_FL); + if (r < 0) + log_warning_errno(r, "Failed to set file attributes on %s: %m", i->temp_path); + + return 0; +} + +static int raw_pull_job_on_open_disk_settings(PullJob *j) { + RawPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + assert(i->settings_job == j); + assert(!i->settings_path); + assert(!i->settings_temp_path); + + r = pull_make_path(j->url, j->etag, i->image_root, ".settings-", NULL, &i->settings_path); + if (r < 0) + return log_oom(); + + r = tempfn_random(i->settings_path, NULL, &i->settings_temp_path); + if (r < 0) + return log_oom(); + + mkdir_parents_label(i->settings_temp_path, 0700); + + j->disk_fd = open(i->settings_temp_path, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (j->disk_fd < 0) + return log_error_errno(errno, "Failed to create %s: %m", i->settings_temp_path); + + return 0; +} + +static void raw_pull_job_on_progress(PullJob *j) { + RawPull *i; + + assert(j); + assert(j->userdata); + + i = j->userdata; + + raw_pull_report_progress(i, RAW_DOWNLOADING); +} + +int raw_pull_start( + RawPull *i, + const char *url, + const char *local, + bool force_local, + ImportVerify verify, + bool settings) { + + int r; + + assert(i); + assert(verify < _IMPORT_VERIFY_MAX); + assert(verify >= 0); + + if (!http_url_is_valid(url)) + return -EINVAL; + + if (local && !machine_name_is_valid(local)) + return -EINVAL; + + if (i->raw_job) + return -EBUSY; + + r = free_and_strdup(&i->local, local); + if (r < 0) + return r; + + i->force_local = force_local; + i->verify = verify; + i->settings = settings; + + /* Queue job for the image itself */ + r = pull_job_new(&i->raw_job, url, i->glue, i); + if (r < 0) + return r; + + i->raw_job->on_finished = raw_pull_job_on_finished; + i->raw_job->on_open_disk = raw_pull_job_on_open_disk_raw; + i->raw_job->on_progress = raw_pull_job_on_progress; + i->raw_job->calc_checksum = verify != IMPORT_VERIFY_NO; + i->raw_job->grow_machine_directory = i->grow_machine_directory; + + r = pull_find_old_etags(url, i->image_root, DT_REG, ".raw-", ".raw", &i->raw_job->old_etags); + if (r < 0) + return r; + + if (settings) { + r = pull_make_settings_job(&i->settings_job, url, i->glue, raw_pull_job_on_finished, i); + if (r < 0) + return r; + + i->settings_job->on_open_disk = raw_pull_job_on_open_disk_settings; + i->settings_job->on_progress = raw_pull_job_on_progress; + i->settings_job->calc_checksum = verify != IMPORT_VERIFY_NO; + + r = pull_find_old_etags(i->settings_job->url, i->image_root, DT_REG, ".settings-", NULL, &i->settings_job->old_etags); + if (r < 0) + return r; + } + + r = pull_make_verification_jobs(&i->checksum_job, &i->signature_job, verify, url, i->glue, raw_pull_job_on_finished, i); + if (r < 0) + return r; + + r = pull_job_begin(i->raw_job); + if (r < 0) + return r; + + if (i->settings_job) { + r = pull_job_begin(i->settings_job); + if (r < 0) + return r; + } + + if (i->checksum_job) { + i->checksum_job->on_progress = raw_pull_job_on_progress; + + r = pull_job_begin(i->checksum_job); + if (r < 0) + return r; + } + + if (i->signature_job) { + i->signature_job->on_progress = raw_pull_job_on_progress; + + r = pull_job_begin(i->signature_job); + if (r < 0) + return r; + } + + return 0; +} diff --git a/src/grp-machine/grp-import/systemd-pull/pull-raw.h b/src/grp-machine/grp-import/systemd-pull/pull-raw.h new file mode 100644 index 0000000000..0940cff93b --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-raw.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "systemd-basic/macro.h" +#include "systemd-shared/import-util.h" + +typedef struct RawPull RawPull; +typedef void (*RawPullFinished)(RawPull *pull, int error, void *userdata); + +int raw_pull_new(RawPull **pull, sd_event *event, const char *image_root, RawPullFinished on_finished, void *userdata); +RawPull* raw_pull_unref(RawPull *pull); + +DEFINE_TRIVIAL_CLEANUP_FUNC(RawPull*, raw_pull_unref); + +int raw_pull_start(RawPull *pull, const char *url, const char *local, bool force_local, ImportVerify verify, bool settings); diff --git a/src/grp-machine/grp-import/systemd-pull/pull-tar.c b/src/grp-machine/grp-import/systemd-pull/pull-tar.c new file mode 100644 index 0000000000..167c7fade5 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-tar.c @@ -0,0 +1,562 @@ +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <curl/curl.h> +#include <sys/prctl.h> + +#include <systemd/sd-daemon.h> + +#include "import-common.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/macro.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/rm-rf.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/utf8.h" +#include "systemd-basic/util.h" +#include "systemd-basic/web-util.h" +#include "systemd-shared/import-util.h" + +#include "curl-util.h" +#include "pull-common.h" +#include "pull-job.h" +#include "pull-tar.h" + +typedef enum TarProgress { + TAR_DOWNLOADING, + TAR_VERIFYING, + TAR_FINALIZING, + TAR_COPYING, +} TarProgress; + +struct TarPull { + sd_event *event; + CurlGlue *glue; + + char *image_root; + + PullJob *tar_job; + PullJob *settings_job; + PullJob *checksum_job; + PullJob *signature_job; + + TarPullFinished on_finished; + void *userdata; + + char *local; + bool force_local; + bool grow_machine_directory; + bool settings; + + pid_t tar_pid; + + char *final_path; + char *temp_path; + + char *settings_path; + char *settings_temp_path; + + ImportVerify verify; +}; + +TarPull* tar_pull_unref(TarPull *i) { + if (!i) + return NULL; + + if (i->tar_pid > 1) { + (void) kill_and_sigcont(i->tar_pid, SIGKILL); + (void) wait_for_terminate(i->tar_pid, NULL); + } + + pull_job_unref(i->tar_job); + pull_job_unref(i->settings_job); + pull_job_unref(i->checksum_job); + pull_job_unref(i->signature_job); + + curl_glue_unref(i->glue); + sd_event_unref(i->event); + + if (i->temp_path) { + (void) rm_rf(i->temp_path, REMOVE_ROOT|REMOVE_PHYSICAL|REMOVE_SUBVOLUME); + free(i->temp_path); + } + + if (i->settings_temp_path) { + (void) unlink(i->settings_temp_path); + free(i->settings_temp_path); + } + + free(i->final_path); + free(i->settings_path); + free(i->image_root); + free(i->local); + return mfree(i); +} + +int tar_pull_new( + TarPull **ret, + sd_event *event, + const char *image_root, + TarPullFinished on_finished, + void *userdata) { + + _cleanup_(tar_pull_unrefp) TarPull *i = NULL; + int r; + + assert(ret); + + i = new0(TarPull, 1); + if (!i) + return -ENOMEM; + + i->on_finished = on_finished; + i->userdata = userdata; + + i->image_root = strdup(image_root ?: "/var/lib/machines"); + if (!i->image_root) + return -ENOMEM; + + i->grow_machine_directory = path_startswith(i->image_root, "/var/lib/machines"); + + if (event) + i->event = sd_event_ref(event); + else { + r = sd_event_default(&i->event); + if (r < 0) + return r; + } + + r = curl_glue_new(&i->glue, i->event); + if (r < 0) + return r; + + i->glue->on_finished = pull_job_curl_on_finished; + i->glue->userdata = i; + + *ret = i; + i = NULL; + + return 0; +} + +static void tar_pull_report_progress(TarPull *i, TarProgress p) { + unsigned percent; + + assert(i); + + switch (p) { + + case TAR_DOWNLOADING: { + unsigned remain = 85; + + percent = 0; + + if (i->settings_job) { + percent += i->settings_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->checksum_job) { + percent += i->checksum_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->signature_job) { + percent += i->signature_job->progress_percent * 5 / 100; + remain -= 5; + } + + if (i->tar_job) + percent += i->tar_job->progress_percent * remain / 100; + break; + } + + case TAR_VERIFYING: + percent = 85; + break; + + case TAR_FINALIZING: + percent = 90; + break; + + case TAR_COPYING: + percent = 95; + break; + + default: + assert_not_reached("Unknown progress state"); + } + + sd_notifyf(false, "X_IMPORT_PROGRESS=%u", percent); + log_debug("Combined progress %u%%", percent); +} + +static int tar_pull_make_local_copy(TarPull *i) { + int r; + + assert(i); + assert(i->tar_job); + + if (!i->local) + return 0; + + if (!i->final_path) { + r = pull_make_path(i->tar_job->url, i->tar_job->etag, i->image_root, ".tar-", NULL, &i->final_path); + if (r < 0) + return log_oom(); + } + + r = pull_make_local_copy(i->final_path, i->image_root, i->local, i->force_local); + if (r < 0) + return r; + + if (i->settings) { + const char *local_settings; + assert(i->settings_job); + + if (!i->settings_path) { + r = pull_make_path(i->settings_job->url, i->settings_job->etag, i->image_root, ".settings-", NULL, &i->settings_path); + if (r < 0) + return log_oom(); + } + + local_settings = strjoina(i->image_root, "/", i->local, ".nspawn"); + + r = copy_file_atomic(i->settings_path, local_settings, 0664, i->force_local, 0); + if (r == -EEXIST) + log_warning_errno(r, "Settings file %s already exists, not replacing.", local_settings); + else if (r == -ENOENT) + log_debug_errno(r, "Skipping creation of settings file, since none was found."); + else if (r < 0) + log_warning_errno(r, "Failed to copy settings files %s, ignoring: %m", local_settings); + else + log_info("Created new settings file %s.", local_settings); + } + + return 0; +} + +static bool tar_pull_is_done(TarPull *i) { + assert(i); + assert(i->tar_job); + + if (!PULL_JOB_IS_COMPLETE(i->tar_job)) + return false; + if (i->settings_job && !PULL_JOB_IS_COMPLETE(i->settings_job)) + return false; + if (i->checksum_job && !PULL_JOB_IS_COMPLETE(i->checksum_job)) + return false; + if (i->signature_job && !PULL_JOB_IS_COMPLETE(i->signature_job)) + return false; + + return true; +} + +static void tar_pull_job_on_finished(PullJob *j) { + TarPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + + if (j == i->settings_job) { + if (j->error != 0) + log_info_errno(j->error, "Settings file could not be retrieved, proceeding without."); + } else if (j->error != 0) { + if (j == i->checksum_job) + log_error_errno(j->error, "Failed to retrieve SHA256 checksum, cannot verify. (Try --verify=no?)"); + else if (j == i->signature_job) + log_error_errno(j->error, "Failed to retrieve signature file, cannot verify. (Try --verify=no?)"); + else + log_error_errno(j->error, "Failed to retrieve image file. (Wrong URL?)"); + + r = j->error; + goto finish; + } + + /* This is invoked if either the download completed + * successfully, or the download was skipped because we + * already have the etag. */ + + if (!tar_pull_is_done(i)) + return; + + i->tar_job->disk_fd = safe_close(i->tar_job->disk_fd); + if (i->settings_job) + i->settings_job->disk_fd = safe_close(i->settings_job->disk_fd); + + if (i->tar_pid > 0) { + r = wait_for_terminate_and_warn("tar", i->tar_pid, true); + i->tar_pid = 0; + if (r < 0) + goto finish; + if (r > 0) { + r = -EIO; + goto finish; + } + } + + if (!i->tar_job->etag_exists) { + /* This is a new download, verify it, and move it into place */ + + tar_pull_report_progress(i, TAR_VERIFYING); + + r = pull_verify(i->tar_job, i->settings_job, i->checksum_job, i->signature_job); + if (r < 0) + goto finish; + + tar_pull_report_progress(i, TAR_FINALIZING); + + r = import_make_read_only(i->temp_path); + if (r < 0) + goto finish; + + r = rename_noreplace(AT_FDCWD, i->temp_path, AT_FDCWD, i->final_path); + if (r < 0) { + log_error_errno(r, "Failed to rename to final image name: %m"); + goto finish; + } + + i->temp_path = mfree(i->temp_path); + + if (i->settings_job && + i->settings_job->error == 0 && + !i->settings_job->etag_exists) { + + assert(i->settings_temp_path); + assert(i->settings_path); + + /* Also move the settings file into place, if + * it exist. Note that we do so only if we + * also moved the tar file in place, to keep + * things strictly in sync. */ + + r = import_make_read_only(i->settings_temp_path); + if (r < 0) + goto finish; + + r = rename_noreplace(AT_FDCWD, i->settings_temp_path, AT_FDCWD, i->settings_path); + if (r < 0) { + log_error_errno(r, "Failed to rename settings file: %m"); + goto finish; + } + + i->settings_temp_path = mfree(i->settings_temp_path); + } + } + + tar_pull_report_progress(i, TAR_COPYING); + + r = tar_pull_make_local_copy(i); + if (r < 0) + goto finish; + + r = 0; + +finish: + if (i->on_finished) + i->on_finished(i, r, i->userdata); + else + sd_event_exit(i->event, r); +} + +static int tar_pull_job_on_open_disk_tar(PullJob *j) { + TarPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + assert(i->tar_job == j); + assert(!i->final_path); + assert(!i->temp_path); + assert(i->tar_pid <= 0); + + r = pull_make_path(j->url, j->etag, i->image_root, ".tar-", NULL, &i->final_path); + if (r < 0) + return log_oom(); + + r = tempfn_random(i->final_path, NULL, &i->temp_path); + if (r < 0) + return log_oom(); + + mkdir_parents_label(i->temp_path, 0700); + + r = btrfs_subvol_make(i->temp_path); + if (r == -ENOTTY) { + if (mkdir(i->temp_path, 0755) < 0) + return log_error_errno(errno, "Failed to create directory %s: %m", i->temp_path); + } else if (r < 0) + return log_error_errno(r, "Failed to create subvolume %s: %m", i->temp_path); + else + (void) import_assign_pool_quota_and_warn(i->temp_path); + + j->disk_fd = import_fork_tar_x(i->temp_path, &i->tar_pid); + if (j->disk_fd < 0) + return j->disk_fd; + + return 0; +} + +static int tar_pull_job_on_open_disk_settings(PullJob *j) { + TarPull *i; + int r; + + assert(j); + assert(j->userdata); + + i = j->userdata; + assert(i->settings_job == j); + assert(!i->settings_path); + assert(!i->settings_temp_path); + + r = pull_make_path(j->url, j->etag, i->image_root, ".settings-", NULL, &i->settings_path); + if (r < 0) + return log_oom(); + + r = tempfn_random(i->settings_path, NULL, &i->settings_temp_path); + if (r < 0) + return log_oom(); + + mkdir_parents_label(i->settings_temp_path, 0700); + + j->disk_fd = open(i->settings_temp_path, O_RDWR|O_CREAT|O_EXCL|O_NOCTTY|O_CLOEXEC, 0664); + if (j->disk_fd < 0) + return log_error_errno(errno, "Failed to create %s: %m", i->settings_temp_path); + + return 0; +} + +static void tar_pull_job_on_progress(PullJob *j) { + TarPull *i; + + assert(j); + assert(j->userdata); + + i = j->userdata; + + tar_pull_report_progress(i, TAR_DOWNLOADING); +} + +int tar_pull_start( + TarPull *i, + const char *url, + const char *local, + bool force_local, + ImportVerify verify, + bool settings) { + + int r; + + assert(i); + assert(verify < _IMPORT_VERIFY_MAX); + assert(verify >= 0); + + if (!http_url_is_valid(url)) + return -EINVAL; + + if (local && !machine_name_is_valid(local)) + return -EINVAL; + + if (i->tar_job) + return -EBUSY; + + r = free_and_strdup(&i->local, local); + if (r < 0) + return r; + + i->force_local = force_local; + i->verify = verify; + i->settings = settings; + + /* Set up download job for TAR file */ + r = pull_job_new(&i->tar_job, url, i->glue, i); + if (r < 0) + return r; + + i->tar_job->on_finished = tar_pull_job_on_finished; + i->tar_job->on_open_disk = tar_pull_job_on_open_disk_tar; + i->tar_job->on_progress = tar_pull_job_on_progress; + i->tar_job->calc_checksum = verify != IMPORT_VERIFY_NO; + i->tar_job->grow_machine_directory = i->grow_machine_directory; + + r = pull_find_old_etags(url, i->image_root, DT_DIR, ".tar-", NULL, &i->tar_job->old_etags); + if (r < 0) + return r; + + /* Set up download job for the settings file (.nspawn) */ + if (settings) { + r = pull_make_settings_job(&i->settings_job, url, i->glue, tar_pull_job_on_finished, i); + if (r < 0) + return r; + + i->settings_job->on_open_disk = tar_pull_job_on_open_disk_settings; + i->settings_job->on_progress = tar_pull_job_on_progress; + i->settings_job->calc_checksum = verify != IMPORT_VERIFY_NO; + + r = pull_find_old_etags(i->settings_job->url, i->image_root, DT_REG, ".settings-", NULL, &i->settings_job->old_etags); + if (r < 0) + return r; + } + + /* Set up download of checksum/signature files */ + r = pull_make_verification_jobs(&i->checksum_job, &i->signature_job, verify, url, i->glue, tar_pull_job_on_finished, i); + if (r < 0) + return r; + + r = pull_job_begin(i->tar_job); + if (r < 0) + return r; + + if (i->settings_job) { + r = pull_job_begin(i->settings_job); + if (r < 0) + return r; + } + + if (i->checksum_job) { + i->checksum_job->on_progress = tar_pull_job_on_progress; + + r = pull_job_begin(i->checksum_job); + if (r < 0) + return r; + } + + if (i->signature_job) { + i->signature_job->on_progress = tar_pull_job_on_progress; + + r = pull_job_begin(i->signature_job); + if (r < 0) + return r; + } + + return 0; +} diff --git a/src/grp-machine/grp-import/systemd-pull/pull-tar.h b/src/grp-machine/grp-import/systemd-pull/pull-tar.h new file mode 100644 index 0000000000..21d4db5519 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull-tar.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2015 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-event.h> + +#include "systemd-basic/macro.h" +#include "systemd-shared/import-util.h" + +typedef struct TarPull TarPull; +typedef void (*TarPullFinished)(TarPull *pull, int error, void *userdata); + +int tar_pull_new(TarPull **pull, sd_event *event, const char *image_root, TarPullFinished on_finished, void *userdata); +TarPull* tar_pull_unref(TarPull *pull); + +DEFINE_TRIVIAL_CLEANUP_FUNC(TarPull*, tar_pull_unref); + +int tar_pull_start(TarPull *pull, const char *url, const char *local, bool force_local, ImportVerify verify, bool settings); diff --git a/src/grp-machine/grp-import/systemd-pull/pull.c b/src/grp-machine/grp-import/systemd-pull/pull.c new file mode 100644 index 0000000000..a6b5ed47a8 --- /dev/null +++ b/src/grp-machine/grp-import/systemd-pull/pull.c @@ -0,0 +1,338 @@ +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <getopt.h> + +#include <systemd/sd-event.h> + +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/parse-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/verbs.h" +#include "systemd-basic/web-util.h" +#include "systemd-shared/import-util.h" +#include "systemd-shared/machine-image.h" + +#include "pull-raw.h" +#include "pull-tar.h" + +static bool arg_force = false; +static const char *arg_image_root = "/var/lib/machines"; +static ImportVerify arg_verify = IMPORT_VERIFY_SIGNATURE; +static bool arg_settings = true; + +static int interrupt_signal_handler(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) { + log_notice("Transfer aborted."); + sd_event_exit(sd_event_source_get_event(s), EINTR); + return 0; +} + +static void on_tar_finished(TarPull *pull, int error, void *userdata) { + sd_event *event = userdata; + assert(pull); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int pull_tar(int argc, char *argv[], void *userdata) { + _cleanup_(tar_pull_unrefp) TarPull *pull = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + const char *url, *local; + _cleanup_free_ char *l = NULL, *ll = NULL; + int r; + + url = argv[1]; + if (!http_url_is_valid(url)) { + log_error("URL '%s' is not valid.", url); + return -EINVAL; + } + + if (argc >= 3) + local = argv[2]; + else { + r = import_url_last_component(url, &l); + if (r < 0) + return log_error_errno(r, "Failed get final component of URL: %m"); + + local = l; + } + + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = tar_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local image name '%s' is not valid.", local); + return -EINVAL; + } + + if (!arg_force) { + r = image_find(local, NULL); + if (r < 0) + return log_error_errno(r, "Failed to check whether image '%s' exists: %m", local); + else if (r > 0) { + log_error("Image '%s' already exists.", local); + return -EEXIST; + } + } + + log_info("Pulling '%s', saving as '%s'.", url, local); + } else + log_info("Pulling '%s'.", url); + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = tar_pull_new(&pull, event, arg_image_root, on_tar_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate puller: %m"); + + r = tar_pull_start(pull, url, local, arg_force, arg_verify, arg_settings); + if (r < 0) + return log_error_errno(r, "Failed to pull image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static void on_raw_finished(RawPull *pull, int error, void *userdata) { + sd_event *event = userdata; + assert(pull); + + if (error == 0) + log_info("Operation completed successfully."); + + sd_event_exit(event, abs(error)); +} + +static int pull_raw(int argc, char *argv[], void *userdata) { + _cleanup_(raw_pull_unrefp) RawPull *pull = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + const char *url, *local; + _cleanup_free_ char *l = NULL, *ll = NULL; + int r; + + url = argv[1]; + if (!http_url_is_valid(url)) { + log_error("URL '%s' is not valid.", url); + return -EINVAL; + } + + if (argc >= 3) + local = argv[2]; + else { + r = import_url_last_component(url, &l); + if (r < 0) + return log_error_errno(r, "Failed get final component of URL: %m"); + + local = l; + } + + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = raw_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local image name '%s' is not valid.", local); + return -EINVAL; + } + + if (!arg_force) { + r = image_find(local, NULL); + if (r < 0) + return log_error_errno(r, "Failed to check whether image '%s' exists: %m", local); + else if (r > 0) { + log_error("Image '%s' already exists.", local); + return -EEXIST; + } + } + + log_info("Pulling '%s', saving as '%s'.", url, local); + } else + log_info("Pulling '%s'.", url); + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to allocate event loop: %m"); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + (void) sd_event_add_signal(event, NULL, SIGTERM, interrupt_signal_handler, NULL); + (void) sd_event_add_signal(event, NULL, SIGINT, interrupt_signal_handler, NULL); + + r = raw_pull_new(&pull, event, arg_image_root, on_raw_finished, event); + if (r < 0) + return log_error_errno(r, "Failed to allocate puller: %m"); + + r = raw_pull_start(pull, url, local, arg_force, arg_verify, arg_settings); + if (r < 0) + return log_error_errno(r, "Failed to pull image: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + log_info("Exiting."); + return -r; +} + +static int help(int argc, char *argv[], void *userdata) { + + printf("%s [OPTIONS...] {COMMAND} ...\n\n" + "Download container or virtual machine images.\n\n" + " -h --help Show this help\n" + " --version Show package version\n" + " --force Force creation of image\n" + " --verify=MODE Verify downloaded image, one of: 'no',\n" + " 'checksum', 'signature'\n" + " --settings=BOOL Download settings file with image\n" + " --image-root=PATH Image root directory\n\n" + "Commands:\n" + " tar URL [NAME] Download a TAR image\n" + " raw URL [NAME] Download a RAW image\n", + program_invocation_short_name); + + return 0; +} + +static int parse_argv(int argc, char *argv[]) { + + enum { + ARG_VERSION = 0x100, + ARG_FORCE, + ARG_IMAGE_ROOT, + ARG_VERIFY, + ARG_SETTINGS, + }; + + static const struct option options[] = { + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "force", no_argument, NULL, ARG_FORCE }, + { "image-root", required_argument, NULL, ARG_IMAGE_ROOT }, + { "verify", required_argument, NULL, ARG_VERIFY }, + { "settings", required_argument, NULL, ARG_SETTINGS }, + {} + }; + + int c, r; + + assert(argc >= 0); + assert(argv); + + while ((c = getopt_long(argc, argv, "h", options, NULL)) >= 0) + + switch (c) { + + case 'h': + return help(0, NULL, NULL); + + case ARG_VERSION: + return version(); + + case ARG_FORCE: + arg_force = true; + break; + + case ARG_IMAGE_ROOT: + arg_image_root = optarg; + break; + + case ARG_VERIFY: + arg_verify = import_verify_from_string(optarg); + if (arg_verify < 0) { + log_error("Invalid verification setting '%s'", optarg); + return -EINVAL; + } + + break; + + case ARG_SETTINGS: + r = parse_boolean(optarg); + if (r < 0) + return log_error_errno(r, "Failed to parse --settings= parameter '%s'", optarg); + + arg_settings = r; + break; + + case '?': + return -EINVAL; + + default: + assert_not_reached("Unhandled option"); + } + + return 1; +} + +static int pull_main(int argc, char *argv[]) { + + static const Verb verbs[] = { + { "help", VERB_ANY, VERB_ANY, 0, help }, + { "tar", 2, 3, 0, pull_tar }, + { "raw", 2, 3, 0, pull_raw }, + {} + }; + + return dispatch_verb(argc, argv, verbs, NULL); +} + +int main(int argc, char *argv[]) { + int r; + + setlocale(LC_ALL, ""); + log_parse_environment(); + log_open(); + + r = parse_argv(argc, argv); + if (r <= 0) + goto finish; + + (void) ignore_signals(SIGPIPE, -1); + + r = pull_main(argc, argv); + +finish: + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/libmachine-core/GNUmakefile b/src/grp-machine/libmachine-core/GNUmakefile new file mode 120000 index 0000000000..95e5924740 --- /dev/null +++ b/src/grp-machine/libmachine-core/GNUmakefile @@ -0,0 +1 @@ +../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/libmachine-core/Makefile b/src/grp-machine/libmachine-core/Makefile new file mode 100644 index 0000000000..7c64e6af5f --- /dev/null +++ b/src/grp-machine/libmachine-core/Makefile @@ -0,0 +1,29 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +nested.subdirs += src +nested.subdirs += test + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/libmachine-core/include/machine-core/image-dbus.h b/src/grp-machine/libmachine-core/include/machine-core/image-dbus.h new file mode 100644 index 0000000000..b62da996c6 --- /dev/null +++ b/src/grp-machine/libmachine-core/include/machine-core/image-dbus.h @@ -0,0 +1,35 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "machined.h" + +extern const sd_bus_vtable image_vtable[]; + +char *image_bus_path(const char *name); + +int image_object_find(sd_bus *bus, const char *path, const char *interface, void *userdata, void **found, sd_bus_error *error); +int image_node_enumerator(sd_bus *bus, const char *path, void *userdata, char ***nodes, sd_bus_error *error); + +int bus_image_method_remove(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_image_method_rename(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_image_method_clone(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_image_method_mark_read_only(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_image_method_set_limit(sd_bus_message *message, void *userdata, sd_bus_error *error); diff --git a/src/grp-machine/libmachine-core/include/machine-core/machine-dbus.h b/src/grp-machine/libmachine-core/include/machine-core/machine-dbus.h new file mode 100644 index 0000000000..d3faf5cb07 --- /dev/null +++ b/src/grp-machine/libmachine-core/include/machine-core/machine-dbus.h @@ -0,0 +1,44 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2013 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <systemd/sd-bus.h> + +#include "machine.h" + +extern const sd_bus_vtable machine_vtable[]; + +char *machine_bus_path(Machine *s); +int machine_object_find(sd_bus *bus, const char *path, const char *interface, void *userdata, void **found, sd_bus_error *error); +int machine_node_enumerator(sd_bus *bus, const char *path, void *userdata, char ***nodes, sd_bus_error *error); + +int bus_machine_method_terminate(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_kill(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_get_addresses(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_get_os_release(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_open_pty(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_open_login(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_open_shell(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_copy(sd_bus_message *message, void *userdata, sd_bus_error *error); +int bus_machine_method_open_root_directory(sd_bus_message *message, void *userdata, sd_bus_error *error); + +int machine_send_signal(Machine *m, bool new_machine); +int machine_send_create_reply(Machine *m, sd_bus_error *error); diff --git a/src/grp-machine/libmachine-core/include/machine-core/machine.h b/src/grp-machine/libmachine-core/include/machine-core/machine.h new file mode 100644 index 0000000000..361451b1c8 --- /dev/null +++ b/src/grp-machine/libmachine-core/include/machine-core/machine.h @@ -0,0 +1,111 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2013 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "systemd-basic/list.h" + +typedef enum KillWho KillWho; +typedef struct Machine Machine; + +#include "machined.h" +#include "operation.h" + +typedef enum MachineState { + MACHINE_OPENING, /* Machine is being registered */ + MACHINE_RUNNING, /* Machine is running */ + MACHINE_CLOSING, /* Machine is terminating */ + _MACHINE_STATE_MAX, + _MACHINE_STATE_INVALID = -1 +} MachineState; + +typedef enum MachineClass { + MACHINE_CONTAINER, + MACHINE_VM, + MACHINE_HOST, + _MACHINE_CLASS_MAX, + _MACHINE_CLASS_INVALID = -1 +} MachineClass; + +enum KillWho { + KILL_LEADER, + KILL_ALL, + _KILL_WHO_MAX, + _KILL_WHO_INVALID = -1 +}; + +struct Machine { + Manager *manager; + + char *name; + sd_id128_t id; + + MachineClass class; + + char *state_file; + char *service; + char *root_directory; + + char *unit; + char *scope_job; + + pid_t leader; + + dual_timestamp timestamp; + + bool in_gc_queue:1; + bool started:1; + bool stopping:1; + + sd_bus_message *create_message; + + int *netif; + unsigned n_netif; + + LIST_HEAD(Operation, operations); + + LIST_FIELDS(Machine, gc_queue); +}; + +Machine* machine_new(Manager *manager, MachineClass class, const char *name); +void machine_free(Machine *m); +bool machine_check_gc(Machine *m, bool drop_not_started); +void machine_add_to_gc_queue(Machine *m); +int machine_start(Machine *m, sd_bus_message *properties, sd_bus_error *error); +int machine_stop(Machine *m); +int machine_finalize(Machine *m); +int machine_save(Machine *m); +int machine_load(Machine *m); +int machine_kill(Machine *m, KillWho who, int signo); + +void machine_release_unit(Machine *m); + +MachineState machine_get_state(Machine *u); + +const char* machine_class_to_string(MachineClass t) _const_; +MachineClass machine_class_from_string(const char *s) _pure_; + +const char* machine_state_to_string(MachineState t) _const_; +MachineState machine_state_from_string(const char *s) _pure_; + +const char *kill_who_to_string(KillWho k) _const_; +KillWho kill_who_from_string(const char *s) _pure_; + +int machine_openpt(Machine *m, int flags); +int machine_open_terminal(Machine *m, const char *path, int mode); diff --git a/src/grp-machine/libmachine-core/include/machine-core/machined.h b/src/grp-machine/libmachine-core/include/machine-core/machined.h new file mode 100644 index 0000000000..8994097db0 --- /dev/null +++ b/src/grp-machine/libmachine-core/include/machine-core/machined.h @@ -0,0 +1,82 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2013 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <stdbool.h> + +#include <systemd/sd-bus.h> +#include <systemd/sd-event.h> + +#include "systemd-basic/hashmap.h" +#include "systemd-basic/list.h" + +typedef struct Manager Manager; + +#include "image-dbus.h" +#include "machine-dbus.h" +#include "machine.h" +#include "operation.h" + +struct Manager { + sd_event *event; + sd_bus *bus; + + Hashmap *machines; + Hashmap *machine_units; + Hashmap *machine_leaders; + + Hashmap *polkit_registry; + + Hashmap *image_cache; + sd_event_source *image_cache_defer_event; + + LIST_HEAD(Machine, machine_gc_queue); + + Machine *host_machine; + + LIST_HEAD(Operation, operations); + unsigned n_operations; +}; + +Manager *manager_new(void); +void manager_free(Manager *m); + +int manager_add_machine(Manager *m, const char *name, Machine **_machine); +int manager_enumerate_machines(Manager *m); + +int manager_startup(Manager *m); +int manager_run(Manager *m); + +void manager_gc(Manager *m, bool drop_not_started); + +int manager_get_machine_by_pid(Manager *m, pid_t pid, Machine **machine); + +extern const sd_bus_vtable manager_vtable[]; + +int match_reloading(sd_bus_message *message, void *userdata, sd_bus_error *error); +int match_unit_removed(sd_bus_message *message, void *userdata, sd_bus_error *error); +int match_properties_changed(sd_bus_message *message, void *userdata, sd_bus_error *error); +int match_job_removed(sd_bus_message *message, void *userdata, sd_bus_error *error); + +int manager_start_scope(Manager *manager, const char *scope, pid_t pid, const char *slice, const char *description, sd_bus_message *more_properties, sd_bus_error *error, char **job); +int manager_stop_unit(Manager *manager, const char *unit, sd_bus_error *error, char **job); +int manager_kill_unit(Manager *manager, const char *unit, int signo, sd_bus_error *error); +int manager_unit_is_active(Manager *manager, const char *unit); +int manager_job_is_active(Manager *manager, const char *path); diff --git a/src/grp-machine/libmachine-core/include/machine-core/operation.h b/src/grp-machine/libmachine-core/include/machine-core/operation.h new file mode 100644 index 0000000000..5eed62bd6f --- /dev/null +++ b/src/grp-machine/libmachine-core/include/machine-core/operation.h @@ -0,0 +1,49 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2016 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <sys/types.h> + +#include <systemd/sd-bus.h> +#include <systemd/sd-event.h> + +#include "systemd-basic/list.h" + +typedef struct Operation Operation; + +#include "machined.h" + +#define OPERATIONS_MAX 64 + +struct Operation { + Manager *manager; + Machine *machine; + pid_t pid; + sd_bus_message *message; + int errno_fd; + int extra_fd; + sd_event_source *event_source; + int (*done)(Operation *o, int ret, sd_bus_error *error); + LIST_FIELDS(Operation, operations); + LIST_FIELDS(Operation, operations_by_machine); +}; + +int operation_new(Manager *manager, Machine *machine, pid_t child, sd_bus_message *message, int errno_fd, Operation **ret); +Operation *operation_free(Operation *o); diff --git a/src/grp-machine/libmachine-core/src/GNUmakefile b/src/grp-machine/libmachine-core/src/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/libmachine-core/src/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/libmachine-core/src/Makefile b/src/grp-machine/libmachine-core/src/Makefile new file mode 100644 index 0000000000..ac7cdc0e2d --- /dev/null +++ b/src/grp-machine/libmachine-core/src/Makefile @@ -0,0 +1,43 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +libmachine_core_la_SOURCES = \ + src/machine/machine.c \ + src/machine/machine.h \ + src/machine/machined-dbus.c \ + src/machine/machine-dbus.c \ + src/machine/machine-dbus.h \ + src/machine/image-dbus.c \ + src/machine/image-dbus.h \ + src/machine/operation.c \ + src/machine/operation.h + +libmachine_core_la_LIBADD = \ + libsystemd-shared.la + +noinst_LTLIBRARIES += \ + libmachine-core.la + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/libmachine-core/src/image-dbus.c b/src/grp-machine/libmachine-core/src/image-dbus.c new file mode 100644 index 0000000000..f3467bf193 --- /dev/null +++ b/src/grp-machine/libmachine-core/src/image-dbus.c @@ -0,0 +1,422 @@ +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "machine-core/image-dbus.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/bus-label.