diff options
Diffstat (limited to 'src')
69 files changed, 2365 insertions, 2391 deletions
diff --git a/src/basic/cgroup-util.c b/src/basic/cgroup-util.c index 6099dc8721..95fc2b9e5d 100644 --- a/src/basic/cgroup-util.c +++ b/src/basic/cgroup-util.c @@ -2207,6 +2207,46 @@ bool cg_is_legacy_wanted(void) { return !cg_is_unified_wanted(); } +int cg_cpu_shares_parse(const char *s, uint64_t *ret) { + uint64_t u; + int r; + + if (isempty(s)) { + *ret = CGROUP_CPU_SHARES_INVALID; + return 0; + } + + r = safe_atou64(s, &u); + if (r < 0) + return r; + + if (u < CGROUP_CPU_SHARES_MIN || u > CGROUP_CPU_SHARES_MAX) + return -ERANGE; + + *ret = u; + return 0; +} + +int cg_blkio_weight_parse(const char *s, uint64_t *ret) { + uint64_t u; + int r; + + if (isempty(s)) { + *ret = CGROUP_BLKIO_WEIGHT_INVALID; + return 0; + } + + r = safe_atou64(s, &u); + if (r < 0) + return r; + + if (u < CGROUP_BLKIO_WEIGHT_MIN || u > CGROUP_BLKIO_WEIGHT_MAX) + return -ERANGE; + + *ret = u; + return 0; +} + static const char *cgroup_controller_table[_CGROUP_CONTROLLER_MAX] = { [CGROUP_CONTROLLER_CPU] = "cpu", [CGROUP_CONTROLLER_CPUACCT] = "cpuacct", @@ -2214,6 +2254,7 @@ static const char *cgroup_controller_table[_CGROUP_CONTROLLER_MAX] = { [CGROUP_CONTROLLER_MEMORY] = "memory", [CGROUP_CONTROLLER_DEVICES] = "devices", [CGROUP_CONTROLLER_PIDS] = "pids", + [CGROUP_CONTROLLER_NET_CLS] = "net_cls", }; DEFINE_STRING_TABLE_LOOKUP(cgroup_controller, CGroupController); diff --git a/src/basic/cgroup-util.h b/src/basic/cgroup-util.h index 7e1b4f9c72..01359fa7cb 100644 --- a/src/basic/cgroup-util.h +++ b/src/basic/cgroup-util.h @@ -36,6 +36,7 @@ typedef enum CGroupController { CGROUP_CONTROLLER_MEMORY, CGROUP_CONTROLLER_DEVICES, CGROUP_CONTROLLER_PIDS, + CGROUP_CONTROLLER_NET_CLS, _CGROUP_CONTROLLER_MAX, _CGROUP_CONTROLLER_INVALID = -1, } CGroupController; @@ -50,9 +51,34 @@ typedef enum CGroupMask { CGROUP_MASK_MEMORY = CGROUP_CONTROLLER_TO_MASK(CGROUP_CONTROLLER_MEMORY), CGROUP_MASK_DEVICES = CGROUP_CONTROLLER_TO_MASK(CGROUP_CONTROLLER_DEVICES), CGROUP_MASK_PIDS = CGROUP_CONTROLLER_TO_MASK(CGROUP_CONTROLLER_PIDS), + CGROUP_MASK_NET_CLS = CGROUP_CONTROLLER_TO_MASK(CGROUP_CONTROLLER_NET_CLS), _CGROUP_MASK_ALL = CGROUP_CONTROLLER_TO_MASK(_CGROUP_CONTROLLER_MAX) - 1 } CGroupMask; +/* Special values for the cpu.shares attribute */ +#define CGROUP_CPU_SHARES_INVALID ((uint64_t) -1) +#define CGROUP_CPU_SHARES_MIN UINT64_C(2) +#define CGROUP_CPU_SHARES_MAX UINT64_C(262144) +#define CGROUP_CPU_SHARES_DEFAULT UINT64_C(1024) + +static inline bool CGROUP_CPU_SHARES_IS_OK(uint64_t x) { + return + x == CGROUP_CPU_SHARES_INVALID || + (x >= CGROUP_CPU_SHARES_MIN && x <= CGROUP_CPU_SHARES_MAX); +} + +/* Special values for the blkio.weight attribute */ +#define CGROUP_BLKIO_WEIGHT_INVALID ((uint64_t) -1) +#define CGROUP_BLKIO_WEIGHT_MIN UINT64_C(10) +#define CGROUP_BLKIO_WEIGHT_MAX UINT64_C(1000) +#define CGROUP_BLKIO_WEIGHT_DEFAULT UINT64_C(500) + +static inline bool CGROUP_BLKIO_WEIGHT_IS_OK(uint64_t x) { + return + x == CGROUP_BLKIO_WEIGHT_INVALID || + (x >= CGROUP_BLKIO_WEIGHT_MIN && x <= CGROUP_BLKIO_WEIGHT_MAX); +} + /* * General rules: * @@ -161,3 +187,6 @@ bool cg_is_legacy_wanted(void); const char* cgroup_controller_to_string(CGroupController c) _const_; CGroupController cgroup_controller_from_string(const char *s) _pure_; + +int cg_cpu_shares_parse(const char *s, uint64_t *ret); +int cg_blkio_weight_parse(const char *s, uint64_t *ret); diff --git a/src/basic/macro.h b/src/basic/macro.h index cbc3ca97b8..248f7a86dd 100644 --- a/src/basic/macro.h +++ b/src/basic/macro.h @@ -123,8 +123,11 @@ static inline unsigned long ALIGN_POWER2(unsigned long u) { return 1UL << (sizeof(u) * 8 - __builtin_clzl(u - 1UL)); } -#define ELEMENTSOF(x) (sizeof(x)/sizeof((x)[0])) - +#define ELEMENTSOF(x) \ + __extension__ (__builtin_choose_expr( \ + !__builtin_types_compatible_p(typeof(x), typeof(&*(x))), \ + sizeof(x)/sizeof((x)[0]), \ + (void)0)) /* * container_of - cast a member of a structure out to the containing structure * @ptr: the pointer to the member. @@ -213,18 +216,20 @@ static inline unsigned long ALIGN_POWER2(unsigned long u) { (__x / __y + !!(__x % __y)); \ }) -#define assert_se(expr) \ +#define assert_message_se(expr, message) \ do { \ if (_unlikely_(!(expr))) \ - log_assert_failed(#expr, __FILE__, __LINE__, __PRETTY_FUNCTION__); \ - } while (false) \ + log_assert_failed(message, __FILE__, __LINE__, __PRETTY_FUNCTION__); \ + } while (false) + +#define assert_se(expr) assert_message_se(expr, #expr) /* We override the glibc assert() here. */ #undef assert #ifdef NDEBUG #define assert(expr) do {} while(false) #else -#define assert(expr) assert_se(expr) +#define assert(expr) assert_message_se(expr, #expr) #endif #define assert_not_reached(t) \ @@ -249,19 +254,19 @@ static inline unsigned long ALIGN_POWER2(unsigned long u) { REENABLE_WARNING #endif -#define assert_log(expr) ((_likely_(expr)) \ - ? (true) \ - : (log_assert_failed_return(#expr, __FILE__, __LINE__, __PRETTY_FUNCTION__), false)) +#define assert_log(expr, message) ((_likely_(expr)) \ + ? (true) \ + : (log_assert_failed_return(message, __FILE__, __LINE__, __PRETTY_FUNCTION__), false)) #define assert_return(expr, r) \ do { \ - if (!assert_log(expr)) \ + if (!assert_log(expr, #expr)) \ return (r); \ } while (false) #define assert_return_errno(expr, r, err) \ do { \ - if (!assert_log(expr)) { \ + if (!assert_log(expr, #expr)) { \ errno = err; \ return (r); \ } \ diff --git a/src/basic/ring.c b/src/basic/ring.c deleted file mode 100644 index 6814918464..0000000000 --- a/src/basic/ring.c +++ /dev/null @@ -1,209 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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 <stdlib.h> -#include <string.h> -#include <sys/uio.h> -#include "macro.h" -#include "ring.h" - -#define RING_MASK(_r, _v) ((_v) & ((_r)->size - 1)) - -void ring_flush(Ring *r) { - assert(r); - - r->start = 0; - r->used = 0; -} - -void ring_clear(Ring *r) { - assert(r); - - free(r->buf); - zero(*r); -} - -/* - * Get data pointers for current ring-buffer data. @vec must be an array of 2 - * iovec objects. They are filled according to the data available in the - * ring-buffer. 0, 1 or 2 is returned according to the number of iovec objects - * that were filled (0 meaning buffer is empty). - * - * Hint: "struct iovec" is defined in <sys/uio.h> and looks like this: - * struct iovec { - * void *iov_base; - * size_t iov_len; - * }; - */ -size_t ring_peek(Ring *r, struct iovec *vec) { - assert(r); - - if (r->used == 0) { - return 0; - } else if (r->start + r->used <= r->size) { - if (vec) { - vec[0].iov_base = &r->buf[r->start]; - vec[0].iov_len = r->used; - } - return 1; - } else { - if (vec) { - vec[0].iov_base = &r->buf[r->start]; - vec[0].iov_len = r->size - r->start; - vec[1].iov_base = r->buf; - vec[1].iov_len = r->used - (r->size - r->start); - } - return 2; - } -} - -/* - * Copy data from the ring buffer into the linear external buffer @buf. Copy - * at most @size bytes. If the ring buffer size is smaller, copy less bytes and - * return the number of bytes copied. - */ -size_t ring_copy(Ring *r, void *buf, size_t size) { - size_t l; - - assert(r); - assert(buf); - - if (size > r->used) - size = r->used; - - if (size > 0) { - l = r->size - r->start; - if (size <= l) { - memcpy(buf, &r->buf[r->start], size); - } else { - memcpy(buf, &r->buf[r->start], l); - memcpy((uint8_t*)buf + l, r->buf, size - l); - } - } - - return size; -} - -/* - * Resize ring-buffer to size @nsize. @nsize must be a power-of-2, otherwise - * ring operations will behave incorrectly. - */ -static int ring_resize(Ring *r, size_t nsize) { - uint8_t *buf; - size_t l; - - assert(r); - assert(nsize > 0); - - buf = malloc(nsize); - if (!buf) - return -ENOMEM; - - if (r->used > 0) { - l = r->size - r->start; - if (r->used <= l) { - memcpy(buf, &r->buf[r->start], r->used); - } else { - memcpy(buf, &r->buf[r->start], l); - memcpy(&buf[l], r->buf, r->used - l); - } - } - - free(r->buf); - r->buf = buf; - r->size = nsize; - r->start = 0; - - return 0; -} - -/* - * Resize ring-buffer to provide enough room for @add bytes of new data. This - * resizes the buffer if it is too small. It returns -ENOMEM on OOM and 0 on - * success. - */ -static int ring_grow(Ring *r, size_t add) { - size_t need; - - assert(r); - - if (r->size - r->used >= add) - return 0; - - need = r->used + add; - if (need <= r->used) - return -ENOMEM; - else if (need < 4096) - need = 4096; - - need = ALIGN_POWER2(need); - if (need == 0) - return -ENOMEM; - - return ring_resize(r, need); -} - -/* - * Push @len bytes from @u8 into the ring buffer. The buffer is resized if it - * is too small. -ENOMEM is returned on OOM, 0 on success. - */ -int ring_push(Ring *r, const void *u8, size_t size) { - int err; - size_t pos, l; - - assert(r); - assert(u8); - - if (size == 0) - return 0; - - err = ring_grow(r, size); - if (err < 0) - return err; - - pos = RING_MASK(r, r->start + r->used); - l = r->size - pos; - if (l >= size) { - memcpy(&r->buf[pos], u8, size); - } else { - memcpy(&r->buf[pos], u8, l); - memcpy(r->buf, (const uint8_t*)u8 + l, size - l); - } - - r->used += size; - - return 0; -} - -/* - * Remove @len bytes from the start of the ring-buffer. Note that we protect - * against overflows so removing more bytes than available is safe. - */ -void ring_pull(Ring *r, size_t size) { - assert(r); - - if (size > r->used) - size = r->used; - - r->start = RING_MASK(r, r->start + size); - r->used -= size; -} diff --git a/src/basic/ring.h b/src/basic/ring.h deleted file mode 100644 index dbd6296384..0000000000 --- a/src/basic/ring.h +++ /dev/null @@ -1,55 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -#pragma once - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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/>. -***/ - - -typedef struct Ring Ring; - -struct Ring { - uint8_t *buf; /* buffer or NULL */ - size_t size; /* actual size of @buf */ - size_t start; /* start position of ring */ - size_t used; /* number of actually used bytes */ -}; - -/* flush buffer so it is empty again */ -void ring_flush(Ring *r); - -/* flush buffer, free allocated data and reset to initial state */ -void ring_clear(Ring *r); - -/* get pointers to buffer data and their length */ -size_t ring_peek(Ring *r, struct iovec *vec); - -/* copy data into external linear buffer */ -size_t ring_copy(Ring *r, void *buf, size_t size); - -/* push data to the end of the buffer */ -int ring_push(Ring *r, const void *u8, size_t size); - -/* pull data from the front of the buffer */ -void ring_pull(Ring *r, size_t size); - -/* return size of occupied buffer in bytes */ -static inline size_t ring_get_size(Ring *r) { - return r->used; -} diff --git a/src/basic/time-util.h b/src/basic/time-util.h index de881e8fe1..1af01541fc 100644 --- a/src/basic/time-util.h +++ b/src/basic/time-util.h @@ -112,6 +112,8 @@ bool timezone_is_valid(const char *name); clockid_t clock_boottime_or_monotonic(void); -#define xstrftime(buf, fmt, tm) assert_se(strftime(buf, ELEMENTSOF(buf), fmt, tm) > 0) +#define xstrftime(buf, fmt, tm) \ + assert_message_se(strftime(buf, ELEMENTSOF(buf), fmt, tm) > 0, \ + "xstrftime: " #buf "[] must be big enough") int get_timezone(char **timezone); diff --git a/src/basic/util.h b/src/basic/util.h index c7dff9a86d..8abaa740b2 100644 --- a/src/basic/util.h +++ b/src/basic/util.h @@ -374,7 +374,9 @@ DEFINE_TRIVIAL_CLEANUP_FUNC(cpu_set_t*, CPU_FREE); cpu_set_t* cpu_set_malloc(unsigned *ncpus); -#define xsprintf(buf, fmt, ...) assert_se((size_t) snprintf(buf, ELEMENTSOF(buf), fmt, __VA_ARGS__) < ELEMENTSOF(buf)) +#define xsprintf(buf, fmt, ...) \ + assert_message_se((size_t) snprintf(buf, ELEMENTSOF(buf), fmt, __VA_ARGS__) < ELEMENTSOF(buf), \ + "xsprintf: " #buf "[] must be big enough") int files_same(const char *filea, const char *fileb); diff --git a/src/cgls/cgls.c b/src/cgls/cgls.c index 4fb642e7b3..ec4215f741 100644 --- a/src/cgls/cgls.c +++ b/src/cgls/cgls.c @@ -165,6 +165,13 @@ static int get_cgroup_root(char **ret) { return 0; } +static void show_cg_info(const char *controller, const char *path) { + if (cg_unified() <= 0) + printf("Controller %s; ", controller); + printf("Control group %s:\n", isempty(path) ? "/" : path); + fflush(stdout); +} + int main(int argc, char *argv[]) { int r, output_flags; @@ -225,11 +232,7 @@ int main(int argc, char *argv[]) { } else path = root; - if (cg_unified() > 0) - printf("Control group %s:\n", path); - else - printf("Controller %s; control group %s:\n", controller, path); - fflush(stdout); + show_cg_info(controller, path); q = show_cgroup(controller, path, NULL, 0, arg_kernel_threads, output_flags); } @@ -266,8 +269,7 @@ int main(int argc, char *argv[]) { if (r < 0) goto finish; - printf("Control group %s:\n", isempty(root) ? "/" : root); - fflush(stdout); + show_cg_info(SYSTEMD_CGROUP_CONTROLLER, root); r = show_cgroup(SYSTEMD_CGROUP_CONTROLLER, root, NULL, 0, arg_kernel_threads, output_flags); } diff --git a/src/cgtop/cgtop.c b/src/cgtop/cgtop.c index 4786a155da..1307a34ab7 100644 --- a/src/cgtop/cgtop.c +++ b/src/cgtop/cgtop.c @@ -36,6 +36,10 @@ #include "cgroup-util.h" #include "build.h" #include "fileio.h" +#include "sd-bus.h" +#include "bus-util.h" +#include "bus-error.h" +#include "unit-name.h" typedef struct Group { char *path; @@ -65,6 +69,7 @@ static unsigned arg_iterations = (unsigned) -1; static bool arg_batch = false; static bool arg_raw = false; static usec_t arg_delay = 1*USEC_PER_SEC; +static char* arg_machine = NULL; enum { COUNT_PIDS, @@ -645,6 +650,7 @@ static void help(void) { " -n --iterations=N Run for N iterations before exiting\n" " -b --batch Run in batch mode, accepting no input\n" " --depth=DEPTH Maximum traversal depth (default: %u)\n" + " -M --machine= Show container\n" , program_invocation_short_name, arg_depth); } @@ -669,6 +675,7 @@ static int parse_argv(int argc, char *argv[]) { { "cpu", optional_argument, NULL, ARG_CPU_TYPE }, { "order", required_argument, NULL, ARG_ORDER }, { "recursive", required_argument, NULL, ARG_RECURSIVE }, + { "machine", required_argument, NULL, 'M' }, {} }; @@ -678,7 +685,7 @@ static int parse_argv(int argc, char *argv[]) { assert(argc >= 1); assert(argv); - while ((c = getopt_long(argc, argv, "hptcmin:brd:kP", options, NULL)) >= 0) + while ((c = getopt_long(argc, argv, "hptcmin:brd:kPM:", options, NULL)) >= 0) switch (c) { @@ -797,6 +804,10 @@ static int parse_argv(int argc, char *argv[]) { recursive_unset = r == 0; break; + case 'M': + arg_machine = optarg; + break; + case '?': return -EINVAL; @@ -826,6 +837,48 @@ static const char* counting_what(void) { return "userspace processes (excl. kernel)"; } +static int get_cgroup_root(char **ret) { + _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_bus_flush_close_unref_ sd_bus *bus = NULL; + _cleanup_free_ char *unit = NULL, *path = NULL; + const char *m; + int r; + + if (!arg_machine) { + r = cg_get_root_path(ret); + if (r < 0) + return log_error_errno(r, "Failed to get root control group path: %m"); + + return 0; + } + + m = strjoina("/run/systemd/machines/", arg_machine); + r = parse_env_file(m, NEWLINE, "SCOPE", &unit, NULL); + if (r < 0) + return log_error_errno(r, "Failed to load machine data: %m"); + + path = unit_dbus_path_from_name(unit); + if (!path) + return log_oom(); + + r = bus_open_transport(BUS_TRANSPORT_LOCAL, NULL, false, &bus); + if (r < 0) + return log_error_errno(r, "Failed to create bus connection: %m"); + + r = sd_bus_get_property_string( + bus, + "org.freedesktop.systemd1", + path, + unit_dbus_interface_from_name(unit), + "ControlGroup", + &error, + ret); + if (r < 0) + return log_error_errno(r, "Failed to query unit control group path: %s", bus_error_message(&error, r)); + + return 0; +} + int main(int argc, char *argv[]) { int r; Hashmap *a = NULL, *b = NULL; @@ -850,7 +903,7 @@ int main(int argc, char *argv[]) { if (r <= 0) goto finish; - r = cg_get_root_path(&root); + r = get_cgroup_root(&root); if (r < 0) { log_error_errno(r, "Failed to get root control group path: %m"); goto finish; diff --git a/src/core/cgroup.c b/src/core/cgroup.c index 48000d4e6e..0c790c33da 100644 --- a/src/core/cgroup.c +++ b/src/core/cgroup.c @@ -37,14 +37,18 @@ void cgroup_context_init(CGroupContext *c) { /* Initialize everything to the kernel defaults, assuming the * structure is preinitialized to 0 */ - c->cpu_shares = (unsigned long) -1; - c->startup_cpu_shares = (unsigned long) -1; + c->cpu_shares = CGROUP_CPU_SHARES_INVALID; + c->startup_cpu_shares = CGROUP_CPU_SHARES_INVALID; + c->cpu_quota_per_sec_usec = USEC_INFINITY; + c->memory_limit = (uint64_t) -1; - c->blockio_weight = (unsigned long) -1; - c->startup_blockio_weight = (unsigned long) -1; + + c->blockio_weight = CGROUP_BLKIO_WEIGHT_INVALID; + c->startup_blockio_weight = CGROUP_BLKIO_WEIGHT_INVALID; + c->tasks_max = (uint64_t) -1; - c->cpu_quota_per_sec_usec = USEC_INFINITY; + c->netclass_type = CGROUP_NETCLASS_TYPE_NONE; } void cgroup_context_free_device_allow(CGroupContext *c, CGroupDeviceAllow *a) { @@ -102,11 +106,12 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) { "%sCPUAccounting=%s\n" "%sBlockIOAccounting=%s\n" "%sMemoryAccounting=%s\n" - "%sCPUShares=%lu\n" - "%sStartupCPUShares=%lu\n" + "%sTasksAccounting=%s\n" + "%sCPUShares=%" PRIu64 "\n" + "%sStartupCPUShares=%" PRIu64 "\n" "%sCPUQuotaPerSecSec=%s\n" - "%sBlockIOWeight=%lu\n" - "%sStartupBlockIOWeight=%lu\n" + "%sBlockIOWeight=%" PRIu64 "\n" + "%sStartupBlockIOWeight=%" PRIu64 "\n" "%sMemoryLimit=%" PRIu64 "\n" "%sTasksMax=%" PRIu64 "\n" "%sDevicePolicy=%s\n" @@ -114,6 +119,7 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) { prefix, yes_no(c->cpu_accounting), prefix, yes_no(c->blockio_accounting), prefix, yes_no(c->memory_accounting), + prefix, yes_no(c->tasks_accounting), prefix, c->cpu_shares, prefix, c->startup_cpu_shares, prefix, format_timespan(u, sizeof(u), c->cpu_quota_per_sec_usec, 1), @@ -133,7 +139,7 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) { LIST_FOREACH(device_weights, w, c->blockio_device_weights) fprintf(f, - "%sBlockIODeviceWeight=%s %lu", + "%sBlockIODeviceWeight=%s %" PRIu64, prefix, w->path, w->weight); @@ -287,7 +293,7 @@ fail: return -errno; } -void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, ManagerState state) { +void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, uint32_t netclass, ManagerState state) { bool is_root; int r; @@ -309,11 +315,11 @@ void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, M * and missing cgroups, i.e. EROFS and ENOENT. */ if ((mask & CGROUP_MASK_CPU) && !is_root) { - char buf[MAX(DECIMAL_STR_MAX(unsigned long), DECIMAL_STR_MAX(usec_t)) + 1]; + char buf[MAX(DECIMAL_STR_MAX(uint64_t), DECIMAL_STR_MAX(usec_t)) + 1]; - sprintf(buf, "%lu\n", - IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_cpu_shares != (unsigned long) -1 ? c->startup_cpu_shares : - c->cpu_shares != (unsigned long) -1 ? c->cpu_shares : 1024); + sprintf(buf, "%" PRIu64 "\n", + IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_cpu_shares != CGROUP_CPU_SHARES_INVALID ? c->startup_cpu_shares : + c->cpu_shares != CGROUP_CPU_SHARES_INVALID ? c->cpu_shares : CGROUP_CPU_SHARES_DEFAULT); r = cg_set_attribute("cpu", path, "cpu.shares", buf); if (r < 0) log_full_errno(IN_SET(r, -ENOENT, -EROFS) ? LOG_DEBUG : LOG_WARNING, r, @@ -336,15 +342,15 @@ void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, M } if (mask & CGROUP_MASK_BLKIO) { - char buf[MAX3(DECIMAL_STR_MAX(unsigned long)+1, - DECIMAL_STR_MAX(dev_t)*2+2+DECIMAL_STR_MAX(unsigned long)*1, - DECIMAL_STR_MAX(dev_t)*2+2+DECIMAL_STR_MAX(uint64_t)+1)]; + char buf[MAX(DECIMAL_STR_MAX(uint64_t)+1, + DECIMAL_STR_MAX(dev_t)*2+2+DECIMAL_STR_MAX(uint64_t)+1)]; CGroupBlockIODeviceWeight *w; CGroupBlockIODeviceBandwidth *b; if (!is_root) { - sprintf(buf, "%lu\n", IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_blockio_weight != (unsigned long) -1 ? c->startup_blockio_weight : - c->blockio_weight != (unsigned long) -1 ? c->blockio_weight : 1000); + sprintf(buf, "%" PRIu64 "\n", + IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ? c->startup_blockio_weight : + c->blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ? c->blockio_weight : CGROUP_BLKIO_WEIGHT_DEFAULT); r = cg_set_attribute("blkio", path, "blkio.weight", buf); if (r < 0) log_full_errno(IN_SET(r, -ENOENT, -EROFS) ? LOG_DEBUG : LOG_WARNING, r, @@ -358,7 +364,7 @@ void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, M if (r < 0) continue; - sprintf(buf, "%u:%u %lu", major(dev), minor(dev), w->weight); + sprintf(buf, "%u:%u %" PRIu64 "\n", major(dev), minor(dev), w->weight); r = cg_set_attribute("blkio", path, "blkio.weight_device", buf); if (r < 0) log_full_errno(IN_SET(r, -ENOENT, -EROFS) ? LOG_DEBUG : LOG_WARNING, r, @@ -485,6 +491,17 @@ void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, M log_full_errno(IN_SET(r, -ENOENT, -EROFS) ? LOG_DEBUG : LOG_WARNING, r, "Failed to set pids.max on %s: %m", path); } + + if (mask & CGROUP_MASK_NET_CLS) { + char buf[DECIMAL_STR_MAX(uint32_t)]; + + sprintf(buf, "%" PRIu32, netclass); + + r = cg_set_attribute("net_cls", path, "net_cls.classid", buf); + if (r < 0) + log_full_errno(IN_SET(r, -ENOENT, -EROFS) ? LOG_DEBUG : LOG_WARNING, r, + "Failed to set net_cls.classid on %s: %m", path); + } } CGroupMask cgroup_context_get_mask(CGroupContext *c) { @@ -493,14 +510,14 @@ CGroupMask cgroup_context_get_mask(CGroupContext *c) { /* Figure out which controllers we need */ if (c->cpu_accounting || - c->cpu_shares != (unsigned long) -1 || - c->startup_cpu_shares != (unsigned long) -1 || + c->cpu_shares != CGROUP_CPU_SHARES_INVALID || + c->startup_cpu_shares != CGROUP_CPU_SHARES_INVALID || c->cpu_quota_per_sec_usec != USEC_INFINITY) mask |= CGROUP_MASK_CPUACCT | CGROUP_MASK_CPU; if (c->blockio_accounting || - c->blockio_weight != (unsigned long) -1 || - c->startup_blockio_weight != (unsigned long) -1 || + c->blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID || + c->startup_blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID || c->blockio_device_weights || c->blockio_device_bandwidths) mask |= CGROUP_MASK_BLKIO; @@ -517,6 +534,9 @@ CGroupMask cgroup_context_get_mask(CGroupContext *c) { c->tasks_max != (uint64_t) -1) mask |= CGROUP_MASK_PIDS; + if (c->netclass_type != CGROUP_NETCLASS_TYPE_NONE) + mask |= CGROUP_MASK_NET_CLS; + return mask; } @@ -884,6 +904,103 @@ static bool unit_has_mask_realized(Unit *u, CGroupMask target_mask) { return u->cgroup_realized && u->cgroup_realized_mask == target_mask; } +static int unit_find_free_netclass_cgroup(Unit *u, uint32_t *ret) { + + uint32_t start, i; + Manager *m; + + assert(u); + + m = u->manager; + + i = start = m->cgroup_netclass_registry_last; + + do { + i++; + + if (!hashmap_get(m->cgroup_netclass_registry, UINT_TO_PTR(i))) { + m->cgroup_netclass_registry_last = i; + *ret = i; + return 0; + } + + if (i == UINT32_MAX) + i = CGROUP_NETCLASS_FIXED_MAX; + + } while (i != start); + + return -ENOBUFS; +} + +int unit_add_to_netclass_cgroup(Unit *u) { + + CGroupContext *cc; + Unit *first; + void *key; + int r; + + assert(u); + + cc = unit_get_cgroup_context(u); + if (!cc) + return 0; + + switch (cc->netclass_type) { + case CGROUP_NETCLASS_TYPE_NONE: + return 0; + + case CGROUP_NETCLASS_TYPE_FIXED: + u->cgroup_netclass_id = cc->netclass_id; + break; + + case CGROUP_NETCLASS_TYPE_AUTO: + /* Allocate a new ID in case it was requested and not done yet */ + if (u->cgroup_netclass_id == 0) { + r = unit_find_free_netclass_cgroup(u, &u->cgroup_netclass_id); + if (r < 0) + return r; + + log_debug("Dynamically assigned netclass cgroup id %" PRIu32 " to %s", u->cgroup_netclass_id, u->id); + } + + break; + } + + r = hashmap_ensure_allocated(&u->manager->cgroup_netclass_registry, &trivial_hash_ops); + if (r < 0) + return r; + + key = UINT32_TO_PTR(u->cgroup_netclass_id); + first = hashmap_get(u->manager->cgroup_netclass_registry, key); + + if (first) { + LIST_PREPEND(cgroup_netclass, first, u); + return hashmap_replace(u->manager->cgroup_netclass_registry, key, u); + } + + return hashmap_put(u->manager->cgroup_netclass_registry, key, u); +} + +int unit_remove_from_netclass_cgroup(Unit *u) { + + Unit *head; + void *key; + + assert(u); + + key = UINT32_TO_PTR(u->cgroup_netclass_id); + + LIST_FIND_HEAD(cgroup_netclass, u, head); + LIST_REMOVE(cgroup_netclass, head, u); + + if (head) + return hashmap_replace(u->manager->cgroup_netclass_registry, key, head); + + hashmap_remove(u->manager->cgroup_netclass_registry, key); + + return 0; +} + /* Check if necessary controllers and attributes for a unit are in place. * * If so, do nothing. @@ -919,7 +1036,7 @@ static int unit_realize_cgroup_now(Unit *u, ManagerState state) { return r; /* Finally, apply the necessary attributes. */ - cgroup_context_apply(unit_get_cgroup_context(u), target_mask, u->cgroup_path, state); + cgroup_context_apply(unit_get_cgroup_context(u), target_mask, u->cgroup_path, u->cgroup_netclass_id, state); return 