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/user-util.h" +#include "systemd-shared/machine-image.h" + +static BUS_DEFINE_PROPERTY_GET_ENUM(property_get_type, image_type, ImageType); + +int bus_image_method_remove( + sd_bus_message *message, + void *userdata, + sd_bus_error *error) { + + _cleanup_close_pair_ int errno_pipe_fd[2] = { -1, -1 }; + Image *image = userdata; + Manager *m = image->userdata; + pid_t child; + int r; + + assert(message); + assert(image); + + if (m->n_operations >= OPERATIONS_MAX) + return sd_bus_error_setf(error, SD_BUS_ERROR_LIMITS_EXCEEDED, "Too many ongoing operations."); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-images", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + if (pipe2(errno_pipe_fd, O_CLOEXEC|O_NONBLOCK) < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to create pipe: %m"); + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + if (child == 0) { + errno_pipe_fd[0] = safe_close(errno_pipe_fd[0]); + + r = image_remove(image); + if (r < 0) { + (void) write(errno_pipe_fd[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + _exit(EXIT_SUCCESS); + } + + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + r = operation_new(m, NULL, child, message, errno_pipe_fd[0], NULL); + if (r < 0) { + (void) sigkill_wait(child); + return r; + } + + errno_pipe_fd[0] = -1; + + return 1; +} + +int bus_image_method_rename( + sd_bus_message *message, + void *userdata, + sd_bus_error *error) { + + Image *image = userdata; + Manager *m = image->userdata; + const char *new_name; + int r; + + assert(message); + assert(image); + + r = sd_bus_message_read(message, "s", &new_name); + if (r < 0) + return r; + + if (!image_name_is_valid(new_name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", new_name); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-images", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = image_rename(image, new_name); + if (r < 0) + return r; + + return sd_bus_reply_method_return(message, NULL); +} + +int bus_image_method_clone( + sd_bus_message *message, + void *userdata, + sd_bus_error *error) { + + _cleanup_close_pair_ int errno_pipe_fd[2] = { -1, -1 }; + Image *image = userdata; + Manager *m = image->userdata; + const char *new_name; + int r, read_only; + pid_t child; + + assert(message); + assert(image); + assert(m); + + if (m->n_operations >= OPERATIONS_MAX) + return sd_bus_error_setf(error, SD_BUS_ERROR_LIMITS_EXCEEDED, "Too many ongoing operations."); + + r = sd_bus_message_read(message, "sb", &new_name, &read_only); + if (r < 0) + return r; + + if (!image_name_is_valid(new_name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", new_name); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-images", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + if (pipe2(errno_pipe_fd, O_CLOEXEC|O_NONBLOCK) < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to create pipe: %m"); + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + if (child == 0) { + errno_pipe_fd[0] = safe_close(errno_pipe_fd[0]); + + r = image_clone(image, new_name, read_only); + if (r < 0) { + (void) write(errno_pipe_fd[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + _exit(EXIT_SUCCESS); + } + + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + r = operation_new(m, NULL, child, message, errno_pipe_fd[0], NULL); + if (r < 0) { + (void) sigkill_wait(child); + return r; + } + + errno_pipe_fd[0] = -1; + + return 1; +} + +int bus_image_method_mark_read_only( + sd_bus_message *message, + void *userdata, + sd_bus_error *error) { + + Image *image = userdata; + Manager *m = image->userdata; + int r, read_only; + + assert(message); + + r = sd_bus_message_read(message, "b", &read_only); + if (r < 0) + return r; + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-images", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = image_read_only(image, read_only); + if (r < 0) + return r; + + return sd_bus_reply_method_return(message, NULL); +} + +int bus_image_method_set_limit( + sd_bus_message *message, + void *userdata, + sd_bus_error *error) { + + Image *image = userdata; + Manager *m = image->userdata; + uint64_t limit; + int r; + + assert(message); + + r = sd_bus_message_read(message, "t", &limit); + if (r < 0) + return r; + if (!FILE_SIZE_VALID_OR_INFINITY(limit)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "New limit out of range"); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-images", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = image_set_limit(image, limit); + if (r < 0) + return r; + + return sd_bus_reply_method_return(message, NULL); +} + +const sd_bus_vtable image_vtable[] = { + SD_BUS_VTABLE_START(0), + SD_BUS_PROPERTY("Name", "s", NULL, offsetof(Image, name), 0), + SD_BUS_PROPERTY("Path", "s", NULL, offsetof(Image, path), 0), + SD_BUS_PROPERTY("Type", "s", property_get_type, offsetof(Image, type), 0), + SD_BUS_PROPERTY("ReadOnly", "b", bus_property_get_bool, offsetof(Image, read_only), 0), + SD_BUS_PROPERTY("CreationTimestamp", "t", NULL, offsetof(Image, crtime), 0), + SD_BUS_PROPERTY("ModificationTimestamp", "t", NULL, offsetof(Image, mtime), 0), + SD_BUS_PROPERTY("Usage", "t", NULL, offsetof(Image, usage), 0), + SD_BUS_PROPERTY("Limit", "t", NULL, offsetof(Image, limit), 0), + SD_BUS_PROPERTY("UsageExclusive", "t", NULL, offsetof(Image, usage_exclusive), 0), + SD_BUS_PROPERTY("LimitExclusive", "t", NULL, offsetof(Image, limit_exclusive), 0), + SD_BUS_METHOD("Remove", NULL, NULL, bus_image_method_remove, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("Rename", "s", NULL, bus_image_method_rename, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("Clone", "sb", NULL, bus_image_method_clone, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MarkReadOnly", "b", NULL, bus_image_method_mark_read_only, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("SetLimit", "t", NULL, bus_image_method_set_limit, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_VTABLE_END +}; + +static int image_flush_cache(sd_event_source *s, void *userdata) { + Manager *m = userdata; + Image *i; + + assert(s); + assert(m); + + while ((i = hashmap_steal_first(m->image_cache))) + image_unref(i); + + return 0; +} + +int image_object_find(sd_bus *bus, const char *path, const char *interface, void *userdata, void **found, sd_bus_error *error) { + _cleanup_free_ char *e = NULL; + Manager *m = userdata; + Image *image = NULL; + const char *p; + int r; + + assert(bus); + assert(path); + assert(interface); + assert(found); + + p = startswith(path, "/org/freedesktop/machine1/image/"); + if (!p) + return 0; + + e = bus_label_unescape(p); + if (!e) + return -ENOMEM; + + image = hashmap_get(m->image_cache, e); + if (image) { + *found = image; + return 1; + } + + r = hashmap_ensure_allocated(&m->image_cache, &string_hash_ops); + if (r < 0) + return r; + + if (!m->image_cache_defer_event) { + r = sd_event_add_defer(m->event, &m->image_cache_defer_event, image_flush_cache, m); + if (r < 0) + return r; + + r = sd_event_source_set_priority(m->image_cache_defer_event, SD_EVENT_PRIORITY_IDLE); + if (r < 0) + return r; + } + + r = sd_event_source_set_enabled(m->image_cache_defer_event, SD_EVENT_ONESHOT); + if (r < 0) + return r; + + r = image_find(e, &image); + if (r <= 0) + return r; + + image->userdata = m; + + r = hashmap_put(m->image_cache, image->name, image); + if (r < 0) { + image_unref(image); + return r; + } + + *found = image; + return 1; +} + +char *image_bus_path(const char *name) { + _cleanup_free_ char *e = NULL; + + assert(name); + + e = bus_label_escape(name); + if (!e) + return NULL; + + return strappend("/org/freedesktop/machine1/image/", e); +} + +int image_node_enumerator(sd_bus *bus, const char *path, void *userdata, char ***nodes, sd_bus_error *error) { + _cleanup_(image_hashmap_freep) Hashmap *images = NULL; + _cleanup_strv_free_ char **l = NULL; + Image *image; + Iterator i; + int r; + + assert(bus); + assert(path); + assert(nodes); + + images = hashmap_new(&string_hash_ops); + if (!images) + return -ENOMEM; + + r = image_discover(images); + if (r < 0) + return r; + + HASHMAP_FOREACH(image, images, i) { + char *p; + + p = image_bus_path(image->name); + if (!p) + return -ENOMEM; + + r = strv_consume(&l, p); + if (r < 0) + return r; + } + + *nodes = l; + l = NULL; + + return 1; +} diff --git a/src/grp-machine/libmachine-core/src/machine-dbus.c b/src/grp-machine/libmachine-core/src/machine-dbus.c new file mode 100644 index 0000000000..b7363054b3 --- /dev/null +++ b/src/grp-machine/libmachine-core/src/machine-dbus.c @@ -0,0 +1,1454 @@ +/*** + This file is part of systemd. + + Copyright 2011 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <errno.h> +#include <string.h> +#include <sys/mount.h> +#include <sys/wait.h> + +/* When we include libgen.h because we need dirname() we immediately + * undefine basename() since libgen.h defines it as a macro to the POSIX + * version which is really broken. We prefer GNU basename(). */ +#include <libgen.h> +#undef basename + +#include "machine-core/machine-dbus.h" +#include "machine-core/machine.h" +#include "sd-bus/bus-common-errors.h" +#include "sd-bus/bus-internal.h" +#include "sd-bus/bus-util.h" +#include "sd-netlink/local-addresses.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/bus-label.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/env-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/formats-util.h" +#include "systemd-basic/fs-util.h" +#include "systemd-basic/in-addr-util.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/terminal-util.h" +#include "systemd-basic/user-util.h" + +static int property_get_state( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + Machine *m = userdata; + const char *state; + int r; + + assert(bus); + assert(reply); + assert(m); + + state = machine_state_to_string(machine_get_state(m)); + + r = sd_bus_message_append_basic(reply, 's', state); + if (r < 0) + return r; + + return 1; +} + +static int property_get_netif( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + Machine *m = userdata; + + assert(bus); + assert(reply); + assert(m); + + assert_cc(sizeof(int) == sizeof(int32_t)); + + return sd_bus_message_append_array(reply, 'i', m->netif, m->n_netif * sizeof(int)); +} + +static BUS_DEFINE_PROPERTY_GET_ENUM(property_get_class, machine_class, MachineClass); + +int bus_machine_method_terminate(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Machine *m = userdata; + int r; + + assert(message); + assert(m); + + r = bus_verify_polkit_async( + message, + CAP_KILL, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = machine_stop(m); + if (r < 0) + return r; + + return sd_bus_reply_method_return(message, NULL); +} + +int bus_machine_method_kill(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Machine *m = userdata; + const char *swho; + int32_t signo; + KillWho who; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "si", &swho, &signo); + if (r < 0) + return r; + + if (isempty(swho)) + who = KILL_ALL; + else { + who = kill_who_from_string(swho); + if (who < 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid kill parameter '%s'", swho); + } + + if (!SIGNAL_VALID(signo)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid signal %i", signo); + + r = bus_verify_polkit_async( + message, + CAP_KILL, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + r = machine_kill(m, who, signo); + if (r < 0) + return r; + + return sd_bus_reply_method_return(message, NULL); +} + +int bus_machine_method_get_addresses(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + Machine *m = userdata; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_open_container(reply, 'a', "(iay)"); + if (r < 0) + return r; + + switch (m->class) { + + case MACHINE_HOST: { + _cleanup_free_ struct local_address *addresses = NULL; + struct local_address *a; + int n, i; + + n = local_addresses(NULL, 0, AF_UNSPEC, &addresses); + if (n < 0) + return n; + + for (a = addresses, i = 0; i < n; a++, i++) { + + r = sd_bus_message_open_container(reply, 'r', "iay"); + if (r < 0) + return r; + + r = sd_bus_message_append(reply, "i", addresses[i].family); + if (r < 0) + return r; + + r = sd_bus_message_append_array(reply, 'y', &addresses[i].address, FAMILY_ADDRESS_SIZE(addresses[i].family)); + if (r < 0) + return r; + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + } + + break; + } + + case MACHINE_CONTAINER: { + _cleanup_close_pair_ int pair[2] = { -1, -1 }; + _cleanup_free_ char *us = NULL, *them = NULL; + _cleanup_close_ int netns_fd = -1; + const char *p; + siginfo_t si; + pid_t child; + + r = readlink_malloc("/proc/self/ns/net", &us); + if (r < 0) + return r; + + p = procfs_file_alloca(m->leader, "ns/net"); + r = readlink_malloc(p, &them); + if (r < 0) + return r; + + if (streq(us, them)) + return sd_bus_error_setf(error, BUS_ERROR_NO_PRIVATE_NETWORKING, "Machine %s does not use private networking", m->name); + + r = namespace_open(m->leader, NULL, NULL, &netns_fd, NULL, NULL); + if (r < 0) + return r; + + if (socketpair(AF_UNIX, SOCK_SEQPACKET, 0, pair) < 0) + return -errno; + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + + if (child == 0) { + _cleanup_free_ struct local_address *addresses = NULL; + struct local_address *a; + int i, n; + + pair[0] = safe_close(pair[0]); + + r = namespace_enter(-1, -1, netns_fd, -1, -1); + if (r < 0) + _exit(EXIT_FAILURE); + + n = local_addresses(NULL, 0, AF_UNSPEC, &addresses); + if (n < 0) + _exit(EXIT_FAILURE); + + for (a = addresses, i = 0; i < n; a++, i++) { + struct iovec iov[2] = { + { .iov_base = &a->family, .iov_len = sizeof(a->family) }, + { .iov_base = &a->address, .iov_len = FAMILY_ADDRESS_SIZE(a->family) }, + }; + + r = writev(pair[1], iov, 2); + if (r < 0) + _exit(EXIT_FAILURE); + } + + pair[1] = safe_close(pair[1]); + + _exit(EXIT_SUCCESS); + } + + pair[1] = safe_close(pair[1]); + + for (;;) { + int family; + ssize_t n; + union in_addr_union in_addr; + struct iovec iov[2]; + struct msghdr mh = { + .msg_iov = iov, + .msg_iovlen = 2, + }; + + iov[0] = (struct iovec) { .iov_base = &family, .iov_len = sizeof(family) }; + iov[1] = (struct iovec) { .iov_base = &in_addr, .iov_len = sizeof(in_addr) }; + + n = recvmsg(pair[0], &mh, 0); + if (n < 0) + return -errno; + if ((size_t) n < sizeof(family)) + break; + + r = sd_bus_message_open_container(reply, 'r', "iay"); + if (r < 0) + return r; + + r = sd_bus_message_append(reply, "i", family); + if (r < 0) + return r; + + switch (family) { + + case AF_INET: + if (n != sizeof(struct in_addr) + sizeof(family)) + return -EIO; + + r = sd_bus_message_append_array(reply, 'y', &in_addr.in, sizeof(in_addr.in)); + break; + + case AF_INET6: + if (n != sizeof(struct in6_addr) + sizeof(family)) + return -EIO; + + r = sd_bus_message_append_array(reply, 'y', &in_addr.in6, sizeof(in_addr.in6)); + break; + } + if (r < 0) + return r; + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + } + + r = wait_for_terminate(child, &si); + if (r < 0) + return sd_bus_error_set_errnof(error, r, "Failed to wait for child: %m"); + if (si.si_code != CLD_EXITED || si.si_status != EXIT_SUCCESS) + return sd_bus_error_setf(error, SD_BUS_ERROR_FAILED, "Child died abnormally."); + break; + } + + default: + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Requesting IP address data is only supported on container machines."); + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +int bus_machine_method_get_os_release(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_strv_free_ char **l = NULL; + Machine *m = userdata; + char **k, **v; + int r; + + assert(message); + assert(m); + + switch (m->class) { + + case MACHINE_HOST: + r = load_env_file_pairs(NULL, "/etc/os-release", NULL, &l); + if (r < 0) + return r; + + break; + + case MACHINE_CONTAINER: { + _cleanup_close_ int mntns_fd = -1, root_fd = -1; + _cleanup_close_pair_ int pair[2] = { -1, -1 }; + _cleanup_fclose_ FILE *f = NULL; + siginfo_t si; + pid_t child; + + r = namespace_open(m->leader, NULL, &mntns_fd, NULL, NULL, &root_fd); + if (r < 0) + return r; + + if (socketpair(AF_UNIX, SOCK_SEQPACKET, 0, pair) < 0) + return -errno; + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + + if (child == 0) { + _cleanup_close_ int fd = -1; + + pair[0] = safe_close(pair[0]); + + r = namespace_enter(-1, mntns_fd, -1, -1, root_fd); + if (r < 0) + _exit(EXIT_FAILURE); + + fd = open("/etc/os-release", O_RDONLY|O_CLOEXEC); + if (fd < 0) { + fd = open("/usr/lib/os-release", O_RDONLY|O_CLOEXEC); + if (fd < 0) + _exit(EXIT_FAILURE); + } + + r = copy_bytes(fd, pair[1], (uint64_t) -1, false); + if (r < 0) + _exit(EXIT_FAILURE); + + _exit(EXIT_SUCCESS); + } + + pair[1] = safe_close(pair[1]); + + f = fdopen(pair[0], "re"); + if (!f) + return -errno; + + pair[0] = -1; + + r = load_env_file_pairs(f, "/etc/os-release", NULL, &l); + if (r < 0) + return r; + + r = wait_for_terminate(child, &si); + if (r < 0) + return sd_bus_error_set_errnof(error, r, "Failed to wait for child: %m"); + if (si.si_code != CLD_EXITED || si.si_status != EXIT_SUCCESS) + return sd_bus_error_setf(error, SD_BUS_ERROR_FAILED, "Child died abnormally."); + + break; + } + + default: + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Requesting OS release data is only supported on container machines."); + } + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_open_container(reply, 'a', "{ss}"); + if (r < 0) + return r; + + STRV_FOREACH_PAIR(k, v, l) { + r = sd_bus_message_append(reply, "{ss}", *k, *v); + if (r < 0) + return r; + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +int bus_machine_method_open_pty(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *pty_name = NULL; + _cleanup_close_ int master = -1; + Machine *m = userdata; + int r; + + assert(message); + assert(m); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + m->class == MACHINE_HOST ? "org.freedesktop.machine1.host-open-pty" : "org.freedesktop.machine1.open-pty", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + master = machine_openpt(m, O_RDWR|O_NOCTTY|O_CLOEXEC); + if (master < 0) + return master; + + r = ptsname_namespace(master, &pty_name); + if (r < 0) + return r; + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_append(reply, "hs", master, pty_name); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +static int container_bus_new(Machine *m, sd_bus_error *error, sd_bus **ret) { + int r; + + assert(m); + assert(ret); + + switch (m->class) { + + case MACHINE_HOST: + *ret = NULL; + break; + + case MACHINE_CONTAINER: { + _cleanup_(sd_bus_unrefp) sd_bus *bus = NULL; + char *address; + + r = sd_bus_new(&bus); + if (r < 0) + return r; + + if (asprintf(&address, "x-machine-kernel:pid=%1$" PID_PRI ";x-machine-unix:pid=%1$" PID_PRI, m->leader) < 0) + return -ENOMEM; + + bus->address = address; + bus->bus_client = true; + bus->trusted = false; + bus->is_system = true; + + r = sd_bus_start(bus); + if (r == -ENOENT) + return sd_bus_error_set_errnof(error, r, "There is no system bus in container %s.", m->name); + if (r < 0) + return r; + + *ret = bus; + bus = NULL; + break; + } + + default: + return -EOPNOTSUPP; + } + + return 0; +} + +int bus_machine_method_open_login(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *pty_name = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *allocated_bus = NULL; + _cleanup_close_ int master = -1; + sd_bus *container_bus = NULL; + Machine *m = userdata; + const char *p, *getty; + int r; + + assert(message); + assert(m); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + m->class == MACHINE_HOST ? "org.freedesktop.machine1.host-login" : "org.freedesktop.machine1.login", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + master = machine_openpt(m, O_RDWR|O_NOCTTY|O_CLOEXEC); + if (master < 0) + return master; + + r = ptsname_namespace(master, &pty_name); + if (r < 0) + return r; + + p = path_startswith(pty_name, "/dev/pts/"); + if (!p) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "PTS name %s is invalid", pty_name); + + r = container_bus_new(m, error, &allocated_bus); + if (r < 0) + return r; + + container_bus = allocated_bus ?: m->manager->bus; + + getty = strjoina("container-getty@", p, ".service"); + + r = sd_bus_call_method( + container_bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "StartUnit", + error, NULL, + "ss", getty, "replace"); + if (r < 0) + return r; + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_append(reply, "hs", master, pty_name); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +int bus_machine_method_open_shell(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL, *tm = NULL; + _cleanup_free_ char *pty_name = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *allocated_bus = NULL; + sd_bus *container_bus = NULL; + _cleanup_close_ int master = -1, slave = -1; + _cleanup_strv_free_ char **env = NULL, **args = NULL; + Machine *m = userdata; + const char *p, *unit, *user, *path, *description, *utmp_id; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "ss", &user, &path); + if (r < 0) + return r; + user = empty_to_null(user); + if (isempty(path)) + path = "/bin/sh"; + if (!path_is_absolute(path)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Specified path '%s' is not absolute", path); + + r = sd_bus_message_read_strv(message, &args); + if (r < 0) + return r; + if (strv_isempty(args)) { + args = strv_free(args); + + args = strv_new(path, NULL); + if (!args) + return -ENOMEM; + + args[0][0] = '-'; /* Tell /bin/sh that this shall be a login shell */ + } + + r = sd_bus_message_read_strv(message, &env); + if (r < 0) + return r; + if (!strv_env_is_valid(env)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid environment assignments"); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + m->class == MACHINE_HOST ? "org.freedesktop.machine1.host-shell" : "org.freedesktop.machine1.shell", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + master = machine_openpt(m, O_RDWR|O_NOCTTY|O_CLOEXEC); + if (master < 0) + return master; + + r = ptsname_namespace(master, &pty_name); + if (r < 0) + return r; + + p = path_startswith(pty_name, "/dev/pts/"); + assert(p); + + slave = machine_open_terminal(m, pty_name, O_RDWR|O_NOCTTY|O_CLOEXEC); + if (slave < 0) + return slave; + + utmp_id = path_startswith(pty_name, "/dev/"); + assert(utmp_id); + + r = container_bus_new(m, error, &allocated_bus); + if (r < 0) + return r; + + container_bus = allocated_bus ?: m->manager->bus; + + r = sd_bus_message_new_method_call( + container_bus, + &tm, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "StartTransientUnit"); + if (r < 0) + return r; + + /* Name and mode */ + unit = strjoina("container-shell@", p, ".service"); + r = sd_bus_message_append(tm, "ss", unit, "fail"); + if (r < 0) + return r; + + /* Properties */ + r = sd_bus_message_open_container(tm, 'a', "(sv)"); + if (r < 0) + return r; + + description = strjoina("Shell for User ", isempty(user) ? "root" : user); + r = sd_bus_message_append(tm, + "(sv)(sv)(sv)(sv)(sv)(sv)(sv)(sv)(sv)(sv)(sv)(sv)", + "Description", "s", description, + "StandardInputFileDescriptor", "h", slave, + "StandardOutputFileDescriptor", "h", slave, + "StandardErrorFileDescriptor", "h", slave, + "SendSIGHUP", "b", true, + "IgnoreSIGPIPE", "b", false, + "KillMode", "s", "mixed", + "TTYReset", "b", true, + "UtmpIdentifier", "s", utmp_id, + "UtmpMode", "s", "user", + "PAMName", "s", "login", + "WorkingDirectory", "s", "-~"); + if (r < 0) + return r; + + r = sd_bus_message_append(tm, "(sv)", "User", "s", isempty(user) ? "root" : user); + if (r < 0) + return r; + + if (!strv_isempty(env)) { + r = sd_bus_message_open_container(tm, 'r', "sv"); + if (r < 0) + return r; + + r = sd_bus_message_append(tm, "s", "Environment"); + if (r < 0) + return r; + + r = sd_bus_message_open_container(tm, 'v', "as"); + if (r < 0) + return r; + + r = sd_bus_message_append_strv(tm, env); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + } + + /* Exec container */ + r = sd_bus_message_open_container(tm, 'r', "sv"); + if (r < 0) + return r; + + r = sd_bus_message_append(tm, "s", "ExecStart"); + if (r < 0) + return r; + + r = sd_bus_message_open_container(tm, 'v', "a(sasb)"); + if (r < 0) + return r; + + r = sd_bus_message_open_container(tm, 'a', "(sasb)"); + if (r < 0) + return r; + + r = sd_bus_message_open_container(tm, 'r', "sasb"); + if (r < 0) + return r; + + r = sd_bus_message_append(tm, "s", path); + if (r < 0) + return r; + + r = sd_bus_message_append_strv(tm, args); + if (r < 0) + return r; + + r = sd_bus_message_append(tm, "b", true); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + r = sd_bus_message_close_container(tm); + if (r < 0) + return r; + + /* Auxiliary units */ + r = sd_bus_message_append(tm, "a(sa(sv))", 0); + if (r < 0) + return r; + + r = sd_bus_call(container_bus, tm, 0, error, NULL); + if (r < 0) + return r; + + slave = safe_close(slave); + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_append(reply, "hs", master, pty_name); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_close_pair_ int errno_pipe_fd[2] = { -1, -1 }; + char mount_slave[] = "/tmp/propagate.XXXXXX", *mount_tmp, *mount_outside, *p; + bool mount_slave_created = false, mount_slave_mounted = false, + mount_tmp_created = false, mount_tmp_mounted = false, + mount_outside_created = false, mount_outside_mounted = false; + const char *dest, *src; + Machine *m = userdata; + int read_only, make_directory; + pid_t child; + siginfo_t si; + int r; + + assert(message); + assert(m); + + if (m->class != MACHINE_CONTAINER) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Bind mounting is only supported on container machines."); + + r = sd_bus_message_read(message, "ssbb", &src, &dest, &read_only, &make_directory); + if (r < 0) + return r; + + if (!path_is_absolute(src) || !path_is_safe(src)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Source path must be absolute and not contain ../."); + + if (isempty(dest)) + dest = src; + else if (!path_is_absolute(dest) || !path_is_safe(dest)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Destination path must be absolute and not contain ../."); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + /* One day, when bind mounting /proc/self/fd/n works across + * namespace boundaries we should rework this logic to make + * use of it... */ + + p = strjoina("/run/systemd/nspawn/propagate/", m->name, "/"); + if (laccess(p, F_OK) < 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Container does not allow propagation of mount points."); + + /* Our goal is to install a new bind mount into the container, + possibly read-only. This is irritatingly complex + unfortunately, currently. + + First, we start by creating a private playground in /tmp, + that we can mount MS_SLAVE. (Which is necessary, since + MS_MOVE cannot be applied to mounts with MS_SHARED parent + mounts.) */ + + if (!mkdtemp(mount_slave)) + return sd_bus_error_set_errnof(error, errno, "Failed to create playground %s: %m", mount_slave); + + mount_slave_created = true; + + if (mount(mount_slave, mount_slave, NULL, MS_BIND, NULL) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to make bind mount %s: %m", mount_slave); + goto finish; + } + + mount_slave_mounted = true; + + if (mount(NULL, mount_slave, NULL, MS_SLAVE, NULL) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to remount slave %s: %m", mount_slave); + goto finish; + } + + /* Second, we mount the source directory to a directory inside + of our MS_SLAVE playground. */ + mount_tmp = strjoina(mount_slave, "/mount"); + if (mkdir(mount_tmp, 0700) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to create temporary mount point %s: %m", mount_tmp); + goto finish; + } + + mount_tmp_created = true; + + if (mount(src, mount_tmp, NULL, MS_BIND, NULL) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to overmount %s: %m", mount_tmp); + goto finish; + } + + mount_tmp_mounted = true; + + /* Third, we remount the new bind mount read-only if requested. */ + if (read_only) + if (mount(NULL, mount_tmp, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to remount read-only %s: %m", mount_tmp); + goto finish; + } + + /* Fourth, we move the new bind mount into the propagation + * directory. This way it will appear there read-only + * right-away. */ + + mount_outside = strjoina("/run/systemd/nspawn/propagate/", m->name, "/XXXXXX"); + if (!mkdtemp(mount_outside)) { + r = sd_bus_error_set_errnof(error, errno, "Cannot create propagation directory %s: %m", mount_outside); + goto finish; + } + + mount_outside_created = true; + + if (mount(mount_tmp, mount_outside, NULL, MS_MOVE, NULL) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to move %s to %s: %m", mount_tmp, mount_outside); + goto finish; + } + + mount_outside_mounted = true; + mount_tmp_mounted = false; + + (void) rmdir(mount_tmp); + mount_tmp_created = false; + + (void) umount(mount_slave); + mount_slave_mounted = false; + + (void) rmdir(mount_slave); + mount_slave_created = false; + + if (pipe2(errno_pipe_fd, O_CLOEXEC|O_NONBLOCK) < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to create pipe: %m"); + goto finish; + } + + child = fork(); + if (child < 0) { + r = sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + goto finish; + } + + if (child == 0) { + const char *mount_inside; + int mntfd; + const char *q; + + errno_pipe_fd[0] = safe_close(errno_pipe_fd[0]); + + q = procfs_file_alloca(m->leader, "ns/mnt"); + mntfd = open(q, O_RDONLY|O_NOCTTY|O_CLOEXEC); + if (mntfd < 0) { + r = log_error_errno(errno, "Failed to open mount namespace of leader: %m"); + goto child_fail; + } + + if (setns(mntfd, CLONE_NEWNS) < 0) { + r = log_error_errno(errno, "Failed to join namespace of leader: %m"); + goto child_fail; + } + + if (make_directory) + (void) mkdir_p(dest, 0755); + + /* Fifth, move the mount to the right place inside */ + mount_inside = strjoina("/run/systemd/nspawn/incoming/", basename(mount_outside)); + if (mount(mount_inside, dest, NULL, MS_MOVE, NULL) < 0) { + r = log_error_errno(errno, "Failed to mount: %m"); + goto child_fail; + } + + _exit(EXIT_SUCCESS); + + child_fail: + (void) write(errno_pipe_fd[1], &r, sizeof(r)); + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + _exit(EXIT_FAILURE); + } + + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + r = wait_for_terminate(child, &si); + if (r < 0) { + r = sd_bus_error_set_errnof(error, r, "Failed to wait for child: %m"); + goto finish; + } + if (si.si_code != CLD_EXITED) { + r = sd_bus_error_setf(error, SD_BUS_ERROR_FAILED, "Child died abnormally."); + goto finish; + } + if (si.si_status != EXIT_SUCCESS) { + + if (read(errno_pipe_fd[0], &r, sizeof(r)) == sizeof(r)) + r = sd_bus_error_set_errnof(error, r, "Failed to mount: %m"); + else + r = sd_bus_error_setf(error, SD_BUS_ERROR_FAILED, "Child failed."); + goto finish; + } + + r = sd_bus_reply_method_return(message, NULL); + +finish: + if (mount_outside_mounted) + umount(mount_outside); + if (mount_outside_created) + rmdir(mount_outside); + + if (mount_tmp_mounted) + umount(mount_tmp); + if (mount_tmp_created) + rmdir(mount_tmp); + + if (mount_slave_mounted) + umount(mount_slave); + if (mount_slave_created) + rmdir(mount_slave); + + return r; +} + +int bus_machine_method_copy(sd_bus_message *message, void *userdata, sd_bus_error *error) { + const char *src, *dest, *host_path, *container_path, *host_basename, *host_dirname, *container_basename, *container_dirname; + _cleanup_close_pair_ int errno_pipe_fd[2] = { -1, -1 }; + _cleanup_close_ int hostfd = -1; + Machine *m = userdata; + bool copy_from; + pid_t child; + char *t; + int r; + + assert(message); + assert(m); + + if (m->manager->n_operations >= OPERATIONS_MAX) + return sd_bus_error_setf(error, SD_BUS_ERROR_LIMITS_EXCEEDED, "Too many ongoing copies."); + + if (m->class != MACHINE_CONTAINER) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Copying files is only supported on container machines."); + + r = sd_bus_message_read(message, "ss", &src, &dest); + if (r < 0) + return r; + + if (!path_is_absolute(src)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Source path must be absolute."); + + if (isempty(dest)) + dest = src; + else if (!path_is_absolute(dest)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Destination path must be absolute."); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + copy_from = strstr(sd_bus_message_get_member(message), "CopyFrom"); + + if (copy_from) { + container_path = src; + host_path = dest; + } else { + host_path = src; + container_path = dest; + } + + host_basename = basename(host_path); + t = strdupa(host_path); + host_dirname = dirname(t); + + container_basename = basename(container_path); + t = strdupa(container_path); + container_dirname = dirname(t); + + hostfd = open(host_dirname, O_CLOEXEC|O_RDONLY|O_NOCTTY|O_DIRECTORY); + if (hostfd < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to open host directory %s: %m", host_dirname); + + if (pipe2(errno_pipe_fd, O_CLOEXEC|O_NONBLOCK) < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to create pipe: %m"); + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + + if (child == 0) { + int containerfd; + const char *q; + int mntfd; + + errno_pipe_fd[0] = safe_close(errno_pipe_fd[0]); + + q = procfs_file_alloca(m->leader, "ns/mnt"); + mntfd = open(q, O_RDONLY|O_NOCTTY|O_CLOEXEC); + if (mntfd < 0) { + r = log_error_errno(errno, "Failed to open mount namespace of leader: %m"); + goto child_fail; + } + + if (setns(mntfd, CLONE_NEWNS) < 0) { + r = log_error_errno(errno, "Failed to join namespace of leader: %m"); + goto child_fail; + } + + containerfd = open(container_dirname, O_CLOEXEC|O_RDONLY|O_NOCTTY|O_DIRECTORY); + if (containerfd < 0) { + r = log_error_errno(errno, "Failed top open destination directory: %m"); + goto child_fail; + } + + if (copy_from) + r = copy_tree_at(containerfd, container_basename, hostfd, host_basename, true); + else + r = copy_tree_at(hostfd, host_basename, containerfd, container_basename, true); + + hostfd = safe_close(hostfd); + containerfd = safe_close(containerfd); + + if (r < 0) { + r = log_error_errno(r, "Failed to copy tree: %m"); + goto child_fail; + } + + _exit(EXIT_SUCCESS); + + child_fail: + (void) write(errno_pipe_fd[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + /* Copying might take a while, hence install a watch on the child, and return */ + + r = operation_new(m->manager, m, child, message, errno_pipe_fd[0], NULL); + if (r < 0) { + (void) sigkill_wait(child); + return r; + } + errno_pipe_fd[0] = -1; + + return 1; +} + +int bus_machine_method_open_root_directory(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_close_ int fd = -1; + Machine *m = userdata; + int r; + + assert(message); + assert(m); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->manager->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + switch (m->class) { + + case MACHINE_HOST: + fd = open("/", O_RDONLY|O_CLOEXEC|O_DIRECTORY); + if (fd < 0) + return -errno; + + break; + + case MACHINE_CONTAINER: { + _cleanup_close_ int mntns_fd = -1, root_fd = -1; + _cleanup_close_pair_ int pair[2] = { -1, -1 }; + siginfo_t si; + pid_t child; + + r = namespace_open(m->leader, NULL, &mntns_fd, NULL, NULL, &root_fd); + if (r < 0) + return r; + + if (socketpair(AF_UNIX, SOCK_DGRAM, 0, pair) < 0) + return -errno; + + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + + if (child == 0) { + _cleanup_close_ int dfd = -1; + + pair[0] = safe_close(pair[0]); + + r = namespace_enter(-1, mntns_fd, -1, -1, root_fd); + if (r < 0) + _exit(EXIT_FAILURE); + + dfd = open("/", O_RDONLY|O_CLOEXEC|O_DIRECTORY); + if (dfd < 0) + _exit(EXIT_FAILURE); + + r = send_one_fd(pair[1], dfd, 0); + dfd = safe_close(dfd); + if (r < 0) + _exit(EXIT_FAILURE); + + _exit(EXIT_SUCCESS); + } + + pair[1] = safe_close(pair[1]); + + r = wait_for_terminate(child, &si); + if (r < 0) + return sd_bus_error_set_errnof(error, r, "Failed to wait for child: %m"); + if (si.si_code != CLD_EXITED || si.si_status != EXIT_SUCCESS) + return sd_bus_error_setf(error, SD_BUS_ERROR_FAILED, "Child died abnormally."); + + fd = receive_one_fd(pair[0], MSG_DONTWAIT); + if (fd < 0) + return fd; + + break; + } + + default: + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Opening the root directory is only supported on container machines."); + } + + return sd_bus_reply_method_return(message, "h", fd); +} + +const sd_bus_vtable machine_vtable[] = { + SD_BUS_VTABLE_START(0), + SD_BUS_PROPERTY("Name", "s", NULL, offsetof(Machine, name), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Id", "ay", bus_property_get_id128, offsetof(Machine, id), SD_BUS_VTABLE_PROPERTY_CONST), + BUS_PROPERTY_DUAL_TIMESTAMP("Timestamp", offsetof(Machine, timestamp), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Service", "s", NULL, offsetof(Machine, service), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Unit", "s", NULL, offsetof(Machine, unit), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Scope", "s", NULL, offsetof(Machine, unit), SD_BUS_VTABLE_PROPERTY_CONST|SD_BUS_VTABLE_HIDDEN), + SD_BUS_PROPERTY("Leader", "u", NULL, offsetof(Machine, leader), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("Class", "s", property_get_class, offsetof(Machine, class), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("RootDirectory", "s", NULL, offsetof(Machine, root_directory), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("NetworkInterfaces", "ai", property_get_netif, 0, SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("State", "s", property_get_state, 0, 0), + SD_BUS_METHOD("Terminate", NULL, NULL, bus_machine_method_terminate, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("Kill", "si", NULL, bus_machine_method_kill, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetAddresses", NULL, "a(iay)", bus_machine_method_get_addresses, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetOSRelease", NULL, "a{ss}", bus_machine_method_get_os_release, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenPTY", NULL, "hs", bus_machine_method_open_pty, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenLogin", NULL, "hs", bus_machine_method_open_login, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenShell", "ssasas", "hs", bus_machine_method_open_shell, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("BindMount", "ssbb", NULL, bus_machine_method_bind_mount, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CopyFrom", "ss", NULL, bus_machine_method_copy, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CopyTo", "ss", NULL, bus_machine_method_copy, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenRootDirectory", NULL, "h", bus_machine_method_open_root_directory, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_VTABLE_END +}; + +int machine_object_find(sd_bus *bus, const char *path, const char *interface, void *userdata, void **found, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + int r; + + assert(bus); + assert(path); + assert(interface); + assert(found); + assert(m); + + if (streq(path, "/org/freedesktop/machine1/machine/self")) { + _cleanup_(sd_bus_creds_unrefp) sd_bus_creds *creds = NULL; + sd_bus_message *message; + pid_t pid; + + message = sd_bus_get_current_message(bus); + if (!message) + return 0; + + r = sd_bus_query_sender_creds(message, SD_BUS_CREDS_PID, &creds); + if (r < 0) + return r; + + r = sd_bus_creds_get_pid(creds, &pid); + if (r < 0) + return r; + + r = manager_get_machine_by_pid(m, pid, &machine); + if (r <= 0) + return 0; + } else { + _cleanup_free_ char *e = NULL; + const char *p; + + p = startswith(path, "/org/freedesktop/machine1/machine/"); + if (!p) + return 0; + + e = bus_label_unescape(p); + if (!e) + return -ENOMEM; + + machine = hashmap_get(m->machines, e); + if (!machine) + return 0; + } + + *found = machine; + return 1; +} + +char *machine_bus_path(Machine *m) { + _cleanup_free_ char *e = NULL; + + assert(m); + + e = bus_label_escape(m->name); + if (!e) + return NULL; + + return strappend("/org/freedesktop/machine1/machine/", e); +} + +int machine_node_enumerator(sd_bus *bus, const char *path, void *userdata, char ***nodes, sd_bus_error *error) { + _cleanup_strv_free_ char **l = NULL; + Machine *machine = NULL; + Manager *m = userdata; + Iterator i; + int r; + + assert(bus); + assert(path); + assert(nodes); + + HASHMAP_FOREACH(machine, m->machines, i) { + char *p; + + p = machine_bus_path(machine); + if (!p) + return -ENOMEM; + + r = strv_consume(&l, p); + if (r < 0) + return r; + } + + *nodes = l; + l = NULL; + + return 1; +} + +int machine_send_signal(Machine *m, bool new_machine) { + _cleanup_free_ char *p = NULL; + + assert(m); + + p = machine_bus_path(m); + if (!p) + return -ENOMEM; + + return sd_bus_emit_signal( + m->manager->bus, + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + new_machine ? "MachineNew" : "MachineRemoved", + "so", m->name, p); +} + +int machine_send_create_reply(Machine *m, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *c = NULL; + _cleanup_free_ char *p = NULL; + + assert(m); + + if (!m->create_message) + return 0; + + c = m->create_message; + m->create_message = NULL; + + if (error) + return sd_bus_reply_method_error(c, error); + + /* Update the machine state file before we notify the client + * about the result. */ + machine_save(m); + + p = machine_bus_path(m); + if (!p) + return -ENOMEM; + + return sd_bus_reply_method_return(c, "o", p); +} diff --git a/src/grp-machine/libmachine-core/src/machine.c b/src/grp-machine/libmachine-core/src/machine.c new file mode 100644 index 0000000000..67a2afe061 --- /dev/null +++ b/src/grp-machine/libmachine-core/src/machine.c @@ -0,0 +1,628 @@ +/*** + This file is part of systemd. + + Copyright 2011 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <errno.h> +#include <string.h> +#include <unistd.h> + +#include <systemd/sd-messages.h> + +#include "machine-core/machine-dbus.h" +#include "machine-core/machine.h" +#include "sd-bus/bus-error.