0; } @@ -1577,6 +1694,32 @@ bool unit_cgroup_delegate(Unit *u) { return c->delegate; } +void unit_invalidate_cgroup(Unit *u, CGroupMask m) { + assert(u); + + if (!UNIT_HAS_CGROUP_CONTEXT(u)) + return; + + if (m == 0) + return; + + if ((u->cgroup_realized_mask & m) == 0) + return; + + u->cgroup_realized_mask &= ~m; + unit_add_to_cgroup_queue(u); +} + +void manager_invalidate_startup_units(Manager *m) { + Iterator i; + Unit *u; + + assert(m); + + SET_FOREACH(u, m->startup_units, i) + unit_invalidate_cgroup(u, CGROUP_MASK_CPU|CGROUP_MASK_BLKIO); +} + static const char* const cgroup_device_policy_table[_CGROUP_DEVICE_POLICY_MAX] = { [CGROUP_AUTO] = "auto", [CGROUP_CLOSED] = "closed", diff --git a/src/core/cgroup.h b/src/core/cgroup.h index 3ba09d56a4..457544b49f 100644 --- a/src/core/cgroup.h +++ b/src/core/cgroup.h @@ -26,6 +26,11 @@ #include "list.h" #include "time-util.h" +/* Maximum value for fixed (manual) net class ID assignment, + * and also the value at which the range of automatic assignments starts + */ +#define CGROUP_NETCLASS_FIXED_MAX UINT32_C(65535) + typedef struct CGroupContext CGroupContext; typedef struct CGroupDeviceAllow CGroupDeviceAllow; typedef struct CGroupBlockIODeviceWeight CGroupBlockIODeviceWeight; @@ -47,6 +52,17 @@ typedef enum CGroupDevicePolicy { _CGROUP_DEVICE_POLICY_INVALID = -1 } CGroupDevicePolicy; +typedef enum CGroupNetClassType { + /* Default - do not assign a net class */ + CGROUP_NETCLASS_TYPE_NONE, + + /* Automatically assign a net class */ + CGROUP_NETCLASS_TYPE_AUTO, + + /* Assign the net class that was provided by the user */ + CGROUP_NETCLASS_TYPE_FIXED, +} CGroupNetClassType; + struct CGroupDeviceAllow { LIST_FIELDS(CGroupDeviceAllow, device_allow); char *path; @@ -58,7 +74,7 @@ struct CGroupDeviceAllow { struct CGroupBlockIODeviceWeight { LIST_FIELDS(CGroupBlockIODeviceWeight, device_weights); char *path; - unsigned long weight; + uint64_t weight; }; struct CGroupBlockIODeviceBandwidth { @@ -74,12 +90,12 @@ struct CGroupContext { bool memory_accounting; bool tasks_accounting; - unsigned long cpu_shares; - unsigned long startup_cpu_shares; + uint64_t cpu_shares; + uint64_t startup_cpu_shares; usec_t cpu_quota_per_sec_usec; - unsigned long blockio_weight; - unsigned long startup_blockio_weight; + uint64_t blockio_weight; + uint64_t startup_blockio_weight; LIST_HEAD(CGroupBlockIODeviceWeight, blockio_device_weights); LIST_HEAD(CGroupBlockIODeviceBandwidth, blockio_device_bandwidths); @@ -88,9 +104,12 @@ struct CGroupContext { CGroupDevicePolicy device_policy; LIST_HEAD(CGroupDeviceAllow, device_allow); - bool delegate; + CGroupNetClassType netclass_type; + uint32_t netclass_id; uint64_t tasks_max; + + bool delegate; }; #include "unit.h" @@ -99,7 +118,7 @@ struct CGroupContext { void cgroup_context_init(CGroupContext *c); void cgroup_context_done(CGroupContext *c); void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix); -void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, ManagerState state); +void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, uint32_t netclass_id, ManagerState state); CGroupMask cgroup_context_get_mask(CGroupContext *c); @@ -127,6 +146,9 @@ int unit_watch_cgroup(Unit *u); int unit_attach_pids_to_cgroup(Unit *u); +int unit_add_to_netclass_cgroup(Unit *u); +int unit_remove_from_netclass_cgroup(Unit *u); + int manager_setup_cgroup(Manager *m); void manager_shutdown_cgroup(Manager *m, bool delete); @@ -149,5 +171,9 @@ bool unit_cgroup_delegate(Unit *u); int unit_notify_cgroup_empty(Unit *u); int manager_notify_cgroup_empty(Manager *m, const char *group); +void unit_invalidate_cgroup(Unit *u, CGroupMask m); + +void manager_invalidate_startup_units(Manager *m); + const char* cgroup_device_policy_to_string(CGroupDevicePolicy i) _const_; CGroupDevicePolicy cgroup_device_policy_from_string(const char *s) _pure_; diff --git a/src/core/dbus-cgroup.c b/src/core/dbus-cgroup.c index 94de92c3dd..f334dc928d 100644 --- a/src/core/dbus-cgroup.c +++ b/src/core/dbus-cgroup.c @@ -133,34 +133,16 @@ static int property_get_device_allow( return sd_bus_message_close_container(reply); } -static int property_get_ulong_as_u64( - sd_bus *bus, - const char *path, - const char *interface, - const char *property, - sd_bus_message *reply, - void *userdata, - sd_bus_error *error) { - - unsigned long *ul = userdata; - - assert(bus); - assert(reply); - assert(ul); - - return sd_bus_message_append(reply, "t", *ul == (unsigned long) -1 ? (uint64_t) -1 : (uint64_t) *ul); -} - const sd_bus_vtable bus_cgroup_vtable[] = { SD_BUS_VTABLE_START(0), SD_BUS_PROPERTY("Delegate", "b", bus_property_get_bool, offsetof(CGroupContext, delegate), 0), SD_BUS_PROPERTY("CPUAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, cpu_accounting), 0), - SD_BUS_PROPERTY("CPUShares", "t", property_get_ulong_as_u64, offsetof(CGroupContext, cpu_shares), 0), - SD_BUS_PROPERTY("StartupCPUShares", "t", property_get_ulong_as_u64, offsetof(CGroupContext, startup_cpu_shares), 0), + SD_BUS_PROPERTY("CPUShares", "t", NULL, offsetof(CGroupContext, cpu_shares), 0), + SD_BUS_PROPERTY("StartupCPUShares", "t", NULL, offsetof(CGroupContext, startup_cpu_shares), 0), SD_BUS_PROPERTY("CPUQuotaPerSecUSec", "t", bus_property_get_usec, offsetof(CGroupContext, cpu_quota_per_sec_usec), 0), SD_BUS_PROPERTY("BlockIOAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, blockio_accounting), 0), - SD_BUS_PROPERTY("BlockIOWeight", "t", property_get_ulong_as_u64, offsetof(CGroupContext, blockio_weight), 0), - SD_BUS_PROPERTY("StartupBlockIOWeight", "t", property_get_ulong_as_u64, offsetof(CGroupContext, startup_blockio_weight), 0), + SD_BUS_PROPERTY("BlockIOWeight", "t", NULL, offsetof(CGroupContext, blockio_weight), 0), + SD_BUS_PROPERTY("StartupBlockIOWeight", "t", NULL, offsetof(CGroupContext, startup_blockio_weight), 0), SD_BUS_PROPERTY("BlockIODeviceWeight", "a(st)", property_get_blockio_device_weight, 0, 0), SD_BUS_PROPERTY("BlockIOReadBandwidth", "a(st)", property_get_blockio_device_bandwidths, 0, 0), SD_BUS_PROPERTY("BlockIOWriteBandwidth", "a(st)", property_get_blockio_device_bandwidths, 0, 0), @@ -230,56 +212,52 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->cpu_accounting = b; - u->cgroup_realized_mask &= ~CGROUP_MASK_CPUACCT; + unit_invalidate_cgroup(u, CGROUP_MASK_CPUACCT|CGROUP_MASK_CPU); unit_write_drop_in_private(u, mode, name, b ? "CPUAccounting=yes" : "CPUAccounting=no"); } return 1; } else if (streq(name, "CPUShares")) { - uint64_t u64; - unsigned long ul; + uint64_t shares; - r = sd_bus_message_read(message, "t", &u64); + r = sd_bus_message_read(message, "t", &shares); if (r < 0) return r; - if (u64 == (uint64_t) -1) - ul = (unsigned long) -1; - else { - ul = (unsigned long) u64; - if (ul <= 0 || (uint64_t) ul != u64) - return sd_bus_error_set_errnof(error, EINVAL, "CPUShares value out of range"); - } + if (!CGROUP_CPU_SHARES_IS_OK(shares)) + return sd_bus_error_set_errnof(error, EINVAL, "CPUShares value out of range"); if (mode != UNIT_CHECK) { - c->cpu_shares = ul; - u->cgroup_realized_mask &= ~CGROUP_MASK_CPU; - unit_write_drop_in_private_format(u, mode, name, "CPUShares=%lu", ul); + c->cpu_shares = shares; + unit_invalidate_cgroup(u, CGROUP_MASK_CPU); + + if (shares == CGROUP_CPU_SHARES_INVALID) + unit_write_drop_in_private(u, mode, name, "CPUShares="); + else + unit_write_drop_in_private_format(u, mode, name, "CPUShares=%" PRIu64, shares); } return 1; } else if (streq(name, "StartupCPUShares")) { - uint64_t u64; - unsigned long ul; + uint64_t shares; - r = sd_bus_message_read(message, "t", &u64); + r = sd_bus_message_read(message, "t", &shares); if (r < 0) return r; - if (u64 == (uint64_t) -1) - ul = (unsigned long) -1; - else { - ul = (unsigned long) u64; - if (ul <= 0 || (uint64_t) ul != u64) - return sd_bus_error_set_errnof(error, EINVAL, "StartupCPUShares value out of range"); - } + if (!CGROUP_CPU_SHARES_IS_OK(shares)) + return sd_bus_error_set_errnof(error, EINVAL, "StartupCPUShares value out of range"); if (mode != UNIT_CHECK) { - c->startup_cpu_shares = ul; - u->cgroup_realized_mask &= ~CGROUP_MASK_CPU; - unit_write_drop_in_private_format(u, mode, name, "StartupCPUShares=%lu", ul); + c->startup_cpu_shares = shares; + unit_invalidate_cgroup(u, CGROUP_MASK_CPU); + + if (shares == CGROUP_CPU_SHARES_INVALID) + unit_write_drop_in_private(u, mode, name, "StartupCPUShares="); + else + unit_write_drop_in_private_format(u, mode, name, "StartupCPUShares=%" PRIu64, shares); } return 1; @@ -296,7 +274,7 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->cpu_quota_per_sec_usec = u64; - u->cgroup_realized_mask &= ~CGROUP_MASK_CPU; + unit_invalidate_cgroup(u, CGROUP_MASK_CPU); unit_write_drop_in_private_format(u, mode, "CPUQuota", "CPUQuota=%0.f%%", (double) (c->cpu_quota_per_sec_usec / 10000)); } @@ -311,56 +289,52 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->blockio_accounting = b; - u->cgroup_realized_mask &= ~CGROUP_MASK_BLKIO; + unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO); unit_write_drop_in_private(u, mode, name, b ? "BlockIOAccounting=yes" : "BlockIOAccounting=no"); } return 1; } else if (streq(name, "BlockIOWeight")) { - uint64_t u64; - unsigned long ul; + uint64_t weight; - r = sd_bus_message_read(message, "t", &u64); + r = sd_bus_message_read(message, "t", &weight); if (r < 0) return r; - if (u64 == (uint64_t) -1) - ul = (unsigned long) -1; - else { - ul = (unsigned long) u64; - if (ul < 10 || ul > 1000) - return sd_bus_error_set_errnof(error, EINVAL, "BlockIOWeight value out of range"); - } + if (!CGROUP_BLKIO_WEIGHT_IS_OK(weight)) + return sd_bus_error_set_errnof(error, EINVAL, "BlockIOWeight value out of range"); if (mode != UNIT_CHECK) { - c->blockio_weight = ul; - u->cgroup_realized_mask &= ~CGROUP_MASK_BLKIO; - unit_write_drop_in_private_format(u, mode, name, "BlockIOWeight=%lu", ul); + c->blockio_weight = weight; + unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO); + + if (weight == CGROUP_BLKIO_WEIGHT_INVALID) + unit_write_drop_in_private(u, mode, name, "BlockIOWeight="); + else + unit_write_drop_in_private_format(u, mode, name, "BlockIOWeight=%" PRIu64, weight); } return 1; } else if (streq(name, "StartupBlockIOWeight")) { - uint64_t u64; - unsigned long ul; + uint64_t weight; - r = sd_bus_message_read(message, "t", &u64); + r = sd_bus_message_read(message, "t", &weight); if (r < 0) return r; - if (u64 == (uint64_t) -1) - ul = (unsigned long) -1; - else { - ul = (unsigned long) u64; - if (ul < 10 || ul > 1000) - return sd_bus_error_set_errnof(error, EINVAL, "StartupBlockIOWeight value out of range"); - } + if (CGROUP_BLKIO_WEIGHT_IS_OK(weight)) + return sd_bus_error_set_errnof(error, EINVAL, "StartupBlockIOWeight value out of range"); if (mode != UNIT_CHECK) { - c->startup_blockio_weight = ul; - u->cgroup_realized_mask &= ~CGROUP_MASK_BLKIO; - unit_write_drop_in_private_format(u, mode, name, "StartupBlockIOWeight=%lu", ul); + c->startup_blockio_weight = weight; + unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO); + + if (weight == CGROUP_BLKIO_WEIGHT_INVALID) + unit_write_drop_in_private(u, mode, name, "StartupBlockIOWeight="); + else + unit_write_drop_in_private_format(u, mode, name, "StartupBlockIOWeight=%" PRIu64, weight); } return 1; @@ -429,7 +403,7 @@ int bus_cgroup_set_property( cgroup_context_free_blockio_device_bandwidth(c, a); } - u->cgroup_realized_mask &= ~CGROUP_MASK_BLKIO; + unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO); f = open_memstream(&buf, &size); if (!f) @@ -455,17 +429,16 @@ int bus_cgroup_set_property( } else if (streq(name, "BlockIODeviceWeight")) { const char *path; - uint64_t u64; + uint64_t weight; unsigned n = 0; r = sd_bus_message_enter_container(message, 'a', "(st)"); if (r < 0) return r; - while ((r = sd_bus_message_read(message, "(st)", &path, &u64)) > 0) { - unsigned long ul = u64; + while ((r = sd_bus_message_read(message, "(st)", &path, &weight)) > 0) { - if (ul < 10 || ul > 1000) + if (!CGROUP_BLKIO_WEIGHT_IS_OK(weight) || weight == CGROUP_BLKIO_WEIGHT_INVALID) return sd_bus_error_set_errnof(error, EINVAL, "BlockIODeviceWeight out of range"); if (mode != UNIT_CHECK) { @@ -491,7 +464,7 @@ int bus_cgroup_set_property( LIST_PREPEND(device_weights,c->blockio_device_weights, a); } - a->weight = ul; + a->weight = weight; } n++; @@ -512,7 +485,7 @@ int bus_cgroup_set_property( cgroup_context_free_blockio_device_weight(c, c->blockio_device_weights); } - u->cgroup_realized_mask &= ~CGROUP_MASK_BLKIO; + unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO); f = open_memstream(&buf, &size); if (!f) @@ -520,7 +493,7 @@ int bus_cgroup_set_property( fputs("BlockIODeviceWeight=\n", f); LIST_FOREACH(device_weights, a, c->blockio_device_weights) - fprintf(f, "BlockIODeviceWeight=%s %lu\n", a->path, a->weight); + fprintf(f, "BlockIODeviceWeight=%s %" PRIu64 "\n", a->path, a->weight); fflush(f); unit_write_drop_in_private(u, mode, name, buf); @@ -537,7 +510,7 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->memory_accounting = b; - u->cgroup_realized_mask &= ~CGROUP_MASK_MEMORY; + unit_invalidate_cgroup(u, CGROUP_MASK_MEMORY); unit_write_drop_in_private(u, mode, name, b ? "MemoryAccounting=yes" : "MemoryAccounting=no"); } @@ -552,7 +525,7 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->memory_limit = limit; - u->cgroup_realized_mask &= ~CGROUP_MASK_MEMORY; + unit_invalidate_cgroup(u, CGROUP_MASK_MEMORY); if (limit == (uint64_t) -1) unit_write_drop_in_private(u, mode, name, "MemoryLimit=infinity"); @@ -578,7 +551,7 @@ int bus_cgroup_set_property( char *buf; c->device_policy = p; - u->cgroup_realized_mask &= ~CGROUP_MASK_DEVICES; + unit_invalidate_cgroup(u, CGROUP_MASK_DEVICES); buf = strjoina("DevicePolicy=", policy); unit_write_drop_in_private(u, mode, name, buf); @@ -657,7 +630,7 @@ int bus_cgroup_set_property( cgroup_context_free_device_allow(c, c->device_allow); } - u->cgroup_realized_mask &= ~CGROUP_MASK_DEVICES; + unit_invalidate_cgroup(u, CGROUP_MASK_DEVICES); f = open_memstream(&buf, &size); if (!f) @@ -682,7 +655,7 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->tasks_accounting = b; - u->cgroup_realized_mask &= ~CGROUP_MASK_PIDS; + unit_invalidate_cgroup(u, CGROUP_MASK_PIDS); unit_write_drop_in_private(u, mode, name, b ? "TasksAccounting=yes" : "TasksAccounting=no"); } @@ -697,7 +670,7 @@ int bus_cgroup_set_property( if (mode != UNIT_CHECK) { c->tasks_max = limit; - u->cgroup_realized_mask &= ~CGROUP_MASK_PIDS; + unit_invalidate_cgroup(u, CGROUP_MASK_PIDS); if (limit == (uint64_t) -1) unit_write_drop_in_private(u, mode, name, "TasksMax=infinity"); diff --git a/src/core/dbus-execute.c b/src/core/dbus-execute.c index ed55fcfca2..fd13c6d019 100644 --- a/src/core/dbus-execute.c +++ b/src/core/dbus-execute.c @@ -933,7 +933,10 @@ int bus_exec_context_set_transient_property( return 1; - } else if (streq(name, "IgnoreSIGPIPE")) { + } else if (STR_IN_SET(name, + "IgnoreSIGPIPE", "TTYVHangup", "TTYReset", + "PrivateTmp", "PrivateDevices", "PrivateNetwork", + "NoNewPrivileges")) { int b; r = sd_bus_message_read(message, "b", &b); @@ -941,39 +944,22 @@ int bus_exec_context_set_transient_property( return r; if (mode != UNIT_CHECK) { - c->ignore_sigpipe = b; - - unit_write_drop_in_private_format(u, mode, name, "IgnoreSIGPIPE=%s\n", yes_no(b)); - } - - return 1; - - } else if (streq(name, "TTYVHangup")) { - int b; - - r = sd_bus_message_read(message, "b", &b); - if (r < 0) - return r; - - if (mode != UNIT_CHECK) { - c->tty_vhangup = b; - - unit_write_drop_in_private_format(u, mode, name, "TTYVHangup=%s\n", yes_no(b)); - } - - return 1; - - } else if (streq(name, "TTYReset")) { - int b; - - r = sd_bus_message_read(message, "b", &b); - if (r < 0) - return r; - - if (mode != UNIT_CHECK) { - c->tty_reset = b; - - unit_write_drop_in_private_format(u, mode, name, "TTYReset=%s\n", yes_no(b)); + if (streq(name, "IgnoreSIGPIPE")) + c->ignore_sigpipe = b; + else if (streq(name, "TTYVHangup")) + c->tty_vhangup = b; + else if (streq(name, "TTYReset")) + c->tty_reset = b; + else if (streq(name, "PrivateTmp")) + c->private_tmp = b; + else if (streq(name, "PrivateDevices")) + c->private_devices = b; + else if (streq(name, "PrivateNetwork")) + c->private_network = b; + else if (streq(name, "NoNewPrivileges")) + c->no_new_privileges = b; + + unit_write_drop_in_private_format(u, mode, name, "%s=%s\n", name, yes_no(b)); } return 1; diff --git a/src/core/dbus-manager.c b/src/core/dbus-manager.c index 4e5d67fc19..561b6f8bfa 100644 --- a/src/core/dbus-manager.c +++ b/src/core/dbus-manager.c @@ -1201,8 +1201,10 @@ static int method_exit(sd_bus_message *message, void *userdata, sd_bus_error *er if (r < 0) return r; - if (m->running_as == MANAGER_SYSTEM) - return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "Exit is only supported for user service managers."); + /* Exit() (in contrast to SetExitCode()) is actually allowed even if + * we are running on the host. It will fall back on reboot() in + * systemd-shutdown if it cannot do the exit() because it isn't a + * container. */ m->exit_code = MANAGER_EXIT; @@ -1450,6 +1452,30 @@ static int method_unset_and_set_environment(sd_bus_message *message, void *userd return sd_bus_reply_method_return(message, NULL); } +static int method_set_exit_code(sd_bus_message *message, void *userdata, sd_bus_error *error) { + uint8_t code; + Manager *m = userdata; + int r; + + assert(message); + assert(m); + + r = mac_selinux_access_check(message, "exit", error); + if (r < 0) + return r; + + r = sd_bus_message_read_basic(message, 'y', &code); + if (r < 0) + return r; + + if (m->running_as == MANAGER_SYSTEM && detect_container() <= 0) + return sd_bus_error_setf(error, SD_BUS_ERROR_NOT_SUPPORTED, "ExitCode can only be set for user service managers or in containers."); + + m->return_value = code; + + return sd_bus_reply_method_return(message, NULL); +} + static int method_list_unit_files(sd_bus_message *message, void *userdata, sd_bus_error *error) { _cleanup_bus_message_unref_ sd_bus_message *reply = NULL; Manager *m = userdata; @@ -1933,6 +1959,7 @@ const sd_bus_vtable bus_manager_vtable[] = { SD_BUS_WRITABLE_PROPERTY("ShutdownWatchdogUSec", "t", bus_property_get_usec, bus_property_set_usec, offsetof(Manager, shutdown_watchdog), 0), SD_BUS_PROPERTY("ControlGroup", "s", NULL, offsetof(Manager, cgroup_root), 0), SD_BUS_PROPERTY("SystemState", "s", property_get_system_state, 0, 0), + SD_BUS_PROPERTY("ExitCode", "y", bus_property_get_unsigned, offsetof(Manager, return_value), 0), SD_BUS_METHOD("GetUnit", "s", "o", method_get_unit, SD_BUS_VTABLE_UNPRIVILEGED), SD_BUS_METHOD("GetUnitByPID", "u", "o", method_get_unit_by_pid, SD_BUS_VTABLE_UNPRIVILEGED), @@ -1986,6 +2013,7 @@ const sd_bus_vtable bus_manager_vtable[] = { SD_BUS_METHOD("GetDefaultTarget", NULL, "s", method_get_default_target, SD_BUS_VTABLE_UNPRIVILEGED), SD_BUS_METHOD("PresetAllUnitFiles", "sbb", "a(sss)", method_preset_all_unit_files, SD_BUS_VTABLE_UNPRIVILEGED), SD_BUS_METHOD("AddDependencyUnitFiles", "asssbb", "a(sss)", method_add_dependency_unit_files, SD_BUS_VTABLE_UNPRIVILEGED), + SD_BUS_METHOD("SetExitCode", "y", NULL, method_set_exit_code, SD_BUS_VTABLE_UNPRIVILEGED), SD_BUS_SIGNAL("UnitNew", "so", 0), SD_BUS_SIGNAL("UnitRemoved", "so", 0), diff --git a/src/core/dbus-unit.c b/src/core/dbus-unit.c index f7e9795928..cd88a87340 100644 --- a/src/core/dbus-unit.c +++ b/src/core/dbus-unit.c @@ -679,6 +679,7 @@ const sd_bus_vtable bus_unit_vtable[] = { SD_BUS_PROPERTY("Asserts", "a(sbbsi)", property_get_conditions, offsetof(Unit, asserts), 0), SD_BUS_PROPERTY("LoadError", "(ss)", property_get_load_error, 0, SD_BUS_VTABLE_PROPERTY_CONST), SD_BUS_PROPERTY("Transient", "b", bus_property_get_bool, offsetof(Unit, transient), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("NetClass", "u", bus_property_get_unsigned, offsetof(Unit, cgroup_netclass_id), 0), SD_BUS_METHOD("Start", "s", "o", method_start, SD_BUS_VTABLE_UNPRIVILEGED), SD_BUS_METHOD("Stop", "s", "o", method_stop, SD_BUS_VTABLE_UNPRIVILEGED), diff --git a/src/core/execute.c b/src/core/execute.c index d1acda6682..6e14848cd4 100644 --- a/src/core/execute.c +++ b/src/core/execute.c @@ -629,15 +629,6 @@ static int enforce_groups(const ExecContext *context, const char *username, gid_ * we avoid NSS lookups for gid=0. */ if (context->group || username) { - - if (context->group) { - const char *g = context->group; - - r = get_group_creds(&g, &gid); - if (r < 0) - return r; - } - /* First step, initialize groups from /etc/groups */ if (username && gid != 0) { if (initgroups(username, gid) < 0) @@ -1160,8 +1151,8 @@ static void do_idle_pipe_dance(int idle_pipe[4]) { assert(idle_pipe); - safe_close(idle_pipe[1]); - safe_close(idle_pipe[2]); + idle_pipe[1] = safe_close(idle_pipe[1]); + idle_pipe[2] = safe_close(idle_pipe[2]); if (idle_pipe[0] >= 0) { int r; @@ -1169,18 +1160,20 @@ static void do_idle_pipe_dance(int idle_pipe[4]) { r = fd_wait_for_event(idle_pipe[0], POLLHUP, IDLE_TIMEOUT_USEC); if (idle_pipe[3] >= 0 && r == 0 /* timeout */) { + ssize_t n; + /* Signal systemd that we are bored and want to continue. */ - r = write(idle_pipe[3], "x", 1); - if (r > 0) + n = write(idle_pipe[3], "x", 1); + if (n > 0) /* Wait for systemd to react to the signal above. */ fd_wait_for_event(idle_pipe[0], POLLHUP, IDLE_TIMEOUT2_USEC); } - safe_close(idle_pipe[0]); + idle_pipe[0] = safe_close(idle_pipe[0]); } - safe_close(idle_pipe[3]); + idle_pipe[3] = safe_close(idle_pipe[3]); } static int build_environment( @@ -1412,6 +1405,17 @@ static int exec_child( } } + if (context->group) { + const char *g = context->group; + + r = get_group_creds(&g, &gid); + if (r < 0) { + *exit_status = EXIT_GROUP; + return r; + } + } + + /* If a socket is connected to STDIN/STDOUT/STDERR, we * must sure to drop O_NONBLOCK */ if (socket_fd >= 0) diff --git a/src/core/kill.c b/src/core/kill.c index 2de71c6bf9..bddfa4460f 100644 --- a/src/core/kill.c +++ b/src/core/kill.c @@ -60,7 +60,10 @@ DEFINE_STRING_TABLE_LOOKUP(kill_mode, KillMode); static const char* const kill_who_table[_KILL_WHO_MAX] = { [KILL_MAIN] = "main", [KILL_CONTROL] = "control", - [KILL_ALL] = "all" + [KILL_ALL] = "all", + [KILL_MAIN_FAIL] = "main-fail", + [KILL_CONTROL_FAIL] = "control-fail", + [KILL_ALL_FAIL] = "all-fail" }; DEFINE_STRING_TABLE_LOOKUP(kill_who, KillWho); diff --git a/src/core/kill.h b/src/core/kill.h index d5f125fa41..5d97abb104 100644 --- a/src/core/kill.h +++ b/src/core/kill.h @@ -50,6 +50,9 @@ typedef enum KillWho { KILL_MAIN, KILL_CONTROL, KILL_ALL, + KILL_MAIN_FAIL, + KILL_CONTROL_FAIL, + KILL_ALL_FAIL, _KILL_WHO_MAX, _KILL_WHO_INVALID = -1 } KillWho; diff --git a/src/core/load-fragment-gperf.gperf.m4 b/src/core/load-fragment-gperf.gperf.m4 index e056fd863c..ec744214c1 100644 --- a/src/core/load-fragment-gperf.gperf.m4 +++ b/src/core/load-fragment-gperf.gperf.m4 @@ -126,7 +126,8 @@ $1.BlockIOReadBandwidth, config_parse_blockio_bandwidth, 0, $1.BlockIOWriteBandwidth, config_parse_blockio_bandwidth, 0, offsetof($1, cgroup_context) $1.TasksAccounting, config_parse_bool, 0, offsetof($1, cgroup_context.tasks_accounting) $1.TasksMax, config_parse_tasks_max, 0, offsetof($1, cgroup_context) -$1.Delegate, config_parse_bool, 0, offsetof($1, cgroup_context.delegate)' +$1.Delegate, config_parse_bool, 0, offsetof($1, cgroup_context.delegate) +$1.NetClass, config_parse_netclass, 0, offsetof($1, cgroup_context)' )m4_dnl Unit.Description, config_parse_unit_string_printf, 0, offsetof(Unit, description) Unit.Documentation, config_parse_documentation, 0, offsetof(Unit, documentation) diff --git a/src/core/load-fragment.c b/src/core/load-fragment.c index f7a8539910..fcf863c5c7 100644 --- a/src/core/load-fragment.c +++ b/src/core/load-fragment.c @@ -2605,26 +2605,19 @@ int config_parse_cpu_shares( void *data, void *userdata) { - unsigned long *shares = data, lu; + uint64_t *shares = data; int r; assert(filename); assert(lvalue); assert(rvalue); - if (isempty(rvalue)) { - *shares = (unsigned long) -1; - return 0; - } - - r = safe_atolu(rvalue, &lu); - if (r < 0 || lu <= 0) { - log_syntax(unit, LOG_ERR, filename, line, EINVAL, - "CPU shares '%s' invalid. Ignoring.", rvalue); + r = cg_cpu_shares_parse(rvalue, shares); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "CPU shares '%s' invalid. Ignoring.", rvalue); return 0; } - *shares = lu; return 0; } @@ -2805,26 +2798,19 @@ int config_parse_blockio_weight( void *data, void *userdata) { - unsigned long *weight = data, lu; + uint64_t *weight = data; int r; assert(filename); assert(lvalue); assert(rvalue); - if (isempty(rvalue)) { - *weight = (unsigned long) -1; - return 0; - } - - r = safe_atolu(rvalue, &lu); - if (r < 0 || lu < 10 || lu > 1000) { - log_syntax(unit, LOG_ERR, filename, line, EINVAL, - "Block IO weight '%s' invalid. Ignoring.", rvalue); + r = cg_blkio_weight_parse(rvalue, weight); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Block IO weight '%s' invalid. Ignoring.", rvalue); return 0; } - *weight = lu; return 0; } @@ -2843,8 +2829,8 @@ int config_parse_blockio_device_weight( _cleanup_free_ char *path = NULL; CGroupBlockIODeviceWeight *w; CGroupContext *c = data; - unsigned long lu; const char *weight; + uint64_t u; size_t n; int r; @@ -2861,9 +2847,10 @@ int config_parse_blockio_device_weight( n = strcspn(rvalue, WHITESPACE); weight = rvalue + n; - if (!*weight) { - log_syntax(unit, LOG_ERR, filename, line, EINVAL, - "Expected block device and device weight. Ignoring."); + weight += strspn(weight, WHITESPACE); + + if (isempty(weight)) { + log_syntax(unit, LOG_ERR, filename, line, EINVAL, "Expected block device and device weight. Ignoring."); return 0; } @@ -2872,19 +2859,18 @@ int config_parse_blockio_device_weight( return log_oom(); if (!path_startswith(path, "/dev")) { - log_syntax(unit, LOG_ERR, filename, line, EINVAL, - "Invalid device node path '%s'. Ignoring.", path); + log_syntax(unit, LOG_ERR, filename, line, EINVAL, "Invalid device node path '%s'. Ignoring.", path); return 0; } - weight += strspn(weight, WHITESPACE); - r = safe_atolu(weight, &lu); - if (r < 0 || lu < 10 || lu > 1000) { - log_syntax(unit, LOG_ERR, filename, line, EINVAL, - "Block IO weight '%s' invalid. Ignoring.", rvalue); + r = cg_blkio_weight_parse(weight, &u); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Block IO weight '%s' invalid. Ignoring.", weight); return 0; } + assert(u != CGROUP_BLKIO_WEIGHT_INVALID); + w = new0(CGroupBlockIODeviceWeight, 1); if (!w) return log_oom(); @@ -2892,7 +2878,7 @@ int config_parse_blockio_device_weight( w->path = path; path = NULL; - w->weight = lu; + w->weight = u; LIST_PREPEND(device_weights, c->blockio_device_weights, w); return 0; @@ -2976,6 +2962,48 @@ int config_parse_blockio_bandwidth( return 0; } +int config_parse_netclass( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + CGroupContext *c = data; + unsigned v; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + + if (streq(rvalue, "auto")) { + c->netclass_type = CGROUP_NETCLASS_TYPE_AUTO; + return 0; + } + + r = safe_atou32(rvalue, &v); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, EINVAL, + "Netclass '%s' invalid. Ignoring.", rvalue); + return 0; + } + + if (v > CGROUP_NETCLASS_FIXED_MAX) + log_syntax(unit, LOG_ERR, filename, line, EINVAL, + "Fixed netclass %" PRIu32 " out of allowed range (0-%d). Applying anyway.", v, (uint32_t) CGROUP_NETCLASS_FIXED_MAX); + + c->netclass_id = v; + c->netclass_type = CGROUP_NETCLASS_TYPE_FIXED; + + return 0; +} + DEFINE_CONFIG_PARSE_ENUM(config_parse_job_mode, job_mode, JobMode, "Failed to parse job mode"); int config_parse_job_mode_isolate( @@ -3021,6 +3049,7 @@ int config_parse_runtime_directory( void *userdata) { char***rt = data; + Unit *u = userdata; const char *word, *state; size_t l; int r; @@ -3037,12 +3066,19 @@ int config_parse_runtime_directory( } FOREACH_WORD_QUOTED(word, l, rvalue, state) { - _cleanup_free_ char *n; + _cleanup_free_ char *t = NULL, *n = NULL; - n = strndup(word, l); - if (!n) + t = strndup(word, l); + if (!t) return log_oom(); + r = unit_name_printf(u, t, &n); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, -r, + "Failed to resolve specifiers, ignoring: %s", strerror(-r)); + continue; + } + if (!filename_is_valid(n)) { log_syntax(unit, LOG_ERR, filename, line, EINVAL, "Runtime directory is not valid, ignoring assignment: %s", rvalue); diff --git a/src/core/load-fragment.h b/src/core/load-fragment.h index 638b343a6e..5d0a09249f 100644 --- a/src/core/load-fragment.h +++ b/src/core/load-fragment.h @@ -90,6 +90,7 @@ int config_parse_device_allow(const char *unit, const char *filename, unsigned l int config_parse_blockio_weight(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_blockio_device_weight(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_blockio_bandwidth(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_netclass(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_job_mode(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_job_mode_isolate(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_exec_selinux_context(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); diff --git a/src/core/main.c b/src/core/main.c index 200fe740da..9b59648279 100644 --- a/src/core/main.c +++ b/src/core/main.c @@ -1254,6 +1254,7 @@ int main(int argc, char *argv[]) { char *switch_root_dir = NULL, *switch_root_init = NULL; struct rlimit saved_rlimit_nofile = RLIMIT_MAKE_CONST(0); const char *error_message = NULL; + uint8_t shutdown_exit_code = 0; #ifdef HAVE_SYSV_COMPAT if (getpid() != 1 && strstr(program_invocation_short_name, "init")) { @@ -1764,11 +1765,6 @@ int main(int argc, char *argv[]) { switch (m->exit_code) { - case MANAGER_EXIT: - retval = EXIT_SUCCESS; - log_debug("Exit."); - goto finish; - case MANAGER_RELOAD: log_info("Reloading."); @@ -1810,11 +1806,13 @@ int main(int argc, char *argv[]) { log_notice("Switching root."); goto finish; + case MANAGER_EXIT: case MANAGER_REBOOT: case MANAGER_POWEROFF: case MANAGER_HALT: case MANAGER_KEXEC: { static const char * const table[_MANAGER_EXIT_CODE_MAX] = { + [MANAGER_EXIT] = "exit", [MANAGER_REBOOT] = "reboot", [MANAGER_POWEROFF] = "poweroff", [MANAGER_HALT] = "halt", @@ -1836,8 +1834,10 @@ int main(int argc, char *argv[]) { finish: pager_close(); - if (m) + if (m) { arg_shutdown_watchdog = m->shutdown_watchdog; + shutdown_exit_code = m->return_value; + } m = manager_free(m); for (j = 0; j < ELEMENTSOF(arg_default_rlimit); j++) @@ -1978,7 +1978,8 @@ finish: if (shutdown_verb) { char log_level[DECIMAL_STR_MAX(int) + 1]; - const char* command_line[9] = { + char exit_code[DECIMAL_STR_MAX(uint8_t) + 1]; + const char* command_line[11] = { SYSTEMD_SHUTDOWN_BINARY_PATH, shutdown_verb, "--log-level", log_level, @@ -2015,6 +2016,12 @@ finish: if (log_get_show_location()) command_line[pos++] = "--log-location"; + if (streq(shutdown_verb, "exit")) { + command_line[pos++] = "--exit-code"; + command_line[pos++] = exit_code; + xsprintf(exit_code, "%d", shutdown_exit_code); + } + assert(pos < ELEMENTSOF(command_line)); if (arm_reboot_watchdog && arg_shutdown_watchdog > 0) { diff --git a/src/core/manager.c b/src/core/manager.c index d918007bb8..4e672a8c48 100644 --- a/src/core/manager.c +++ b/src/core/manager.c @@ -317,6 +317,8 @@ static int manager_watch_idle_pipe(Manager *m) { static void manager_close_idle_pipe(Manager *m) { assert(m); + m->idle_pipe_event_source = sd_event_source_unref(m->idle_pipe_event_source); + safe_close_pair(m->idle_pipe); safe_close_pair(m->idle_pipe + 2); } @@ -577,6 +579,8 @@ int manager_new(ManagerRunningAs running_as, bool test_run, Manager **_m) { m->have_ask_password = -EINVAL; /* we don't know */ m->first_boot = -1; + m->cgroup_netclass_registry_last = CGROUP_NETCLASS_FIXED_MAX; + m->test_run = test_run; /* Reboot immediately if the user hits C-A-D more often than 7x per 2s */ @@ -602,14 +606,6 @@ int manager_new(ManagerRunningAs running_as, bool test_run, Manager **_m) { if (r < 0) goto fail; - r = set_ensure_allocated(&m->startup_units, NULL); - if (r < 0) - goto fail; - - r = set_ensure_allocated(&m->failed_units, NULL); - if (r < 0) - goto fail; - r = sd_event_default(&m->event); if (r < 0) goto fail; @@ -944,7 +940,6 @@ Manager* manager_free(Manager *m) { sd_event_source_unref(m->notify_event_source); sd_event_source_unref(m->time_change_event_source); sd_event_source_unref(m->jobs_in_progress_event_source); - sd_event_source_unref(m->idle_pipe_event_source); sd_event_source_unref(m->run_queue_event_source); safe_close(m->signal_fd); @@ -967,6 +962,8 @@ Manager* manager_free(Manager *m) { hashmap_free(m->cgroup_unit); set_free_free(m->unit_path_cache); + hashmap_free(m->cgroup_netclass_registry); + free(m->switch_root); free(m->switch_root_init); @@ -1962,7 +1959,6 @@ static int manager_dispatch_idle_pipe_fd(sd_event_source *source, int fd, uint32 m->no_console_output = m->n_on_console > 0; - m->idle_pipe_event_source = sd_event_source_unref(m->idle_pipe_event_source); manager_close_idle_pipe(m); return 0; @@ -2675,9 +2671,6 @@ static void manager_notify_finished(Manager *m) { } void manager_check_finished(Manager *m) { - Unit *u = NULL; - Iterator i; - assert(m); if (m->n_reloading > 0) @@ -2690,11 +2683,9 @@ void manager_check_finished(Manager *m) { return; if (hashmap_size(m->jobs) > 0) { - if (m->jobs_in_progress_event_source) /* Ignore any failure, this is only for feedback */ - (void) sd_event_source_set_time(m->jobs_in_progress_event_source, - now(CLOCK_MONOTONIC) + JOBS_IN_PROGRESS_WAIT_USEC); + (void) sd_event_source_set_time(m->jobs_in_progress_event_source, now(CLOCK_MONOTONIC) + JOBS_IN_PROGRESS_WAIT_USEC); return; } @@ -2702,7 +2693,6 @@ void manager_check_finished(Manager *m) { manager_flip_auto_status(m, false); /* Notify Type=idle units that we are done now */ - m->idle_pipe_event_source = sd_event_source_unref(m->idle_pipe_event_source); manager_close_idle_pipe(m); /* Turn off confirm spawn now */ @@ -2721,9 +2711,7 @@ void manager_check_finished(Manager *m) { manager_notify_finished(m); - SET_FOREACH(u, m->startup_units, i) - if (u->cgroup_path) - cgroup_context_apply(unit_get_cgroup_context(u), unit_get_own_mask(u), u->cgroup_path, manager_state(m)); + manager_invalidate_startup_units(m); } static int create_generator_dir(Manager *m, char **generator, const char *name) { @@ -3069,8 +3057,9 @@ const char *manager_get_runtime_prefix(Manager *m) { getenv("XDG_RUNTIME_DIR"); } -void manager_update_failed_units(Manager *m, Unit *u, bool failed) { +int manager_update_failed_units(Manager *m, Unit *u, bool failed) { unsigned size; + int r; assert(m); assert(u->manager == m); @@ -3078,13 +3067,19 @@ void manager_update_failed_units(Manager *m, Unit *u, bool failed) { size = set_size(m->failed_units); if (failed) { + r = set_ensure_allocated(&m->failed_units, NULL); + if (r < 0) + return log_oom(); + if (set_put(m->failed_units, u) < 0) - log_oom(); + return log_oom(); } else - set_remove(m->failed_units, u); + (void) set_remove(m->failed_units, u); if (set_size(m->failed_units) != size) bus_manager_send_change_signal(m); + + return 0; } ManagerState manager_state(Manager *m) { diff --git a/src/core/manager.h b/src/core/manager.h index 78a0e50a33..1384eb33a4 100644 --- a/src/core/manager.h +++ b/src/core/manager.h @@ -242,6 +242,11 @@ struct Manager { bool test_run:1; + /* If non-zero, exit with the following value when the systemd + * process terminate. Useful for containers: systemd-nspawn could get + * the return value. */ + uint8_t return_value; + ShowStatus show_status; bool confirm_spawn; bool no_console_output; @@ -303,6 +308,10 @@ struct Manager { const char *unit_log_format_string; int first_boot; + + /* Used for NetClass=auto units */ + Hashmap *cgroup_netclass_registry; + uint32_t cgroup_netclass_registry_last; }; int manager_new(ManagerRunningAs running_as, bool test_run, Manager **m); @@ -369,7 +378,7 @@ const char *manager_get_runtime_prefix(Manager *m); ManagerState manager_state(Manager *m); -void manager_update_failed_units(Manager *m, Unit *u, bool failed); +int manager_update_failed_units(Manager *m, Unit *u, bool failed); const char *manager_state_to_string(ManagerState m) _const_; ManagerState manager_state_from_string(const char *s) _pure_; diff --git a/src/core/service.c b/src/core/service.c index 248a9e8c62..fc28ba4d07 100644 --- a/src/core/service.c +++ b/src/core/service.c @@ -2737,6 +2737,8 @@ static int service_dispatch_timer(sd_event_source *source, usec_t usec, void *us case SERVICE_RELOAD: log_unit_warning(UNIT(s), "Reload operation timed out. Stopping."); + service_unwatch_control_pid(s); + service_kill_control_processes(s); s->reload_result = SERVICE_FAILURE_TIMEOUT; service_enter_running(s, SERVICE_SUCCESS); break; diff --git a/src/core/shutdown.c b/src/core/shutdown.c index 8cc6efc5b8..5296efce1d 100644 --- a/src/core/shutdown.c +++ b/src/core/shutdown.c @@ -48,6 +48,7 @@ #define FINALIZE_ATTEMPTS 50 static char* arg_verb; +static uint8_t arg_exit_code; static int parse_argv(int argc, char *argv[]) { enum { @@ -55,6 +56,7 @@ static int parse_argv(int argc, char *argv[]) { ARG_LOG_TARGET, ARG_LOG_COLOR, ARG_LOG_LOCATION, + ARG_EXIT_CODE, }; static const struct option options[] = { @@ -62,6 +64,7 @@ static int parse_argv(int argc, char *argv[]) { { "log-target", required_argument, NULL, ARG_LOG_TARGET }, { "log-color", optional_argument, NULL, ARG_LOG_COLOR }, { "log-location", optional_argument, NULL, ARG_LOG_LOCATION }, + { "exit-code", required_argument, NULL, ARG_EXIT_CODE }, {} }; @@ -110,6 +113,13 @@ static int parse_argv(int argc, char *argv[]) { break; + case ARG_EXIT_CODE: + r = safe_atou8(optarg, &arg_exit_code); + if (r < 0) + log_error("Failed to parse exit code %s, ignoring", optarg); + + break; + case '\001': if (!arg_verb) arg_verb = optarg; @@ -183,6 +193,8 @@ int main(int argc, char *argv[]) { cmd = RB_HALT_SYSTEM; else if (streq(arg_verb, "kexec")) cmd = LINUX_REBOOT_CMD_KEXEC; + else if (streq(arg_verb, "exit")) + cmd = 0; /* ignored, just checking that arg_verb is valid */ else { r = -EINVAL; log_error("Unknown action '%s'.", arg_verb); @@ -339,6 +351,16 @@ int main(int argc, char *argv[]) { if (!in_container) sync(); + if (streq(arg_verb, "exit")) { + if (in_container) + exit(arg_exit_code); + else { + /* We cannot exit() on the host, fallback on another + * method. */ + cmd = RB_POWER_OFF; + } + } + switch (cmd) { case LINUX_REBOOT_CMD_KEXEC: diff --git a/src/core/transaction.c b/src/core/transaction.c index b8f69ec6f3..2d120af4b5 100644 --- a/src/core/transaction.c +++ b/src/core/transaction.c @@ -736,8 +736,8 @@ int transaction_activate(Transaction *tr, Manager *m, JobMode mode, sd_bus_error if (m->idle_pipe[0] < 0 && m->idle_pipe[1] < 0 && m->idle_pipe[2] < 0 && m->idle_pipe[3] < 0) { - pipe2(m->idle_pipe, O_NONBLOCK|O_CLOEXEC); - pipe2(m->idle_pipe + 2, O_NONBLOCK|O_CLOEXEC); + (void) pipe2(m->idle_pipe, O_NONBLOCK|O_CLOEXEC); + (void) pipe2(m->idle_pipe + 2, O_NONBLOCK|O_CLOEXEC); } } diff --git a/src/core/unit.c b/src/core/unit.c index 2ebfb09a7a..3a6313e4a2 100644 --- a/src/core/unit.c +++ b/src/core/unit.c @@ -452,6 +452,7 @@ static void unit_free_requires_mounts_for(Unit *u) { static void unit_done(Unit *u) { ExecContext *ec; CGroupContext *cc; + int r; assert(u); @@ -468,6 +469,10 @@ static void unit_done(Unit *u) { cc = unit_get_cgroup_context(u); if (cc) cgroup_context_done(cc); + + r = unit_remove_from_netclass_cgroup(u); + if (r < 0) + log_warning_errno(r, "Unable to remove unit from netclass group: %m"); } void unit_free(Unit *u) { @@ -528,7 +533,7 @@ void unit_free(Unit *u) { unit_release_cgroup(u); - manager_update_failed_units(u->manager, u, false); + (void) manager_update_failed_units(u->manager, u, false); set_remove(u->manager->startup_units, u); free(u->description); @@ -1172,15 +1177,20 @@ static int unit_add_mount_dependencies(Unit *u) { static int unit_add_startup_units(Unit *u) { CGroupContext *c; + int r; c = unit_get_cgroup_context(u); if (!c) return 0; - if (c->startup_cpu_shares == (unsigned long) -1 && - c->startup_blockio_weight == (unsigned long) -1) + if (c->startup_cpu_shares == CGROUP_CPU_SHARES_INVALID && + c->startup_blockio_weight == CGROUP_BLKIO_WEIGHT_INVALID) return 0; + r = set_ensure_allocated(&u->manager->startup_units, NULL); + if (r < 0) + return r; + return set_put(u->manager->startup_units, u); } @@ -1236,6 +1246,14 @@ int unit_load(Unit *u) { } unit_update_cgroup_members_masks(u); + + /* If we are reloading, we need to wait for the deserializer + * to restore the net_cls ids that have been set previously */ + if (u->manager->n_reloading <= 0) { + r = unit_add_to_netclass_cgroup(u); + if (r < 0) + return r; + } } assert((u->load_state != UNIT_MERGED) == !u->merged_into); @@ -1807,7 +1825,7 @@ void unit_notify(Unit *u, UnitActiveState os, UnitActiveState ns, bool reload_su } /* Keep track of failed units */ - manager_update_failed_units(u->manager, u, ns == UNIT_FAILED); + (void) manager_update_failed_units(u->manager, u, ns == UNIT_FAILED); /* Make sure the cgroup is always removed when we become inactive */ if (UNIT_IS_INACTIVE_OR_FAILED(ns)) @@ -2586,6 +2604,9 @@ int unit_serialize(Unit *u, FILE *f, FDSet *fds, bool serialize_jobs) { unit_serialize_item(u, f, "cgroup", u->cgroup_path); unit_serialize_item(u, f, "cgroup-realized", yes_no(u->cgroup_realized)); + if (u->cgroup_netclass_id) + unit_serialize_item_format(u, f, "netclass-id", "%" PRIu32, u->cgroup_netclass_id); + if (serialize_jobs) { if (u->job) { fprintf(f, "job\n"); @@ -2773,6 +2794,17 @@ int unit_deserialize(Unit *u, FILE *f, FDSet *fds) { u->cgroup_realized = b; continue; + } else if (streq(l, "netclass-id")) { + r = safe_atou32(v, &u->cgroup_netclass_id); + if (r < 0) + log_unit_debug(u, "Failed to parse netclass ID %s, ignoring.", v); + else { + r = unit_add_to_netclass_cgroup(u); + if (r < 0) + log_unit_debug_errno(u, r, "Failed to add unit to netclass cgroup, ignoring: %m"); + } + + continue; } if (unit_can_serialize(u)) { @@ -3032,32 +3064,39 @@ int unit_kill_common( sd_bus_error *error) { int r = 0; + bool killed = false; - if (who == KILL_MAIN) { + if (IN_SET(who, KILL_MAIN, KILL_MAIN_FAIL)) { if (main_pid < 0) return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_PROCESS, "%s units have no main processes", unit_type_to_string(u->type)); else if (main_pid == 0) return sd_bus_error_set_const(error, BUS_ERROR_NO_SUCH_PROCESS, "No main process to kill"); } - if (who == KILL_CONTROL) { + if (IN_SET(who, KILL_CONTROL, KILL_CONTROL_FAIL)) { if (control_pid < 0) return sd_bus_error_setf(error, BUS_ERROR_NO_SUCH_PROCESS, "%s units have no control processes", unit_type_to_string(u->type)); else if (control_pid == 0) return sd_bus_error_set_const(error, BUS_ERROR_NO_SUCH_PROCESS, "No control process to kill"); } - if (who == KILL_CONTROL || who == KILL_ALL) - if (control_pid > 0) + if (IN_SET(who, KILL_CONTROL, KILL_CONTROL_FAIL, KILL_ALL, KILL_ALL_FAIL)) + if (control_pid > 0) { if (kill(control_pid, signo) < 0) r = -errno; + else + killed = true; + } - if (who == KILL_MAIN || who == KILL_ALL) - if (main_pid > 0) + if (IN_SET(who, KILL_MAIN, KILL_MAIN_FAIL, KILL_ALL, KILL_ALL_FAIL)) + if (main_pid > 0) { if (kill(main_pid, signo) < 0) r = -errno; + else + killed = true; + } - if (who == KILL_ALL && u->cgroup_path) { + if (IN_SET(who, KILL_ALL, KILL_ALL_FAIL) && u->cgroup_path) { _cleanup_set_free_ Set *pid_set = NULL; int q; @@ -3069,8 +3108,13 @@ int unit_kill_common( q = cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, signo, false, false, false, pid_set); if (q < 0 && q != -EAGAIN && q != -ESRCH && q != -ENOENT) r = q; + else + killed = true; } + if (r == 0 && !killed && IN_SET(who, KILL_ALL_FAIL, KILL_CONTROL_FAIL, KILL_ALL_FAIL)) + return -ESRCH; + return r; } diff --git a/src/core/unit.h b/src/core/unit.h index 3c7684411b..c868d75c79 100644 --- a/src/core/unit.h +++ b/src/core/unit.h @@ -161,6 +161,9 @@ struct Unit { /* CGroup realize members queue */ LIST_FIELDS(Unit, cgroup_queue); + /* Units with the same CGroup netclass */ + LIST_FIELDS(Unit, cgroup_netclass); + /* PIDs we keep an eye on. Note that a unit might have many * more, but these are the ones we care enough about to * process SIGCHLD for */ @@ -189,6 +192,8 @@ struct Unit { CGroupMask cgroup_members_mask; int cgroup_inotify_wd; + uint32_t cgroup_netclass_id; + /* How to start OnFailure units */ JobMode on_failure_job_mode; diff --git a/src/libsystemd-network/arp-util.c b/src/libsystemd-network/arp-util.c new file mode 100644 index 0000000000..2f5b9b3731 --- /dev/null +++ b/src/libsystemd-network/arp-util.c @@ -0,0 +1,153 @@ +/*** + This file is part of systemd. + + Copyright (C) 2014 Axis Communications AB. All rights reserved. + Copyright (C) 2015 Tom Gundersen + + 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/filter.h> +#include <arpa/inet.h> + +#include "util.h" +#include "arp-util.h" + +int arp_network_bind_raw_socket(int ifindex, be32_t address, const struct ether_addr *eth_mac) { + struct sock_filter filter[] = { + BPF_STMT(BPF_LD + BPF_W + BPF_LEN, 0), /* A <- packet length */ + BPF_JUMP(BPF_JMP + BPF_JGE + BPF_K, sizeof(struct ether_arp), 1, 0), /* packet >= arp packet ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_hrd)), /* A <- header */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPHRD_ETHER, 1, 0), /* header == ethernet ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_pro)), /* A <- protocol */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ETHERTYPE_IP, 1, 0), /* protocol == IP ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + BPF_STMT(BPF_LD + BPF_B + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_hln)), /* A <- hardware address length */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, sizeof(struct ether_addr), 1, 0), /* length == sizeof(ether_addr)? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + BPF_STMT(BPF_LD + BPF_B + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_pln)), /* A <- protocol address length */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, sizeof(struct in_addr), 1, 0), /* length == sizeof(in_addr) ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_op)), /* A <- operation */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPOP_REQUEST, 2, 0), /* protocol == request ? */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPOP_REPLY, 1, 0), /* protocol == reply ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + /* Sender Hardware Address must be different from our own */ + BPF_STMT(BPF_LD + BPF_IMM, htobe32(*((uint32_t *) eth_mac))), /* A <- 4 bytes of client's MAC */ + BPF_STMT(BPF_MISC + BPF_TAX, 0), /* X <- A */ + BPF_STMT(BPF_LD + BPF_W + BPF_ABS, offsetof(struct ether_arp, arp_sha)), /* A <- 4 bytes of SHA */ + BPF_STMT(BPF_ALU + BPF_XOR + BPF_X, 0), /* A xor X */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, 0, 0, 6), /* A == 0 ? */ + BPF_STMT(BPF_LD + BPF_IMM, htobe16(*((uint16_t *) (((char *) eth_mac) + 4)))), /* A <- remainder of client's MAC */ + BPF_STMT(BPF_MISC + BPF_TAX, 0), /* X <- A */ + BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, arp_sha) + 4), /* A <- remainder of SHA */ + BPF_STMT(BPF_ALU + BPF_XOR + BPF_X, 0), /* A xor X */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, 0, 0, 1), /* A == 0 ? */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + /* Sender Protocol Address or Target Protocol Address must be equal to the one we care about*/ + BPF_STMT(BPF_LD + BPF_IMM, htobe32(address)), /* A <- clients IP */ + BPF_STMT(BPF_MISC + BPF_TAX, 0), /* X <- A */ + BPF_STMT(BPF_LD + BPF_W + BPF_ABS, offsetof(struct ether_arp, arp_spa)), /* A <- SPA */ + BPF_STMT(BPF_ALU + BPF_XOR + BPF_X, 0), /* X xor A */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, 0, 0, 1), /* A == 0 ? */ + BPF_STMT(BPF_RET + BPF_K, 65535), /* return all */ + BPF_STMT(BPF_LD + BPF_IMM, htobe32(address)), /* A <- clients IP */ + BPF_STMT(BPF_MISC + BPF_TAX, 0), /* X <- A */ + BPF_STMT(BPF_LD + BPF_W + BPF_ABS, offsetof(struct ether_arp, arp_tpa)), /* A <- TPA */ + BPF_STMT(BPF_ALU + BPF_XOR + BPF_X, 0), /* X xor A */ + BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, 0, 0, 1), /* A == 0 ? */ + BPF_STMT(BPF_RET + BPF_K, 65535), /* return all */ + BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ + }; + struct sock_fprog fprog = { + .len = ELEMENTSOF(filter), + .filter = (struct sock_filter*) filter + }; + union sockaddr_union link = { + .ll.sll_family = AF_PACKET, + .ll.sll_protocol = htons(ETH_P_ARP), + .ll.sll_ifindex = ifindex, + .ll.sll_halen = ETH_ALEN, + .ll.sll_addr = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }, + }; + _cleanup_close_ int s = -1; + int r; + + assert(ifindex > 0); + + s = socket(PF_PACKET, SOCK_DGRAM | SOCK_CLOEXEC | SOCK_NONBLOCK, 0); + if (s < 0) + return -errno; + + r = setsockopt(s, SOL_SOCKET, SO_ATTACH_FILTER, &fprog, sizeof(fprog)); + if (r < 0) + return -errno; + + r = bind(s, &link.sa, sizeof(link.ll)); + if (r < 0) + return -errno; + + r = s; + s = -1; + + return r; +} + +static int arp_send_packet(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha, + bool announce) { + union sockaddr_union link = { + .ll.sll_family = AF_PACKET, + .ll.sll_protocol = htons(ETH_P_ARP), + .ll.sll_ifindex = ifindex, + .ll.sll_halen = ETH_ALEN, + .ll.sll_addr = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }, + }; + struct ether_arp arp = { + .ea_hdr.ar_hrd = htons(ARPHRD_ETHER), /* HTYPE */ + .ea_hdr.ar_pro = htons(ETHERTYPE_IP), /* PTYPE */ + .ea_hdr.ar_hln = ETH_ALEN, /* HLEN */ + .ea_hdr.ar_pln = sizeof(be32_t), /* PLEN */ + .ea_hdr.ar_op = htons(ARPOP_REQUEST), /* REQUEST */ + }; + int r; + + assert(fd >= 0); + assert(pa != 0); + assert(ha); + + memcpy(&arp.arp_sha, ha, ETH_ALEN); + memcpy(&arp.arp_tpa, &pa, sizeof(pa)); + + if (announce) + memcpy(&arp.arp_spa, &pa, sizeof(pa)); + + r = sendto(fd, &arp, sizeof(struct ether_arp), 0, &link.sa, sizeof(link.ll)); + if (r < 0) + return -errno; + + return 0; +} + +int