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/escape.h" +#include "systemd-basic/extract-word.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/formats-util.h" +#include "systemd-basic/hashmap.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/parse-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/special.h" +#include "systemd-basic/string-table.h" +#include "systemd-basic/terminal-util.h" +#include "systemd-basic/unit-name.h" +#include "systemd-basic/util.h" + +Machine* machine_new(Manager *manager, MachineClass class, const char *name) { + Machine *m; + + assert(manager); + assert(class < _MACHINE_CLASS_MAX); + assert(name); + + /* Passing class == _MACHINE_CLASS_INVALID here is fine. It + * means as much as "we don't know yet", and that we'll figure + * it out later when loading the state file. */ + + m = new0(Machine, 1); + if (!m) + return NULL; + + m->name = strdup(name); + if (!m->name) + goto fail; + + if (class != MACHINE_HOST) { + m->state_file = strappend("/run/systemd/machines/", m->name); + if (!m->state_file) + goto fail; + } + + m->class = class; + + if (hashmap_put(manager->machines, m->name, m) < 0) + goto fail; + + m->manager = manager; + + return m; + +fail: + free(m->state_file); + free(m->name); + return mfree(m); +} + +void machine_free(Machine *m) { + assert(m); + + while (m->operations) + operation_free(m->operations); + + if (m->in_gc_queue) + LIST_REMOVE(gc_queue, m->manager->machine_gc_queue, m); + + machine_release_unit(m); + + free(m->scope_job); + + (void) hashmap_remove(m->manager->machines, m->name); + + if (m->manager->host_machine == m) + m->manager->host_machine = NULL; + + if (m->leader > 0) + (void) hashmap_remove_value(m->manager->machine_leaders, PID_TO_PTR(m->leader), m); + + sd_bus_message_unref(m->create_message); + + free(m->name); + free(m->state_file); + free(m->service); + free(m->root_directory); + free(m->netif); + free(m); +} + +int machine_save(Machine *m) { + _cleanup_free_ char *temp_path = NULL; + _cleanup_fclose_ FILE *f = NULL; + int r; + + assert(m); + + if (!m->state_file) + return 0; + + if (!m->started) + return 0; + + r = mkdir_safe_label("/run/systemd/machines", 0755, 0, 0); + if (r < 0) + goto fail; + + r = fopen_temporary(m->state_file, &f, &temp_path); + if (r < 0) + goto fail; + + (void) fchmod(fileno(f), 0644); + + fprintf(f, + "# This is private data. Do not parse.\n" + "NAME=%s\n", + m->name); + + if (m->unit) { + _cleanup_free_ char *escaped; + + escaped = cescape(m->unit); + if (!escaped) { + r = -ENOMEM; + goto fail; + } + + fprintf(f, "SCOPE=%s\n", escaped); /* We continue to call this "SCOPE=" because it is internal only, and we want to stay compatible with old files */ + } + + if (m->scope_job) + fprintf(f, "SCOPE_JOB=%s\n", m->scope_job); + + if (m->service) { + _cleanup_free_ char *escaped; + + escaped = cescape(m->service); + if (!escaped) { + r = -ENOMEM; + goto fail; + } + fprintf(f, "SERVICE=%s\n", escaped); + } + + if (m->root_directory) { + _cleanup_free_ char *escaped; + + escaped = cescape(m->root_directory); + if (!escaped) { + r = -ENOMEM; + goto fail; + } + fprintf(f, "ROOT=%s\n", escaped); + } + + if (!sd_id128_is_null(m->id)) + fprintf(f, "ID=" SD_ID128_FORMAT_STR "\n", SD_ID128_FORMAT_VAL(m->id)); + + if (m->leader != 0) + fprintf(f, "LEADER="PID_FMT"\n", m->leader); + + if (m->class != _MACHINE_CLASS_INVALID) + fprintf(f, "CLASS=%s\n", machine_class_to_string(m->class)); + + if (dual_timestamp_is_set(&m->timestamp)) + fprintf(f, + "REALTIME="USEC_FMT"\n" + "MONOTONIC="USEC_FMT"\n", + m->timestamp.realtime, + m->timestamp.monotonic); + + if (m->n_netif > 0) { + unsigned i; + + fputs("NETIF=", f); + + for (i = 0; i < m->n_netif; i++) { + if (i != 0) + fputc(' ', f); + + fprintf(f, "%i", m->netif[i]); + } + + fputc('\n', f); + } + + r = fflush_and_check(f); + if (r < 0) + goto fail; + + if (rename(temp_path, m->state_file) < 0) { + r = -errno; + goto fail; + } + + if (m->unit) { + char *sl; + + /* Create a symlink from the unit name to the machine + * name, so that we can quickly find the machine for + * each given unit. Ignore error. */ + sl = strjoina("/run/systemd/machines/unit:", m->unit); + (void) symlink(m->name, sl); + } + + return 0; + +fail: + (void) unlink(m->state_file); + + if (temp_path) + (void) unlink(temp_path); + + return log_error_errno(r, "Failed to save machine data %s: %m", m->state_file); +} + +static void machine_unlink(Machine *m) { + assert(m); + + if (m->unit) { + + char *sl; + + sl = strjoina("/run/systemd/machines/unit:", m->unit); + (void) unlink(sl); + } + + if (m->state_file) + (void) unlink(m->state_file); +} + +int machine_load(Machine *m) { + _cleanup_free_ char *realtime = NULL, *monotonic = NULL, *id = NULL, *leader = NULL, *class = NULL, *netif = NULL; + int r; + + assert(m); + + if (!m->state_file) + return 0; + + r = parse_env_file(m->state_file, NEWLINE, + "SCOPE", &m->unit, + "SCOPE_JOB", &m->scope_job, + "SERVICE", &m->service, + "ROOT", &m->root_directory, + "ID", &id, + "LEADER", &leader, + "CLASS", &class, + "REALTIME", &realtime, + "MONOTONIC", &monotonic, + "NETIF", &netif, + NULL); + if (r < 0) { + if (r == -ENOENT) + return 0; + + return log_error_errno(r, "Failed to read %s: %m", m->state_file); + } + + if (id) + sd_id128_from_string(id, &m->id); + + if (leader) + parse_pid(leader, &m->leader); + + if (class) { + MachineClass c; + + c = machine_class_from_string(class); + if (c >= 0) + m->class = c; + } + + if (realtime) + timestamp_deserialize(realtime, &m->timestamp.realtime); + if (monotonic) + timestamp_deserialize(monotonic, &m->timestamp.monotonic); + + if (netif) { + size_t allocated = 0, nr = 0; + const char *p; + int *ni = NULL; + + p = netif; + for (;;) { + _cleanup_free_ char *word = NULL; + int ifi; + + r = extract_first_word(&p, &word, NULL, 0); + if (r == 0) + break; + if (r == -ENOMEM) + return log_oom(); + if (r < 0) { + log_warning_errno(r, "Failed to parse NETIF: %s", netif); + break; + } + + if (parse_ifindex(word, &ifi) < 0) + continue; + + if (!GREEDY_REALLOC(ni, allocated, nr+1)) { + free(ni); + return log_oom(); + } + + ni[nr++] = ifi; + } + + free(m->netif); + m->netif = ni; + m->n_netif = nr; + } + + return r; +} + +static int machine_start_scope(Machine *m, sd_bus_message *properties, sd_bus_error *error) { + int r = 0; + + assert(m); + assert(m->class != MACHINE_HOST); + + if (!m->unit) { + _cleanup_free_ char *escaped = NULL; + char *scope, *description, *job = NULL; + + escaped = unit_name_escape(m->name); + if (!escaped) + return log_oom(); + + scope = strjoin("machine-", escaped, ".scope", NULL); + if (!scope) + return log_oom(); + + description = strjoina(m->class == MACHINE_VM ? "Virtual Machine " : "Container ", m->name); + + r = manager_start_scope(m->manager, scope, m->leader, SPECIAL_MACHINE_SLICE, description, properties, error, &job); + if (r < 0) { + log_error("Failed to start machine scope: %s", bus_error_message(error, r)); + free(scope); + return r; + } else { + m->unit = scope; + + free(m->scope_job); + m->scope_job = job; + } + } + + if (m->unit) + hashmap_put(m->manager->machine_units, m->unit, m); + + return r; +} + +int machine_start(Machine *m, sd_bus_message *properties, sd_bus_error *error) { + int r; + + assert(m); + + if (!IN_SET(m->class, MACHINE_CONTAINER, MACHINE_VM)) + return -EOPNOTSUPP; + + if (m->started) + return 0; + + r = hashmap_put(m->manager->machine_leaders, PID_TO_PTR(m->leader), m); + if (r < 0) + return r; + + /* Create cgroup */ + r = machine_start_scope(m, properties, error); + if (r < 0) + return r; + + log_struct(LOG_INFO, + LOG_MESSAGE_ID(SD_MESSAGE_MACHINE_START), + "NAME=%s", m->name, + "LEADER="PID_FMT, m->leader, + LOG_MESSAGE("New machine %s.", m->name), + NULL); + + if (!dual_timestamp_is_set(&m->timestamp)) + dual_timestamp_get(&m->timestamp); + + m->started = true; + + /* Save new machine data */ + machine_save(m); + + machine_send_signal(m, true); + + return 0; +} + +static int machine_stop_scope(Machine *m) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + char *job = NULL; + int r; + + assert(m); + assert(m->class != MACHINE_HOST); + + if (!m->unit) + return 0; + + r = manager_stop_unit(m->manager, m->unit, &error, &job); + if (r < 0) { + log_error("Failed to stop machine scope: %s", bus_error_message(&error, r)); + return r; + } + + free(m->scope_job); + m->scope_job = job; + + return 0; +} + +int machine_stop(Machine *m) { + int r; + assert(m); + + if (!IN_SET(m->class, MACHINE_CONTAINER, MACHINE_VM)) + return -EOPNOTSUPP; + + r = machine_stop_scope(m); + + m->stopping = true; + + machine_save(m); + + return r; +} + +int machine_finalize(Machine *m) { + assert(m); + + if (m->started) + log_struct(LOG_INFO, + LOG_MESSAGE_ID(SD_MESSAGE_MACHINE_STOP), + "NAME=%s", m->name, + "LEADER="PID_FMT, m->leader, + LOG_MESSAGE("Machine %s terminated.", m->name), + NULL); + + machine_unlink(m); + machine_add_to_gc_queue(m); + + if (m->started) { + machine_send_signal(m, false); + m->started = false; + } + + return 0; +} + +bool machine_check_gc(Machine *m, bool drop_not_started) { + assert(m); + + if (m->class == MACHINE_HOST) + return true; + + if (drop_not_started && !m->started) + return false; + + if (m->scope_job && manager_job_is_active(m->manager, m->scope_job)) + return true; + + if (m->unit && manager_unit_is_active(m->manager, m->unit)) + return true; + + return false; +} + +void machine_add_to_gc_queue(Machine *m) { + assert(m); + + if (m->in_gc_queue) + return; + + LIST_PREPEND(gc_queue, m->manager->machine_gc_queue, m); + m->in_gc_queue = true; +} + +MachineState machine_get_state(Machine *s) { + assert(s); + + if (s->class == MACHINE_HOST) + return MACHINE_RUNNING; + + if (s->stopping) + return MACHINE_CLOSING; + + if (s->scope_job) + return MACHINE_OPENING; + + return MACHINE_RUNNING; +} + +int machine_kill(Machine *m, KillWho who, int signo) { + assert(m); + + if (!IN_SET(m->class, MACHINE_VM, MACHINE_CONTAINER)) + return -EOPNOTSUPP; + + if (!m->unit) + return -ESRCH; + + if (who == KILL_LEADER) { + /* If we shall simply kill the leader, do so directly */ + + if (kill(m->leader, signo) < 0) + return -errno; + + return 0; + } + + /* Otherwise, make PID 1 do it for us, for the entire cgroup */ + return manager_kill_unit(m->manager, m->unit, signo, NULL); +} + +int machine_openpt(Machine *m, int flags) { + assert(m); + + switch (m->class) { + + case MACHINE_HOST: { + int fd; + + fd = posix_openpt(flags); + if (fd < 0) + return -errno; + + if (unlockpt(fd) < 0) + return -errno; + + return fd; + } + + case MACHINE_CONTAINER: + if (m->leader <= 0) + return -EINVAL; + + return openpt_in_namespace(m->leader, flags); + + default: + return -EOPNOTSUPP; + } +} + +int machine_open_terminal(Machine *m, const char *path, int mode) { + assert(m); + + switch (m->class) { + + case MACHINE_HOST: + return open_terminal(path, mode); + + case MACHINE_CONTAINER: + if (m->leader <= 0) + return -EINVAL; + + return open_terminal_in_namespace(m->leader, path, mode); + + default: + return -EOPNOTSUPP; + } +} + +void machine_release_unit(Machine *m) { + assert(m); + + if (!m->unit) + return; + + (void) hashmap_remove(m->manager->machine_units, m->unit); + m->unit = mfree(m->unit); +} + +static const char* const machine_class_table[_MACHINE_CLASS_MAX] = { + [MACHINE_CONTAINER] = "container", + [MACHINE_VM] = "vm", + [MACHINE_HOST] = "host", +}; + +DEFINE_STRING_TABLE_LOOKUP(machine_class, MachineClass); + +static const char* const machine_state_table[_MACHINE_STATE_MAX] = { + [MACHINE_OPENING] = "opening", + [MACHINE_RUNNING] = "running", + [MACHINE_CLOSING] = "closing" +}; + +DEFINE_STRING_TABLE_LOOKUP(machine_state, MachineState); + +static const char* const kill_who_table[_KILL_WHO_MAX] = { + [KILL_LEADER] = "leader", + [KILL_ALL] = "all" +}; + +DEFINE_STRING_TABLE_LOOKUP(kill_who, KillWho); diff --git a/src/grp-machine/libmachine-core/src/machined-dbus.c b/src/grp-machine/libmachine-core/src/machined-dbus.c new file mode 100644 index 0000000000..bc1bbd4891 --- /dev/null +++ b/src/grp-machine/libmachine-core/src/machined-dbus.c @@ -0,0 +1,1806 @@ +/*** + This file is part of systemd. + + Copyright 2011 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <errno.h> +#include <string.h> +#include <unistd.h> + +#include <systemd/sd-id128.h> + +#include "machine-core/image-dbus.h" +#include "machine-core/machine-dbus.h" +#include "machine-core/machined.h" +#include "sd-bus/bus-common-errors.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/btrfs-util.h" +#include "systemd-basic/cgroup-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/fileio.h" +#include "systemd-basic/formats-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/io-util.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/stdio-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/unit-name.h" +#include "systemd-basic/user-util.h" +#include "systemd-shared/machine-image.h" +#include "systemd-shared/machine-pool.h" + +static int property_get_pool_path( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + assert(bus); + assert(reply); + + return sd_bus_message_append(reply, "s", "/var/lib/machines"); +} + +static int property_get_pool_usage( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + _cleanup_close_ int fd = -1; + uint64_t usage = (uint64_t) -1; + struct stat st; + + assert(bus); + assert(reply); + + /* We try to read the quota info from /var/lib/machines, as + * well as the usage of the loopback file + * /var/lib/machines.raw, and pick the larger value. */ + + fd = open("/var/lib/machines", O_RDONLY|O_CLOEXEC|O_DIRECTORY); + if (fd >= 0) { + BtrfsQuotaInfo q; + + if (btrfs_subvol_get_subtree_quota_fd(fd, 0, &q) >= 0) + usage = q.referenced; + } + + if (stat("/var/lib/machines.raw", &st) >= 0) { + if (usage == (uint64_t) -1 || st.st_blocks * 512ULL > usage) + usage = st.st_blocks * 512ULL; + } + + return sd_bus_message_append(reply, "t", usage); +} + +static int property_get_pool_limit( + sd_bus *bus, + const char *path, + const char *interface, + const char *property, + sd_bus_message *reply, + void *userdata, + sd_bus_error *error) { + + _cleanup_close_ int fd = -1; + uint64_t size = (uint64_t) -1; + struct stat st; + + assert(bus); + assert(reply); + + /* We try to read the quota limit from /var/lib/machines, as + * well as the size of the loopback file + * /var/lib/machines.raw, and pick the smaller value. */ + + fd = open("/var/lib/machines", O_RDONLY|O_CLOEXEC|O_DIRECTORY); + if (fd >= 0) { + BtrfsQuotaInfo q; + + if (btrfs_subvol_get_subtree_quota_fd(fd, 0, &q) >= 0) + size = q.referenced_max; + } + + if (stat("/var/lib/machines.raw", &st) >= 0) { + if (size == (uint64_t) -1 || (uint64_t) st.st_size < size) + size = st.st_size; + } + + return sd_bus_message_append(reply, "t", size); +} + +static int method_get_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_free_ char *p = NULL; + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + p = machine_bus_path(machine); + if (!p) + return -ENOMEM; + + return sd_bus_reply_method_return(message, "o", p); +} + +static int method_get_image(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_free_ char *p = NULL; + Manager *m = userdata; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + r = image_find(name, NULL); + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", name); + if (r < 0) + return r; + + p = image_bus_path(name); + if (!p) + return -ENOMEM; + + return sd_bus_reply_method_return(message, "o", p); +} + +static int method_get_machine_by_pid(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_free_ char *p = NULL; + Manager *m = userdata; + Machine *machine = NULL; + pid_t pid; + int r; + + assert(message); + assert(m); + + assert_cc(sizeof(pid_t) == sizeof(uint32_t)); + + r = sd_bus_message_read(message, "u", &pid); + if (r < 0) + return r; + + if (pid < 0) + return -EINVAL; + + if (pid == 0) { + _cleanup_(sd_bus_creds_unrefp) sd_bus_creds *creds = NULL; + + r = sd_bus_query_sender_creds(message, SD_BUS_CREDS_PID, &creds); + if (r < 0) + return r; + + r = sd_bus_creds_get_pid(creds, &pid); + if (r < 0) + return r; + } + + r = manager_get_machine_by_pid(m, pid, &machine); + if (r < 0) + return r; + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_MACHINE_FOR_PID, "PID "PID_FMT" does not belong to any known machine", pid); + + p = machine_bus_path(machine); + if (!p) + return -ENOMEM; + + return sd_bus_reply_method_return(message, "o", p); +} + +static int method_list_machines(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + Manager *m = userdata; + Machine *machine; + Iterator i; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + r = sd_bus_message_open_container(reply, 'a', "(ssso)"); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + HASHMAP_FOREACH(machine, m->machines, i) { + _cleanup_free_ char *p = NULL; + + p = machine_bus_path(machine); + if (!p) + return -ENOMEM; + + r = sd_bus_message_append(reply, "(ssso)", + machine->name, + strempty(machine_class_to_string(machine->class)), + machine->service, + p); + if (r < 0) + return sd_bus_error_set_errno(error, r); + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + return sd_bus_send(NULL, reply, NULL); +} + +static int method_create_or_register_machine(Manager *manager, sd_bus_message *message, bool read_network, Machine **_m, sd_bus_error *error) { + const char *name, *service, *class, *root_directory; + const int32_t *netif = NULL; + MachineClass c; + uint32_t leader; + sd_id128_t id; + const void *v; + Machine *m; + size_t n, n_netif = 0; + int r; + + assert(manager); + assert(message); + assert(_m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + if (!machine_name_is_valid(name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid machine name"); + + r = sd_bus_message_read_array(message, 'y', &v, &n); + if (r < 0) + return r; + if (n == 0) + id = SD_ID128_NULL; + else if (n == 16) + memcpy(&id, v, n); + else + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid machine ID parameter"); + + r = sd_bus_message_read(message, "ssus", &service, &class, &leader, &root_directory); + if (r < 0) + return r; + + if (read_network) { + size_t i; + + r = sd_bus_message_read_array(message, 'i', (const void**) &netif, &n_netif); + if (r < 0) + return r; + + n_netif /= sizeof(int32_t); + + for (i = 0; i < n_netif; i++) { + if (netif[i] <= 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid network interface index %i", netif[i]); + } + } + + if (isempty(class)) + c = _MACHINE_CLASS_INVALID; + else { + c = machine_class_from_string(class); + if (c < 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid machine class parameter"); + } + + if (leader == 1) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid leader PID"); + + if (!isempty(root_directory) && !path_is_absolute(root_directory)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Root directory must be empty or an absolute path"); + + if (leader == 0) { + _cleanup_(sd_bus_creds_unrefp) sd_bus_creds *creds = NULL; + + r = sd_bus_query_sender_creds(message, SD_BUS_CREDS_PID, &creds); + if (r < 0) + return r; + + assert_cc(sizeof(uint32_t) == sizeof(pid_t)); + + r = sd_bus_creds_get_pid(creds, (pid_t*) &leader); + if (r < 0) + return r; + } + + if (hashmap_get(manager->machines, name)) + return sd_bus_error_setf(error, BUS_ERROR_MACHINE_EXISTS, "Machine '%s' already exists", name); + + r = manager_add_machine(manager, name, &m); + if (r < 0) + return r; + + m->leader = leader; + m->class = c; + m->id = id; + + if (!isempty(service)) { + m->service = strdup(service); + if (!m->service) { + r = -ENOMEM; + goto fail; + } + } + + if (!isempty(root_directory)) { + m->root_directory = strdup(root_directory); + if (!m->root_directory) { + r = -ENOMEM; + goto fail; + } + } + + if (n_netif > 0) { + assert_cc(sizeof(int32_t) == sizeof(int)); + m->netif = memdup(netif, sizeof(int32_t) * n_netif); + if (!m->netif) { + r = -ENOMEM; + goto fail; + } + + m->n_netif = n_netif; + } + + *_m = m; + + return 1; + +fail: + machine_add_to_gc_queue(m); + return r; +} + +static int method_create_machine_internal(sd_bus_message *message, bool read_network, void *userdata, sd_bus_error *error) { + Manager *manager = userdata; + Machine *m = NULL; + int r; + + assert(message); + assert(manager); + + r = method_create_or_register_machine(manager, message, read_network, &m, error); + if (r < 0) + return r; + + r = sd_bus_message_enter_container(message, 'a', "(sv)"); + if (r < 0) + goto fail; + + r = machine_start(m, message, error); + if (r < 0) + goto fail; + + m->create_message = sd_bus_message_ref(message); + return 1; + +fail: + machine_add_to_gc_queue(m); + return r; +} + +static int method_create_machine_with_network(sd_bus_message *message, void *userdata, sd_bus_error *error) { + return method_create_machine_internal(message, true, userdata, error); +} + +static int method_create_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + return method_create_machine_internal(message, false, userdata, error); +} + +static int method_register_machine_internal(sd_bus_message *message, bool read_network, void *userdata, sd_bus_error *error) { + Manager *manager = userdata; + _cleanup_free_ char *p = NULL; + Machine *m = NULL; + int r; + + assert(message); + assert(manager); + + r = method_create_or_register_machine(manager, message, read_network, &m, error); + if (r < 0) + return r; + + r = cg_pid_get_unit(m->leader, &m->unit); + if (r < 0) { + r = sd_bus_error_set_errnof(error, r, + "Failed to determine unit of process "PID_FMT" : %m", + m->leader); + goto fail; + } + + r = machine_start(m, NULL, error); + if (r < 0) + goto fail; + + p = machine_bus_path(m); + if (!p) { + r = -ENOMEM; + goto fail; + } + + return sd_bus_reply_method_return(message, "o", p); + +fail: + machine_add_to_gc_queue(m); + return r; +} + +static int method_register_machine_with_network(sd_bus_message *message, void *userdata, sd_bus_error *error) { + return method_register_machine_internal(message, true, userdata, error); +} + +static int method_register_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + return method_register_machine_internal(message, false, userdata, error); +} + +static int method_terminate_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_terminate(message, machine, error); +} + +static int method_kill_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_kill(message, machine, error); +} + +static int method_get_machine_addresses(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_get_addresses(message, machine, error); +} + +static int method_get_machine_os_release(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_get_os_release(message, machine, error); +} + +static int method_list_images(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_(image_hashmap_freep) Hashmap *images = NULL; + Manager *m = userdata; + Image *image; + Iterator i; + int r; + + assert(message); + assert(m); + + images = hashmap_new(&string_hash_ops); + if (!images) + return -ENOMEM; + + r = image_discover(images); + if (r < 0) + return r; + + r = sd_bus_message_new_method_return(message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_open_container(reply, 'a', "(ssbttto)"); + if (r < 0) + return r; + + HASHMAP_FOREACH(image, images, i) { + _cleanup_free_ char *p = NULL; + + p = image_bus_path(image->name); + if (!p) + return -ENOMEM; + + r = sd_bus_message_append(reply, "(ssbttto)", + image->name, + image_type_to_string(image->type), + image->read_only, + image->crtime, + image->mtime, + image->usage, + p); + if (r < 0) + return r; + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +static int method_open_machine_pty(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return sd_bus_error_set_errno(error, r); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_open_pty(message, machine, error); +} + +static int method_open_machine_login(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_open_login(message, machine, error); +} + +static int method_open_machine_shell(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_open_shell(message, machine, error); +} + +static int method_bind_mount_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_bind_mount(message, machine, error); +} + +static int method_copy_machine(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_copy(message, machine, error); +} + +static int method_open_machine_root_directory(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + const char *name; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + return bus_machine_method_open_root_directory(message, machine, error); +} + +static int method_remove_image(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(image_unrefp) Image* i = NULL; + const char *name; + int r; + + assert(message); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + if (!image_name_is_valid(name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", name); + + r = image_find(name, &i); + if (r < 0) + return r; + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", name); + + i->userdata = userdata; + return bus_image_method_remove(message, i, error); +} + +static int method_rename_image(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(image_unrefp) Image* i = NULL; + const char *old_name; + int r; + + assert(message); + + r = sd_bus_message_read(message, "s", &old_name); + if (r < 0) + return r; + + if (!image_name_is_valid(old_name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", old_name); + + r = image_find(old_name, &i); + if (r < 0) + return r; + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", old_name); + + i->userdata = userdata; + return bus_image_method_rename(message, i, error); +} + +static int method_clone_image(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(image_unrefp) Image *i = NULL; + const char *old_name; + int r; + + assert(message); + + r = sd_bus_message_read(message, "s", &old_name); + if (r < 0) + return r; + + if (!image_name_is_valid(old_name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", old_name); + + r = image_find(old_name, &i); + if (r < 0) + return r; + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", old_name); + + i->userdata = userdata; + return bus_image_method_clone(message, i, error); +} + +static int method_mark_image_read_only(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(image_unrefp) Image *i = NULL; + const char *name; + int r; + + assert(message); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + if (!image_name_is_valid(name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", name); + + r = image_find(name, &i); + if (r < 0) + return r; + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", name); + + i->userdata = userdata; + return bus_image_method_mark_read_only(message, i, error); +} + +static int clean_pool_done(Operation *operation, int ret, sd_bus_error *error) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_fclose_ FILE *f = NULL; + bool success; + size_t n; + int r; + + assert(operation); + assert(operation->extra_fd >= 0); + + if (lseek(operation->extra_fd, 0, SEEK_SET) == (off_t) -1) + return -errno; + + f = fdopen(operation->extra_fd, "re"); + if (!f) + return -errno; + + operation->extra_fd = -1; + + /* The resulting temporary file starts with a boolean value that indicates success or not. */ + errno = 0; + n = fread(&success, 1, sizeof(success), f); + if (n != sizeof(success)) + return ret < 0 ? ret : (errno != 0 ? -errno : -EIO); + + if (ret < 0) { + _cleanup_free_ char *name = NULL; + + /* The clean-up operation failed. In this case the resulting temporary file should contain a boolean + * set to false followed by the name of the failed image. Let's try to read this and use it for the + * error message. If we can't read it, don't mind, and return the naked error. */ + + if (success) /* The resulting temporary file could not be updated, ignore it. */ + return ret; + + r = read_nul_string(f, &name); + if (r < 0 || isempty(name)) /* Same here... */ + return ret; + + return sd_bus_error_set_errnof(error, ret, "Failed to remove image %s: %m", name); + } + + assert(success); + + r = sd_bus_message_new_method_return(operation->message, &reply); + if (r < 0) + return r; + + r = sd_bus_message_open_container(reply, 'a', "(st)"); + if (r < 0) + return r; + + /* On success the resulting temporary file will contain a list of image names that were removed followed by + * their size on disk. Let's read that and turn it into a bus message. */ + for (;;) { + _cleanup_free_ char *name = NULL; + uint64_t size; + + r = read_nul_string(f, &name); + if (r < 0) + return r; + if (isempty(name)) /* reached the end */ + break; + + errno = 0; + n = fread(&size, 1, sizeof(size), f); + if (n != sizeof(size)) + return errno != 0 ? -errno : -EIO; + + r = sd_bus_message_append(reply, "(st)", name, size); + if (r < 0) + return r; + } + + r = sd_bus_message_close_container(reply); + if (r < 0) + return r; + + return sd_bus_send(NULL, reply, NULL); +} + +static int method_clean_pool(sd_bus_message *message, void *userdata, sd_bus_error *error) { + enum { + REMOVE_ALL, + REMOVE_HIDDEN, + } mode; + + _cleanup_close_pair_ int errno_pipe_fd[2] = { -1, -1 }; + _cleanup_close_ int result_fd = -1; + Manager *m = userdata; + Operation *operation; + const char *mm; + pid_t child; + int r; + + assert(message); + + if (m->n_operations >= OPERATIONS_MAX) + return sd_bus_error_setf(error, SD_BUS_ERROR_LIMITS_EXCEEDED, "Too many ongoing operations."); + + r = sd_bus_message_read(message, "s", &mm); + if (r < 0) + return r; + + if (streq(mm, "all")) + mode = REMOVE_ALL; + else if (streq(mm, "hidden")) + mode = REMOVE_HIDDEN; + else + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Unknown mode '%s'.", mm); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + if (pipe2(errno_pipe_fd, O_CLOEXEC|O_NONBLOCK) < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to create pipe: %m"); + + /* Create a temporary file we can dump information about deleted images into. We use a temporary file for this + * instead of a pipe or so, since this might grow quit large in theory and we don't want to process this + * continuously */ + result_fd = open_tmpfile_unlinkable(NULL, O_RDWR|O_CLOEXEC); + if (result_fd < 0) + return -errno; + + /* This might be a slow operation, run it asynchronously in a background process */ + child = fork(); + if (child < 0) + return sd_bus_error_set_errnof(error, errno, "Failed to fork(): %m"); + + if (child == 0) { + _cleanup_(image_hashmap_freep) Hashmap *images = NULL; + bool success = true; + Image *image; + Iterator i; + ssize_t l; + + errno_pipe_fd[0] = safe_close(errno_pipe_fd[0]); + + images = hashmap_new(&string_hash_ops); + if (!images) { + r = -ENOMEM; + goto child_fail; + } + + r = image_discover(images); + if (r < 0) + goto child_fail; + + l = write(result_fd, &success, sizeof(success)); + if (l < 0) { + r = -errno; + goto child_fail; + } + + HASHMAP_FOREACH(image, images, i) { + + /* We can't remove vendor images (i.e. those in /usr) */ + if (IMAGE_IS_VENDOR(image)) + continue; + + if (IMAGE_IS_HOST(image)) + continue; + + if (mode == REMOVE_HIDDEN && !IMAGE_IS_HIDDEN(image)) + continue; + + r = image_remove(image); + if (r == -EBUSY) /* keep images that are currently being used. */ + continue; + if (r < 0) { + /* If the operation failed, let's override everything we wrote, and instead write there at which image we failed. */ + success = false; + (void) ftruncate(result_fd, 0); + (void) lseek(result_fd, 0, SEEK_SET); + (void) write(result_fd, &success, sizeof(success)); + (void) write(result_fd, image->name, strlen(image->name)+1); + goto child_fail; + } + + l = write(result_fd, image->name, strlen(image->name)+1); + if (l < 0) { + r = -errno; + goto child_fail; + } + + l = write(result_fd, &image->usage_exclusive, sizeof(image->usage_exclusive)); + if (l < 0) { + r = -errno; + goto child_fail; + } + } + + result_fd = safe_close(result_fd); + _exit(EXIT_SUCCESS); + + child_fail: + (void) write(errno_pipe_fd[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + errno_pipe_fd[1] = safe_close(errno_pipe_fd[1]); + + /* The clean-up might take a while, hence install a watch on the child and return */ + + r = operation_new(m, NULL, child, message, errno_pipe_fd[0], &operation); + if (r < 0) { + (void) sigkill_wait(child); + return r; + } + + operation->extra_fd = result_fd; + operation->done = clean_pool_done; + + result_fd = -1; + errno_pipe_fd[0] = -1; + + return 1; +} + +static int method_set_pool_limit(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + uint64_t limit; + int r; + + assert(message); + + r = sd_bus_message_read(message, "t", &limit); + if (r < 0) + return r; + if (!FILE_SIZE_VALID_OR_INFINITY(limit)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "New limit out of range"); + + r = bus_verify_polkit_async( + message, + CAP_SYS_ADMIN, + "org.freedesktop.machine1.manage-machines", + NULL, + false, + UID_INVALID, + &m->polkit_registry, + error); + if (r < 0) + return r; + if (r == 0) + return 1; /* Will call us back */ + + /* Set up the machine directory if necessary */ + r = setup_machine_directory(limit, error); + if (r < 0) + return r; + + /* Resize the backing loopback device, if there is one, except if we asked to drop any limit */ + if (limit != (uint64_t) -1) { + r = btrfs_resize_loopback("/var/lib/machines", limit, false); + if (r == -ENOTTY) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Quota is only supported on btrfs."); + if (r < 0 && r != -ENODEV) /* ignore ENODEV, as that's what is returned if the file system is not on loopback */ + return sd_bus_error_set_errnof(error, r, "Failed to adjust loopback limit: %m"); + } + + (void) btrfs_qgroup_set_limit("/var/lib/machines", 0, limit); + + r = btrfs_subvol_set_subtree_quota_limit("/var/lib/machines", 0, limit); + if (r == -ENOTTY) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Quota is only supported on btrfs."); + if (r < 0) + return sd_bus_error_set_errnof(error, r, "Failed to adjust quota limit: %m"); + + return sd_bus_reply_method_return(message, NULL); +} + +static int method_set_image_limit(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_(image_unrefp) Image *i = NULL; + const char *name; + int r; + + assert(message); + + r = sd_bus_message_read(message, "s", &name); + if (r < 0) + return r; + + if (!image_name_is_valid(name)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Image name '%s' is invalid.", name); + + r = image_find(name, &i); + if (r < 0) + return r; + if (r == 0) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_IMAGE, "No image '%s' known", name); + + i->userdata = userdata; + return bus_image_method_set_limit(message, i, error); +} + +static int method_map_from_machine_user(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_fclose_ FILE *f = NULL; + Manager *m = userdata; + const char *name, *p; + Machine *machine; + uint32_t uid; + int r; + + r = sd_bus_message_read(message, "su", &name, &uid); + if (r < 0) + return r; + + if (!uid_is_valid(uid)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid user ID " UID_FMT, uid); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + if (machine->class != MACHINE_CONTAINER) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Not supported for non-container machines."); + + p = procfs_file_alloca(machine->leader, "uid_map"); + f = fopen(p, "re"); + if (!f) + return -errno; + + for (;;) { + uid_t uid_base, uid_shift, uid_range, converted; + int k; + + errno = 0; + k = fscanf(f, UID_FMT " " UID_FMT " " UID_FMT, &uid_base, &uid_shift, &uid_range); + if (k < 0 && feof(f)) + break; + if (k != 3) { + if (ferror(f) && errno > 0) + return -errno; + + return -EIO; + } + + if (uid < uid_base || uid >= uid_base + uid_range) + continue; + + converted = uid - uid_base + uid_shift; + if (!uid_is_valid(converted)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid user ID " UID_FMT, uid); + + return sd_bus_reply_method_return(message, "u", (uint32_t) converted); + } + + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_USER_MAPPING, "Machine '%s' has no matching user mappings.", name); +} + +static int method_map_to_machine_user(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + uid_t uid; + Iterator i; + int r; + + r = sd_bus_message_read(message, "u", &uid); + if (r < 0) + return r; + if (!uid_is_valid(uid)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid user ID " UID_FMT, uid); + if (uid < 0x10000) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_USER_MAPPING, "User " UID_FMT " belongs to host UID range", uid); + + HASHMAP_FOREACH(machine, m->machines, i) { + _cleanup_fclose_ FILE *f = NULL; + char p[strlen("/proc//uid_map") + DECIMAL_STR_MAX(pid_t) + 1]; + + if (machine->class != MACHINE_CONTAINER) + continue; + + xsprintf(p, "/proc/" UID_FMT "/uid_map", machine->leader); + f = fopen(p, "re"); + if (!f) { + log_warning_errno(errno, "Failed top open %s, ignoring,", p); + continue; + } + + for (;;) { + _cleanup_free_ char *o = NULL; + uid_t uid_base, uid_shift, uid_range, converted; + int k; + + errno = 0; + k = fscanf(f, UID_FMT " " UID_FMT " " UID_FMT, &uid_base, &uid_shift, &uid_range); + if (k < 0 && feof(f)) + break; + if (k != 3) { + if (ferror(f) && errno > 0) + return -errno; + + return -EIO; + } + + if (uid < uid_shift || uid >= uid_shift + uid_range) + continue; + + converted = (uid - uid_shift + uid_base); + if (!uid_is_valid(converted)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid user ID " UID_FMT, uid); + + o = machine_bus_path(machine); + if (!o) + return -ENOMEM; + + return sd_bus_reply_method_return(message, "sou", machine->name, o, (uint32_t) converted); + } + } + + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_USER_MAPPING, "No matching user mapping for " UID_FMT ".", uid); +} + +static int method_map_from_machine_group(sd_bus_message *message, void *groupdata, sd_bus_error *error) { + _cleanup_fclose_ FILE *f = NULL; + Manager *m = groupdata; + const char *name, *p; + Machine *machine; + uint32_t gid; + int r; + + r = sd_bus_message_read(message, "su", &name, &gid); + if (r < 0) + return r; + + if (!gid_is_valid(gid)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid group ID " GID_FMT, gid); + + machine = hashmap_get(m->machines, name); + if (!machine) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_MACHINE, "No machine '%s' known", name); + + if (machine->class != MACHINE_CONTAINER) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Not supported for non-container machines."); + + p = procfs_file_alloca(machine->leader, "gid_map"); + f = fopen(p, "re"); + if (!f) + return -errno; + + for (;;) { + gid_t gid_base, gid_shift, gid_range, converted; + int k; + + errno = 0; + k = fscanf(f, GID_FMT " " GID_FMT " " GID_FMT, &gid_base, &gid_shift, &gid_range); + if (k < 0 && feof(f)) + break; + if (k != 3) { + if (ferror(f) && errno > 0) + return -errno; + + return -EIO; + } + + if (gid < gid_base || gid >= gid_base + gid_range) + continue; + + converted = gid - gid_base + gid_shift; + if (!gid_is_valid(converted)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid group ID " GID_FMT, gid); + + return sd_bus_reply_method_return(message, "u", (uint32_t) converted); + } + + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_GROUP_MAPPING, "Machine '%s' has no matching group mappings.", name); +} + +static int method_map_to_machine_group(sd_bus_message *message, void *groupdata, sd_bus_error *error) { + Manager *m = groupdata; + Machine *machine; + gid_t gid; + Iterator i; + int r; + + r = sd_bus_message_read(message, "u", &gid); + if (r < 0) + return r; + if (!gid_is_valid(gid)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid group ID " GID_FMT, gid); + if (gid < 0x10000) + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_GROUP_MAPPING, "Group " GID_FMT " belongs to host GID range", gid); + + HASHMAP_FOREACH(machine, m->machines, i) { + _cleanup_fclose_ FILE *f = NULL; + char p[strlen("/proc//gid_map") + DECIMAL_STR_MAX(pid_t) + 1]; + + if (machine->class != MACHINE_CONTAINER) + continue; + + xsprintf(p, "/proc/" GID_FMT "/gid_map", machine->leader); + f = fopen(p, "re"); + if (!f) { + log_warning_errno(errno, "Failed top open %s, ignoring,", p); + continue; + } + + for (;;) { + _cleanup_free_ char *o = NULL; + gid_t gid_base, gid_shift, gid_range, converted; + int k; + + errno = 0; + k = fscanf(f, GID_FMT " " GID_FMT " " GID_FMT, &gid_base, &gid_shift, &gid_range); + if (k < 0 && feof(f)) + break; + if (k != 3) { + if (ferror(f) && errno > 0) + return -errno; + + return -EIO; + } + + if (gid < gid_shift || gid >= gid_shift + gid_range) + continue; + + converted = (gid - gid_shift + gid_base); + if (!gid_is_valid(converted)) + return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid group ID " GID_FMT, gid); + + o = machine_bus_path(machine); + if (!o) + return -ENOMEM; + + return sd_bus_reply_method_return(message, "sou", machine->name, o, (uint32_t) converted); + } + } + + return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_GROUP_MAPPING, "No matching group mapping for " GID_FMT ".", gid); +} + +const sd_bus_vtable manager_vtable[] = { + SD_BUS_VTABLE_START(0), + SD_BUS_PROPERTY("PoolPath", "s", property_get_pool_path, 0, 0), + SD_BUS_PROPERTY("PoolUsage", "t", property_get_pool_usage, 0, 0), + SD_BUS_PROPERTY("PoolLimit", "t", property_get_pool_limit, 0, 0), + SD_BUS_METHOD("GetMachine", "s", "o", method_get_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetImage", "s", "o", method_get_image, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetMachineByPID", "u", "o", method_get_machine_by_pid, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ListMachines", NULL, "a(ssso)", method_list_machines, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("ListImages", NULL, "a(ssbttto)", method_list_images, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CreateMachine", "sayssusa(sv)", "o", method_create_machine, 0), + SD_BUS_METHOD("CreateMachineWithNetwork", "sayssusaia(sv)", "o", method_create_machine_with_network, 0), + SD_BUS_METHOD("RegisterMachine", "sayssus", "o", method_register_machine, 0), + SD_BUS_METHOD("RegisterMachineWithNetwork", "sayssusai", "o", method_register_machine_with_network, 0), + SD_BUS_METHOD("TerminateMachine", "s", NULL, method_terminate_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("KillMachine", "ssi", NULL, method_kill_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetMachineAddresses", "s", "a(iay)", method_get_machine_addresses, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("GetMachineOSRelease", "s", "a{ss}", method_get_machine_os_release, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenMachinePTY", "s", "hs", method_open_machine_pty, 0), + SD_BUS_METHOD("OpenMachineLogin", "s", "hs", method_open_machine_login, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenMachineShell", "sssasas", "hs", method_open_machine_shell, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("BindMountMachine", "sssbb", NULL, method_bind_mount_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CopyFromMachine", "sss", NULL, method_copy_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CopyToMachine", "sss", NULL, method_copy_machine, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("OpenMachineRootDirectory", "s", "h", method_open_machine_root_directory, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("RemoveImage", "s", NULL, method_remove_image, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("RenameImage", "ss", NULL, method_rename_image, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CloneImage", "ssb", NULL, method_clone_image, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MarkImageReadOnly", "sb", NULL, method_mark_image_read_only, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("SetPoolLimit", "t", NULL, method_set_pool_limit, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("SetImageLimit", "st", NULL, method_set_image_limit, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("CleanPool", "s", "a(st)", method_clean_pool, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MapFromMachineUser", "su", "u", method_map_from_machine_user, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MapToMachineUser", "u", "sou", method_map_to_machine_user, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MapFromMachineGroup", "su", "u", method_map_from_machine_group, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("MapToMachineGroup", "u", "sou", method_map_to_machine_group, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_SIGNAL("MachineNew", "so", 0), + SD_BUS_SIGNAL("MachineRemoved", "so", 0), + SD_BUS_VTABLE_END +}; + +int match_job_removed(sd_bus_message *message, void *userdata, sd_bus_error *error) { + const char *path, *result, *unit; + Manager *m = userdata; + Machine *machine; + uint32_t id; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "uoss", &id, &path, &unit, &result); + if (r < 0) { + bus_log_parse_error(r); + return 0; + } + + machine = hashmap_get(m->machine_units, unit); + if (!machine) + return 0; + + if (streq_ptr(path, machine->scope_job)) { + machine->scope_job = mfree(machine->scope_job); + + if (machine->started) { + if (streq(result, "done")) + machine_send_create_reply(machine, NULL); + else { + _cleanup_(sd_bus_error_free) sd_bus_error e = SD_BUS_ERROR_NULL; + + sd_bus_error_setf(&e, BUS_ERROR_JOB_FAILED, "Start job for unit %s failed with '%s'", unit, result); + + machine_send_create_reply(machine, &e); + } + } + + machine_save(machine); + } + + machine_add_to_gc_queue(machine); + return 0; +} + +int match_properties_changed(sd_bus_message *message, void *userdata, sd_bus_error *error) { + _cleanup_free_ char *unit = NULL; + const char *path; + Manager *m = userdata; + Machine *machine; + int r; + + assert(message); + assert(m); + + path = sd_bus_message_get_path(message); + if (!path) + return 0; + + r = unit_name_from_dbus_path(path, &unit); + if (r == -EINVAL) /* not for a unit */ + return 0; + if (r < 0) { + log_oom(); + return 0; + } + + machine = hashmap_get(m->machine_units, unit); + if (!machine) + return 0; + + machine_add_to_gc_queue(machine); + return 0; +} + +int match_unit_removed(sd_bus_message *message, void *userdata, sd_bus_error *error) { + const char *path, *unit; + Manager *m = userdata; + Machine *machine; + int r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "so", &unit, &path); + if (r < 0) { + bus_log_parse_error(r); + return 0; + } + + machine = hashmap_get(m->machine_units, unit); + if (!machine) + return 0; + + machine_add_to_gc_queue(machine); + return 0; +} + +int match_reloading(sd_bus_message *message, void *userdata, sd_bus_error *error) { + Manager *m = userdata; + Machine *machine; + Iterator i; + int b, r; + + assert(message); + assert(m); + + r = sd_bus_message_read(message, "b", &b); + if (r < 0) { + bus_log_parse_error(r); + return 0; + } + if (b) + return 0; + + /* systemd finished reloading, let's recheck all our machines */ + log_debug("System manager has been reloaded, rechecking machines..."); + + HASHMAP_FOREACH(machine, m->machines, i) + machine_add_to_gc_queue(machine); + + return 0; +} + +int manager_start_scope( + Manager *manager, + const char *scope, + pid_t pid, + const char *slice, + const char *description, + sd_bus_message *more_properties, + sd_bus_error *error, + char **job) { + + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL, *reply = NULL; + int r; + + assert(manager); + assert(scope); + assert(pid > 1); + + r = sd_bus_message_new_method_call( + manager->bus, + &m, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "StartTransientUnit"); + if (r < 0) + return r; + + r = sd_bus_message_append(m, "ss", strempty(scope), "fail"); + if (r < 0) + return r; + + r = sd_bus_message_open_container(m, 'a', "(sv)"); + if (r < 0) + return r; + + if (!isempty(slice)) { + r = sd_bus_message_append(m, "(sv)", "Slice", "s", slice); + if (r < 0) + return r; + } + + if (!isempty(description)) { + r = sd_bus_message_append(m, "(sv)", "Description", "s", description); + if (r < 0) + return r; + } + + r = sd_bus_message_append(m, "(sv)", "PIDs", "au", 1, pid); + if (r < 0) + return r; + + r = sd_bus_message_append(m, "(sv)", "Delegate", "b", 1); + if (r < 0) + return r; + + r = sd_bus_message_append(m, "(sv)", "TasksMax", "t", UINT64_C(16384)); + if (r < 0) + return bus_log_create_error(r); + + if (more_properties) { + r = sd_bus_message_copy(m, more_properties, true); + if (r < 0) + return r; + } + + r = sd_bus_message_close_container(m); + if (r < 0) + return r; + + r = sd_bus_message_append(m, "a(sa(sv))", 0); + if (r < 0) + return r; + + r = sd_bus_call(manager->bus, m, 0, error, &reply); + if (r < 0) + return r; + + if (job) { + const char *j; + char *copy; + + r = sd_bus_message_read(reply, "o", &j); + if (r < 0) + return r; + + copy = strdup(j); + if (!copy) + return -ENOMEM; + + *job = copy; + } + + return 1; +} + +int manager_stop_unit(Manager *manager, const char *unit, sd_bus_error *error, char **job) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + int r; + + assert(manager); + assert(unit); + + r = sd_bus_call_method( + manager->bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "StopUnit", + error, + &reply, + "ss", unit, "fail"); + if (r < 0) { + if (sd_bus_error_has_name(error, BUS_ERROR_NO_SUCH_UNIT) || + sd_bus_error_has_name(error, BUS_ERROR_LOAD_FAILED)) { + + if (job) + *job = NULL; + + sd_bus_error_free(error); + return 0; + } + + return r; + } + + if (job) { + const char *j; + char *copy; + + r = sd_bus_message_read(reply, "o", &j); + if (r < 0) + return r; + + copy = strdup(j); + if (!copy) + return -ENOMEM; + + *job = copy; + } + + return 1; +} + +int manager_kill_unit(Manager *manager, const char *unit, int signo, sd_bus_error *error) { + assert(manager); + assert(unit); + + return sd_bus_call_method( + manager->bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "KillUnit", + error, + NULL, + "ssi", unit, "all", signo); +} + +int manager_unit_is_active(Manager *manager, const char *unit) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *path = NULL; + const char *state; + int r; + + assert(manager); + assert(unit); + + path = unit_dbus_path_from_name(unit); + if (!path) + return -ENOMEM; + + r = sd_bus_get_property( + manager->bus, + "org.freedesktop.systemd1", + path, + "org.freedesktop.systemd1.Unit", + "ActiveState", + &error, + &reply, + "s"); + if (r < 0) { + if (sd_bus_error_has_name(&error, SD_BUS_ERROR_NO_REPLY) || + sd_bus_error_has_name(&error, SD_BUS_ERROR_DISCONNECTED)) + return true; + + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_UNIT) || + sd_bus_error_has_name(&error, BUS_ERROR_LOAD_FAILED)) + return false; + + return r; + } + + r = sd_bus_message_read(reply, "s", &state); + if (r < 0) + return -EINVAL; + + return !STR_IN_SET(state, "inactive", "failed"); +} + +int manager_job_is_active(Manager *manager, const char *path) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + int r; + + assert(manager); + assert(path); + + r = sd_bus_get_property( + manager->bus, + "org.freedesktop.systemd1", + path, + "org.freedesktop.systemd1.Job", + "State", + &error, + &reply, + "s"); + if (r < 0) { + if (sd_bus_error_has_name(&error, SD_BUS_ERROR_NO_REPLY) || + sd_bus_error_has_name(&error, SD_BUS_ERROR_DISCONNECTED)) + return true; + + if (sd_bus_error_has_name(&error, SD_BUS_ERROR_UNKNOWN_OBJECT)) + return false; + + return r; + } + + /* We don't actually care about the state really. The fact + * that we could read the job state is enough for us */ + + return true; +} + +int manager_get_machine_by_pid(Manager *m, pid_t pid, Machine **machine) { + Machine *mm; + int r; + + assert(m); + assert(pid >= 1); + assert(machine); + + mm = hashmap_get(m->machine_leaders, PID_TO_PTR(pid)); + if (!mm) { + _cleanup_free_ char *unit = NULL; + + r = cg_pid_get_unit(pid, &unit); + if (r >= 0) + mm = hashmap_get(m->machine_units, unit); + } + if (!mm) + return 0; + + *machine = mm; + return 1; +} + +int manager_add_machine(Manager *m, const char *name, Machine **_machine) { + Machine *machine; + + assert(m); + assert(name); + + machine = hashmap_get(m->machines, name); + if (!machine) { + machine = machine_new(m, _MACHINE_CLASS_INVALID, name); + if (!machine) + return -ENOMEM; + } + + if (_machine) + *_machine = machine; + + return 0; +} diff --git a/src/grp-machine/libmachine-core/src/operation.c b/src/grp-machine/libmachine-core/src/operation.c new file mode 100644 index 0000000000..ce37eee98b --- /dev/null +++ b/src/grp-machine/libmachine-core/src/operation.c @@ -0,0 +1,151 @@ +/*** + This file is part of systemd. + + Copyright 2016 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "machine-core/operation.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/process-util.h" + +static int operation_done(sd_event_source *s, const siginfo_t *si, void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + Operation *o = userdata; + int r; + + assert(o); + assert(si); + + log_debug("Operating " PID_FMT " is now complete with code=%s status=%i", + o->pid, + sigchld_code_to_string(si->si_code), si->si_status); + + o->pid = 0; + + if (si->si_code != CLD_EXITED) { + r = sd_bus_error_setf(&error, SD_BUS_ERROR_FAILED, "Child died abnormally."); + goto fail; + } + + if (si->si_status == EXIT_SUCCESS) + r = 0; + else if (read(o->errno_fd, &r, sizeof(r)) != sizeof(r)) { /* Try to acquire error code for failed operation */ + r = sd_bus_error_setf(&error, SD_BUS_ERROR_FAILED, "Child failed."); + goto fail; + } + + if (o->done) { + /* A completion routine is set for this operation, call it. */ + r = o->done(o, r, &error); + if (r < 0) { + if (!sd_bus_error_is_set(&error)) + sd_bus_error_set_errno(&error, r); + + goto fail; + } + + } else { + /* The default operation when done is to simply return an error on failure or an empty success + * message on success. */ + if (r < 0) + goto fail; + + r = sd_bus_reply_method_return(o->message, NULL); + if (r < 0) + log_error_errno(r, "Failed to reply to message: %m"); + } + + operation_free(o); + return 0; + +fail: + r = sd_bus_reply_method_error(o->message, &error); + if (r < 0) + log_error_errno(r, "Failed to reply to message: %m"); + + operation_free(o); + return 0; +} + +int operation_new(Manager *manager, Machine *machine, pid_t child, sd_bus_message *message, int errno_fd, Operation **ret) { + Operation *o; + int r; + + assert(manager); + assert(child > 1); + assert(message); + assert(errno_fd >= 0); + + o = new0(Operation, 1); + if (!o) + return -ENOMEM; + + o->extra_fd = -1; + + r = sd_event_add_child(manager->event, &o->event_source, child, WEXITED, operation_done, o); + if (r < 0) { + free(o); + return r; + } + + o->pid = child; + o->message = sd_bus_message_ref(message); + o->errno_fd = errno_fd; + + LIST_PREPEND(operations, manager->operations, o); + manager->n_operations++; + o->manager = manager; + + if (machine) { + LIST_PREPEND(operations_by_machine, machine->operations, o); + o->machine = machine; + } + + log_debug("Started new operation " PID_FMT ".", child); + + /* At this point we took ownership of both the child and the errno file descriptor! */ + + if (ret) + *ret = o; + + return 0; +} + +Operation *operation_free(Operation *o) { + if (!o) + return NULL; + + sd_event_source_unref(o->event_source); + + safe_close(o->errno_fd); + safe_close(o->extra_fd); + + if (o->pid > 1) + (void) sigkill_wait(o->pid); + + sd_bus_message_unref(o->message); + + if (o->manager) { + LIST_REMOVE(operations, o->manager->operations, o); + o->manager->n_operations--; + } + + if (o->machine) + LIST_REMOVE(operations_by_machine, o->machine->operations, o); + + return mfree(o); +} diff --git a/src/grp-machine/libmachine-core/test/GNUmakefile b/src/grp-machine/libmachine-core/test/GNUmakefile new file mode 120000 index 0000000000..13308a50cd --- /dev/null +++ b/src/grp-machine/libmachine-core/test/GNUmakefile @@ -0,0 +1 @@ +../../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/libmachine-core/test/Makefile b/src/grp-machine/libmachine-core/test/Makefile new file mode 100644 index 0000000000..a75e11f3ac --- /dev/null +++ b/src/grp-machine/libmachine-core/test/Makefile @@ -0,0 +1,35 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +test_machine_tables_SOURCES = \ + src/machine/test-machine-tables.c + +test_machine_tables_LDADD = \ + libmachine-core.la + +tests += \ + test-machine-tables + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/libmachine-core/test/test-machine-tables.c b/src/grp-machine/libmachine-core/test/test-machine-tables.c new file mode 100644 index 0000000000..5f50c3856b --- /dev/null +++ b/src/grp-machine/libmachine-core/test/test-machine-tables.c @@ -0,0 +1,29 @@ +/*** + This file is part of systemd + + Copyright 2013 Zbigniew Jędrzejewski-Szmek + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "machine-core/machine.h" +#include "systemd-shared/test-tables.h" + +int main(int argc, char **argv) { + test_table(machine_class, MACHINE_CLASS); + test_table(machine_state, MACHINE_STATE); + test_table(kill_who, KILL_WHO); + + return EXIT_SUCCESS; +} diff --git a/src/grp-machine/machinectl/GNUmakefile b/src/grp-machine/machinectl/GNUmakefile new file mode 120000 index 0000000000..95e5924740 --- /dev/null +++ b/src/grp-machine/machinectl/GNUmakefile @@ -0,0 +1 @@ +../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/machinectl/Makefile b/src/grp-machine/machinectl/Makefile new file mode 100644 index 0000000000..f6760f3174 --- /dev/null +++ b/src/grp-machine/machinectl/Makefile @@ -0,0 +1,42 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +machinectl_SOURCES = \ + src/machine/machinectl.c + +machinectl_LDADD = \ + libsystemd-shared.la + +rootbin_PROGRAMS += \ + machinectl + +dist_bashcompletion_data += \ + shell-completion/bash/machinectl + +dist_zshcompletion_data += \ + shell-completion/zsh/_machinectl \ + shell-completion/zsh/_sd_machines + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/machinectl/machinectl.c b/src/grp-machine/machinectl/machinectl.c new file mode 100644 index 0000000000..733d73435e --- /dev/null +++ b/src/grp-machine/machinectl/machinectl.c @@ -0,0 +1,3052 @@ +/*** + This file is part of systemd. + + Copyright 2013 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <arpa/inet.h> +#include <errno.h> +#include <fcntl.h> +#include <getopt.h> +#include <locale.h> +#include <net/if.h> +#include <netinet/in.h> +#include <string.h> +#include <sys/mount.h> +#include <sys/socket.h> +#include <unistd.h> + +#include <systemd/sd-bus.h> + +#include "sd-bus/bus-common-errors.h" +#include "sd-bus/bus-error.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/cgroup-util.h" +#include "systemd-basic/copy.h" +#include "systemd-basic/env-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/log.h" +#include "systemd-basic/macro.h" +#include "systemd-basic/mkdir.h" +#include "systemd-basic/parse-util.h" +#include "systemd-basic/path-util.h" +#include "systemd-basic/process-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/stdio-util.h" +#include "systemd-basic/strv.h" +#include "systemd-basic/terminal-util.h" +#include "systemd-basic/unit-name.h" +#include "systemd-basic/util.h" +#include "systemd-basic/verbs.h" +#include "systemd-basic/web-util.h" +#include "systemd-shared/bus-unit-util.h" +#include "systemd-shared/cgroup-show.h" +#include "systemd-shared/import-util.h" +#include "systemd-shared/logs-show.h" +#include "systemd-shared/pager.h" +#include "systemd-shared/ptyfwd.h" +#include "systemd-shared/spawn-polkit-agent.h" + +#define ALL_IP_ADDRESSES -1 + +static char **arg_property = NULL; +static bool arg_all = false; +static bool arg_value = false; +static bool arg_full = false; +static bool arg_no_pager = false; +static bool arg_legend = true; +static const char *arg_kill_who = NULL; +static int arg_signal = SIGTERM; +static BusTransport arg_transport = BUS_TRANSPORT_LOCAL; +static char *arg_host = NULL; +static bool arg_read_only = false; +static bool arg_mkdir = false; +static bool arg_quiet = false; +static bool arg_ask_password = true; +static unsigned arg_lines = 10; +static OutputMode arg_output = OUTPUT_SHORT; +static bool arg_force = false; +static ImportVerify arg_verify = IMPORT_VERIFY_SIGNATURE; +static const char* arg_format = NULL; +static const char *arg_uid = NULL; +static char **arg_setenv = NULL; +static int arg_addrs = 1; + +static int print_addresses(sd_bus *bus, const char *name, int, const char *pr1, const char *pr2, int n_addr); + +static void polkit_agent_open_if_enabled(void) { + + /* Open the polkit agent as a child process if necessary */ + + if (!arg_ask_password) + return; + + if (arg_transport != BUS_TRANSPORT_LOCAL) + return; + + polkit_agent_open(); +} + +static OutputFlags get_output_flags(void) { + return + arg_all * OUTPUT_SHOW_ALL | + arg_full * OUTPUT_FULL_WIDTH | + (!on_tty() || pager_have()) * OUTPUT_FULL_WIDTH | + colors_enabled() * OUTPUT_COLOR | + !arg_quiet * OUTPUT_WARN_CUTOFF; +} + +typedef struct MachineInfo { + const char *name; + const char *class; + const char *service; + char *os; + char *version_id; +} MachineInfo; + +static int compare_machine_info(const void *a, const void *b) { + const MachineInfo *x = a, *y = b; + + return strcmp(x->name, y->name); +} + +static void clean_machine_info(MachineInfo *machines, size_t n_machines) { + size_t i; + + if (!machines || n_machines == 0) + return; + + for (i = 0; i < n_machines; i++) { + free(machines[i].os); + free(machines[i].version_id); + } + free(machines); +} + +static int get_os_release_property(sd_bus *bus, const char *name, const char *query, ...) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + const char *k, *v, *iter, **query_res = NULL; + size_t count = 0, awaited_args = 0; + va_list ap; + int r; + + assert(bus); + assert(name); + assert(query); + + NULSTR_FOREACH(iter, query) + awaited_args++; + query_res = newa0(const char *, awaited_args); + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetMachineOSRelease", + NULL, &reply, "s", name); + if (r < 0) + return r; + + r = sd_bus_message_enter_container(reply, 'a', "{ss}"); + if (r < 0) + return bus_log_parse_error(r); + + while ((r = sd_bus_message_read(reply, "{ss}", &k, &v)) > 0) { + count = 0; + NULSTR_FOREACH(iter, query) { + if (streq(k, iter)) { + query_res[count] = v; + break; + } + count++; + } + } + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + va_start(ap, query); + for (count = 0; count < awaited_args; count++) { + char *val, **out; + + out = va_arg(ap, char **); + assert(out); + if (query_res[count]) { + val = strdup(query_res[count]); + if (!val) { + va_end(ap); + return log_oom(); + } + *out = val; + } + } + va_end(ap); + + return 0; +} + +static int list_machines(int argc, char *argv[], void *userdata) { + + size_t max_name = strlen("MACHINE"), max_class = strlen("CLASS"), + max_service = strlen("SERVICE"), max_os = strlen("OS"), max_version_id = strlen("VERSION"); + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *prefix = NULL; + MachineInfo *machines = NULL; + const char *name, *class, *service, *object; + size_t n_machines = 0, n_allocated = 0, j; + sd_bus *bus = userdata; + int r; + + assert(bus); + + pager_open(arg_no_pager, false); + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "ListMachines", + &error, + &reply, + NULL); + if (r < 0) { + log_error("Could not get machines: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_enter_container(reply, 'a', "(ssso)"); + if (r < 0) + return bus_log_parse_error(r); + while ((r = sd_bus_message_read(reply, "(ssso)", &name, &class, &service, &object)) > 0) { + size_t l; + + if (name[0] == '.' && !arg_all) + continue; + + if (!GREEDY_REALLOC(machines, n_allocated, n_machines + 1)) { + r = log_oom(); + goto out; + } + + machines[n_machines].os = NULL; + machines[n_machines].version_id = NULL; + r = get_os_release_property(bus, name, + "ID\0" "VERSION_ID\0", + &machines[n_machines].os, + &machines[n_machines].version_id); + if (r < 0) + goto out; + + machines[n_machines].name = name; + machines[n_machines].class = class; + machines[n_machines].service = service; + + l = strlen(name); + if (l > max_name) + max_name = l; + + l = strlen(class); + if (l > max_class) + max_class = l; + + l = strlen(service); + if (l > max_service) + max_service = l; + + l = machines[n_machines].os ? strlen(machines[n_machines].os) : 1; + if (l > max_os) + max_os = l; + + l = machines[n_machines].version_id ? strlen(machines[n_machines].version_id) : 1; + if (l > max_version_id) + max_version_id = l; + + n_machines++; + } + if (r < 0) { + r = bus_log_parse_error(r); + goto out; + } + + r = sd_bus_message_exit_container(reply); + if (r < 0) { + r = bus_log_parse_error(r); + goto out; + } + + qsort_safe(machines, n_machines, sizeof(MachineInfo), compare_machine_info); + + /* Allocate for prefix max characters for all fields + spaces between them + strlen(",\n") */ + r = asprintf(&prefix, "%-*s", + (int) (max_name + + max_class + + max_service + + max_os + + max_version_id + 5 + strlen(",\n")), + ",\n"); + if (r < 0) { + r = log_oom(); + goto out; + } + + if (arg_legend && n_machines > 0) + printf("%-*s %-*s %-*s %-*s %-*s %s\n", + (int) max_name, "MACHINE", + (int) max_class, "CLASS", + (int) max_service, "SERVICE", + (int) max_os, "OS", + (int) max_version_id, "VERSION", + "ADDRESSES"); + + for (j = 0; j < n_machines; j++) { + printf("%-*s %-*s %-*s %-*s %-*s ", + (int) max_name, machines[j].name, + (int) max_class, machines[j].class, + (int) max_service, strdash_if_empty(machines[j].service), + (int) max_os, strdash_if_empty(machines[j].os), + (int) max_version_id, strdash_if_empty(machines[j].version_id)); + + r = print_addresses(bus, machines[j].name, 0, "", prefix, arg_addrs); + if (r == -ENOSYS) + printf("-\n"); + } + + if (arg_legend && n_machines > 0) + printf("\n%zu machines listed.\n", n_machines); + else + printf("No machines.\n"); + +out: + clean_machine_info(machines, n_machines); + return r; +} + +typedef struct ImageInfo { + const char *name; + const char *type; + bool read_only; + usec_t crtime; + usec_t mtime; + uint64_t size; +} ImageInfo; + +static int compare_image_info(const void *a, const void *b) { + const ImageInfo *x = a, *y = b; + + return strcmp(x->name, y->name); +} + +static int list_images(int argc, char *argv[], void *userdata) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + size_t max_name = strlen("NAME"), max_type = strlen("TYPE"), max_size = strlen("USAGE"), max_crtime = strlen("CREATED"), max_mtime = strlen("MODIFIED"); + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ ImageInfo *images = NULL; + size_t n_images = 0, n_allocated = 0, j; + const char *name, *type, *object; + sd_bus *bus = userdata; + uint64_t crtime, mtime, size; + int read_only, r; + + assert(bus); + + pager_open(arg_no_pager, false); + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "ListImages", + &error, + &reply, + ""); + if (r < 0) { + log_error("Could not get images: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_enter_container(reply, SD_BUS_TYPE_ARRAY, "(ssbttto)"); + if (r < 0) + return bus_log_parse_error(r); + + while ((r = sd_bus_message_read(reply, "(ssbttto)", &name, &type, &read_only, &crtime, &mtime, &size, &object)) > 0) { + char buf[MAX(FORMAT_TIMESTAMP_MAX, FORMAT_BYTES_MAX)]; + size_t l; + + if (name[0] == '.' && !arg_all) + continue; + + if (!GREEDY_REALLOC(images, n_allocated, n_images + 1)) + return log_oom(); + + images[n_images].name = name; + images[n_images].type = type; + images[n_images].read_only = read_only; + images[n_images].crtime = crtime; + images[n_images].mtime = mtime; + images[n_images].size = size; + + l = strlen(name); + if (l > max_name) + max_name = l; + + l = strlen(type); + if (l > max_type) + max_type = l; + + if (crtime != 0) { + l = strlen(strna(format_timestamp(buf, sizeof(buf), crtime))); + if (l > max_crtime) + max_crtime = l; + } + + if (mtime != 0) { + l = strlen(strna(format_timestamp(buf, sizeof(buf), mtime))); + if (l > max_mtime) + max_mtime = l; + } + + if (size != (uint64_t) -1) { + l = strlen(strna(format_bytes(buf, sizeof(buf), size))); + if (l > max_size) + max_size = l; + } + + n_images++; + } + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + qsort_safe(images, n_images, sizeof(ImageInfo), compare_image_info); + + if (arg_legend && n_images > 0) + printf("%-*s %-*s %-3s %-*s %-*s %-*s\n", + (int) max_name, "NAME", + (int) max_type, "TYPE", + "RO", + (int) max_size, "USAGE", + (int) max_crtime, "CREATED", + (int) max_mtime, "MODIFIED"); + + for (j = 0; j < n_images; j++) { + char crtime_buf[FORMAT_TIMESTAMP_MAX], mtime_buf[FORMAT_TIMESTAMP_MAX], size_buf[FORMAT_BYTES_MAX]; + + printf("%-*s %-*s %s%-3s%s %-*s %-*s %-*s\n", + (int) max_name, images[j].name, + (int) max_type, images[j].type, + images[j].read_only ? ansi_highlight_red() : "", yes_no(images[j].read_only), images[j].read_only ? ansi_normal() : "", + (int) max_size, strna(format_bytes(size_buf, sizeof(size_buf), images[j].size)), + (int) max_crtime, strna(format_timestamp(crtime_buf, sizeof(crtime_buf), images[j].crtime)), + (int) max_mtime, strna(format_timestamp(mtime_buf, sizeof(mtime_buf), images[j].mtime))); + } + + if (arg_legend && n_images > 0) + printf("\n%zu images listed.\n", n_images); + else + printf("No images.\n"); + + return 0; +} + +static int show_unit_cgroup(sd_bus *bus, const char *unit, pid_t leader) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *path = NULL; + const char *cgroup; + int r; + unsigned c; + + assert(bus); + assert(unit); + + path = unit_dbus_path_from_name(unit); + if (!path) + return log_oom(); + + r = sd_bus_get_property( + bus, + "org.freedesktop.systemd1", + path, + unit_dbus_interface_from_name(unit), + "ControlGroup", + &error, + &reply, + "s"); + if (r < 0) + return log_error_errno(r, "Failed to query ControlGroup: %s", bus_error_message(&error, r)); + + r = sd_bus_message_read(reply, "s", &cgroup); + if (r < 0) + return bus_log_parse_error(r); + + if (isempty(cgroup)) + return 0; + + c = columns(); + if (c > 18) + c -= 18; + else + c = 0; + + r = unit_show_processes(bus, unit, cgroup, "\t\t ", c, get_output_flags(), &error); + if (r == -EBADR) { + + if (arg_transport == BUS_TRANSPORT_REMOTE) + return 0; + + /* Fallback for older systemd versions where the GetUnitProcesses() call is not yet available */ + + if (cg_is_empty_recursive(SYSTEMD_CGROUP_CONTROLLER, cgroup) != 0 && leader <= 0) + return 0; + + show_cgroup_and_extra(SYSTEMD_CGROUP_CONTROLLER, cgroup, "\t\t ", c, &leader, leader > 0, get_output_flags()); + } else if (r < 0) + return log_error_errno(r, "Failed to dump process list: %s", bus_error_message(&error, r)); + + return 0; +} + +static int print_addresses(sd_bus *bus, const char *name, int ifi, const char *prefix, const char *prefix2, int n_addr) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *addresses = NULL; + bool truncate = false; + int r; + + assert(bus); + assert(name); + assert(prefix); + assert(prefix2); + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetMachineAddresses", + NULL, + &reply, + "s", name); + if (r < 0) + return r; + + addresses = strdup(prefix); + if (!addresses) + return log_oom(); + prefix = ""; + + r = sd_bus_message_enter_container(reply, 'a', "(iay)"); + if (r < 0) + return bus_log_parse_error(r); + + while ((r = sd_bus_message_enter_container(reply, 'r', "iay")) > 0) { + int family; + const void *a; + size_t sz; + char buf_ifi[DECIMAL_STR_MAX(int) + 2], buffer[MAX(INET6_ADDRSTRLEN, INET_ADDRSTRLEN)]; + + r = sd_bus_message_read(reply, "i", &family); + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_read_array(reply, 'y', &a, &sz); + if (r < 0) + return bus_log_parse_error(r); + + if (n_addr != 0) { + if (family == AF_INET6 && ifi > 0) + xsprintf(buf_ifi, "%%%i", ifi); + else + strcpy(buf_ifi, ""); + + if(!strextend(&addresses, prefix, inet_ntop(family, a, buffer, sizeof(buffer)), buf_ifi, NULL)) + return log_oom(); + } else + truncate = true; + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + if (prefix != prefix2) + prefix = prefix2; + + if (n_addr > 0) + n_addr -= 1; + } + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + fprintf(stdout, "%s%s\n", addresses, truncate ? "..." : ""); + return 0; +} + +static int print_os_release(sd_bus *bus, const char *name, const char *prefix) { + _cleanup_free_ char *pretty = NULL; + int r; + + assert(bus); + assert(name); + assert(prefix); + + r = get_os_release_property(bus, name, "PRETTY_NAME\0", &pretty, NULL); + if (r < 0) + return r; + + if (pretty) + printf("%s%s\n", prefix, pretty); + + return 0; +} + +typedef struct MachineStatusInfo { + char *name; + sd_id128_t id; + char *class; + char *service; + char *unit; + char *root_directory; + pid_t leader; + struct dual_timestamp timestamp; + int *netif; + unsigned n_netif; +} MachineStatusInfo; + +static void machine_status_info_clear(MachineStatusInfo *info) { + if (info) { + free(info->name); + free(info->class); + free(info->service); + free(info->unit); + free(info->root_directory); + free(info->netif); + zero(*info); + } +} + +static void print_machine_status_info(sd_bus *bus, MachineStatusInfo *i) { + char since1[FORMAT_TIMESTAMP_RELATIVE_MAX], *s1; + char since2[FORMAT_TIMESTAMP_MAX], *s2; + int ifi = -1; + + assert(bus); + assert(i); + + fputs(strna(i->name), stdout); + + if (!sd_id128_is_null(i->id)) + printf("(" SD_ID128_FORMAT_STR ")\n", SD_ID128_FORMAT_VAL(i->id)); + else + putchar('\n'); + + s1 = format_timestamp_relative(since1, sizeof(since1), i->timestamp.realtime); + s2 = format_timestamp(since2, sizeof(since2), i->timestamp.realtime); + + if (s1) + printf("\t Since: %s; %s\n", s2, s1); + else if (s2) + printf("\t Since: %s\n", s2); + + if (i->leader > 0) { + _cleanup_free_ char *t = NULL; + + printf("\t Leader: %u", (unsigned) i->leader); + + get_process_comm(i->leader, &t); + if (t) + printf(" (%s)", t); + + putchar('\n'); + } + + if (i->service) { + printf("\t Service: %s", i->service); + + if (i->class) + printf("; class %s", i->class); + + putchar('\n'); + } else if (i->class) + printf("\t Class: %s\n", i->class); + + if (i->root_directory) + printf("\t Root: %s\n", i->root_directory); + + if (i->n_netif > 0) { + unsigned c; + + fputs("\t Iface:", stdout); + + for (c = 0; c < i->n_netif; c++) { + char name[IF_NAMESIZE+1] = ""; + + if (if_indextoname(i->netif[c], name)) { + fputc(' ', stdout); + fputs(name, stdout); + + if (ifi < 0) + ifi = i->netif[c]; + else + ifi = 0; + } else + printf(" %i", i->netif[c]); + } + + fputc('\n', stdout); + } + + print_addresses(bus, i->name, ifi, + "\t Address: ", + "\n\t ", + ALL_IP_ADDRESSES); + + print_os_release(bus, i->name, "\t OS: "); + + if (i->unit) { + printf("\t Unit: %s\n", i->unit); + show_unit_cgroup(bus, i->unit, i->leader); + + if (arg_transport == BUS_TRANSPORT_LOCAL) + + show_journal_by_unit( + stdout, + i->unit, + arg_output, + 0, + i->timestamp.monotonic, + arg_lines, + 0, + get_output_flags() | OUTPUT_BEGIN_NEWLINE, + SD_JOURNAL_LOCAL_ONLY, + true, + NULL); + } +} + +static int map_netif(sd_bus *bus, const char *member, sd_bus_message *m, sd_bus_error *error, void *userdata) { + MachineStatusInfo *i = userdata; + size_t l; + const void *v; + int r; + + assert_cc(sizeof(int32_t) == sizeof(int)); + r = sd_bus_message_read_array(m, SD_BUS_TYPE_INT32, &v, &l); + if (r < 0) + return r; + if (r == 0) + return -EBADMSG; + + i->n_netif = l / sizeof(int32_t); + i->netif = memdup(v, l); + if (!i->netif) + return -ENOMEM; + + return 0; +} + +static int show_machine_info(const char *verb, sd_bus *bus, const char *path, bool *new_line) { + + static const struct bus_properties_map map[] = { + { "Name", "s", NULL, offsetof(MachineStatusInfo, name) }, + { "Class", "s", NULL, offsetof(MachineStatusInfo, class) }, + { "Service", "s", NULL, offsetof(MachineStatusInfo, service) }, + { "Unit", "s", NULL, offsetof(MachineStatusInfo, unit) }, + { "RootDirectory", "s", NULL, offsetof(MachineStatusInfo, root_directory) }, + { "Leader", "u", NULL, offsetof(MachineStatusInfo, leader) }, + { "Timestamp", "t", NULL, offsetof(MachineStatusInfo, timestamp.realtime) }, + { "TimestampMonotonic", "t", NULL, offsetof(MachineStatusInfo, timestamp.monotonic) }, + { "Id", "ay", bus_map_id128, offsetof(MachineStatusInfo, id) }, + { "NetworkInterfaces", "ai", map_netif, 0 }, + {} + }; + + _cleanup_(machine_status_info_clear) MachineStatusInfo info = {}; + int r; + + assert(verb); + assert(bus); + assert(path); + assert(new_line); + + r = bus_map_all_properties(bus, + "org.freedesktop.machine1", + path, + map, + &info); + if (r < 0) + return log_error_errno(r, "Could not get properties: %m"); + + if (*new_line) + printf("\n"); + *new_line = true; + + print_machine_status_info(bus, &info); + + return r; +} + +static int show_machine_properties(sd_bus *bus, const char *path, bool *new_line) { + int r; + + assert(bus); + assert(path); + assert(new_line); + + if (*new_line) + printf("\n"); + + *new_line = true; + + r = bus_print_all_properties(bus, "org.freedesktop.machine1", path, arg_property, arg_value, arg_all); + if (r < 0) + log_error_errno(r, "Could not get properties: %m"); + + return r; +} + +static int show_machine(int argc, char *argv[], void *userdata) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + bool properties, new_line = false; + sd_bus *bus = userdata; + int r = 0, i; + + assert(bus); + + properties = !strstr(argv[0], "status"); + + pager_open(arg_no_pager, false); + + if (properties && argc <= 1) { + + /* If no argument is specified, inspect the manager + * itself */ + r = show_machine_properties(bus, "/org/freedesktop/machine1", &new_line); + if (r < 0) + return r; + } + + for (i = 1; i < argc; i++) { + const char *path = NULL; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetMachine", + &error, + &reply, + "s", argv[i]); + if (r < 0) { + log_error("Could not get path to machine: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "o", &path); + if (r < 0) + return bus_log_parse_error(r); + + if (properties) + r = show_machine_properties(bus, path, &new_line); + else + r = show_machine_info(argv[0], bus, path, &new_line); + } + + return r; +} + +typedef struct ImageStatusInfo { + char *name; + char *path; + char *type; + int read_only; + usec_t crtime; + usec_t mtime; + uint64_t usage; + uint64_t limit; + uint64_t usage_exclusive; + uint64_t limit_exclusive; +} ImageStatusInfo; + +static void image_status_info_clear(ImageStatusInfo *info) { + if (info) { + free(info->name); + free(info->path); + free(info->type); + zero(*info); + } +} + +static void print_image_status_info(sd_bus *bus, ImageStatusInfo *i) { + char ts_relative[FORMAT_TIMESTAMP_RELATIVE_MAX], *s1; + char ts_absolute[FORMAT_TIMESTAMP_MAX], *s2; + char bs[FORMAT_BYTES_MAX], *s3; + char bs_exclusive[FORMAT_BYTES_MAX], *s4; + + assert(bus); + assert(i); + + if (i->name) { + fputs(i->name, stdout); + putchar('\n'); + } + + if (i->type) + printf("\t Type: %s\n", i->type); + + if (i->path) + printf("\t Path: %s\n", i->path); + + printf("\t RO: %s%s%s\n", + i->read_only ? ansi_highlight_red() : "", + i->read_only ? "read-only" : "writable", + i->read_only ? ansi_normal() : ""); + + s1 = format_timestamp_relative(ts_relative, sizeof(ts_relative), i->crtime); + s2 = format_timestamp(ts_absolute, sizeof(ts_absolute), i->crtime); + if (s1 && s2) + printf("\t Created: %s; %s\n", s2, s1); + else if (s2) + printf("\t Created: %s\n", s2); + + s1 = format_timestamp_relative(ts_relative, sizeof(ts_relative), i->mtime); + s2 = format_timestamp(ts_absolute, sizeof(ts_absolute), i->mtime); + if (s1 && s2) + printf("\tModified: %s; %s\n", s2, s1); + else if (s2) + printf("\tModified: %s\n", s2); + + s3 = format_bytes(bs, sizeof(bs), i->usage); + s4 = i->usage_exclusive != i->usage ? format_bytes(bs_exclusive, sizeof(bs_exclusive), i->usage_exclusive) : NULL; + if (s3 && s4) + printf("\t Usage: %s (exclusive: %s)\n", s3, s4); + else if (s3) + printf("\t Usage: %s\n", s3); + + s3 = format_bytes(bs, sizeof(bs), i->limit); + s4 = i->limit_exclusive != i->limit ? format_bytes(bs_exclusive, sizeof(bs_exclusive), i->limit_exclusive) : NULL; + if (s3 && s4) + printf("\t Limit: %s (exclusive: %s)\n", s3, s4); + else if (s3) + printf("\t Limit: %s\n", s3); +} + +static int show_image_info(sd_bus *bus, const char *path, bool *new_line) { + + static const struct bus_properties_map map[] = { + { "Name", "s", NULL, offsetof(ImageStatusInfo, name) }, + { "Path", "s", NULL, offsetof(ImageStatusInfo, path) }, + { "Type", "s", NULL, offsetof(ImageStatusInfo, type) }, + { "ReadOnly", "b", NULL, offsetof(ImageStatusInfo, read_only) }, + { "CreationTimestamp", "t", NULL, offsetof(ImageStatusInfo, crtime) }, + { "ModificationTimestamp", "t", NULL, offsetof(ImageStatusInfo, mtime) }, + { "Usage", "t", NULL, offsetof(ImageStatusInfo, usage) }, + { "Limit", "t", NULL, offsetof(ImageStatusInfo, limit) }, + { "UsageExclusive", "t", NULL, offsetof(ImageStatusInfo, usage_exclusive) }, + { "LimitExclusive", "t", NULL, offsetof(ImageStatusInfo, limit_exclusive) }, + {} + }; + + _cleanup_(image_status_info_clear) ImageStatusInfo info = {}; + int r; + + assert(bus); + assert(path); + assert(new_line); + + r = bus_map_all_properties(bus, + "org.freedesktop.machine1", + path, + map, + &info); + if (r < 0) + return log_error_errno(r, "Could not get properties: %m"); + + if (*new_line) + printf("\n"); + *new_line = true; + + print_image_status_info(bus, &info); + + return r; +} + +typedef struct PoolStatusInfo { + char *path; + uint64_t usage; + uint64_t limit; +} PoolStatusInfo; + +static void pool_status_info_clear(PoolStatusInfo *info) { + if (info) { + free(info->path); + zero(*info); + info->usage = -1; + info->limit = -1; + } +} + +static void print_pool_status_info(sd_bus *bus, PoolStatusInfo *i) { + char bs[FORMAT_BYTES_MAX], *s; + + if (i->path) + printf("\t Path: %s\n", i->path); + + s = format_bytes(bs, sizeof(bs), i->usage); + if (s) + printf("\t Usage: %s\n", s); + + s = format_bytes(bs, sizeof(bs), i->limit); + if (s) + printf("\t Limit: %s\n", s); +} + +static int show_pool_info(sd_bus *bus) { + + static const struct bus_properties_map map[] = { + { "PoolPath", "s", NULL, offsetof(PoolStatusInfo, path) }, + { "PoolUsage", "t", NULL, offsetof(PoolStatusInfo, usage) }, + { "PoolLimit", "t", NULL, offsetof(PoolStatusInfo, limit) }, + {} + }; + + _cleanup_(pool_status_info_clear) PoolStatusInfo info = { + .usage = (uint64_t) -1, + .limit = (uint64_t) -1, + }; + int r; + + assert(bus); + + r = bus_map_all_properties(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + map, + &info); + if (r < 0) + return log_error_errno(r, "Could not get properties: %m"); + + print_pool_status_info(bus, &info); + + return 0; +} + + +static int show_image_properties(sd_bus *bus, const char *path, bool *new_line) { + int r; + + assert(bus); + assert(path); + assert(new_line); + + if (*new_line) + printf("\n"); + + *new_line = true; + + r = bus_print_all_properties(bus, "org.freedesktop.machine1", path, arg_property, arg_value, arg_all); + if (r < 0) + log_error_errno(r, "Could not get properties: %m"); + + return r; +} + +static int show_image(int argc, char *argv[], void *userdata) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + bool properties, new_line = false; + sd_bus *bus = userdata; + int r = 0, i; + + assert(bus); + + properties = !strstr(argv[0], "status"); + + pager_open(arg_no_pager, false); + + if (argc <= 1) { + + /* If no argument is specified, inspect the manager + * itself */ + + if (properties) + r = show_image_properties(bus, "/org/freedesktop/machine1", &new_line); + else + r = show_pool_info(bus); + if (r < 0) + return r; + } + + for (i = 1; i < argc; i++) { + const char *path = NULL; + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetImage", + &error, + &reply, + "s", argv[i]); + if (r < 0) { + log_error("Could not get path to image: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "o", &path); + if (r < 0) + return bus_log_parse_error(r); + + if (properties) + r = show_image_properties(bus, path, &new_line); + else + r = show_image_info(bus, path, &new_line); + } + + return r; +} + +static int kill_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + if (!arg_kill_who) + arg_kill_who = "all"; + + for (i = 1; i < argc; i++) { + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "KillMachine", + &error, + NULL, + "ssi", argv[i], arg_kill_who, arg_signal); + if (r < 0) { + log_error("Could not kill machine: %s", bus_error_message(&error, -r)); + return r; + } + } + + return 0; +} + +static int reboot_machine(int argc, char *argv[], void *userdata) { + arg_kill_who = "leader"; + arg_signal = SIGINT; /* sysvinit + systemd */ + + return kill_machine(argc, argv, userdata); +} + +static int poweroff_machine(int argc, char *argv[], void *userdata) { + arg_kill_who = "leader"; + arg_signal = SIGRTMIN+4; /* only systemd */ + + return kill_machine(argc, argv, userdata); +} + +static int terminate_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + for (i = 1; i < argc; i++) { + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "TerminateMachine", + &error, + NULL, + "s", argv[i]); + if (r < 0) { + log_error("Could not terminate machine: %s", bus_error_message(&error, -r)); + return r; + } + } + + return 0; +} + +static int copy_files(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_free_ char *abs_host_path = NULL; + char *dest, *host_path, *container_path; + sd_bus *bus = userdata; + bool copy_from; + int r; + + assert(bus); + + polkit_agent_open_if_enabled(); + + copy_from = streq(argv[0], "copy-from"); + dest = argv[3] ?: argv[2]; + host_path = copy_from ? dest : argv[2]; + container_path = copy_from ? argv[2] : dest; + + if (!path_is_absolute(host_path)) { + r = path_make_absolute_cwd(host_path, &abs_host_path); + if (r < 0) + return log_error_errno(r, "Failed to make path absolute: %m"); + + host_path = abs_host_path; + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + copy_from ? "CopyFromMachine" : "CopyToMachine"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "sss", + argv[1], + copy_from ? container_path : host_path, + copy_from ? host_path : container_path); + if (r < 0) + return bus_log_create_error(r); + + /* This is a slow operation, hence turn off any method call timeouts */ + r = sd_bus_call(bus, m, USEC_INFINITY, &error, NULL); + if (r < 0) + return log_error_errno(r, "Failed to copy: %s", bus_error_message(&error, r)); + + return 0; +} + +static int bind_mount(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int r; + + assert(bus); + + polkit_agent_open_if_enabled(); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "BindMountMachine", + &error, + NULL, + "sssbb", + argv[1], + argv[2], + argv[3], + arg_read_only, + arg_mkdir); + if (r < 0) { + log_error("Failed to bind mount: %s", bus_error_message(&error, -r)); + return r; + } + + return 0; +} + +static int on_machine_removed(sd_bus_message *m, void *userdata, sd_bus_error *ret_error) { + PTYForward ** forward = (PTYForward**) userdata; + int r; + + assert(m); + assert(forward); + + if (*forward) { + /* If the forwarder is already initialized, tell it to + * exit on the next vhangup(), so that we still flush + * out what might be queued and exit then. */ + + r = pty_forward_set_ignore_vhangup(*forward, false); + if (r >= 0) + return 0; + + log_error_errno(r, "Failed to set ignore_vhangup flag: %m"); + } + + /* On error, or when the forwarder is not initialized yet, quit immediately */ + sd_event_exit(sd_bus_get_event(sd_bus_message_get_bus(m)), EXIT_FAILURE); + return 0; +} + +static int process_forward(sd_event *event, PTYForward **forward, int master, PTYForwardFlags flags, const char *name) { + char last_char = 0; + bool machine_died; + int ret = 0, r; + + assert(event); + assert(master >= 0); + assert(name); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGWINCH, SIGTERM, SIGINT, -1) >= 0); + + if (!arg_quiet) { + if (streq(name, ".host")) + log_info("Connected to the local host. Press ^] three times within 1s to exit session."); + else + log_info("Connected to machine %s. Press ^] three times within 1s to exit session.", name); + } + + sd_event_add_signal(event, NULL, SIGINT, NULL, NULL); + sd_event_add_signal(event, NULL, SIGTERM, NULL, NULL); + + r = pty_forward_new(event, master, flags, forward); + if (r < 0) + return log_error_errno(r, "Failed to create PTY forwarder: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + pty_forward_get_last_char(*forward, &last_char); + + machine_died = + (flags & PTY_FORWARD_IGNORE_VHANGUP) && + pty_forward_get_ignore_vhangup(*forward) == 0; + + *forward = pty_forward_free(*forward); + + if (last_char != '\n') + fputc('\n', stdout); + + if (!arg_quiet) { + if (machine_died) + log_info("Machine %s terminated.", name); + else if (streq(name, ".host")) + log_info("Connection to the local host terminated."); + else + log_info("Connection to machine %s terminated.", name); + } + + sd_event_get_exit_code(event, &ret); + return ret; +} + +static int parse_machine_uid(const char *spec, const char **machine, char **uid) { + /* + * Whatever is specified in the spec takes priority over global arguments. + */ + char *_uid = NULL; + const char *_machine = NULL; + + if (spec) { + const char *at; + + at = strchr(spec, '@'); + if (at) { + if (at == spec) + /* Do the same as ssh and refuse "@host". */ + return -EINVAL; + + _machine = at + 1; + _uid = strndup(spec, at - spec); + if (!_uid) + return -ENOMEM; + } else + _machine = spec; + }; + + if (arg_uid && !_uid) { + _uid = strdup(arg_uid); + if (!_uid) + return -ENOMEM; + } + + *uid = _uid; + *machine = isempty(_machine) ? ".host" : _machine; + return 0; +} + +static int login_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(pty_forward_freep) PTYForward *forward = NULL; + _cleanup_(sd_bus_slot_unrefp) sd_bus_slot *slot = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + int master = -1, r; + sd_bus *bus = userdata; + const char *pty, *match, *machine; + + assert(bus); + + if (!strv_isempty(arg_setenv) || arg_uid) { + log_error("--setenv= and --uid= are not supported for 'login'. Use 'shell' instead."); + return -EINVAL; + } + + if (arg_transport != BUS_TRANSPORT_LOCAL && + arg_transport != BUS_TRANSPORT_MACHINE) { + log_error("Login only supported on local machines."); + return -EOPNOTSUPP; + } + + polkit_agent_open_if_enabled(); + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to get event loop: %m"); + + r = sd_bus_attach_event(bus, event, 0); + if (r < 0) + return log_error_errno(r, "Failed to attach bus to event loop: %m"); + + machine = argc < 2 || isempty(argv[1]) ? ".host" : argv[1]; + + match = strjoina("type='signal'," + "sender='org.freedesktop.machine1'," + "path='/org/freedesktop/machine1',", + "interface='org.freedesktop.machine1.Manager'," + "member='MachineRemoved'," + "arg0='", machine, "'"); + + r = sd_bus_add_match(bus, &slot, match, on_machine_removed, &forward); + if (r < 0) + return log_error_errno(r, "Failed to add machine removal match: %m"); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "OpenMachineLogin", + &error, + &reply, + "s", machine); + if (r < 0) { + log_error("Failed to get login PTY: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "hs", &master, &pty); + if (r < 0) + return bus_log_parse_error(r); + + return process_forward(event, &forward, master, PTY_FORWARD_IGNORE_VHANGUP, machine); +} + +static int shell_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL, *m = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(pty_forward_freep) PTYForward *forward = NULL; + _cleanup_(sd_bus_slot_unrefp) sd_bus_slot *slot = NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + int master = -1, r; + sd_bus *bus = userdata; + const char *pty, *match, *machine, *path; + _cleanup_free_ char *uid = NULL; + + assert(bus); + + if (arg_transport != BUS_TRANSPORT_LOCAL && + arg_transport != BUS_TRANSPORT_MACHINE) { + log_error("Shell only supported on local machines."); + return -EOPNOTSUPP; + } + + /* Pass $TERM to shell session, if not explicitly specified. */ + if (!strv_find_prefix(arg_setenv, "TERM=")) { + const char *t; + + t = strv_find_prefix(environ, "TERM="); + if (t) { + if (strv_extend(&arg_setenv, t) < 0) + return log_oom(); + } + } + + polkit_agent_open_if_enabled(); + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to get event loop: %m"); + + r = sd_bus_attach_event(bus, event, 0); + if (r < 0) + return log_error_errno(r, "Failed to attach bus to event loop: %m"); + + r = parse_machine_uid(argc >= 2 ? argv[1] : NULL, &machine, &uid); + if (r < 0) + return log_error_errno(r, "Failed to parse machine specification: %m"); + + match = strjoina("type='signal'," + "sender='org.freedesktop.machine1'," + "path='/org/freedesktop/machine1',", + "interface='org.freedesktop.machine1.Manager'," + "member='MachineRemoved'," + "arg0='", machine, "'"); + + r = sd_bus_add_match(bus, &slot, match, on_machine_removed, &forward); + if (r < 0) + return log_error_errno(r, "Failed to add machine removal match: %m"); + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "OpenMachineShell"); + if (r < 0) + return bus_log_create_error(r); + + path = argc < 3 || isempty(argv[2]) ? NULL : argv[2]; + + r = sd_bus_message_append(m, "sss", machine, uid, path); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append_strv(m, strv_length(argv) <= 3 ? NULL : argv + 2); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append_strv(m, arg_setenv); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_call(bus, m, 0, &error, &reply); + if (r < 0) { + log_error("Failed to get shell PTY: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "hs", &master, &pty); + if (r < 0) + return bus_log_parse_error(r); + + return process_forward(event, &forward, master, 0, machine); +} + +static int remove_image(int argc, char *argv[], void *userdata) { + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + for (i = 1; i < argc; i++) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "RemoveImage"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append(m, "s", argv[i]); + if (r < 0) + return bus_log_create_error(r); + + /* This is a slow operation, hence turn off any method call timeouts */ + r = sd_bus_call(bus, m, USEC_INFINITY, &error, NULL); + if (r < 0) + return log_error_errno(r, "Could not remove image: %s", bus_error_message(&error, r)); + } + + return 0; +} + +static int rename_image(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int r; + + polkit_agent_open_if_enabled(); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "RenameImage", + &error, + NULL, + "ss", argv[1], argv[2]); + if (r < 0) { + log_error("Could not rename image: %s", bus_error_message(&error, -r)); + return r; + } + + return 0; +} + +static int clone_image(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + sd_bus *bus = userdata; + int r; + + polkit_agent_open_if_enabled(); + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "CloneImage"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append(m, "ssb", argv[1], argv[2], arg_read_only); + if (r < 0) + return bus_log_create_error(r); + + /* This is a slow operation, hence turn off any method call timeouts */ + r = sd_bus_call(bus, m, USEC_INFINITY, &error, NULL); + if (r < 0) + return log_error_errno(r, "Could not clone image: %s", bus_error_message(&error, r)); + + return 0; +} + +static int read_only_image(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int b = true, r; + + if (argc > 2) { + b = parse_boolean(argv[2]); + if (b < 0) { + log_error("Failed to parse boolean argument: %s", argv[2]); + return -EINVAL; + } + } + + polkit_agent_open_if_enabled(); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "MarkImageReadOnly", + &error, + NULL, + "sb", argv[1], b); + if (r < 0) { + log_error("Could not mark image read-only: %s", bus_error_message(&error, -r)); + return r; + } + + return 0; +} + +static int image_exists(sd_bus *bus, const char *name) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + int r; + + assert(bus); + assert(name); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetImage", + &error, + NULL, + "s", name); + if (r < 0) { + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_IMAGE)) + return 0; + + return log_error_errno(r, "Failed to check whether image %s exists: %s", name, bus_error_message(&error, -r)); + } + + return 1; +} + +static int make_service_name(const char *name, char **ret) { + int r; + + assert(name); + assert(ret); + + if (!machine_name_is_valid(name)) { + log_error("Invalid machine name %s.", name); + return -EINVAL; + } + + r = unit_name_build("systemd-nspawn", name, ".service", ret); + if (r < 0) + return log_error_errno(r, "Failed to build unit name: %m"); + + return 0; +} + +static int start_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(bus_wait_for_jobs_freep) BusWaitForJobs *w = NULL; + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + r = bus_wait_for_jobs_new(bus, &w); + if (r < 0) + return log_oom(); + + for (i = 1; i < argc; i++) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_free_ char *unit = NULL; + const char *object; + + r = make_service_name(argv[i], &unit); + if (r < 0) + return r; + + r = image_exists(bus, argv[i]); + if (r < 0) + return r; + if (r == 0) { + log_error("Machine image '%s' does not exist.", argv[1]); + return -ENXIO; + } + + r = sd_bus_call_method( + bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "StartUnit", + &error, + &reply, + "ss", unit, "fail"); + if (r < 0) { + log_error("Failed to start unit: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "o", &object); + if (r < 0) + return bus_log_parse_error(r); + + r = bus_wait_for_jobs_add(w, object); + if (r < 0) + return log_oom(); + } + + r = bus_wait_for_jobs(w, arg_quiet, NULL); + if (r < 0) + return r; + + return 0; +} + +static int enable_machine(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL, *reply = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + UnitFileChange *changes = NULL; + unsigned n_changes = 0; + int carries_install_info = 0; + const char *method = NULL; + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + method = streq(argv[0], "enable") ? "EnableUnitFiles" : "DisableUnitFiles"; + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + method); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_open_container(m, 'a', "s"); + if (r < 0) + return bus_log_create_error(r); + + for (i = 1; i < argc; i++) { + _cleanup_free_ char *unit = NULL; + + r = make_service_name(argv[i], &unit); + if (r < 0) + return r; + + r = image_exists(bus, argv[i]); + if (r < 0) + return r; + if (r == 0) { + log_error("Machine image '%s' does not exist.", argv[1]); + return -ENXIO; + } + + r = sd_bus_message_append(m, "s", unit); + if (r < 0) + return bus_log_create_error(r); + } + + r = sd_bus_message_close_container(m); + if (r < 0) + return bus_log_create_error(r); + + if (streq(argv[0], "enable")) + r = sd_bus_message_append(m, "bb", false, false); + else + r = sd_bus_message_append(m, "b", false); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_call(bus, m, 0, &error, &reply); + if (r < 0) { + log_error("Failed to enable or disable unit: %s", bus_error_message(&error, -r)); + return r; + } + + if (streq(argv[0], "enable")) { + r = sd_bus_message_read(reply, "b", carries_install_info); + if (r < 0) + return bus_log_parse_error(r); + } + + r = bus_deserialize_and_dump_unit_file_changes(reply, arg_quiet, &changes, &n_changes); + if (r < 0) + goto finish; + + r = sd_bus_call_method( + bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "Reload", + &error, + NULL, + NULL); + if (r < 0) { + log_error("Failed to reload daemon: %s", bus_error_message(&error, -r)); + goto finish; + } + + r = 0; + +finish: + unit_file_changes_free(changes, n_changes); + + return r; +} + +static int match_log_message(sd_bus_message *m, void *userdata, sd_bus_error *error) { + const char **our_path = userdata, *line; + unsigned priority; + int r; + + assert(m); + assert(our_path); + + r = sd_bus_message_read(m, "us", &priority, &line); + if (r < 0) { + bus_log_parse_error(r); + return 0; + } + + if (!streq_ptr(*our_path, sd_bus_message_get_path(m))) + return 0; + + if (arg_quiet && LOG_PRI(priority) >= LOG_INFO) + return 0; + + log_full(priority, "%s", line); + return 0; +} + +static int match_transfer_removed(sd_bus_message *m, void *userdata, sd_bus_error *error) { + const char **our_path = userdata, *path, *result; + uint32_t id; + int r; + + assert(m); + assert(our_path); + + r = sd_bus_message_read(m, "uos", &id, &path, &result); + if (r < 0) { + bus_log_parse_error(r); + return 0; + } + + if (!streq_ptr(*our_path, path)) + return 0; + + sd_event_exit(sd_bus_get_event(sd_bus_message_get_bus(m)), !streq_ptr(result, "done")); + return 0; +} + +static int transfer_signal_handler(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) { + assert(s); + assert(si); + + if (!arg_quiet) + log_info("Continuing download in the background. Use \"machinectl cancel-transfer %" PRIu32 "\" to abort transfer.", PTR_TO_UINT32(userdata)); + + sd_event_exit(sd_event_source_get_event(s), EINTR); + return 0; +} + +static int transfer_image_common(sd_bus *bus, sd_bus_message *m) { + _cleanup_(sd_bus_slot_unrefp) sd_bus_slot *slot_job_removed = NULL, *slot_log_message = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_(sd_event_unrefp) sd_event* event = NULL; + const char *path = NULL; + uint32_t id; + int r; + + assert(bus); + assert(m); + + polkit_agent_open_if_enabled(); + + r = sd_event_default(&event); + if (r < 0) + return log_error_errno(r, "Failed to get event loop: %m"); + + r = sd_bus_attach_event(bus, event, 0); + if (r < 0) + return log_error_errno(r, "Failed to attach bus to event loop: %m"); + + r = sd_bus_add_match( + bus, + &slot_job_removed, + "type='signal'," + "sender='org.freedesktop.import1'," + "interface='org.freedesktop.import1.Manager'," + "member='TransferRemoved'," + "path='/org/freedesktop/import1'", + match_transfer_removed, &path); + if (r < 0) + return log_error_errno(r, "Failed to install match: %m"); + + r = sd_bus_add_match( + bus, + &slot_log_message, + "type='signal'," + "sender='org.freedesktop.import1'," + "interface='org.freedesktop.import1.Transfer'," + "member='LogMessage'", + match_log_message, &path); + if (r < 0) + return log_error_errno(r, "Failed to install match: %m"); + + r = sd_bus_call(bus, m, 0, &error, &reply); + if (r < 0) { + log_error("Failed to transfer image: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_read(reply, "uo", &id, &path); + if (r < 0) + return bus_log_parse_error(r); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGTERM, SIGINT, -1) >= 0); + + if (!arg_quiet) + log_info("Enqueued transfer job %u. Press C-c to continue download in background.", id); + + sd_event_add_signal(event, NULL, SIGINT, transfer_signal_handler, UINT32_TO_PTR(id)); + sd_event_add_signal(event, NULL, SIGTERM, transfer_signal_handler, UINT32_TO_PTR(id)); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + return -r; +} + +static int import_tar(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_free_ char *ll = NULL; + _cleanup_close_ int fd = -1; + const char *local = NULL, *path = NULL; + sd_bus *bus = userdata; + int r; + + assert(bus); + + if (argc >= 2) + path = argv[1]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (argc >= 3) + local = argv[2]; + else if (path) + local = basename(path); + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (!local) { + log_error("Need either path or local name."); + return -EINVAL; + } + + r = tar_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local name %s is not a suitable machine name.", local); + return -EINVAL; + } + + if (path) { + fd = open(path, O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (fd < 0) + return log_error_errno(errno, "Failed to open %s: %m", path); + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "ImportTar"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "hsbb", + fd >= 0 ? fd : STDIN_FILENO, + local, + arg_force, + arg_read_only); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +static int import_raw(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_free_ char *ll = NULL; + _cleanup_close_ int fd = -1; + const char *local = NULL, *path = NULL; + sd_bus *bus = userdata; + int r; + + assert(bus); + + if (argc >= 2) + path = argv[1]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (argc >= 3) + local = argv[2]; + else if (path) + local = basename(path); + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (!local) { + log_error("Need either path or local name."); + return -EINVAL; + } + + r = raw_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local name %s is not a suitable machine name.", local); + return -EINVAL; + } + + if (path) { + fd = open(path, O_RDONLY|O_CLOEXEC|O_NOCTTY); + if (fd < 0) + return log_error_errno(errno, "Failed to open %s: %m", path); + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "ImportRaw"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "hsbb", + fd >= 0 ? fd : STDIN_FILENO, + local, + arg_force, + arg_read_only); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +static void determine_compression_from_filename(const char *p) { + if (arg_format) + return; + + if (!p) + return; + + if (endswith(p, ".xz")) + arg_format = "xz"; + else if (endswith(p, ".gz")) + arg_format = "gzip"; + else if (endswith(p, ".bz2")) + arg_format = "bzip2"; +} + +static int export_tar(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_close_ int fd = -1; + const char *local = NULL, *path = NULL; + sd_bus *bus = userdata; + int r; + + assert(bus); + + local = argv[1]; + if (!machine_name_is_valid(local)) { + log_error("Machine name %s is not valid.", local); + return -EINVAL; + } + + if (argc >= 3) + path = argv[2]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (path) { + determine_compression_from_filename(path); + + fd = open(path, O_WRONLY|O_CREAT|O_TRUNC|O_CLOEXEC|O_NOCTTY, 0666); + if (fd < 0) + return log_error_errno(errno, "Failed to open %s: %m", path); + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "ExportTar"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "shs", + local, + fd >= 0 ? fd : STDOUT_FILENO, + arg_format); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +static int export_raw(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_close_ int fd = -1; + const char *local = NULL, *path = NULL; + sd_bus *bus = userdata; + int r; + + assert(bus); + + local = argv[1]; + if (!machine_name_is_valid(local)) { + log_error("Machine name %s is not valid.", local); + return -EINVAL; + } + + if (argc >= 3) + path = argv[2]; + if (isempty(path) || streq(path, "-")) + path = NULL; + + if (path) { + determine_compression_from_filename(path); + + fd = open(path, O_WRONLY|O_CREAT|O_TRUNC|O_CLOEXEC|O_NOCTTY, 0666); + if (fd < 0) + return log_error_errno(errno, "Failed to open %s: %m", path); + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "ExportRaw"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "shs", + local, + fd >= 0 ? fd : STDOUT_FILENO, + arg_format); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +static int pull_tar(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_free_ char *l = NULL, *ll = NULL; + const char *local, *remote; + sd_bus *bus = userdata; + int r; + + assert(bus); + + remote = argv[1]; + if (!http_url_is_valid(remote)) { + log_error("URL '%s' is not valid.", remote); + return -EINVAL; + } + + if (argc >= 3) + local = argv[2]; + else { + r = import_url_last_component(remote, &l); + if (r < 0) + return log_error_errno(r, "Failed to get final component of URL: %m"); + + local = l; + } + + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = tar_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local name %s is not a suitable machine name.", local); + return -EINVAL; + } + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "PullTar"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "sssb", + remote, + local, + import_verify_to_string(arg_verify), + arg_force); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +static int pull_raw(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; + _cleanup_free_ char *l = NULL, *ll = NULL; + const char *local, *remote; + sd_bus *bus = userdata; + int r; + + assert(bus); + + remote = argv[1]; + if (!http_url_is_valid(remote)) { + log_error("URL '%s' is not valid.", remote); + return -EINVAL; + } + + if (argc >= 3) + local = argv[2]; + else { + r = import_url_last_component(remote, &l); + if (r < 0) + return log_error_errno(r, "Failed to get final component of URL: %m"); + + local = l; + } + + if (isempty(local) || streq(local, "-")) + local = NULL; + + if (local) { + r = raw_strip_suffixes(local, &ll); + if (r < 0) + return log_oom(); + + local = ll; + + if (!machine_name_is_valid(local)) { + log_error("Local name %s is not a suitable machine name.", local); + return -EINVAL; + } + } + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "PullRaw"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append( + m, + "sssb", + remote, + local, + import_verify_to_string(arg_verify), + arg_force); + if (r < 0) + return bus_log_create_error(r); + + return transfer_image_common(bus, m); +} + +typedef struct TransferInfo { + uint32_t id; + const char *type; + const char *remote; + const char *local; + double progress; +} TransferInfo; + +static int compare_transfer_info(const void *a, const void *b) { + const TransferInfo *x = a, *y = b; + + return strcmp(x->local, y->local); +} + +static int list_transfers(int argc, char *argv[], void *userdata) { + size_t max_type = strlen("TYPE"), max_local = strlen("LOCAL"), max_remote = strlen("REMOTE"); + _cleanup_(sd_bus_message_unrefp) sd_bus_message *reply = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_free_ TransferInfo *transfers = NULL; + size_t n_transfers = 0, n_allocated = 0, j; + const char *type, *remote, *local, *object; + sd_bus *bus = userdata; + uint32_t id, max_id = 0; + double progress; + int r; + + pager_open(arg_no_pager, false); + + r = sd_bus_call_method(bus, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "ListTransfers", + &error, + &reply, + NULL); + if (r < 0) { + log_error("Could not get transfers: %s", bus_error_message(&error, -r)); + return r; + } + + r = sd_bus_message_enter_container(reply, 'a', "(usssdo)"); + if (r < 0) + return bus_log_parse_error(r); + + while ((r = sd_bus_message_read(reply, "(usssdo)", &id, &type, &remote, &local, &progress, &object)) > 0) { + size_t l; + + if (!GREEDY_REALLOC(transfers, n_allocated, n_transfers + 1)) + return log_oom(); + + transfers[n_transfers].id = id; + transfers[n_transfers].type = type; + transfers[n_transfers].remote = remote; + transfers[n_transfers].local = local; + transfers[n_transfers].progress = progress; + + l = strlen(type); + if (l > max_type) + max_type = l; + + l = strlen(remote); + if (l > max_remote) + max_remote = l; + + l = strlen(local); + if (l > max_local) + max_local = l; + + if (id > max_id) + max_id = id; + + n_transfers++; + } + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + qsort_safe(transfers, n_transfers, sizeof(TransferInfo), compare_transfer_info); + + if (arg_legend && n_transfers > 0) + printf("%-*s %-*s %-*s %-*s %-*s\n", + (int) MAX(2U, DECIMAL_STR_WIDTH(max_id)), "ID", + (int) 7, "PERCENT", + (int) max_type, "TYPE", + (int) max_local, "LOCAL", + (int) max_remote, "REMOTE"); + + for (j = 0; j < n_transfers; j++) + printf("%*" PRIu32 " %*u%% %-*s %-*s %-*s\n", + (int) MAX(2U, DECIMAL_STR_WIDTH(max_id)), transfers[j].id, + (int) 6, (unsigned) (transfers[j].progress * 100), + (int) max_type, transfers[j].type, + (int) max_local, transfers[j].local, + (int) max_remote, transfers[j].remote); + + if (arg_legend && n_transfers > 0) + printf("\n%zu transfers listed.\n", n_transfers); + else + printf("No transfers.\n"); + + return 0; +} + +static int cancel_transfer(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + int r, i; + + assert(bus); + + polkit_agent_open_if_enabled(); + + for (i = 1; i < argc; i++) { + uint32_t id; + + r = safe_atou32(argv[i], &id); + if (r < 0) + return log_error_errno(r, "Failed to parse transfer id: %s", argv[i]); + + r = sd_bus_call_method( + bus, + "org.freedesktop.import1", + "/org/freedesktop/import1", + "org.freedesktop.import1.Manager", + "CancelTransfer", + &error, + NULL, + "u", id); + if (r < 0) { + log_error("Could not cancel transfer: %s", bus_error_message(&error, -r)); + return r; + } + } + + return 0; +} + +static int set_limit(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + sd_bus *bus = userdata; + uint64_t limit; + int r; + + if (STR_IN_SET(argv[argc-1], "-", "none", "infinity")) + limit = (uint64_t) -1; + else { + r = parse_size(argv[argc-1], 1024, &limit); + if (r < 0) + return log_error("Failed to parse size: %s", argv[argc-1]); + } + + if (argc > 2) + /* With two arguments changes the quota limit of the + * specified image */ + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "SetImageLimit", + &error, + NULL, + "st", argv[1], limit); + else + /* With one argument changes the pool quota limit */ + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "SetPoolLimit", + &error, + NULL, + "t", limit); + + if (r < 0) { + log_error("Could not set limit: %s", bus_error_message(&error, -r)); + return r; + } + + return 0; +} + +static int clean_images(int argc, char *argv[], void *userdata) { + _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL, *reply = NULL; + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + uint64_t usage, total = 0; + char fb[FORMAT_BYTES_MAX]; + sd_bus *bus = userdata; + const char *name; + unsigned c = 0; + int r; + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "CleanPool"); + if (r < 0) + return bus_log_create_error(r); + + r = sd_bus_message_append(m, "s", arg_all ? "all" : "hidden"); + if (r < 0) + return bus_log_create_error(r); + + /* This is a slow operation, hence permit a longer time for completion. */ + r = sd_bus_call(bus, m, USEC_INFINITY, &error, &reply); + if (r < 0) + return log_error_errno(r, "Could not clean pool: %s", bus_error_message(&error, r)); + + r = sd_bus_message_enter_container(reply, 'a', "(st)"); + if (r < 0) + return bus_log_parse_error(r); + + while ((r = sd_bus_message_read(reply, "(st)", &name, &usage)) > 0) { + log_info("Removed image '%s'. Freed exclusive disk space: %s", + name, format_bytes(fb, sizeof(fb), usage)); + + total += usage; + c++; + } + + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + log_info("Removed %u images in total. Total freed exclusive disk space %s.", + c, format_bytes(fb, sizeof(fb), total)); + + return 0; +} + +static int help(int argc, char *argv[], void *userdata) { + pager_open(arg_no_pager, false); + + printf("%s [OPTIONS...] {COMMAND} ...\n\n" + "Send control commands to or query the virtual machine and container\n" + "registration manager.\n\n" + " -h --help Show this help\n" + " --version Show package version\n" + " --no-pager Do not pipe output into a pager\n" + " --no-legend Do not show the headers and footers\n" + " --no-ask-password Do not ask for system passwords\n" + " -H --host=[USER@]HOST Operate on remote host\n" + " -M --machine=CONTAINER Operate on local container\n" + " -p --property=NAME Show only properties by this name\n" + " -q --quiet Suppress output\n" + " -a --all Show all properties, including empty ones\n" + " --value When showing properties, only print the value\n" + " -l --full Do not ellipsize output\n" + " --kill-who=WHO Who to send signal to\n" + " -s --signal=SIGNAL Which signal to send\n" + " --uid=USER Specify user ID to invoke shell as\n" + " -E --setenv=VAR=VALUE Add an environment variable for shell\n" + " --read-only Create read-only bind mount\n" + " --mkdir Create directory before bind mounting, if missing\n" + " -n --lines=INTEGER Number of journal entries to show\n" + " --max-addresses=INTEGER Number of internet addresses to show at most\n" + " -o --output=STRING Change journal output mode (short,\n" + " short-monotonic, verbose, export, json,\n" + " json-pretty, json-sse, cat)\n" + " --verify=MODE Verification mode for downloaded images (no,\n" + " checksum, signature)\n" + " --force Download image even if already exists\n\n" + "Machine Commands:\n" + " list List running VMs and containers\n" + " status NAME... Show VM/container details\n" + " show [NAME...] Show properties of one or more VMs/containers\n" + " start NAME... Start container as a service\n" + " login [NAME] Get a login prompt in a container or on the\n" + " local host\n" + " shell [[USER@]NAME [COMMAND...]]\n" + " Invoke a shell (or other command) in a container\n" + " or on the local host\n" + " enable NAME... Enable automatic container start at boot\n" + " disable NAME... Disable automatic container start at boot\n" + " poweroff NAME... Power off one or more containers\n" + " reboot NAME... Reboot one or more containers\n" + " terminate NAME... Terminate one or more VMs/containers\n" + " kill NAME... Send signal to processes of a VM/container\n" + " copy-to NAME PATH [PATH] Copy files from the host to a container\n" + " copy-from NAME PATH [PATH] Copy files from a container to the host\n" + " bind NAME PATH [PATH] Bind mount a path from the host into a container\n\n" + "Image Commands:\n" + " list-images Show available container and VM images\n" + " image-status [NAME...] Show image details\n" + " show-image [NAME...] Show properties of image\n" + " clone NAME NAME Clone an image\n" + " rename NAME NAME Rename an image\n" + " read-only NAME [BOOL] Mark or unmark image read-only\n" + " remove NAME... Remove an image\n" + " set-limit [NAME] BYTES Set image or pool size limit (disk quota)\n" + " clean Remove hidden (or all) images\n\n" + "Image Transfer Commands:\n" + " pull-tar URL [NAME] Download a TAR container image\n" + " pull-raw URL [NAME] Download a RAW container or VM image\n" + " import-tar FILE [NAME] Import a local TAR container image\n" + " import-raw FILE [NAME] Import a local RAW container or VM image\n" + " export-tar NAME [FILE] Export a TAR container image locally\n" + " export-raw NAME [FILE] Export a RAW container or VM image locally\n" + " list-transfers Show list of downloads in progress\n" + " cancel-transfer Cancel a download\n" + , program_invocation_short_name); + + return 0; +} + +static int parse_argv(int argc, char *argv[]) { + + enum { + ARG_VERSION = 0x100, + ARG_NO_PAGER, + ARG_NO_LEGEND, + ARG_VALUE, + ARG_KILL_WHO, + ARG_READ_ONLY, + ARG_MKDIR, + ARG_NO_ASK_PASSWORD, + ARG_VERIFY, + ARG_FORCE, + ARG_FORMAT, + ARG_UID, + ARG_NUMBER_IPS, + }; + + static const struct option options[] = { + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "property", required_argument, NULL, 'p' }, + { "all", no_argument, NULL, 'a' }, + { "value", no_argument, NULL, ARG_VALUE }, + { "full", no_argument, NULL, 'l' }, + { "no-pager", no_argument, NULL, ARG_NO_PAGER }, + { "no-legend", no_argument, NULL, ARG_NO_LEGEND }, + { "kill-who", required_argument, NULL, ARG_KILL_WHO }, + { "signal", required_argument, NULL, 's' }, + { "host", required_argument, NULL, 'H' }, + { "machine", required_argument, NULL, 'M' }, + { "read-only", no_argument, NULL, ARG_READ_ONLY }, + { "mkdir", no_argument, NULL, ARG_MKDIR }, + { "quiet", no_argument, NULL, 'q' }, + { "lines", required_argument, NULL, 'n' }, + { "output", required_argument, NULL, 'o' }, + { "no-ask-password", no_argument, NULL, ARG_NO_ASK_PASSWORD }, + { "verify", required_argument, NULL, ARG_VERIFY }, + { "force", no_argument, NULL, ARG_FORCE }, + { "format", required_argument, NULL, ARG_FORMAT }, + { "uid", required_argument, NULL, ARG_UID }, + { "setenv", required_argument, NULL, 'E' }, + { "max-addresses", required_argument, NULL, ARG_NUMBER_IPS }, + {} + }; + + bool reorder = false; + int c, r, shell = -1; + + assert(argc >= 0); + assert(argv); + + for (;;) { + static const char option_string[] = "-hp:als:H:M:qn:o:"; + + c = getopt_long(argc, argv, option_string + reorder, options, NULL); + if (c < 0) + break; + + switch (c) { + + case 1: /* getopt_long() returns 1 if "-" was the first character of the option string, and a + * non-option argument was discovered. */ + + assert(!reorder); + + /* We generally are fine with the fact that getopt_long() reorders the command line, and looks + * for switches after the main verb. However, for "shell" we really don't want that, since we + * want that switches specified after the machine name are passed to the program to execute, + * and not processed by us. To make this possible, we'll first invoke getopt_long() with + * reordering disabled (i.e. with the "-" prefix in the option string), looking for the first + * non-option parameter. If it's the verb "shell" we remember its position and continue + * processing options. In this case, as soon as we hit the next non-option argument we found + * the machine name, and stop further processing. If the first non-option argument is any other + * verb than "shell" we switch to normal reordering mode and continue processing arguments + * normally. */ + + if (shell >= 0) { + /* If we already found the "shell" verb on the command line, and now found the next + * non-option argument, then this is the machine name and we should stop processing + * further arguments. */ + optind --; /* don't process this argument, go one step back */ + goto done; + } + if (streq(optarg, "shell")) + /* Remember the position of the "shell" verb, and continue processing normally. */ + shell = optind - 1; + else { + int saved_optind; + + /* OK, this is some other verb. In this case, turn on reordering again, and continue + * processing normally. */ + reorder = true; + + /* We changed the option string. getopt_long() only looks at it again if we invoke it + * at least once with a reset option index. Hence, let's reset the option index here, + * then invoke getopt_long() again (ignoring what it has to say, after all we most + * likely already processed it), and the bump the option index so that we read the + * intended argument again. */ + saved_optind = optind; + optind = 0; + (void) getopt_long(argc, argv, option_string + reorder, options, NULL); + optind = saved_optind - 1; /* go one step back, process this argument again */ + } + + break; + + case 'h': + return help(0, NULL, NULL); + + case ARG_VERSION: + return version(); + + case 'p': + r = strv_extend(&arg_property, optarg); + if (r < 0) + return log_oom(); + + /* If the user asked for a particular + * property, show it to him, even if it is + * empty. */ + arg_all = true; + break; + + case 'a': + arg_all = true; + break; + + case ARG_VALUE: + arg_value = true; + break; + + case 'l': + arg_full = true; + break; + + case 'n': + if (safe_atou(optarg, &arg_lines) < 0) { + log_error("Failed to parse lines '%s'", optarg); + return -EINVAL; + } + break; + + case 'o': + arg_output = output_mode_from_string(optarg); + if (arg_output < 0) { + log_error("Unknown output '%s'.", optarg); + return -EINVAL; + } + break; + + case ARG_NO_PAGER: + arg_no_pager = true; + break; + + case ARG_NO_LEGEND: + arg_legend = false; + break; + + case ARG_KILL_WHO: + arg_kill_who = optarg; + break; + + case 's': + arg_signal = signal_from_string_try_harder(optarg); + if (arg_signal < 0) { + log_error("Failed to parse signal string %s.", optarg); + return -EINVAL; + } + break; + + case ARG_NO_ASK_PASSWORD: + arg_ask_password = false; + break; + + case 'H': + arg_transport = BUS_TRANSPORT_REMOTE; + arg_host = optarg; + break; + + case 'M': + arg_transport = BUS_TRANSPORT_MACHINE; + arg_host = optarg; + break; + + case ARG_READ_ONLY: + arg_read_only = true; + break; + + case ARG_MKDIR: + arg_mkdir = true; + break; + + case 'q': + arg_quiet = true; + break; + + case ARG_VERIFY: + arg_verify = import_verify_from_string(optarg); + if (arg_verify < 0) { + log_error("Failed to parse --verify= setting: %s", optarg); + return -EINVAL; + } + break; + + case ARG_FORCE: + arg_force = true; + break; + + case ARG_FORMAT: + if (!STR_IN_SET(optarg, "uncompressed", "xz", "gzip", "bzip2")) { + log_error("Unknown format: %s", optarg); + return -EINVAL; + } + + arg_format = optarg; + break; + + case ARG_UID: + arg_uid = optarg; + break; + + case 'E': + if (!env_assignment_is_valid(optarg)) { + log_error("Environment assignment invalid: %s", optarg); + return -EINVAL; + } + + r = strv_extend(&arg_setenv, optarg); + if (r < 0) + return log_oom(); + break; + + case ARG_NUMBER_IPS: + if (streq(optarg, "all")) + arg_addrs = ALL_IP_ADDRESSES; + else if (safe_atoi(optarg, &arg_addrs) < 0) { + log_error("Invalid number of IPs"); + return -EINVAL; + } else if (arg_addrs < 0) { + log_error("Number of IPs cannot be negative"); + return -EINVAL; + } + break; + + case '?': + return -EINVAL; + + default: + assert_not_reached("Unhandled option"); + } + } + +done: + if (shell >= 0) { + char *t; + int i; + + /* We found the "shell" verb while processing the argument list. Since we turned off reordering of the + * argument list initially let's readjust it now, and move the "shell" verb to the back. */ + + optind -= 1; /* place the option index where the "shell" verb will be placed */ + + t = argv[shell]; + for (i = shell; i < optind; i++) + argv[i] = argv[i+1]; + argv[optind] = t; + } + + return 1; +} + +static int machinectl_main(int argc, char *argv[], sd_bus *bus) { + + static const Verb verbs[] = { + { "help", VERB_ANY, VERB_ANY, 0, help }, + { "list", VERB_ANY, 1, VERB_DEFAULT, list_machines }, + { "list-images", VERB_ANY, 1, 0, list_images }, + { "status", 2, VERB_ANY, 0, show_machine }, + { "image-status", VERB_ANY, VERB_ANY, 0, show_image }, + { "show", VERB_ANY, VERB_ANY, 0, show_machine }, + { "show-image", VERB_ANY, VERB_ANY, 0, show_image }, + { "terminate", 2, VERB_ANY, 0, terminate_machine }, + { "reboot", 2, VERB_ANY, 0, reboot_machine }, + { "poweroff", 2, VERB_ANY, 0, poweroff_machine }, + { "stop", 2, VERB_ANY, 0, poweroff_machine }, /* Convenience alias */ + { "kill", 2, VERB_ANY, 0, kill_machine }, + { "login", VERB_ANY, 2, 0, login_machine }, + { "shell", VERB_ANY, VERB_ANY, 0, shell_machine }, + { "bind", 3, 4, 0, bind_mount }, + { "copy-to", 3, 4, 0, copy_files }, + { "copy-from", 3, 4, 0, copy_files }, + { "remove", 2, VERB_ANY, 0, remove_image }, + { "rename", 3, 3, 0, rename_image }, + { "clone", 3, 3, 0, clone_image }, + { "read-only", 2, 3, 0, read_only_image }, + { "start", 2, VERB_ANY, 0, start_machine }, + { "enable", 2, VERB_ANY, 0, enable_machine }, + { "disable", 2, VERB_ANY, 0, enable_machine }, + { "import-tar", 2, 3, 0, import_tar }, + { "import-raw", 2, 3, 0, import_raw }, + { "export-tar", 2, 3, 0, export_tar }, + { "export-raw", 2, 3, 0, export_raw }, + { "pull-tar", 2, 3, 0, pull_tar }, + { "pull-raw", 2, 3, 0, pull_raw }, + { "list-transfers", VERB_ANY, 1, 0, list_transfers }, + { "cancel-transfer", 2, VERB_ANY, 0, cancel_transfer }, + { "set-limit", 2, 3, 0, set_limit }, + { "clean", VERB_ANY, 1, 0, clean_images }, + {} + }; + + return dispatch_verb(argc, argv, verbs, bus); +} + +int main(int argc, char*argv[]) { + sd_bus *bus = NULL; + int r; + + setlocale(LC_ALL, ""); + log_parse_environment(); + log_open(); + + r = parse_argv(argc, argv); + if (r <= 0) + goto finish; + + r = bus_connect_transport(arg_transport, arg_host, false, &bus); + if (r < 0) { + log_error_errno(r, "Failed to create bus connection: %m"); + goto finish; + } + + sd_bus_set_allow_interactive_authorization(bus, arg_ask_password); + + r = machinectl_main(argc, argv, bus); + +finish: + sd_bus_flush_close_unref(bus); + pager_close(); + polkit_agent_close(); + + strv_free(arg_property); + strv_free(arg_setenv); + + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/machinectl/machinectl.completion.bash b/src/grp-machine/machinectl/machinectl.completion.bash new file mode 100644 index 0000000000..aebe48304d --- /dev/null +++ b/src/grp-machine/machinectl/machinectl.completion.bash @@ -0,0 +1,99 @@ +# machinectl(1) completion -*- shell-script -*- +# +# This file is part of systemd. +# +# Copyright 2014 Thomas H.P. Andersen +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. + +__contains_word() { + local w word=$1; shift + for w in "$@"; do + [[ $w = "$word" ]] && return + done +} + +__get_machines() { + local a b + (machinectl list-images --no-legend --no-pager; machinectl list --no-legend --no-pager; echo ".host") | \ + { while read a b; do echo " $a"; done; } | sort -u; +} + +_machinectl() { + local cur=${COMP_WORDS[COMP_CWORD]} prev=${COMP_WORDS[COMP_CWORD-1]} + local i verb comps + + local -A OPTS=( + [STANDALONE]='--all -a --full --help -h --no-ask-password --no-legend --no-pager --version' + [ARG]='--host -H --kill-who -M --machine --property -p --signal -s' + ) + + local -A VERBS=( + [STANDALONE]='list list-images pull-tar pull-raw import-tar import-raw export-tar export-raw list-transfers cancel-transfer' + [MACHINES]='status show start stop login shell enable disable poweroff reboot terminate kill copy-to copy-from image-status show-image clone rename read-only remove set-limit' + ) + + _init_completion || return + + for ((i=0; i <= COMP_CWORD; i++)); do + if __contains_word "${COMP_WORDS[i]}" ${VERBS[*]} && + ! __contains_word "${COMP_WORDS[i-1]}" ${OPTS[ARG]}; then + verb=${COMP_WORDS[i]} + break + fi + done + + if __contains_word "$prev" ${OPTS[ARG]}; then + case $prev in + --signal|-s) + _signals + return + ;; + --kill-who) + comps='all leader' + ;; + --host|-H) + comps=$(compgen -A hostname) + ;; + --machine|-M) + comps=$( __get_machines ) + ;; + --property|-p) + comps='' + ;; + esac + COMPREPLY=( $(compgen -W '$comps' -- "$cur") ) + return 0 + fi + + if [[ "$cur" = -* ]]; then + COMPREPLY=( $(compgen -W '${OPTS[*]}' -- "$cur") ) + return 0 + fi + + if [[ -z $verb ]]; then + comps=${VERBS[*]} + + elif __contains_word "$verb" ${VERBS[STANDALONE]}; then + comps='' + + elif __contains_word "$verb" ${VERBS[MACHINES]}; then + comps=$( __get_machines ) + fi + + COMPREPLY=( $(compgen -W '$comps' -- "$cur") ) + return 0 +} + +complete -F _machinectl machinectl diff --git a/src/grp-machine/machinectl/machinectl.completion.zsh b/src/grp-machine/machinectl/machinectl.completion.zsh new file mode 100644 index 0000000000..92d77109a5 --- /dev/null +++ b/src/grp-machine/machinectl/machinectl.completion.zsh @@ -0,0 +1,100 @@ +#compdef machinectl + +__get_available_machines () { + machinectl --no-legend list-images | {while read -r a b; do echo $a; done;} +} + +_available_machines() { + local -a _machines + _machines=("${(fo)$(__get_available_machines)}") + typeset -U _machines + if [[ -n "$_machines" ]]; then + _describe 'machines' _machines + else + _message 'no machines' + fi +} + +(( $+functions[_machinectl_command] )) || _machinectl_command() +{ + local -a _machinectl_cmds + _machinectl_cmds=( + "list:List currently running VMs/containers" + "status:Show VM/container status" + "show:Show properties of one or more VMs/containers" + "start:Start container as a service" + "stop:Stop container (equal to poweroff)" + "login:Get a login prompt on a VM/container" + "enable:Enable automatic container start at boot" + "disable:Disable automatic container start at boot" + "poweroff:Power off one or more VMs/containers" + "reboot:Reboot one or more VMs/containers" + "terminate:Terminate one or more VMs/containers" + "kill:Send signal to process or a VM/container" + "copy-to:Copy files from the host to a container" + "copy-from:Copy files from a container to the host" + "bind:Bind mount a path from the host into a container" + + "list-images:Show available container and VM images" + "image-status:Show image details" + "show-image:Show properties of image" + "clone:Clone an image" + "rename:Rename an image" + "read-only:Mark or unmark image read-only" + "remove:Remove an image" + + "pull-tar:Download a TAR container image" + "pull-raw:Download a RAW container or VM image" + "list-transfers:Show list of downloads in progress" + "cancel-transfer:Cancel a download" + ) + + if (( CURRENT == 1 )); then + _describe -t commands 'machinectl command' _machinectl_cmds || compadd "$@" + else + local curcontext="$curcontext" + cmd="${${_machinectl_cmds[(r)$words[1]:*]%%:*}}" + if (( $#cmd )); then + if (( CURRENT == 2 )); then + case $cmd in + list*|cancel-transfer|pull-tar|pull-raw) + msg="no options" ;; + start) + _available_machines ;; + *) + _sd_machines + esac + else + case $cmd in + copy-to|copy-from|bind) + _files ;; + *) msg="no options" + esac + fi + else + _message "no more options" + fi + fi +} + +_arguments \ + {-h,--help}'[Prints a short help text and exits.]' \ + '--version[Prints a short version string and exits.]' \ + '--no-pager[Do not pipe output into a pager.]' \ + '--no-legend[Do not show the headers and footers.]' \ + '--no-ask-password[Do not ask for system passwords.]' \ + {-H+,--host=}'[Operate on remote host.]:userathost:_sd_hosts_or_user_at_host' \ + {-M+,--machine=}'[Operate on local container.]:machine:_sd_machines' \ + {-p+,--property=}'[Limit output to specified property.]:property:(Name Id Timestamp TimestampMonotonic Service Scope Leader Class State RootDirectory)' \ + {-a,--all}'[Show all proerties.]' \ + {-q,--quiet}'[Suppress output.]' \ + {-l,--full}'[Do not ellipsize cgroup members.]' \ + '--kill-who=[Who to send signal to.]:killwho:(leader all)' \ + {-s+,--signal=}'[Which signal to send.]:signal:_signals' \ + '--read-only[Create read-only bind mount.]' \ + '--mkdir[Create directory before bind mounting, if missing.]' \ + {-n+,--lines=}'[Number of journal entries to show.]:integer' \ + {-o+,--output=}'[Change journal output mode.]:output modes:_sd_outputmodes' \ + '--verify=[Verification mode for downloaded images.]:verify:(no checksum signature)' \ + '--force[Download image even if already exists.]' \ + '*::machinectl command:_machinectl_command' diff --git a/src/grp-machine/machinectl/machinectl.xml b/src/grp-machine/machinectl/machinectl.xml new file mode 100644 index 0000000000..5a6ec294d2 --- /dev/null +++ b/src/grp-machine/machinectl/machinectl.xml @@ -0,0 +1,1029 @@ +<?xml version='1.0'?> <!--*- Mode: nxml; nxml-child-indent: 2; indent-tabs-mode: nil -*--> +<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" + "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"> + +<!-- + This file is part of systemd. + + Copyright 2013 Zbigniew Jędrzejewski-Szmek + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +--> + +<refentry id="machinectl" conditional='ENABLE_MACHINED' + xmlns:xi="http://www.w3.org/2001/XInclude"> + + <refentryinfo> + <title>machinectl</title> + <productname>systemd</productname> + + <authorgroup> + <author> + <contrib>Developer</contrib> + <firstname>Lennart</firstname> + <surname>Poettering</surname> + <email>lennart@poettering.net</email> + </author> + </authorgroup> + </refentryinfo> + + <refmeta> + <refentrytitle>machinectl</refentrytitle> + <manvolnum>1</manvolnum> + </refmeta> + + <refnamediv> + <refname>machinectl</refname> + <refpurpose>Control the systemd machine manager</refpurpose> + </refnamediv> + + <refsynopsisdiv> + <cmdsynopsis> + <command>machinectl</command> + <arg choice="opt" rep="repeat">OPTIONS</arg> + <arg choice="req">COMMAND</arg> + <arg choice="opt" rep="repeat">NAME</arg> + </cmdsynopsis> + </refsynopsisdiv> + + <refsect1> + <title>Description</title> + + <para><command>machinectl</command> may be used to introspect and + control the state of the + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry> + virtual machine and container registration manager + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>.</para> + + <para><command>machinectl</command> may be used to execute + operations on machines and images. Machines in this sense are + considered running instances of:</para> + + <itemizedlist> + <listitem><para>Virtual Machines (VMs) that virtualize hardware + to run full operating system (OS) instances (including their kernels) + in a virtualized environment on top of the host OS.</para></listitem> + + <listitem><para>Containers that share the hardware and + OS kernel with the host OS, in order to run + OS userspace instances on top the host OS.</para></listitem> + + <listitem><para>The host system itself</para></listitem> + </itemizedlist> + + <para>Machines are identified by names that follow the same rules + as UNIX and DNS host names, for details, see below. Machines are + instantiated from disk or file system images that frequently — but not + necessarily — carry the same name as machines running from + them. Images in this sense are considered:</para> + + <itemizedlist> + <listitem><para>Directory trees containing an OS, including its + top-level directories <filename>/usr</filename>, + <filename>/etc</filename>, and so on.</para></listitem> + + <listitem><para>btrfs subvolumes containing OS trees, similar to + normal directory trees.</para></listitem> + + <listitem><para>Binary "raw" disk images containing MBR or GPT + partition tables and Linux file system partitions.</para></listitem> + + <listitem><para>The file system tree of the host OS itself.</para></listitem> + </itemizedlist> + + </refsect1> + + <refsect1> + <title>Options</title> + + <para>The following options are understood:</para> + + <variablelist> + <varlistentry> + <term><option>-p</option></term> + <term><option>--property=</option></term> + + <listitem><para>When showing machine or image properties, + limit the output to certain properties as specified by the + argument. If not specified, all set properties are shown. The + argument should be a property name, such as + <literal>Name</literal>. If specified more than once, all + properties with the specified names are + shown.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>-a</option></term> + <term><option>--all</option></term> + + <listitem><para>When showing machine or image properties, show + all properties regardless of whether they are set or + not.</para> + + <para>When listing VM or container images, do not suppress + images beginning in a dot character + (<literal>.</literal>).</para> + + <para>When cleaning VM or container images, remove all images, not just hidden ones.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--value</option></term> + + <listitem><para>When printing properties with <command>show</command>, only print the value, + and skip the property name and <literal>=</literal>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>-l</option></term> + <term><option>--full</option></term> + + <listitem><para>Do not ellipsize process tree entries.</para> + </listitem> + </varlistentry> + + <varlistentry> + <term><option>--no-ask-password</option></term> + + <listitem><para>Do not query the user for authentication for + privileged operations.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--kill-who=</option></term> + + <listitem><para>When used with <command>kill</command>, choose + which processes to kill. Must be one of + <option>leader</option>, or <option>all</option> to select + whether to kill only the leader process of the machine or all + processes of the machine. If omitted, defaults to + <option>all</option>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>-s</option></term> + <term><option>--signal=</option></term> + + <listitem><para>When used with <command>kill</command>, choose + which signal to send to selected processes. Must be one of the + well-known signal specifiers, such as + <constant>SIGTERM</constant>, <constant>SIGINT</constant> or + <constant>SIGSTOP</constant>. If omitted, defaults to + <constant>SIGTERM</constant>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--uid=</option></term> + + <listitem><para>When used with the <command>shell</command> command, chooses the user ID to + open the interactive shell session as. If the argument to the <command>shell</command> + command also specifies a user name, this option is ignored. If the name is not specified + in either way, <literal>root</literal> will be used by default. Note that this switch is + not supported for the <command>login</command> command (see below).</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>-E <replaceable>NAME</replaceable>=<replaceable>VALUE</replaceable></option></term> + <term><option>--setenv=<replaceable>NAME</replaceable>=<replaceable>VALUE</replaceable></option></term> + + <listitem><para>When used with the <command>shell</command> command, sets an environment + variable to pass to the executed shell. Takes an environment variable name and value, + separated by <literal>=</literal>. This switch may be used multiple times to set multiple + environment variables. Note that this switch is not supported for the + <command>login</command> command (see below).</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--mkdir</option></term> + + <listitem><para>When used with <command>bind</command>, creates + the destination directory before applying the bind + mount.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--read-only</option></term> + + <listitem><para>When used with <command>bind</command>, applies + a read-only bind mount.</para> + + <para>When used with <command>clone</command>, <command>import-raw</command> or <command>import-tar</command> a + read-only container or VM image is created.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>-n</option></term> + <term><option>--lines=</option></term> + + <listitem><para>When used with <command>status</command>, + controls the number of journal lines to show, counting from + the most recent ones. Takes a positive integer argument. + Defaults to 10.</para> + </listitem> + </varlistentry> + + <varlistentry> + <term><option>-o</option></term> + <term><option>--output=</option></term> + + <listitem><para>When used with <command>status</command>, + controls the formatting of the journal entries that are shown. + For the available choices, see + <citerefentry><refentrytitle>journalctl</refentrytitle><manvolnum>1</manvolnum></citerefentry>. + Defaults to <literal>short</literal>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--verify=</option></term> + + <listitem><para>When downloading a container or VM image, + specify whether the image shall be verified before it is made + available. Takes one of <literal>no</literal>, + <literal>checksum</literal> and <literal>signature</literal>. + If <literal>no</literal>, no verification is done. If + <literal>checksum</literal> is specified, the download is + checked for integrity after the transfer is complete, but no + signatures are verified. If <literal>signature</literal> is + specified, the checksum is verified and the image's signature + is checked against a local keyring of trustable vendors. It is + strongly recommended to set this option to + <literal>signature</literal> if the server and protocol + support this. Defaults to + <literal>signature</literal>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--force</option></term> + + <listitem><para>When downloading a container or VM image, and + a local copy by the specified local machine name already + exists, delete it first and replace it by the newly downloaded + image.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--format=</option></term> + + <listitem><para>When used with the <option>export-tar</option> + or <option>export-raw</option> commands, specifies the + compression format to use for the resulting file. Takes one of + <literal>uncompressed</literal>, <literal>xz</literal>, + <literal>gzip</literal>, <literal>bzip2</literal>. By default, + the format is determined automatically from the image file + name passed.</para></listitem> + </varlistentry> + + <varlistentry> + <term><option>--max-addresses=</option></term> + + <listitem><para>When used with the <option>list-machines</option> + command, limits the number of ip addresses output for every machine. + Defaults to 1. All addresses can be requested with <literal>all</literal> + as argument to <option>--max-addresses</option> . If the argument to + <option>--max-addresses</option> is less than the actual number + of addresses,<literal>...</literal>follows the last address. + If multiple addresses are to be written for a given machine, every + address except the first one is on a new line and is followed by + <literal>,</literal> if another address will be output afterwards. </para></listitem> + </varlistentry> + + <xi:include href="user-system-options.xml" xpointer="host" /> + <xi:include href="user-system-options.xml" xpointer="machine" /> + + <xi:include href="standard-options.xml" xpointer="no-pager" /> + <xi:include href="standard-options.xml" xpointer="no-legend" /> + <xi:include href="standard-options.xml" xpointer="help" /> + <xi:include href="standard-options.xml" xpointer="version" /> + </variablelist> + </refsect1> + + <refsect1> + <title>Commands</title> + + <para>The following commands are understood:</para> + + <refsect2><title>Machine Commands</title><variablelist> + + <varlistentry> + <term><command>list</command></term> + + <listitem><para>List currently running (online) virtual + machines and containers. To enumerate machine images that can + be started, use <command>list-images</command> (see + below). Note that this command hides the special + <literal>.host</literal> machine by default. Use the + <option>--all</option> switch to show it.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>status</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Show runtime status information about + one or more virtual machines and containers, followed by the + most recent log data from the journal. This function is + intended to generate human-readable output. If you are looking + for computer-parsable output, use <command>show</command> + instead. Note that the log data shown is reported by the + virtual machine or container manager, and frequently contains + console output of the machine, but not necessarily journal + contents of the machine itself.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>show</command> [<replaceable>NAME</replaceable>...]</term> + + <listitem><para>Show properties of one or more registered virtual machines or containers or the manager + itself. If no argument is specified, properties of the manager will be shown. If a NAME is specified, + properties of this virtual machine or container are shown. By default, empty properties are suppressed. Use + <option>--all</option> to show those too. To select specific properties to show, use + <option>--property=</option>. This command is intended to be used whenever computer-parsable output is + required, and does not print the control group tree or journal entries. Use <command>status</command> if you + are looking for formatted human-readable output.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>start</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Start a container as a system service, using + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry>. + This starts <filename>systemd-nspawn@.service</filename>, + instantiated for the specified machine name, similar to the + effect of <command>systemctl start</command> on the service + name. <command>systemd-nspawn</command> looks for a container + image by the specified name in + <filename>/var/lib/machines/</filename> (and other search + paths, see below) and runs it. Use + <command>list-images</command> (see below) for listing + available container images to start.</para> + + <para>Note that + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry> + also interfaces with a variety of other container and VM + managers, <command>systemd-nspawn</command> is just one + implementation of it. Most of the commands available in + <command>machinectl</command> may be used on containers or VMs + controlled by other managers, not just + <command>systemd-nspawn</command>. Starting VMs and container + images on those managers requires manager-specific + tools.</para> + + <para>To interactively start a container on the command line + with full access to the container's console, please invoke + <command>systemd-nspawn</command> directly. To stop a running + container use <command>machinectl poweroff</command>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>login</command> [<replaceable>NAME</replaceable>]</term> + + <listitem><para>Open an interactive terminal login session in + a container or on the local host. If an argument is supplied, + it refers to the container machine to connect to. If none is + specified, or the container name is specified as the empty + string, or the special machine name <literal>.host</literal> + (see below) is specified, the connection is made to the local + host instead. This will create a TTY connection to a specific + container or the local host and asks for the execution of a + getty on it. Note that this is only supported for containers + running + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry> + as init system.</para> + + <para>This command will open a full login prompt on the + container or the local host, which then asks for username and + password. Use <command>shell</command> (see below) or + <citerefentry><refentrytitle>systemd-run</refentrytitle><manvolnum>1</manvolnum></citerefentry> + with the <option>--machine=</option> switch to directly invoke + a single command, either interactively or in the + background.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>shell</command> [[<replaceable>NAME</replaceable>@]<replaceable>NAME</replaceable> [<replaceable>PATH</replaceable> [<replaceable>ARGUMENTS</replaceable>...]]] </term> + + <listitem><para>Open an interactive shell session in a + container or on the local host. The first argument refers to + the container machine to connect to. If none is specified, or + the machine name is specified as the empty string, or the + special machine name <literal>.host</literal> (see below) is + specified, the connection is made to the local host + instead. This works similar to <command>login</command> but + immediately invokes a user process. This command runs the + specified executable with the specified arguments, or + <filename>/bin/sh</filename> if none is specified. By default, + opens a <literal>root</literal> shell, but by using + <option>--uid=</option>, or by prefixing the machine name with + a username and an <literal>@</literal> character, a different + user may be selected. Use <option>--setenv=</option> to set + environment variables for the executed process.</para> + + <para>When using the <command>shell</command> command without + arguments, (thus invoking the executed shell or command on the + local host), it is in many ways similar to a <citerefentry + project='die-net'><refentrytitle>su</refentrytitle><manvolnum>1</manvolnum></citerefentry> + session, but, unlike <command>su</command>, completely isolates + the new session from the originating session, so that it + shares no process or session properties, and is in a clean and + well-defined state. It will be tracked in a new utmp, login, + audit, security and keyring session, and will not inherit any + environment variables or resource limits, among other + properties.</para> + + <para>Note that + <citerefentry><refentrytitle>systemd-run</refentrytitle><manvolnum>1</manvolnum></citerefentry> + may be used in place of the <command>shell</command> command, + and allows more detailed, low-level configuration of the + invoked unit. However, it is frequently more privileged than + the <command>shell</command> command.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>enable</command> <replaceable>NAME</replaceable>...</term> + <term><command>disable</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Enable or disable a container as a system + service to start at system boot, using + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry>. + This enables or disables + <filename>systemd-nspawn@.service</filename>, instantiated for + the specified machine name, similar to the effect of + <command>systemctl enable</command> or <command>systemctl + disable</command> on the service name.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>poweroff</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Power off one or more containers. This will + trigger a reboot by sending SIGRTMIN+4 to the container's init + process, which causes systemd-compatible init systems to shut + down cleanly. Use <command>stop</command> as alias for <command>poweroff</command>. + This operation does not work on containers that do not run a + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>-compatible + init system, such as sysvinit. Use + <command>terminate</command> (see below) to immediately + terminate a container or VM, without cleanly shutting it + down.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>reboot</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Reboot one or more containers. This will + trigger a reboot by sending SIGINT to the container's init + process, which is roughly equivalent to pressing Ctrl+Alt+Del + on a non-containerized system, and is compatible with + containers running any system manager.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>terminate</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Immediately terminates a virtual machine or + container, without cleanly shutting it down. This kills all + processes of the virtual machine or container and deallocates + all resources attached to that instance. Use + <command>poweroff</command> to issue a clean shutdown + request.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>kill</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Send a signal to one or more processes of the + virtual machine or container. This means processes as seen by + the host, not the processes inside the virtual machine or + container. Use <option>--kill-who=</option> to select which + process to kill. Use <option>--signal=</option> to select the + signal to send.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>bind</command> <replaceable>NAME</replaceable> <replaceable>PATH</replaceable> [<replaceable>PATH</replaceable>]</term> + + <listitem><para>Bind mounts a directory from the host into the + specified container. The first directory argument is the + source directory on the host, the second directory argument + is the destination directory in the container. When the + latter is omitted, the destination path in the container is + the same as the source path on the host. When combined with + the <option>--read-only</option> switch, a ready-only bind + mount is created. When combined with the + <option>--mkdir</option> switch, the destination path is first + created before the mount is applied. Note that this option is + currently only supported for + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + containers.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>copy-to</command> <replaceable>NAME</replaceable> <replaceable>PATH</replaceable> [<replaceable>PATH</replaceable>]</term> + + <listitem><para>Copies files or directories from the host + system into a running container. Takes a container name, + followed by the source path on the host and the destination + path in the container. If the destination path is omitted, the + same as the source path is used.</para></listitem> + </varlistentry> + + + <varlistentry> + <term><command>copy-from</command> <replaceable>NAME</replaceable> <replaceable>PATH</replaceable> [<replaceable>PATH</replaceable>]</term> + + <listitem><para>Copies files or directories from a container + into the host system. Takes a container name, followed by the + source path in the container the destination path on the host. + If the destination path is omitted, the same as the source path + is used.</para></listitem> + </varlistentry> + </variablelist></refsect2> + + <refsect2><title>Image Commands</title><variablelist> + + <varlistentry> + <term><command>list-images</command></term> + + <listitem><para>Show a list of locally installed container and + VM images. This enumerates all raw disk images and container + directories and subvolumes in + <filename>/var/lib/machines/</filename> (and other search + paths, see below). Use <command>start</command> (see above) to + run a container off one of the listed images. Note that, by + default, containers whose name begins with a dot + (<literal>.</literal>) are not shown. To show these too, + specify <option>--all</option>. Note that a special image + <literal>.host</literal> always implicitly exists and refers + to the image the host itself is booted from.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>image-status</command> [<replaceable>NAME</replaceable>...]</term> + + <listitem><para>Show terse status information about one or + more container or VM images. This function is intended to + generate human-readable output. Use + <command>show-image</command> (see below) to generate + computer-parsable output instead.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>show-image</command> [<replaceable>NAME</replaceable>...]</term> + + <listitem><para>Show properties of one or more registered + virtual machine or container images, or the manager itself. If + no argument is specified, properties of the manager will be + shown. If a NAME is specified, properties of this virtual + machine or container image are shown. By default, empty + properties are suppressed. Use <option>--all</option> to show + those too. To select specific properties to show, use + <option>--property=</option>. This command is intended to be + used whenever computer-parsable output is required. Use + <command>image-status</command> if you are looking for + formatted human-readable output.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>clone</command> <replaceable>NAME</replaceable> <replaceable>NAME</replaceable></term> + + <listitem><para>Clones a container or VM image. The arguments specify the name of the image to clone and the + name of the newly cloned image. Note that plain directory container images are cloned into btrfs subvolume + images with this command, if the underlying file system supports this. Note that cloning a container or VM + image is optimized for btrfs file systems, and might not be efficient on others, due to file system + limitations.</para> + + <para>Note that this command leaves host name, machine ID and + all other settings that could identify the instance + unmodified. The original image and the cloned copy will hence + share these credentials, and it might be necessary to manually + change them in the copy.</para> + + <para>If combined with the <option>--read-only</option> switch a read-only cloned image is + created.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>rename</command> <replaceable>NAME</replaceable> <replaceable>NAME</replaceable></term> + + <listitem><para>Renames a container or VM image. The + arguments specify the name of the image to rename and the new + name of the image.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>read-only</command> <replaceable>NAME</replaceable> [<replaceable>BOOL</replaceable>]</term> + + <listitem><para>Marks or (unmarks) a container or VM image + read-only. Takes a VM or container image name, followed by a + boolean as arguments. If the boolean is omitted, positive is + implied, i.e. the image is marked read-only.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>remove</command> <replaceable>NAME</replaceable>...</term> + + <listitem><para>Removes one or more container or VM images. + The special image <literal>.host</literal>, which refers to + the host's own directory tree, may not be + removed.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>set-limit</command> [<replaceable>NAME</replaceable>] <replaceable>BYTES</replaceable></term> + + <listitem><para>Sets the maximum size in bytes that a specific + container or VM image, or all images, may grow up to on disk + (disk quota). Takes either one or two parameters. The first, + optional parameter refers to a container or VM image name. If + specified, the size limit of the specified image is changed. If + omitted, the overall size limit of the sum of all images stored + locally is changed. The final argument specifies the size + limit in bytes, possibly suffixed by the usual K, M, G, T + units. If the size limit shall be disabled, specify + <literal>-</literal> as size.</para> + + <para>Note that per-container size limits are only supported + on btrfs file systems. Also note that, if + <command>set-limit</command> is invoked without an image + parameter, and <filename>/var/lib/machines</filename> is + empty, and the directory is not located on btrfs, a btrfs + loopback file is implicitly created as + <filename>/var/lib/machines.raw</filename> with the given + size, and mounted to + <filename>/var/lib/machines</filename>. The size of the + loopback may later be readjusted with + <command>set-limit</command>, as well. If such a + loopback-mounted <filename>/var/lib/machines</filename> + directory is used, <command>set-limit</command> without an image + name alters both the quota setting within the file system as + well as the loopback file and file system size + itself.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>clean</command></term> + + <listitem><para>Remove hidden VM or container images (or all). This command removes all hidden machine images + from <filename>/var/lib/machines</filename>, i.e. those whose name begins with a dot. Use <command>machinectl + list-images --all</command> to see a list of all machine images, including the hidden ones.</para> + + <para>When combined with the <option>--all</option> switch removes all images, not just hidden ones. This + command effectively empties <filename>/var/lib/machines</filename>.</para> + + <para>Note that commands such as <command>machinectl pull-tar</command> or <command>machinectl + pull-raw</command> usually create hidden, read-only, unmodified machine images from the downloaded image first, + before cloning a writable working copy of it, in order to avoid duplicate downloads in case of images that are + reused multiple times. Use <command>machinectl clean</command> to remove old, hidden images created this + way.</para></listitem> + </varlistentry> + + </variablelist></refsect2> + + <refsect2><title>Image Transfer Commands</title><variablelist> + + <varlistentry> + <term><command>pull-tar</command> <replaceable>URL</replaceable> [<replaceable>NAME</replaceable>]</term> + + <listitem><para>Downloads a <filename>.tar</filename> + container image from the specified URL, and makes it available + under the specified local machine name. The URL must be of + type <literal>http://</literal> or + <literal>https://</literal>, and must refer to a + <filename>.tar</filename>, <filename>.tar.gz</filename>, + <filename>.tar.xz</filename> or <filename>.tar.bz2</filename> + archive file. If the local machine name is omitted, it + is automatically derived from the last component of the URL, + with its suffix removed.</para> + + <para>The image is verified before it is made available, + unless <option>--verify=no</option> is specified. Verification + is done via SHA256SUMS and SHA256SUMS.gpg files that need to + be made available on the same web server, under the same URL + as the <filename>.tar</filename> file, but with the last + component (the filename) of the URL replaced. With + <option>--verify=checksum</option>, only the SHA256 checksum + for the file is verified, based on the + <filename>SHA256SUMS</filename> file. With + <option>--verify=signature</option>, the SHA256SUMS file is + first verified with detached GPG signature file + <filename>SHA256SUMS.gpg</filename>. The public key for this + verification step needs to be available in + <filename>/usr/lib/systemd/import-pubring.gpg</filename> or + <filename>/etc/systemd/import-pubring.gpg</filename>.</para> + + <para>The container image will be downloaded and stored in a + read-only subvolume in + <filename>/var/lib/machines/</filename> that is named after + the specified URL and its HTTP etag. A writable snapshot is + then taken from this subvolume, and named after the specified + local name. This behavior ensures that creating multiple + container instances of the same URL is efficient, as multiple + downloads are not necessary. In order to create only the + read-only image, and avoid creating its writable snapshot, + specify <literal>-</literal> as local machine name.</para> + + <para>Note that the read-only subvolume is prefixed with + <filename>.tar-</filename>, and is thus not shown by + <command>list-images</command>, unless <option>--all</option> + is passed.</para> + + <para>Note that pressing C-c during execution of this command + will not abort the download. Use + <command>cancel-transfer</command>, described + below.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>pull-raw</command> <replaceable>URL</replaceable> [<replaceable>NAME</replaceable>]</term> + + <listitem><para>Downloads a <filename>.raw</filename> + container or VM disk image from the specified URL, and makes + it available under the specified local machine name. The URL + must be of type <literal>http://</literal> or + <literal>https://</literal>. The container image must either + be a <filename>.qcow2</filename> or raw disk image, optionally + compressed as <filename>.gz</filename>, + <filename>.xz</filename>, or <filename>.bz2</filename>. If the + local machine name is omitted, it is automatically + derived from the last component of the URL, with its suffix + removed.</para> + + <para>Image verification is identical for raw and tar images + (see above).</para> + + <para>If the downloaded image is in + <filename>.qcow2</filename> format it is converted into a raw + image file before it is made available.</para> + + <para>Downloaded images of this type will be placed as + read-only <filename>.raw</filename> file in + <filename>/var/lib/machines/</filename>. A local, writable + (reflinked) copy is then made under the specified local + machine name. To omit creation of the local, writable copy + pass <literal>-</literal> as local machine name.</para> + + <para>Similar to the behavior of <command>pull-tar</command>, + the read-only image is prefixed with + <filename>.raw-</filename>, and thus not shown by + <command>list-images</command>, unless <option>--all</option> + is passed.</para> + + <para>Note that pressing C-c during execution of this command + will not abort the download. Use + <command>cancel-transfer</command>, described + below.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>import-tar</command> <replaceable>FILE</replaceable> [<replaceable>NAME</replaceable>]</term> + <term><command>import-raw</command> <replaceable>FILE</replaceable> [<replaceable>NAME</replaceable>]</term> + <listitem><para>Imports a TAR or RAW container or VM image, + and places it under the specified name in + <filename>/var/lib/machines/</filename>. When + <command>import-tar</command> is used, the file specified as + the first argument should be a tar archive, possibly compressed + with xz, gzip or bzip2. It will then be unpacked into its own + subvolume in <filename>/var/lib/machines</filename>. When + <command>import-raw</command> is used, the file should be a + qcow2 or raw disk image, possibly compressed with xz, gzip or + bzip2. If the second argument (the resulting image name) is + not specified, it is automatically derived from the file + name. If the file name is passed as <literal>-</literal>, the + image is read from standard input, in which case the second + argument is mandatory.</para> + + <para>Both <command>pull-tar</command> and <command>pull-raw</command> + will resize <filename>/var/lib/machines.raw</filename> and the + filesystem therein as necessary. Optionally, the + <option>--read-only</option> switch may be used to create a + read-only container or VM image. No cryptographic validation + is done when importing the images.</para> + + <para>Much like image downloads, ongoing imports may be listed + with <command>list-transfers</command> and aborted with + <command>cancel-transfer</command>.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>export-tar</command> <replaceable>NAME</replaceable> [<replaceable>FILE</replaceable>]</term> + <term><command>export-raw</command> <replaceable>NAME</replaceable> [<replaceable>FILE</replaceable>]</term> + <listitem><para>Exports a TAR or RAW container or VM image and + stores it in the specified file. The first parameter should be + a VM or container image name. The second parameter should be a + file path the TAR or RAW image is written to. If the path ends + in <literal>.gz</literal>, the file is compressed with gzip, if + it ends in <literal>.xz</literal>, with xz, and if it ends in + <literal>.bz2</literal>, with bzip2. If the path ends in + neither, the file is left uncompressed. If the second argument + is missing, the image is written to standard output. The + compression may also be explicitly selected with the + <option>--format=</option> switch. This is in particular + useful if the second parameter is left unspecified.</para> + + <para>Much like image downloads and imports, ongoing exports + may be listed with <command>list-transfers</command> and + aborted with + <command>cancel-transfer</command>.</para> + + <para>Note that, currently, only directory and subvolume images + may be exported as TAR images, and only raw disk images as RAW + images.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>list-transfers</command></term> + + <listitem><para>Shows a list of container or VM image + downloads, imports and exports that are currently in + progress.</para></listitem> + </varlistentry> + + <varlistentry> + <term><command>cancel-transfers</command> <replaceable>ID</replaceable>...</term> + + <listitem><para>Aborts a download, import or export of the + container or VM image with the specified ID. To list ongoing + transfers and their IDs, use + <command>list-transfers</command>. </para></listitem> + </varlistentry> + + </variablelist></refsect2> + + </refsect1> + + <refsect1> + <title>Machine and Image Names</title> + + <para>The <command>machinectl</command> tool operates on machines + and images whose names must be chosen following strict + rules. Machine names must be suitable for use as host names + following a conservative subset of DNS and UNIX/Linux + semantics. Specifically, they must consist of one or more + non-empty label strings, separated by dots. No leading or trailing + dots are allowed. No sequences of multiple dots are allowed. The + label strings may only consist of alphanumeric characters as well + as the dash and underscore. The maximum length of a machine name + is 64 characters.</para> + + <para>A special machine with the name <literal>.host</literal> + refers to the running host system itself. This is useful for execution + operations or inspecting the host system as well. Note that + <command>machinectl list</command> will not show this special + machine unless the <option>--all</option> switch is specified.</para> + + <para>Requirements on image names are less strict, however, they must be + valid UTF-8, must be suitable as file names (hence not be the + single or double dot, and not include a slash), and may not + contain control characters. Since many operations search for an + image by the name of a requested machine, it is recommended to name + images in the same strict fashion as machines.</para> + + <para>A special image with the name <literal>.host</literal> + refers to the image of the running host system. It hence + conceptually maps to the special <literal>.host</literal> machine + name described above. Note that <command>machinectl + list-images</command> will not show this special image either, unless + <option>--all</option> is specified.</para> + </refsect1> + + <refsect1> + <title>Files and Directories</title> + + <para>Machine images are preferably stored in + <filename>/var/lib/machines/</filename>, but are also searched for + in <filename>/usr/local/lib/machines/</filename> and + <filename>/usr/lib/machines/</filename>. For compatibility reasons, + the directory <filename>/var/lib/container/</filename> is + searched, too. Note that images stored below + <filename>/usr</filename> are always considered read-only. It is + possible to symlink machines images from other directories into + <filename>/var/lib/machines/</filename> to make them available for + control with <command>machinectl</command>.</para> + + <para>Note that many image operations are only supported, + efficient or atomic on btrfs file systems. Due to this, if the + <command>pull-tar</command>, <command>pull-raw</command>, + <command>import-tar</command>, <command>import-raw</command> and + <command>set-limit</command> commands notice that + <filename>/var/lib/machines</filename> is empty and not located on + btrfs, they will implicitly set up a loopback file + <filename>/var/lib/machines.raw</filename> containing a btrfs file + system that is mounted to + <filename>/var/lib/machines</filename>. The size of this loopback + file may be controlled dynamically with + <command>set-limit</command>.</para> + + <para>Disk images are understood by + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + and <command>machinectl</command> in three formats:</para> + + <itemizedlist> + <listitem><para>A simple directory tree, containing the files + and directories of the container to boot.</para></listitem> + + <listitem><para>Subvolumes (on btrfs file systems), which are + similar to the simple directories, described above. However, + they have additional benefits, such as efficient cloning and + quota reporting.</para></listitem> + + <listitem><para>"Raw" disk images, i.e. binary images of disks + with a GPT or MBR partition table. Images of this type are + regular files with the suffix + <literal>.raw</literal>.</para></listitem> + </itemizedlist> + + <para>See + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + for more information on image formats, in particular its + <option>--directory=</option> and <option>--image=</option> + options.</para> + </refsect1> + + <refsect1> + <title>Examples</title> + <example> + <title>Download an Ubuntu image and open a shell in it</title> + + <programlisting># machinectl pull-tar https://cloud-images.ubuntu.com/trusty/current/trusty-server-cloudimg-amd64-root.tar.gz +# systemd-nspawn -M trusty-server-cloudimg-amd64-root</programlisting> + + <para>This downloads and verifies the specified + <filename>.tar</filename> image, and then uses + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + to open a shell in it.</para> + </example> + + <example> + <title>Download a Fedora image, set a root password in it, start + it as service</title> + + <programlisting># machinectl pull-raw --verify=no https://dl.fedoraproject.org/pub/fedora/linux/releases/23/Cloud/x86_64/Images/Fedora-Cloud-Base-23-20151030.x86_64.raw.xz +# systemd-nspawn -M Fedora-Cloud-Base-23-20151030 +# passwd +# exit +# machinectl start Fedora-Cloud-Base-23-20151030 +# machinectl login Fedora-Cloud-Base-23-20151030</programlisting> + + <para>This downloads the specified <filename>.raw</filename> + image with verification disabled. Then, a shell is opened in it + and a root password is set. Afterwards the shell is left, and + the machine started as system service. With the last command a + login prompt into the container is requested.</para> + </example> + + <example> + <title>Exports a container image as tar file</title> + + <programlisting># machinectl export-tar fedora myfedora.tar.xz</programlisting> + + <para>Exports the container <literal>fedora</literal> as an + xz-compressed tar file <filename>myfedora.tar.xz</filename> into the + current directory.</para> + </example> + + <example> + <title>Create a new shell session</title> + + <programlisting># machinectl shell --uid=lennart</programlisting> + + <para>This creates a new shell session on the local host for + the user ID <literal>lennart</literal>, in a <citerefentry + project='die-net'><refentrytitle>su</refentrytitle><manvolnum>1</manvolnum></citerefentry>-like + fashion.</para> + </example> + + </refsect1> + + <refsect1> + <title>Exit status</title> + + <para>On success, 0 is returned, a non-zero failure code + otherwise.</para> + </refsect1> + + <xi:include href="less-variables.xml" /> + + <refsect1> + <title>See Also</title> + <para> + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd.special</refentrytitle><manvolnum>7</manvolnum></citerefentry>, + <citerefentry project='die-net'><refentrytitle>tar</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry project='die-net'><refentrytitle>xz</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry project='die-net'><refentrytitle>gzip</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry project='die-net'><refentrytitle>bzip2</refentrytitle><manvolnum>1</manvolnum></citerefentry> + </para> + </refsect1> + +</refentry> diff --git a/src/grp-machine/nss-mymachines/GNUmakefile b/src/grp-machine/nss-mymachines/GNUmakefile new file mode 120000 index 0000000000..95e5924740 --- /dev/null +++ b/src/grp-machine/nss-mymachines/GNUmakefile @@ -0,0 +1 @@ +../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/nss-mymachines/Makefile b/src/grp-machine/nss-mymachines/Makefile new file mode 100644 index 0000000000..5f8898b1a0 --- /dev/null +++ b/src/grp-machine/nss-mymachines/Makefile @@ -0,0 +1,45 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +libnss_mymachines_la_SOURCES = \ + src/nss-mymachines/nss-mymachines.sym \ + src/nss-mymachines/nss-mymachines.c + +libnss_mymachines_la_LDFLAGS = \ + -module \ + -export-dynamic \ + -avoid-version \ + -shared \ + -shrext .so.2 \ + -Wl,--version-script=$(srcdir)/nss-mymachines.sym + +libnss_mymachines_la_LIBADD = \ + libsystemd-internal.la \ + libsystemd-basic.la + +rootlib_LTLIBRARIES += \ + libnss_mymachines.la + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/nss-mymachines/nss-mymachines.c b/src/grp-machine/nss-mymachines/nss-mymachines.c new file mode 100644 index 0000000000..18de097746 --- /dev/null +++ b/src/grp-machine/nss-mymachines/nss-mymachines.c @@ -0,0 +1,754 @@ +/*** + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <netdb.h> +#include <nss.h> + +#include <systemd/sd-bus.h> +#include <systemd/sd-login.h> + +#include "sd-bus/bus-common-errors.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/env-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/in-addr-util.h" +#include "systemd-basic/macro.h" +#include "systemd-basic/nss-util.h" +#include "systemd-basic/signal-util.h" +#include "systemd-basic/string-util.h" +#include "systemd-basic/user-util.h" +#include "systemd-basic/util.h" + +NSS_GETHOSTBYNAME_PROTOTYPES(mymachines); +NSS_GETPW_PROTOTYPES(mymachines); +NSS_GETGR_PROTOTYPES(mymachines); + +#define HOST_UID_LIMIT ((uid_t) UINT32_C(0x10000)) +#define HOST_GID_LIMIT ((gid_t) UINT32_C(0x10000)) + +static int count_addresses(sd_bus_message *m, int af, unsigned *ret) { + unsigned c = 0; + int r; + + assert(m); + assert(ret); + + while ((r = sd_bus_message_enter_container(m, 'r', "iay")) > 0) { + int family; + + r = sd_bus_message_read(m, "i", &family); + if (r < 0) + return r; + + r = sd_bus_message_skip(m, "ay"); + if (r < 0) + return r; + + r = sd_bus_message_exit_container(m); + if (r < 0) + return r; + + if (af != AF_UNSPEC && family != af) + continue; + + c++; + } + if (r < 0) + return r; + + r = sd_bus_message_rewind(m, false); + if (r < 0) + return r; + + *ret = c; + return 0; +} + +enum nss_status _nss_mymachines_gethostbyname4_r( + const char *name, + struct gaih_addrtuple **pat, + char *buffer, size_t buflen, + int *errnop, int *h_errnop, + int32_t *ttlp) { + + struct gaih_addrtuple *r_tuple, *r_tuple_first = NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + _cleanup_free_ int *ifindices = NULL; + _cleanup_free_ char *class = NULL; + size_t l, ms, idx; + unsigned i = 0, c = 0; + char *r_name; + int n_ifindices, r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + assert(name); + assert(pat); + assert(buffer); + assert(errnop); + assert(h_errnop); + + r = sd_machine_get_class(name, &class); + if (r < 0) + goto fail; + if (!streq(class, "container")) { + r = -ENOTTY; + goto fail; + } + + n_ifindices = sd_machine_get_ifindices(name, &ifindices); + if (n_ifindices < 0) { + r = n_ifindices; + goto fail; + } + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetMachineAddresses", + NULL, + &reply, + "s", name); + if (r < 0) + goto fail; + + r = sd_bus_message_enter_container(reply, 'a', "(iay)"); + if (r < 0) + goto fail; + + r = count_addresses(reply, AF_UNSPEC, &c); + if (r < 0) + goto fail; + + if (c <= 0) { + *errnop = ESRCH; + *h_errnop = HOST_NOT_FOUND; + return NSS_STATUS_NOTFOUND; + } + + l = strlen(name); + ms = ALIGN(l+1) + ALIGN(sizeof(struct gaih_addrtuple)) * c; + if (buflen < ms) { + *errnop = ENOMEM; + *h_errnop = TRY_AGAIN; + return NSS_STATUS_TRYAGAIN; + } + + /* First, append name */ + r_name = buffer; + memcpy(r_name, name, l+1); + idx = ALIGN(l+1); + + /* Second, append addresses */ + r_tuple_first = (struct gaih_addrtuple*) (buffer + idx); + while ((r = sd_bus_message_enter_container(reply, 'r', "iay")) > 0) { + int family; + const void *a; + size_t sz; + + r = sd_bus_message_read(reply, "i", &family); + if (r < 0) + goto fail; + + r = sd_bus_message_read_array(reply, 'y', &a, &sz); + if (r < 0) + goto fail; + + r = sd_bus_message_exit_container(reply); + if (r < 0) + goto fail; + + if (!IN_SET(family, AF_INET, AF_INET6)) { + r = -EAFNOSUPPORT; + goto fail; + } + + if (sz != FAMILY_ADDRESS_SIZE(family)) { + r = -EINVAL; + goto fail; + } + + r_tuple = (struct gaih_addrtuple*) (buffer + idx); + r_tuple->next = i == c-1 ? NULL : (struct gaih_addrtuple*) ((char*) r_tuple + ALIGN(sizeof(struct gaih_addrtuple))); + r_tuple->name = r_name; + r_tuple->family = family; + r_tuple->scopeid = n_ifindices == 1 ? ifindices[0] : 0; + memcpy(r_tuple->addr, a, sz); + + idx += ALIGN(sizeof(struct gaih_addrtuple)); + i++; + } + + assert(i == c); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + goto fail; + + assert(idx == ms); + + if (*pat) + **pat = *r_tuple_first; + else + *pat = r_tuple_first; + + if (ttlp) + *ttlp = 0; + + /* Explicitly reset all error variables */ + *errnop = 0; + *h_errnop = NETDB_SUCCESS; + h_errno = 0; + + return NSS_STATUS_SUCCESS; + +fail: + *errnop = -r; + *h_errnop = NO_DATA; + return NSS_STATUS_UNAVAIL; +} + +enum nss_status _nss_mymachines_gethostbyname3_r( + const char *name, + int af, + struct hostent *result, + char *buffer, size_t buflen, + int *errnop, int *h_errnop, + int32_t *ttlp, + char **canonp) { + + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + _cleanup_free_ char *class = NULL; + unsigned c = 0, i = 0; + char *r_name, *r_aliases, *r_addr, *r_addr_list; + size_t l, idx, ms, alen; + int r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + assert(name); + assert(result); + assert(buffer); + assert(errnop); + assert(h_errnop); + + if (af == AF_UNSPEC) + af = AF_INET; + + if (af != AF_INET && af != AF_INET6) { + r = -EAFNOSUPPORT; + goto fail; + } + + r = sd_machine_get_class(name, &class); + if (r < 0) + goto fail; + if (!streq(class, "container")) { + r = -ENOTTY; + goto fail; + } + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "GetMachineAddresses", + NULL, + &reply, + "s", name); + if (r < 0) + goto fail; + + r = sd_bus_message_enter_container(reply, 'a', "(iay)"); + if (r < 0) + goto fail; + + r = count_addresses(reply, af, &c); + if (r < 0) + goto fail; + + if (c <= 0) { + *errnop = ENOENT; + *h_errnop = HOST_NOT_FOUND; + return NSS_STATUS_NOTFOUND; + } + + alen = FAMILY_ADDRESS_SIZE(af); + l = strlen(name); + + ms = ALIGN(l+1) + c * ALIGN(alen) + (c+2) * sizeof(char*); + + if (buflen < ms) { + *errnop = ENOMEM; + *h_errnop = NO_RECOVERY; + return NSS_STATUS_TRYAGAIN; + } + + /* First, append name */ + r_name = buffer; + memcpy(r_name, name, l+1); + idx = ALIGN(l+1); + + /* Second, create aliases array */ + r_aliases = buffer + idx; + ((char**) r_aliases)[0] = NULL; + idx += sizeof(char*); + + /* Third, append addresses */ + r_addr = buffer + idx; + while ((r = sd_bus_message_enter_container(reply, 'r', "iay")) > 0) { + int family; + const void *a; + size_t sz; + + r = sd_bus_message_read(reply, "i", &family); + if (r < 0) + goto fail; + + r = sd_bus_message_read_array(reply, 'y', &a, &sz); + if (r < 0) + goto fail; + + r = sd_bus_message_exit_container(reply); + if (r < 0) + goto fail; + + if (family != af) + continue; + + if (sz != alen) { + r = -EINVAL; + goto fail; + } + + memcpy(r_addr + i*ALIGN(alen), a, alen); + i++; + } + + assert(i == c); + idx += c * ALIGN(alen); + + r = sd_bus_message_exit_container(reply); + if (r < 0) + goto fail; + + /* Third, append address pointer array */ + r_addr_list = buffer + idx; + for (i = 0; i < c; i++) + ((char**) r_addr_list)[i] = r_addr + i*ALIGN(alen); + + ((char**) r_addr_list)[i] = NULL; + idx += (c+1) * sizeof(char*); + + assert(idx == ms); + + result->h_name = r_name; + result->h_aliases = (char**) r_aliases; + result->h_addrtype = af; + result->h_length = alen; + result->h_addr_list = (char**) r_addr_list; + + if (ttlp) + *ttlp = 0; + + if (canonp) + *canonp = r_name; + + /* Explicitly reset all error variables */ + *errnop = 0; + *h_errnop = NETDB_SUCCESS; + h_errno = 0; + + return NSS_STATUS_SUCCESS; + +fail: + *errnop = -r; + *h_errnop = NO_DATA; + return NSS_STATUS_UNAVAIL; +} + +NSS_GETHOSTBYNAME_FALLBACKS(mymachines); + +enum nss_status _nss_mymachines_getpwnam_r( + const char *name, + struct passwd *pwd, + char *buffer, size_t buflen, + int *errnop) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + const char *p, *e, *machine; + uint32_t mapped; + uid_t uid; + size_t l; + int r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + assert(name); + assert(pwd); + + p = startswith(name, "vu-"); + if (!p) + goto not_found; + + e = strrchr(p, '-'); + if (!e || e == p) + goto not_found; + + if (e - p > HOST_NAME_MAX - 1) /* -1 for the last dash */ + goto not_found; + + r = parse_uid(e + 1, &uid); + if (r < 0) + goto not_found; + + machine = strndupa(p, e - p); + if (!machine_name_is_valid(machine)) + goto not_found; + + if (getenv_bool("SYSTEMD_NSS_BYPASS_BUS") > 0) + /* Make sure we can't deadlock if we are invoked by dbus-daemon. This way, it won't be able to resolve + * these UIDs, but that should be unproblematic as containers should never be able to connect to a bus + * running on the host. */ + goto not_found; + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "MapFromMachineUser", + &error, + &reply, + "su", + machine, (uint32_t) uid); + if (r < 0) { + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_USER_MAPPING)) + goto not_found; + + goto fail; + } + + r = sd_bus_message_read(reply, "u", &mapped); + if (r < 0) + goto fail; + + /* Refuse to work if the mapped address is in the host UID range, or if there was no mapping at all. */ + if (mapped < HOST_UID_LIMIT || mapped == uid) + goto not_found; + + l = strlen(name); + if (buflen < l+1) { + *errnop = ENOMEM; + return NSS_STATUS_TRYAGAIN; + } + + memcpy(buffer, name, l+1); + + pwd->pw_name = buffer; + pwd->pw_uid = mapped; + pwd->pw_gid = 65534; /* nobody */ + pwd->pw_gecos = buffer; + pwd->pw_passwd = (char*) "*"; /* locked */ + pwd->pw_dir = (char*) "/"; + pwd->pw_shell = (char*) "/sbin/nologin"; + + *errnop = 0; + return NSS_STATUS_SUCCESS; + +not_found: + *errnop = 0; + return NSS_STATUS_NOTFOUND; + +fail: + *errnop = -r; + return NSS_STATUS_UNAVAIL; +} + +enum nss_status _nss_mymachines_getpwuid_r( + uid_t uid, + struct passwd *pwd, + char *buffer, size_t buflen, + int *errnop) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + const char *machine, *object; + uint32_t mapped; + int r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + if (!uid_is_valid(uid)) { + r = -EINVAL; + goto fail; + } + + /* We consider all uids < 65536 host uids */ + if (uid < HOST_UID_LIMIT) + goto not_found; + + if (getenv_bool("SYSTEMD_NSS_BYPASS_BUS") > 0) + goto not_found; + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "MapToMachineUser", + &error, + &reply, + "u", + (uint32_t) uid); + if (r < 0) { + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_USER_MAPPING)) + goto not_found; + + goto fail; + } + + r = sd_bus_message_read(reply, "sou", &machine, &object, &mapped); + if (r < 0) + goto fail; + + if (mapped == uid) + goto not_found; + + if (snprintf(buffer, buflen, "vu-%s-" UID_FMT, machine, (uid_t) mapped) >= (int) buflen) { + *errnop = ENOMEM; + return NSS_STATUS_TRYAGAIN; + } + + pwd->pw_name = buffer; + pwd->pw_uid = uid; + pwd->pw_gid = 65534; /* nobody */ + pwd->pw_gecos = buffer; + pwd->pw_passwd = (char*) "*"; /* locked */ + pwd->pw_dir = (char*) "/"; + pwd->pw_shell = (char*) "/sbin/nologin"; + + *errnop = 0; + return NSS_STATUS_SUCCESS; + +not_found: + *errnop = 0; + return NSS_STATUS_NOTFOUND; + +fail: + *errnop = -r; + return NSS_STATUS_UNAVAIL; +} + +enum nss_status _nss_mymachines_getgrnam_r( + const char *name, + struct group *gr, + char *buffer, size_t buflen, + int *errnop) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + const char *p, *e, *machine; + uint32_t mapped; + uid_t gid; + size_t l; + int r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + assert(name); + assert(gr); + + p = startswith(name, "vg-"); + if (!p) + goto not_found; + + e = strrchr(p, '-'); + if (!e || e == p) + goto not_found; + + if (e - p > HOST_NAME_MAX - 1) /* -1 for the last dash */ + goto not_found; + + r = parse_gid(e + 1, &gid); + if (r < 0) + goto not_found; + + machine = strndupa(p, e - p); + if (!machine_name_is_valid(machine)) + goto not_found; + + if (getenv_bool("SYSTEMD_NSS_BYPASS_BUS") > 0) + goto not_found; + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "MapFromMachineGroup", + &error, + &reply, + "su", + machine, (uint32_t) gid); + if (r < 0) { + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_GROUP_MAPPING)) + goto not_found; + + goto fail; + } + + r = sd_bus_message_read(reply, "u", &mapped); + if (r < 0) + goto fail; + + if (mapped < HOST_GID_LIMIT || mapped == gid) + goto not_found; + + l = sizeof(char*) + strlen(name) + 1; + if (buflen < l) { + *errnop = ENOMEM; + return NSS_STATUS_TRYAGAIN; + } + + memzero(buffer, sizeof(char*)); + strcpy(buffer + sizeof(char*), name); + + gr->gr_name = buffer + sizeof(char*); + gr->gr_gid = gid; + gr->gr_passwd = (char*) "*"; /* locked */ + gr->gr_mem = (char**) buffer; + + *errnop = 0; + return NSS_STATUS_SUCCESS; + +not_found: + *errnop = 0; + return NSS_STATUS_NOTFOUND; + +fail: + *errnop = -r; + return NSS_STATUS_UNAVAIL; +} + +enum nss_status _nss_mymachines_getgrgid_r( + gid_t gid, + struct group *gr, + char *buffer, size_t buflen, + int *errnop) { + + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_(sd_bus_message_unrefp) sd_bus_message* reply = NULL; + _cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL; + const char *machine, *object; + uint32_t mapped; + int r; + + BLOCK_SIGNALS(NSS_SIGNALS_BLOCK); + + if (!gid_is_valid(gid)) { + r = -EINVAL; + goto fail; + } + + /* We consider all gids < 65536 host gids */ + if (gid < HOST_GID_LIMIT) + goto not_found; + + if (getenv_bool("SYSTEMD_NSS_BYPASS_BUS") > 0) + goto not_found; + + r = sd_bus_open_system(&bus); + if (r < 0) + goto fail; + + r = sd_bus_call_method(bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "MapToMachineGroup", + &error, + &reply, + "u", + (uint32_t) gid); + if (r < 0) { + if (sd_bus_error_has_name(&error, BUS_ERROR_NO_SUCH_GROUP_MAPPING)) + goto not_found; + + goto fail; + } + + r = sd_bus_message_read(reply, "sou", &machine, &object, &mapped); + if (r < 0) + goto fail; + + if (mapped == gid) + goto not_found; + + if (buflen < sizeof(char*) + 1) { + *errnop = ENOMEM; + return NSS_STATUS_TRYAGAIN; + } + + memzero(buffer, sizeof(char*)); + if (snprintf(buffer + sizeof(char*), buflen - sizeof(char*), "vg-%s-" GID_FMT, machine, (gid_t) mapped) >= (int) buflen) { + *errnop = ENOMEM; + return NSS_STATUS_TRYAGAIN; + } + + gr->gr_name = buffer + sizeof(char*); + gr->gr_gid = gid; + gr->gr_passwd = (char*) "*"; /* locked */ + gr->gr_mem = (char**) buffer; + + *errnop = 0; + return NSS_STATUS_SUCCESS; + +not_found: + *errnop = 0; + return NSS_STATUS_NOTFOUND; + +fail: + *errnop = -r; + return NSS_STATUS_UNAVAIL; +} diff --git a/src/grp-machine/nss-mymachines/nss-mymachines.sym b/src/grp-machine/nss-mymachines/nss-mymachines.sym new file mode 100644 index 0000000000..0728ac3ba7 --- /dev/null +++ b/src/grp-machine/nss-mymachines/nss-mymachines.sym @@ -0,0 +1,21 @@ +/*** + This file is part of systemd. + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. +***/ + +{ +global: + _nss_mymachines_gethostbyname_r; + _nss_mymachines_gethostbyname2_r; + _nss_mymachines_gethostbyname3_r; + _nss_mymachines_gethostbyname4_r; + _nss_mymachines_getpwnam_r; + _nss_mymachines_getpwuid_r; + _nss_mymachines_getgrnam_r; + _nss_mymachines_getgrgid_r; +local: *; +}; diff --git a/src/grp-machine/nss-mymachines/nss-mymachines.xml b/src/grp-machine/nss-mymachines/nss-mymachines.xml new file mode 100644 index 0000000000..00bcc53ec0 --- /dev/null +++ b/src/grp-machine/nss-mymachines/nss-mymachines.xml @@ -0,0 +1,114 @@ +<?xml version='1.0'?> <!--*-nxml-*--> +<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" + "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"> + +<!-- + This file is part of systemd. + + Copyright 2014 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +--> + +<refentry id="nss-mymachines" conditional='ENABLE_MACHINED'> + + <refentryinfo> + <title>nss-mymachines</title> + <productname>systemd</productname> + + <authorgroup> + <author> + <contrib>Developer</contrib> + <firstname>Lennart</firstname> + <surname>Poettering</surname> + <email>lennart@poettering.net</email> + </author> + </authorgroup> + </refentryinfo> + + <refmeta> + <refentrytitle>nss-mymachines</refentrytitle> + <manvolnum>8</manvolnum> + </refmeta> + + <refnamediv> + <refname>nss-mymachines</refname> + <refname>libnss_mymachines.so.2</refname> + <refpurpose>Provide hostname resolution for local + container instances.</refpurpose> + </refnamediv> + + <refsynopsisdiv> + <para><filename>libnss_mymachines.so.2</filename></para> + </refsynopsisdiv> + + <refsect1> + <title>Description</title> + + <para><command>nss-mymachines</command> is a plug-in module for the GNU Name Service Switch (NSS) functionality of + the GNU C Library (<command>glibc</command>), providing hostname resolution for the names of containers running + locally that are registered with + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>. The + container names are resolved to the IP addresses of the specific container, ordered by their scope. This + functionality only applies to containers using network namespacing.</para> + + <para>The module also resolves user and group IDs used by containers to user and group names indicating the + container name, and back. This functionality only applies to containers using user namespacing.</para> + + <para>To activate the NSS module, add <literal>mymachines</literal> to the lines starting with + <literal>hosts:</literal>, <literal>passwd:</literal> and <literal>group:</literal> in + <filename>/etc/nsswitch.conf</filename>.</para> + + <para>It is recommended to place <literal>mymachines</literal> after the <literal>files</literal> or + <literal>compat</literal> entry of the <filename>/etc/nsswitch.conf</filename> lines to make sure that its mappings + are preferred over other resolvers such as DNS, but so that <filename>/etc/hosts</filename>, + <filename>/etc/passwd</filename> and <filename>/etc/group</filename> based mappings take precedence.</para> + </refsect1> + + <refsect1> + <title>Example</title> + + <para>Here is an example <filename>/etc/nsswitch.conf</filename> file that enables + <command>nss-mymachines</command> correctly:</para> + + <programlisting>passwd: compat <command>mymachines</command> systemd +group: compat <command>mymachines</command> systemd +shadow: compat + +hosts: files <command>mymachines</command> resolve [!UNAVAIL=return] dns myhostname +networks: files + +protocols: db files +services: db files +ethers: db files +rpc: db files + +netgroup: nis</programlisting> + + </refsect1> + + <refsect1> + <title>See Also</title> + <para> + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd-machined.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>nss-systemd</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>nss-resolve</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>nss-myhostname</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry project='man-pages'><refentrytitle>nsswitch.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>, + <citerefentry project='man-pages'><refentrytitle>getent</refentrytitle><manvolnum>1</manvolnum></citerefentry> + </para> + </refsect1> + +</refentry> diff --git a/src/grp-machine/systemd-machined/.gitignore b/src/grp-machine/systemd-machined/.gitignore new file mode 100644 index 0000000000..e1065b5894 --- /dev/null +++ b/src/grp-machine/systemd-machined/.gitignore @@ -0,0 +1 @@ +/org.freedesktop.machine1.policy diff --git a/src/grp-machine/systemd-machined/GNUmakefile b/src/grp-machine/systemd-machined/GNUmakefile new file mode 120000 index 0000000000..95e5924740 --- /dev/null +++ b/src/grp-machine/systemd-machined/GNUmakefile @@ -0,0 +1 @@ +../../../GNUmakefile
\ No newline at end of file diff --git a/src/grp-machine/systemd-machined/Makefile b/src/grp-machine/systemd-machined/Makefile new file mode 100644 index 0000000000..432a792a18 --- /dev/null +++ b/src/grp-machine/systemd-machined/Makefile @@ -0,0 +1,75 @@ +# -*- Mode: makefile; indent-tabs-mode: t -*- +# +# This file is part of systemd. +# +# Copyright 2010-2012 Lennart Poettering +# Copyright 2010-2012 Kay Sievers +# Copyright 2013 Zbigniew Jędrzejewski-Szmek +# Copyright 2013 David Strauss +# Copyright 2016 Luke Shumaker +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. +# +# systemd is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public License +# along with systemd; If not, see <http://www.gnu.org/licenses/>. +include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk +include $(topsrcdir)/build-aux/Makefile.head.mk + +systemd_machined_SOURCES = \ + src/machine/machined.c \ + src/machine/machined.h + +systemd_machined_LDADD = \ + libmachine-core.la + +rootlibexec_PROGRAMS += \ + systemd-machined + +nodist_systemunit_DATA += \ + units/systemd-machined.service + +dist_systemunit_DATA += \ + units/machine.slice + +dist_systemunit_DATA_busnames += \ + units/org.freedesktop.machine1.busname + +dist_dbussystemservice_DATA += \ + src/machine/org.freedesktop.machine1.service + +dist_dbuspolicy_DATA += \ + src/machine/org.freedesktop.machine1.conf + +polkitpolicy_files += \ + src/machine/org.freedesktop.machine1.policy + +SYSTEM_UNIT_ALIASES += \ + systemd-machined.service dbus-org.freedesktop.machine1.service + +BUSNAMES_TARGET_WANTS += \ + org.freedesktop.machine1.busname + +polkitpolicy_in_files += \ + src/machine/org.freedesktop.machine1.policy.in + +EXTRA_DIST += \ + units/systemd-machined.service.in + +# ------------------------------------------------------------------------------ +ifneq ($(ENABLE_IMPORTD),) + +ifneq ($(HAVE_LIBCURL),) +ifneq ($(HAVE_XZ),) +ifneq ($(HAVE_ZLIB),) +ifneq ($(HAVE_BZIP2),) +ifneq ($(HAVE_GCRYPT),) + +include $(topsrcdir)/build-aux/Makefile.tail.mk diff --git a/src/grp-machine/systemd-machined/machine.slice b/src/grp-machine/systemd-machined/machine.slice new file mode 100644 index 0000000000..3d40dfd73b --- /dev/null +++ b/src/grp-machine/systemd-machined/machine.slice @@ -0,0 +1,11 @@ +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[Unit] +Description=Virtual Machine and Container Slice +Documentation=man:systemd.special(7) +Before=slices.target diff --git a/src/grp-machine/systemd-machined/machined.c b/src/grp-machine/systemd-machined/machined.c new file mode 100644 index 0000000000..d6ce7e0c26 --- /dev/null +++ b/src/grp-machine/systemd-machined/machined.c @@ -0,0 +1,415 @@ +/*** + This file is part of systemd. + + Copyright 2013 Lennart Poettering + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <errno.h> +#include <string.h> +#include <unistd.h> + +#include <systemd/sd-daemon.h> + +#include "machine-core/machined.h" +#include "sd-bus/bus-error.h" +#include "sd-bus/bus-util.h" +#include "systemd-basic/alloc-util.h" +#include "systemd-basic/cgroup-util.h" +#include "systemd-basic/dirent-util.h" +#include "systemd-basic/fd-util.h" +#include "systemd-basic/formats-util.h" +#include "systemd-basic/hostname-util.h" +#include "systemd-basic/label.h" +#include "systemd-basic/signal-util.h" +#include "systemd-shared/machine-image.h" + +Manager *manager_new(void) { + Manager *m; + int r; + + m = new0(Manager, 1); + if (!m) + return NULL; + + m->machines = hashmap_new(&string_hash_ops); + m->machine_units = hashmap_new(&string_hash_ops); + m->machine_leaders = hashmap_new(NULL); + + if (!m->machines || !m->machine_units || !m->machine_leaders) { + manager_free(m); + return NULL; + } + + r = sd_event_default(&m->event); + if (r < 0) { + manager_free(m); + return NULL; + } + + sd_event_set_watchdog(m->event, true); + + return m; +} + +void manager_free(Manager *m) { + Machine *machine; + Image *i; + + assert(m); + + while (m->operations) + operation_free(m->operations); + + assert(m->n_operations == 0); + + while ((machine = hashmap_first(m->machines))) + machine_free(machine); + + hashmap_free(m->machines); + hashmap_free(m->machine_units); + hashmap_free(m->machine_leaders); + + while ((i = hashmap_steal_first(m->image_cache))) + image_unref(i); + + hashmap_free(m->image_cache); + + sd_event_source_unref(m->image_cache_defer_event); + + bus_verify_polkit_async_registry_free(m->polkit_registry); + + sd_bus_unref(m->bus); + sd_event_unref(m->event); + + free(m); +} + +static int manager_add_host_machine(Manager *m) { + _cleanup_free_ char *rd = NULL, *unit = NULL; + sd_id128_t mid; + Machine *t; + int r; + + if (m->host_machine) + return 0; + + r = sd_id128_get_machine(&mid); + if (r < 0) + return log_error_errno(r, "Failed to get machine ID: %m"); + + rd = strdup("/"); + if (!rd) + return log_oom(); + + unit = strdup("-.slice"); + if (!unit) + return log_oom(); + + t = machine_new(m, MACHINE_HOST, ".host"); + if (!t) + return log_oom(); + + t->leader = 1; + t->id = mid; + + t->root_directory = rd; + t->unit = unit; + rd = unit = NULL; + + dual_timestamp_from_boottime_or_monotonic(&t->timestamp, 0); + + m->host_machine = t; + + return 0; +} + +int manager_enumerate_machines(Manager *m) { + _cleanup_closedir_ DIR *d = NULL; + struct dirent *de; + int r = 0; + + assert(m); + + r = manager_add_host_machine(m); + if (r < 0) + return r; + + /* Read in machine data stored on disk */ + d = opendir("/run/systemd/machines"); + if (!d) { + if (errno == ENOENT) + return 0; + + return log_error_errno(errno, "Failed to open /run/systemd/machines: %m"); + } + + FOREACH_DIRENT(de, d, return -errno) { + struct Machine *machine; + int k; + + if (!dirent_is_file(de)) + continue; + + /* Ignore symlinks that map the unit name to the machine */ + if (startswith(de->d_name, "unit:")) + continue; + + if (!machine_name_is_valid(de->d_name)) + continue; + + k = manager_add_machine(m, de->d_name, &machine); + if (k < 0) { + r = log_error_errno(k, "Failed to add machine by file name %s: %m", de->d_name); + continue; + } + + machine_add_to_gc_queue(machine); + + k = machine_load(machine); + if (k < 0) + r = k; + } + + return r; +} + +static int manager_connect_bus(Manager *m) { + _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; + int r; + + assert(m); + assert(!m->bus); + + r = sd_bus_default_system(&m->bus); + if (r < 0) + return log_error_errno(r, "Failed to connect to system bus: %m"); + + r = sd_bus_add_object_vtable(m->bus, NULL, "/org/freedesktop/machine1", "org.freedesktop.machine1.Manager", manager_vtable, m); + if (r < 0) + return log_error_errno(r, "Failed to add manager object vtable: %m"); + + r = sd_bus_add_fallback_vtable(m->bus, NULL, "/org/freedesktop/machine1/machine", "org.freedesktop.machine1.Machine", machine_vtable, machine_object_find, m); + if (r < 0) + return log_error_errno(r, "Failed to add machine object vtable: %m"); + + r = sd_bus_add_node_enumerator(m->bus, NULL, "/org/freedesktop/machine1/machine", machine_node_enumerator, m); + if (r < 0) + return log_error_errno(r, "Failed to add machine enumerator: %m"); + + r = sd_bus_add_fallback_vtable(m->bus, NULL, "/org/freedesktop/machine1/image", "org.freedesktop.machine1.Image", image_vtable, image_object_find, m); + if (r < 0) + return log_error_errno(r, "Failed to add image object vtable: %m"); + + r = sd_bus_add_node_enumerator(m->bus, NULL, "/org/freedesktop/machine1/image", image_node_enumerator, m); + if (r < 0) + return log_error_errno(r, "Failed to add image enumerator: %m"); + + r = sd_bus_add_match(m->bus, + NULL, + "type='signal'," + "sender='org.freedesktop.systemd1'," + "interface='org.freedesktop.systemd1.Manager'," + "member='JobRemoved'," + "path='/org/freedesktop/systemd1'", + match_job_removed, + m); + if (r < 0) + return log_error_errno(r, "Failed to add match for JobRemoved: %m"); + + r = sd_bus_add_match(m->bus, + NULL, + "type='signal'," + "sender='org.freedesktop.systemd1'," + "interface='org.freedesktop.systemd1.Manager'," + "member='UnitRemoved'," + "path='/org/freedesktop/systemd1'", + match_unit_removed, + m); + if (r < 0) + return log_error_errno(r, "Failed to add match for UnitRemoved: %m"); + + r = sd_bus_add_match(m->bus, + NULL, + "type='signal'," + "sender='org.freedesktop.systemd1'," + "interface='org.freedesktop.DBus.Properties'," + "member='PropertiesChanged'," + "arg0='org.freedesktop.systemd1.Unit'", + match_properties_changed, + m); + if (r < 0) + return log_error_errno(r, "Failed to add match for PropertiesChanged: %m"); + + r = sd_bus_add_match(m->bus, + NULL, + "type='signal'," + "sender='org.freedesktop.systemd1'," + "interface='org.freedesktop.systemd1.Manager'," + "member='Reloading'," + "path='/org/freedesktop/systemd1'", + match_reloading, + m); + if (r < 0) + return log_error_errno(r, "Failed to add match for Reloading: %m"); + + r = sd_bus_call_method( + m->bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "Subscribe", + &error, + NULL, NULL); + if (r < 0) { + log_error("Failed to enable subscription: %s", bus_error_message(&error, r)); + return r; + } + + r = sd_bus_request_name(m->bus, "org.freedesktop.machine1", 0); + if (r < 0) + return log_error_errno(r, "Failed to register name: %m"); + + r = sd_bus_attach_event(m->bus, m->event, 0); + if (r < 0) + return log_error_errno(r, "Failed to attach bus to event loop: %m"); + + return 0; +} + +void manager_gc(Manager *m, bool drop_not_started) { + Machine *machine; + + assert(m); + + while ((machine = m->machine_gc_queue)) { + LIST_REMOVE(gc_queue, m->machine_gc_queue, machine); + machine->in_gc_queue = false; + + /* First, if we are not closing yet, initiate stopping */ + if (!machine_check_gc(machine, drop_not_started) && + machine_get_state(machine) != MACHINE_CLOSING) + machine_stop(machine); + + /* Now, the stop probably made this referenced + * again, but if it didn't, then it's time to let it + * go entirely. */ + if (!machine_check_gc(machine, drop_not_started)) { + machine_finalize(machine); + machine_free(machine); + } + } +} + +int manager_startup(Manager *m) { + Machine *machine; + Iterator i; + int r; + + assert(m); + + /* Connect to the bus */ + r = manager_connect_bus(m); + if (r < 0) + return r; + + /* Deserialize state */ + manager_enumerate_machines(m); + + /* Remove stale objects before we start them */ + manager_gc(m, false); + + /* And start everything */ + HASHMAP_FOREACH(machine, m->machines, i) + machine_start(machine, NULL, NULL); + + return 0; +} + +static bool check_idle(void *userdata) { + Manager *m = userdata; + + if (m->operations) + return false; + + manager_gc(m, true); + + return hashmap_isempty(m->machines); +} + +int manager_run(Manager *m) { + assert(m); + + return bus_event_loop_with_idle( + m->event, + m->bus, + "org.freedesktop.machine1", + DEFAULT_EXIT_USEC, + check_idle, m); +} + +int main(int argc, char *argv[]) { + Manager *m = NULL; + int r; + + log_set_target(LOG_TARGET_AUTO); + log_set_facility(LOG_AUTH); + log_parse_environment(); + log_open(); + + umask(0022); + + if (argc != 1) { + log_error("This program takes no arguments."); + r = -EINVAL; + goto finish; + } + + /* Always create the directories people can create inotify + * watches in. Note that some applications might check for the + * existence of /run/systemd/machines/ to determine whether + * machined is available, so please always make sure this + * check stays in. */ + mkdir_label("/run/systemd/machines", 0755); + + assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGCHLD, -1) >= 0); + + m = manager_new(); + if (!m) { + r = log_oom(); + goto finish; + } + + r = manager_startup(m); + if (r < 0) { + log_error_errno(r, "Failed to fully start up daemon: %m"); + goto finish; + } + + log_debug("systemd-machined running as pid "PID_FMT, getpid()); + + sd_notify(false, + "READY=1\n" + "STATUS=Processing requests..."); + + r = manager_run(m); + + log_debug("systemd-machined stopped as pid "PID_FMT, getpid()); + +finish: + manager_free(m); + + return r < 0 ? EXIT_FAILURE : EXIT_SUCCESS; +} diff --git a/src/grp-machine/systemd-machined/org.freedesktop.machine1.conf b/src/grp-machine/systemd-machined/org.freedesktop.machine1.conf new file mode 100644 index 0000000000..562b9d3cc0 --- /dev/null +++ b/src/grp-machine/systemd-machined/org.freedesktop.machine1.conf @@ -0,0 +1,198 @@ +<?xml version="1.0"?> <!--*-nxml-*--> +<!DOCTYPE busconfig PUBLIC "-//freedesktop//DTD D-BUS Bus Configuration 1.0//EN" + "http://www.freedesktop.org/standards/dbus/1.0/busconfig.dtd"> + +<!-- + This file is part of systemd. + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. +--> + +<busconfig> + + <policy user="root"> + <allow own="org.freedesktop.machine1"/> + <allow send_destination="org.freedesktop.machine1"/> + <allow receive_sender="org.freedesktop.machine1"/> + </policy> + + <policy context="default"> + <deny send_destination="org.freedesktop.machine1"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.DBus.Introspectable"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.DBus.Peer"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.DBus.Properties" + send_member="Get"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.DBus.Properties" + send_member="GetAll"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="ListMachines"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="ListImages"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="GetMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="GetMachineByPID"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="GetImage"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="GetMachineAddresses"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="GetMachineOSRelease"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="OpenMachineLogin"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="OpenMachineShell"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="TerminateMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="KillMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="BindMountMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="CopyFromMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="CopyToMachine"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="RemoveImage"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="RenameImage"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="CloneImage"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="MarkImageReadOnly"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="SetPoolLimit"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="SetImageLimit"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="CleanPool"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="MapFromMachineUser"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="MapToMachineUser"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="MapFromMachineGroup"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Manager" + send_member="MapToMachineGroup"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="GetAddresses"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="GetOSRelease"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="OpenLogin"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="OpenShell"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="Terminate"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="Kill"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="BindMount"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="CopyFrom"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Machine" + send_member="CopyTo"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Image" + send_member="Remove"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Image" + send_member="Rename"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Image" + send_member="Clone"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Image" + send_member="SetLimit"/> + + <allow send_destination="org.freedesktop.machine1" + send_interface="org.freedesktop.machine1.Image" + send_member="MarkReadOnly"/> + + <allow receive_sender="org.freedesktop.machine1"/> + </policy> + +</busconfig> diff --git a/src/grp-machine/systemd-machined/org.freedesktop.machine1.policy.in b/src/grp-machine/systemd-machined/org.freedesktop.machine1.policy.in new file mode 100644 index 0000000000..69f78a5c25 --- /dev/null +++ b/src/grp-machine/systemd-machined/org.freedesktop.machine1.policy.in @@ -0,0 +1,102 @@ +<?xml version="1.0" encoding="UTF-8"?> <!--*-nxml-*--> +<!DOCTYPE policyconfig PUBLIC "-//freedesktop//DTD PolicyKit Policy Configuration 1.0//EN" + "http://www.freedesktop.org/standards/PolicyKit/1/policyconfig.dtd"> + +<!-- + This file is part of systemd. + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. +--> + +<policyconfig> + + <vendor>The systemd Project</vendor> + <vendor_url>http://www.freedesktop.org/wiki/Software/systemd</vendor_url> + + <action id="org.freedesktop.machine1.login"> + <_description>Log into a local container</_description> + <_message>Authentication is required to log into a local container.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.machine1.host-login"> + <_description>Log into the local host</_description> + <_message>Authentication is required to log into the local host.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>yes</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.machine1.shell"> + <_description>Acquire a shell in a local container</_description> + <_message>Authentication is required to acquire a shell in a local container.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + <annotate key="org.freedesktop.policykit.imply">org.freedesktop.login1.login</annotate> + </action> + + <action id="org.freedesktop.machine1.host-shell"> + <_description>Acquire a shell on the local host</_description> + <_message>Authentication is required to acquire a shell on the local host.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + <annotate key="org.freedesktop.policykit.imply">org.freedesktop.login1.host-login</annotate> + </action> + + <action id="org.freedesktop.machine1.open-pty"> + <_description>Acquire a pseudo TTY in a local container</_description> + <_message>Authentication is required to acquire a pseudo TTY in a local container.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.machine1.host-open-pty"> + <_description>Acquire a pseudo TTY on the local host</_description> + <_message>Authentication is required to acquire a pseudo TTY on the local host.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + + <action id="org.freedesktop.machine1.manage-machines"> + <_description>Manage local virtual machines and containers</_description> + <_message>Authentication is required to manage local virtual machines and containers.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + <annotate key="org.freedesktop.policykit.imply">org.freedesktop.login1.shell org.freedesktop.login1.login</annotate> + </action> + + <action id="org.freedesktop.machine1.manage-images"> + <_description>Manage local virtual machine and container images</_description> + <_message>Authentication is required to manage local virtual machine and container images.</_message> + <defaults> + <allow_any>auth_admin</allow_any> + <allow_inactive>auth_admin</allow_inactive> + <allow_active>auth_admin_keep</allow_active> + </defaults> + </action> + +</policyconfig> diff --git a/src/grp-machine/systemd-machined/org.freedesktop.machine1.service b/src/grp-machine/systemd-machined/org.freedesktop.machine1.service new file mode 100644 index 0000000000..d3dc99852b --- /dev/null +++ b/src/grp-machine/systemd-machined/org.freedesktop.machine1.service @@ -0,0 +1,12 @@ +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[D-BUS Service] +Name=org.freedesktop.machine1 +Exec=/bin/false +User=root +SystemdService=dbus-org.freedesktop.machine1.service diff --git a/src/grp-machine/systemd-machined/systemd-machined.service.in b/src/grp-machine/systemd-machined/systemd-machined.service.in new file mode 100644 index 0000000000..911ead79ee --- /dev/null +++ b/src/grp-machine/systemd-machined/systemd-machined.service.in @@ -0,0 +1,27 @@ +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[Unit] +Description=Virtual Machine and Container Registration Service +Documentation=man:systemd-machined.service(8) +Documentation=http://www.freedesktop.org/wiki/Software/systemd/machined +Wants=machine.slice +After=machine.slice + +[Service] +ExecStart=@rootlibexecdir@/systemd-machined +BusName=org.freedesktop.machine1 +WatchdogSec=3min +CapabilityBoundingSet=CAP_KILL CAP_SYS_PTRACE CAP_SYS_ADMIN CAP_SETGID CAP_SYS_CHROOT CAP_DAC_READ_SEARCH CAP_DAC_OVERRIDE CAP_CHOWN CAP_FOWNER CAP_FSETID CAP_MKNOD +MemoryDenyWriteExecute=yes +RestrictRealtime=yes +RestrictAddressFamilies=AF_UNIX AF_NETLINK AF_INET AF_INET6 +SystemCallFilter=~@clock @cpu-emulation @debug @keyring @module @obsolete @raw-io + +# Note that machined cannot be placed in a mount namespace, since it +# needs access to the host's mount namespace in order to implement the +# "machinectl bind" operation. diff --git a/src/grp-machine/systemd-machined/systemd-machined.service.xml b/src/grp-machine/systemd-machined/systemd-machined.service.xml new file mode 100644 index 0000000000..999aeee1c6 --- /dev/null +++ b/src/grp-machine/systemd-machined/systemd-machined.service.xml @@ -0,0 +1,90 @@ +<?xml version='1.0'?> <!--*-nxml-*--> +<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" + "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"> + +<!-- + This file is part of systemd. + + Copyright 2013 Zbigniew Jędrzejewski-Szmek + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +--> + +<refentry id="systemd-machined.service" conditional='ENABLE_MACHINED'> + + <refentryinfo> + <title>systemd-machined.service</title> + <productname>systemd</productname> + + <authorgroup> + <author> + <contrib>Developer</contrib> + <firstname>Lennart</firstname> + <surname>Poettering</surname> + <email>lennart@poettering.net</email> + </author> + </authorgroup> + </refentryinfo> + + <refmeta> + <refentrytitle>systemd-machined.service</refentrytitle> + <manvolnum>8</manvolnum> + </refmeta> + + <refnamediv> + <refname>systemd-machined.service</refname> + <refname>systemd-machined</refname> + <refpurpose>Virtual machine and container registration manager</refpurpose> + </refnamediv> + + <refsynopsisdiv> + <para><filename>systemd-machined.service</filename></para> + <para><filename>/usr/lib/systemd/systemd-machined</filename></para> + </refsynopsisdiv> + + <refsect1> + <title>Description</title> + + <para><command>systemd-machined</command> is a system service that + keeps track of virtual machines and containers, and processes + belonging to them.</para> + + <para>See + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry> + for some examples on how to run containers with OS tools.</para> + + <para>Use + <citerefentry><refentrytitle>nss-mymachines</refentrytitle><manvolnum>8</manvolnum></citerefentry> + to make the names of local containers known to + <command>systemd-machined</command> locally resolvable as host + names.</para> + + <para>See the + <ulink url="http://www.freedesktop.org/wiki/Software/systemd/machined"> + machined D-Bus API Documentation</ulink> for information about the + APIs <filename>systemd-machined</filename> provides.</para> + </refsect1> + + <refsect1> + <title>See Also</title> + <para> + <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry>, + <citerefentry><refentrytitle>nss-mymachines</refentrytitle><manvolnum>8</manvolnum></citerefentry>, + <citerefentry><refentrytitle>systemd.special</refentrytitle><manvolnum>7</manvolnum></citerefentry> + </para> + </refsect1> + +</refentry> |