arp_send_probe(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha) { + return arp_send_packet(fd, ifindex, pa, ha, false); +} + +int arp_send_announcement(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha) { + return arp_send_packet(fd, ifindex, pa, ha, true); +} diff --git a/src/libsystemd-network/ipv4ll-internal.h b/src/libsystemd-network/arp-util.h index ae0ce43985..44e5c893a7 100644 --- a/src/libsystemd-network/ipv4ll-internal.h +++ b/src/libsystemd-network/arp-util.h @@ -26,13 +26,9 @@ #include "sparse-endian.h" #include "socket-util.h" -int arp_network_bind_raw_socket(int index, union sockaddr_union *link); -int arp_network_send_raw_socket(int fd, const union sockaddr_union *link, - const struct ether_arp *arp); +int arp_network_bind_raw_socket(int index, be32_t address, const struct ether_addr *eth_mac); -void arp_packet_init(struct ether_arp *arp); -void arp_packet_probe(struct ether_arp *arp, be32_t pa, const struct ether_addr *ha); -void arp_packet_announcement(struct ether_arp *arp, be32_t pa, const struct ether_addr *ha); -int arp_packet_verify_headers(struct ether_arp *arp); - -#define log_ipv4ll(ll, fmt, ...) log_internal(LOG_DEBUG, 0, __FILE__, __LINE__, __func__, "IPv4LL: " fmt, ##__VA_ARGS__) +int arp_send_probe(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha); +int arp_send_announcement(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha); diff --git a/src/libsystemd-network/ipv4ll-network.c b/src/libsystemd-network/ipv4ll-network.c deleted file mode 100644 index 93ffed408f..0000000000 --- a/src/libsystemd-network/ipv4ll-network.c +++ /dev/null @@ -1,91 +0,0 @@ -/*** - This file is part of systemd. - - Copyright (C) 2014 Axis Communications AB. All rights reserved. - - 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/filter.h> - -#include "util.h" -#include "ipv4ll-internal.h" - -int arp_network_send_raw_socket(int fd, const union sockaddr_union *link, - const struct ether_arp *arp) { - int r; - - assert(arp); - assert(link); - assert(fd >= 0); - - r = sendto(fd, arp, sizeof(struct ether_arp), 0, &link->sa, sizeof(link->ll)); - if (r < 0) - return -errno; - - return 0; -} - -int arp_network_bind_raw_socket(int ifindex, union sockaddr_union *link) { - - static const struct sock_filter filter[] = { - BPF_STMT(BPF_LD + BPF_W + BPF_LEN, 0), /* A <- packet length */ - BPF_JUMP(BPF_JMP + BPF_JGE + BPF_K, sizeof(struct ether_arp), 1, 0), /* packet >= arp packet ? */ - BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ - BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_hrd)), /* A <- header */ - BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPHRD_ETHER, 1, 0), /* header == ethernet ? */ - BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ - BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_pro)), /* A <- protocol */ - BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ETHERTYPE_IP, 1, 0), /* protocol == IP ? */ - BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ - BPF_STMT(BPF_LD + BPF_H + BPF_ABS, offsetof(struct ether_arp, ea_hdr.ar_op)), /* A <- operation */ - BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPOP_REQUEST, 0, 1), /* protocol == request ? */ - BPF_STMT(BPF_RET + BPF_K, 65535), /* return all */ - BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, ARPOP_REPLY, 0, 1), /* protocol == reply ? */ - BPF_STMT(BPF_RET + BPF_K, 65535), /* return all */ - BPF_STMT(BPF_RET + BPF_K, 0), /* ignore */ - }; - struct sock_fprog fprog = { - .len = ELEMENTSOF(filter), - .filter = (struct sock_filter*) filter - }; - _cleanup_close_ int s = -1; - int r; - - assert(ifindex > 0); - assert(link); - - s = socket(PF_PACKET, SOCK_DGRAM | SOCK_CLOEXEC | SOCK_NONBLOCK, 0); - if (s < 0) - return -errno; - - r = setsockopt(s, SOL_SOCKET, SO_ATTACH_FILTER, &fprog, sizeof(fprog)); - if (r < 0) - return -errno; - - link->ll.sll_family = AF_PACKET; - link->ll.sll_protocol = htons(ETH_P_ARP); - link->ll.sll_ifindex = ifindex; - link->ll.sll_halen = ETH_ALEN; - memset(link->ll.sll_addr, 0xff, ETH_ALEN); - - r = bind(s, &link->sa, sizeof(link->ll)); - if (r < 0) - return -errno; - - r = s; - s = -1; - - return r; -} diff --git a/src/libsystemd-network/ipv4ll-packet.c b/src/libsystemd-network/ipv4ll-packet.c deleted file mode 100644 index 2b6c73ab4b..0000000000 --- a/src/libsystemd-network/ipv4ll-packet.c +++ /dev/null @@ -1,71 +0,0 @@ -/*** - This file is part of systemd. - - Copyright (C) 2014 Axis Communications AB. All rights reserved. - - 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 "util.h" -#include "ipv4ll-internal.h" - -void arp_packet_init(struct ether_arp *arp) { - assert(arp); - - memzero(arp, sizeof(struct ether_arp)); - /* Header */ - arp->ea_hdr.ar_hrd = htons(ARPHRD_ETHER); /* HTYPE */ - arp->ea_hdr.ar_pro = htons(ETHERTYPE_IP); /* PTYPE */ - arp->ea_hdr.ar_hln = ETH_ALEN; /* HLEN */ - arp->ea_hdr.ar_pln = sizeof arp->arp_spa; /* PLEN */ - arp->ea_hdr.ar_op = htons(ARPOP_REQUEST); /* REQUEST */ -} - -void arp_packet_probe(struct ether_arp *arp, be32_t pa, const struct ether_addr *ha) { - assert(ha); - - arp_packet_init(arp); - memcpy(arp->arp_sha, ha, ETH_ALEN); - memcpy(arp->arp_tpa, &pa, sizeof(pa)); -} - -void arp_packet_announcement(struct ether_arp *arp, be32_t pa, const struct ether_addr *ha) { - assert(ha); - - arp_packet_init(arp); - memcpy(arp->arp_sha, ha, ETH_ALEN); - memcpy(arp->arp_tpa, &pa, sizeof(pa)); - memcpy(arp->arp_spa, &pa, sizeof(pa)); -} - -int arp_packet_verify_headers(struct ether_arp *arp) { - assert(arp); - - if (arp->ea_hdr.ar_hrd != htons(ARPHRD_ETHER)) { - log_ipv4ll(NULL, "ignoring packet: header is not ARPHRD_ETHER"); - return -EINVAL; - } - if (arp->ea_hdr.ar_pro != htons(ETHERTYPE_IP)) { - log_ipv4ll(NULL, "ignoring packet: protocol is not ETHERTYPE_IP"); - return -EINVAL; - } - if (arp->ea_hdr.ar_op != htons(ARPOP_REQUEST) && - arp->ea_hdr.ar_op != htons(ARPOP_REPLY)) { - log_ipv4ll(NULL, "ignoring packet: operation is not ARPOP_REQUEST or ARPOP_REPLY"); - return -EINVAL; - } - - return 0; -} diff --git a/src/libsystemd-network/sd-ipv4acd.c b/src/libsystemd-network/sd-ipv4acd.c new file mode 100644 index 0000000000..ee5d13c030 --- /dev/null +++ b/src/libsystemd-network/sd-ipv4acd.c @@ -0,0 +1,529 @@ +/*** + This file is part of systemd. + + Copyright (C) 2014 Axis Communications AB. All rights reserved. + Copyright (C) 2015 Tom Gundersen + + 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 <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include "event-util.h" +#include "in-addr-util.h" +#include "list.h" +#include "refcnt.h" +#include "random-util.h" +#include "siphash24.h" +#include "util.h" + +#include "arp-util.h" +#include "sd-ipv4acd.h" + +/* Constants from the RFC */ +#define PROBE_WAIT 1 +#define PROBE_NUM 3 +#define PROBE_MIN 1 +#define PROBE_MAX 2 +#define ANNOUNCE_WAIT 2 +#define ANNOUNCE_NUM 2 +#define ANNOUNCE_INTERVAL 2 +#define MAX_CONFLICTS 10 +#define RATE_LIMIT_INTERVAL 60 +#define DEFEND_INTERVAL 10 + +#define IPV4ACD_NETWORK 0xA9FE0000L +#define IPV4ACD_NETMASK 0xFFFF0000L + +#define log_ipv4acd_full(ll, level, error, fmt, ...) log_internal(level, error, __FILE__, __LINE__, __func__, "ACD: " fmt, ##__VA_ARGS__) + +#define log_ipv4acd_debug(ll, ...) log_ipv4acd_full(ll, LOG_DEBUG, 0, ##__VA_ARGS__) +#define log_ipv4acd_info(ll, ...) log_ipv4acd_full(ll, LOG_INFO, 0, ##__VA_ARGS__) +#define log_ipv4acd_notice(ll, ...) log_ipv4acd_full(ll, LOG_NOTICE, 0, ##__VA_ARGS__) +#define log_ipv4acd_warning(ll, ...) log_ipv4acd_full(ll, LOG_WARNING, 0, ##__VA_ARGS__) +#define log_ipv4acd_error(ll, ...) log_ipv4acd_full(ll, LOG_ERR, 0, ##__VA_ARGS__) + +#define log_ipv4acd_debug_errno(ll, error, ...) log_ipv4acd_full(ll, LOG_DEBUG, error, ##__VA_ARGS__) +#define log_ipv4acd_info_errno(ll, error, ...) log_ipv4acd_full(ll, LOG_INFO, error, ##__VA_ARGS__) +#define log_ipv4acd_notice_errno(ll, error, ...) log_ipv4acd_full(ll, LOG_NOTICE, error, ##__VA_ARGS__) +#define log_ipv4acd_warning_errno(ll, error, ...) log_ipv4acd_full(ll, LOG_WARNING, error, ##__VA_ARGS__) +#define log_ipv4acd_error_errno(ll, error, ...) log_ipv4acd_full(ll, LOG_ERR, error, ##__VA_ARGS__) + +typedef enum IPv4ACDState { + IPV4ACD_STATE_INIT, + IPV4ACD_STATE_WAITING_PROBE, + IPV4ACD_STATE_PROBING, + IPV4ACD_STATE_WAITING_ANNOUNCE, + IPV4ACD_STATE_ANNOUNCING, + IPV4ACD_STATE_RUNNING, + _IPV4ACD_STATE_MAX, + _IPV4ACD_STATE_INVALID = -1 +} IPv4ACDState; + +struct sd_ipv4acd { + RefCount n_ref; + + IPv4ACDState state; + int index; + int fd; + int iteration; + int conflict; + sd_event_source *receive_message; + sd_event_source *timer; + usec_t defend_window; + be32_t address; + /* External */ + struct ether_addr mac_addr; + sd_event *event; + int event_priority; + sd_ipv4acd_cb_t cb; + void* userdata; +}; + +sd_ipv4acd *sd_ipv4acd_ref(sd_ipv4acd *ll) { + if (ll) + assert_se(REFCNT_INC(ll->n_ref) >= 2); + + return ll; +} + +sd_ipv4acd *sd_ipv4acd_unref(sd_ipv4acd *ll) { + if (!ll || REFCNT_DEC(ll->n_ref) > 0) + return NULL; + + ll->receive_message = sd_event_source_unref(ll->receive_message); + ll->fd = safe_close(ll->fd); + + ll->timer = sd_event_source_unref(ll->timer); + + sd_ipv4acd_detach_event(ll); + + free(ll); + + return NULL; +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(sd_ipv4acd*, sd_ipv4acd_unref); +#define _cleanup_ipv4acd_unref_ _cleanup_(sd_ipv4acd_unrefp) + +int sd_ipv4acd_new(sd_ipv4acd **ret) { + _cleanup_ipv4acd_unref_ sd_ipv4acd *ll = NULL; + + assert_return(ret, -EINVAL); + + ll = new0(sd_ipv4acd, 1); + if (!ll) + return -ENOMEM; + + ll->n_ref = REFCNT_INIT; + ll->state = IPV4ACD_STATE_INIT; + ll->index = -1; + ll->fd = -1; + + *ret = ll; + ll = NULL; + + return 0; +} + +static void ipv4acd_set_state(sd_ipv4acd *ll, IPv4ACDState st, bool reset_counter) { + + assert(ll); + assert(st < _IPV4ACD_STATE_MAX); + + if (st == ll->state && !reset_counter) + ll->iteration++; + else { + ll->state = st; + ll->iteration = 0; + } +} + +static void ipv4acd_client_notify(sd_ipv4acd *ll, int event) { + assert(ll); + + if (ll->cb) + ll->cb(ll, event, ll->userdata); +} + +static void ipv4acd_stop(sd_ipv4acd *ll) { + assert(ll); + + ll->receive_message = sd_event_source_unref(ll->receive_message); + ll->fd = safe_close(ll->fd); + + ll->timer = sd_event_source_unref(ll->timer); + + log_ipv4acd_debug(ll, "STOPPED"); + + ipv4acd_set_state (ll, IPV4ACD_STATE_INIT, true); +} + +int sd_ipv4acd_stop(sd_ipv4acd *ll) { + assert_return(ll, -EINVAL); + + ipv4acd_stop(ll); + + ipv4acd_client_notify(ll, IPV4ACD_EVENT_STOP); + + return 0; +} + +static int ipv4acd_on_timeout(sd_event_source *s, uint64_t usec, void *userdata); + +static int ipv4acd_set_next_wakeup(sd_ipv4acd *ll, int sec, int random_sec) { + _cleanup_event_source_unref_ sd_event_source *timer = NULL; + usec_t next_timeout; + usec_t time_now; + int r; + + assert(sec >= 0); + assert(random_sec >= 0); + assert(ll); + + next_timeout = sec * USEC_PER_SEC; + + if (random_sec) + next_timeout += random_u32() % (random_sec * USEC_PER_SEC); + + assert_se(sd_event_now(ll->event, clock_boottime_or_monotonic(), &time_now) >= 0); + + r = sd_event_add_time(ll->event, &timer, clock_boottime_or_monotonic(), + time_now + next_timeout, 0, ipv4acd_on_timeout, ll); + if (r < 0) + return r; + + r = sd_event_source_set_priority(timer, ll->event_priority); + if (r < 0) + return r; + + r = sd_event_source_set_description(timer, "ipv4acd-timer"); + if (r < 0) + return r; + + ll->timer = sd_event_source_unref(ll->timer); + ll->timer = timer; + timer = NULL; + + return 0; +} + +static bool ipv4acd_arp_conflict(sd_ipv4acd *ll, struct ether_arp *arp) { + assert(ll); + assert(arp); + + /* see the BPF */ + if (memcmp(arp->arp_spa, &ll->address, sizeof(ll->address)) == 0) + return true; + + /* the TPA matched instead of the SPA, this is not a conflict */ + return false; +} + +static int ipv4acd_on_timeout(sd_event_source *s, uint64_t usec, void *userdata) { + sd_ipv4acd *ll = userdata; + int r = 0; + + assert(ll); + + switch (ll->state) { + case IPV4ACD_STATE_INIT: + + ipv4acd_set_state(ll, IPV4ACD_STATE_WAITING_PROBE, true); + + if (ll->conflict >= MAX_CONFLICTS) { + log_ipv4acd_notice(ll, "Max conflicts reached, delaying by %us", RATE_LIMIT_INTERVAL); + r = ipv4acd_set_next_wakeup(ll, RATE_LIMIT_INTERVAL, PROBE_WAIT); + if (r < 0) + goto out; + + ll->conflict = 0; + } else { + r = ipv4acd_set_next_wakeup(ll, 0, PROBE_WAIT); + if (r < 0) + goto out; + } + + break; + case IPV4ACD_STATE_WAITING_PROBE: + case IPV4ACD_STATE_PROBING: + /* Send a probe */ + r = arp_send_probe(ll->fd, ll->index, ll->address, &ll->mac_addr); + if (r < 0) { + log_ipv4acd_error_errno(ll, r, "Failed to send ARP probe: %m"); + goto out; + } else { + _cleanup_free_ char *address = NULL; + union in_addr_union addr = { .in.s_addr = ll->address }; + + r = in_addr_to_string(AF_INET, &addr, &address); + if (r >= 0) + log_ipv4acd_debug(ll, "Probing %s", address); + } + + if (ll->iteration < PROBE_NUM - 2) { + ipv4acd_set_state(ll, IPV4ACD_STATE_PROBING, false); + + r = ipv4acd_set_next_wakeup(ll, PROBE_MIN, (PROBE_MAX-PROBE_MIN)); + if (r < 0) + goto out; + } else { + ipv4acd_set_state(ll, IPV4ACD_STATE_WAITING_ANNOUNCE, true); + + r = ipv4acd_set_next_wakeup(ll, ANNOUNCE_WAIT, 0); + if (r < 0) + goto out; + } + + break; + + case IPV4ACD_STATE_ANNOUNCING: + if (ll->iteration >= ANNOUNCE_NUM - 1) { + ipv4acd_set_state(ll, IPV4ACD_STATE_RUNNING, false); + + break; + } + case IPV4ACD_STATE_WAITING_ANNOUNCE: + /* Send announcement packet */ + r = arp_send_announcement(ll->fd, ll->index, ll->address, &ll->mac_addr); + if (r < 0) { + log_ipv4acd_error_errno(ll, r, "Failed to send ARP announcement: %m"); + goto out; + } else + log_ipv4acd_debug(ll, "ANNOUNCE"); + + ipv4acd_set_state(ll, IPV4ACD_STATE_ANNOUNCING, false); + + r = ipv4acd_set_next_wakeup(ll, ANNOUNCE_INTERVAL, 0); + if (r < 0) + goto out; + + if (ll->iteration == 0) { + ll->conflict = 0; + ipv4acd_client_notify(ll, IPV4ACD_EVENT_BIND); + } + + break; + default: + assert_not_reached("Invalid state."); + } + +out: + if (r < 0) + sd_ipv4acd_stop(ll); + + return 1; +} + +static void ipv4acd_on_conflict(sd_ipv4acd *ll) { + _cleanup_free_ char *address = NULL; + union in_addr_union addr = { .in.s_addr = ll->address }; + int r; + + assert(ll); + + ll->conflict++; + + r = in_addr_to_string(AF_INET, &addr, &address); + if (r >= 0) + log_ipv4acd_debug(ll, "Conflict on %s (%u)", address, ll->conflict); + + ipv4acd_stop(ll); + + ipv4acd_client_notify(ll, IPV4ACD_EVENT_CONFLICT); +} + +static int ipv4acd_on_packet(sd_event_source *s, int fd, + uint32_t revents, void *userdata) { + sd_ipv4acd *ll = userdata; + struct ether_arp packet; + int r; + + assert(ll); + assert(fd >= 0); + + r = read(fd, &packet, sizeof(struct ether_arp)); + if (r < (int) sizeof(struct ether_arp)) + goto out; + + switch (ll->state) { + case IPV4ACD_STATE_ANNOUNCING: + case IPV4ACD_STATE_RUNNING: + if (ipv4acd_arp_conflict(ll, &packet)) { + usec_t ts; + + assert_se(sd_event_now(ll->event, clock_boottime_or_monotonic(), &ts) >= 0); + + /* Defend address */ + if (ts > ll->defend_window) { + ll->defend_window = ts + DEFEND_INTERVAL * USEC_PER_SEC; + r = arp_send_announcement(ll->fd, ll->index, ll->address, &ll->mac_addr); + if (r < 0) { + log_ipv4acd_error_errno(ll, r, "Failed to send ARP announcement: %m"); + goto out; + } else + log_ipv4acd_debug(ll, "DEFEND"); + + } else + ipv4acd_on_conflict(ll); + } + + break; + case IPV4ACD_STATE_WAITING_PROBE: + case IPV4ACD_STATE_PROBING: + case IPV4ACD_STATE_WAITING_ANNOUNCE: + /* BPF ensures this packet indicates a conflict */ + ipv4acd_on_conflict(ll); + + break; + default: + assert_not_reached("Invalid state."); + } + +out: + if (r < 0) + sd_ipv4acd_stop(ll); + + return 1; +} + +int sd_ipv4acd_set_index(sd_ipv4acd *ll, int interface_index) { + assert_return(ll, -EINVAL); + assert_return(interface_index > 0, -EINVAL); + assert_return(ll->state == IPV4ACD_STATE_INIT, -EBUSY); + + ll->index = interface_index; + + return 0; +} + +int sd_ipv4acd_set_mac(sd_ipv4acd *ll, const struct ether_addr *addr) { + assert_return(ll, -EINVAL); + assert_return(addr, -EINVAL); + assert_return(ll->state == IPV4ACD_STATE_INIT, -EBUSY); + + memcpy(&ll->mac_addr, addr, ETH_ALEN); + + return 0; +} + +int sd_ipv4acd_detach_event(sd_ipv4acd *ll) { + assert_return(ll, -EINVAL); + + ll->event = sd_event_unref(ll->event); + + return 0; +} + +int sd_ipv4acd_attach_event(sd_ipv4acd *ll, sd_event *event, int priority) { + int r; + + assert_return(ll, -EINVAL); + assert_return(!ll->event, -EBUSY); + + if (event) + ll->event = sd_event_ref(event); + else { + r = sd_event_default(&ll->event); + if (r < 0) + return r; + } + + ll->event_priority = priority; + + return 0; +} + +int sd_ipv4acd_set_callback(sd_ipv4acd *ll, sd_ipv4acd_cb_t cb, void *userdata) { + assert_return(ll, -EINVAL); + + ll->cb = cb; + ll->userdata = userdata; + + return 0; +} + +int sd_ipv4acd_set_address(sd_ipv4acd *ll, const struct in_addr *address){ + assert_return(ll, -EINVAL); + assert_return(address, -EINVAL); + assert_return(ll->state == IPV4ACD_STATE_INIT, -EBUSY); + + ll->address = address->s_addr; + + return 0; +} + +bool sd_ipv4acd_is_running(sd_ipv4acd *ll) { + assert_return(ll, false); + + return ll->state != IPV4ACD_STATE_INIT; +} + +static bool ether_addr_is_nul(const struct ether_addr *addr) { + const struct ether_addr nul_addr = {}; + + assert(addr); + + return memcmp(addr, &nul_addr, sizeof(struct ether_addr)) == 0; +} + +#define HASH_KEY SD_ID128_MAKE(df,04,22,98,3f,ad,14,52,f9,87,2e,d1,9c,70,e2,f2) + +int sd_ipv4acd_start(sd_ipv4acd *ll) { + int r; + + assert_return(ll, -EINVAL); + assert_return(ll->event, -EINVAL); + assert_return(ll->index > 0, -EINVAL); + assert_return(ll->address != 0, -EINVAL); + assert_return(!ether_addr_is_nul(&ll->mac_addr), -EINVAL); + assert_return(ll->state == IPV4ACD_STATE_INIT, -EBUSY); + + ll->defend_window = 0; + + r = arp_network_bind_raw_socket(ll->index, ll->address, &ll->mac_addr); + if (r < 0) + goto out; + + ll->fd = safe_close(ll->fd); + ll->fd = r; + + r = sd_event_add_io(ll->event, &ll->receive_message, ll->fd, + EPOLLIN, ipv4acd_on_packet, ll); + if (r < 0) + goto out; + + r = sd_event_source_set_priority(ll->receive_message, ll->event_priority); + if (r < 0) + goto out; + + r = sd_event_source_set_description(ll->receive_message, "ipv4acd-receive-message"); + if (r < 0) + goto out; + + r = ipv4acd_set_next_wakeup(ll, 0, 0); + if (r < 0) + goto out; +out: + if (r < 0) { + ipv4acd_stop(ll); + return r; + } + + return 0; +} diff --git a/src/libsystemd-network/sd-ipv4ll.c b/src/libsystemd-network/sd-ipv4ll.c index 14b9444dab..f0230b919c 100644 --- a/src/libsystemd-network/sd-ipv4ll.c +++ b/src/libsystemd-network/sd-ipv4ll.c @@ -2,6 +2,7 @@ This file is part of systemd. Copyright (C) 2014 Axis Communications AB. All rights reserved. + Copyright (C) 2015 Tom Gundersen 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 @@ -23,429 +24,153 @@ #include <stdio.h> #include <arpa/inet.h> -#include "util.h" -#include "siphash24.h" +#include "event-util.h" #include "list.h" #include "random-util.h" +#include "refcnt.h" +#include "siphash24.h" +#include "sparse-endian.h" +#include "util.h" -#include "ipv4ll-internal.h" +#include "sd-ipv4acd.h" #include "sd-ipv4ll.h" -/* Constants from the RFC */ -#define PROBE_WAIT 1 -#define PROBE_NUM 3 -#define PROBE_MIN 1 -#define PROBE_MAX 2 -#define ANNOUNCE_WAIT 2 -#define ANNOUNCE_NUM 2 -#define ANNOUNCE_INTERVAL 2 -#define MAX_CONFLICTS 10 -#define RATE_LIMIT_INTERVAL 60 -#define DEFEND_INTERVAL 10 - #define IPV4LL_NETWORK 0xA9FE0000L #define IPV4LL_NETMASK 0xFFFF0000L -typedef enum IPv4LLTrigger{ - IPV4LL_TRIGGER_NULL, - IPV4LL_TRIGGER_PACKET, - IPV4LL_TRIGGER_TIMEOUT, - _IPV4LL_TRIGGER_MAX, - _IPV4LL_TRIGGER_INVALID = -1 -} IPv4LLTrigger; - -typedef enum IPv4LLState { - IPV4LL_STATE_INIT, - IPV4LL_STATE_WAITING_PROBE, - IPV4LL_STATE_PROBING, - IPV4LL_STATE_WAITING_ANNOUNCE, - IPV4LL_STATE_ANNOUNCING, - IPV4LL_STATE_RUNNING, - IPV4LL_STATE_STOPPED, - _IPV4LL_STATE_MAX, - _IPV4LL_STATE_INVALID = -1 -} IPv4LLState; +#define IPV4LL_DONT_DESTROY(ll) \ + _cleanup_ipv4ll_unref_ _unused_ sd_ipv4ll *_dont_destroy_##ll = sd_ipv4ll_ref(ll) struct sd_ipv4ll { unsigned n_ref; - IPv4LLState state; - int index; - int fd; - union sockaddr_union link; - int iteration; - int conflict; - sd_event_source *receive_message; - sd_event_source *timer; - usec_t next_wakeup; - usec_t defend_window; - int next_wakeup_valid; - be32_t address; + sd_ipv4acd *acd; + be32_t address; /* the address pushed to ACD */ struct random_data *random_data; char *random_data_state; + /* External */ be32_t claimed_address; - struct ether_addr mac_addr; - sd_event *event; - int event_priority; sd_ipv4ll_cb_t cb; void* userdata; }; -static void ipv4ll_run_state_machine(sd_ipv4ll *ll, IPv4LLTrigger trigger, void *trigger_data); - -static void ipv4ll_set_state(sd_ipv4ll *ll, IPv4LLState st, int reset_counter) { - - assert(ll); - assert(st < _IPV4LL_STATE_MAX); - - if (st == ll->state && !reset_counter) { - ll->iteration++; - } else { - ll->state = st; - ll->iteration = 0; - } -} - -static sd_ipv4ll *ipv4ll_client_notify(sd_ipv4ll *ll, int event) { - assert(ll); - - if (ll->cb) { - ll = sd_ipv4ll_ref(ll); - ll->cb(ll, event, ll->userdata); - ll = sd_ipv4ll_unref(ll); - } - - return ll; -} - -static sd_ipv4ll *ipv4ll_stop(sd_ipv4ll *ll, int event) { - assert(ll); - - ll->receive_message = sd_event_source_unref(ll->receive_message); - ll->fd = safe_close(ll->fd); - - ll->timer = sd_event_source_unref(ll->timer); - - log_ipv4ll(ll, "STOPPED"); - - ll = ipv4ll_client_notify(ll, event); +sd_ipv4ll *sd_ipv4ll_ref(sd_ipv4ll *ll) { + if (!ll) + return NULL; - if (ll) { - ll->claimed_address = 0; - ipv4ll_set_state (ll, IPV4LL_STATE_INIT, 1); - } + assert(ll->n_ref >= 1); + ll->n_ref++; return ll; } -static int ipv4ll_pick_address(sd_ipv4ll *ll, be32_t *address) { - be32_t addr; - int r; - int32_t random; - - assert(ll); - assert(address); - assert(ll->random_data); - - do { - r = random_r(ll->random_data, &random); - if (r < 0) - return r; - addr = htonl((random & 0x0000FFFF) | IPV4LL_NETWORK); - } while (addr == ll->address || - (ntohl(addr) & IPV4LL_NETMASK) != IPV4LL_NETWORK || - (ntohl(addr) & 0x0000FF00) == 0x0000 || - (ntohl(addr) & 0x0000FF00) == 0xFF00); - - *address = addr; - return 0; -} - -static int ipv4ll_timer(sd_event_source *s, uint64_t usec, void *userdata) { - sd_ipv4ll *ll = (sd_ipv4ll*)userdata; - - assert(ll); - - ll->next_wakeup_valid = 0; - ipv4ll_run_state_machine(ll, IPV4LL_TRIGGER_TIMEOUT, NULL); - - return 0; -} - -static void ipv4ll_set_next_wakeup(sd_ipv4ll *ll, int sec, int random_sec) { - usec_t next_timeout = 0; - usec_t time_now = 0; - - assert(sec >= 0); - assert(random_sec >= 0); - assert(ll); - - next_timeout = sec * USEC_PER_SEC; - - if (random_sec) - next_timeout += random_u32() % (random_sec * USEC_PER_SEC); - - assert_se(sd_event_now(ll->event, clock_boottime_or_monotonic(), &time_now) >= 0); - - ll->next_wakeup = time_now + next_timeout; - ll->next_wakeup_valid = 1; -} - -static bool ipv4ll_arp_conflict (sd_ipv4ll *ll, struct ether_arp *arp) { - assert(ll); - assert(arp); - - if (memcmp(arp->arp_spa, &ll->address, sizeof(ll->address)) == 0 && - memcmp(arp->arp_sha, &ll->mac_addr, ETH_ALEN) != 0) - return true; +sd_ipv4ll *sd_ipv4ll_unref(sd_ipv4ll *ll) { + if (!ll) + return NULL; - return false; -} + assert(ll->n_ref >= 1); + ll->n_ref--; -static bool ipv4ll_arp_probe_conflict (sd_ipv4ll *ll, struct ether_arp *arp) { - assert(ll); - assert(arp); + if (ll->n_ref > 0) + return NULL; - if (ipv4ll_arp_conflict(ll, arp)) - return true; + sd_ipv4acd_unref(ll->acd); - if (memcmp(arp->arp_tpa, &ll->address, sizeof(ll->address)) == 0 && - memcmp(arp->arp_sha, &ll->mac_addr, ETH_ALEN)) - return true; + free(ll->random_data); + free(ll->random_data_state); + free(ll); - return false; + return NULL; } -static void ipv4ll_run_state_machine(sd_ipv4ll *ll, IPv4LLTrigger trigger, void *trigger_data) { - struct ether_arp out_packet; - int out_packet_ready = 0; - int r = 0; - - assert(ll); - assert(trigger < _IPV4LL_TRIGGER_MAX); - - if (ll->state == IPV4LL_STATE_INIT) { - - log_ipv4ll(ll, "PROBE"); - ipv4ll_set_state(ll, IPV4LL_STATE_WAITING_PROBE, 1); - ipv4ll_set_next_wakeup(ll, 0, PROBE_WAIT); - - } else if ((ll->state == IPV4LL_STATE_WAITING_PROBE && trigger == IPV4LL_TRIGGER_TIMEOUT) || - (ll->state == IPV4LL_STATE_PROBING && trigger == IPV4LL_TRIGGER_TIMEOUT && ll->iteration < PROBE_NUM-2)) { - - /* Send a probe */ - arp_packet_probe(&out_packet, ll->address, &ll->mac_addr); - out_packet_ready = 1; - ipv4ll_set_state(ll, IPV4LL_STATE_PROBING, 0); - - ipv4ll_set_next_wakeup(ll, PROBE_MIN, (PROBE_MAX-PROBE_MIN)); - - } else if (ll->state == IPV4LL_STATE_PROBING && trigger == IPV4LL_TRIGGER_TIMEOUT && ll->iteration >= PROBE_NUM-2) { - - /* Send the last probe */ - arp_packet_probe(&out_packet, ll->address, &ll->mac_addr); - out_packet_ready = 1; - ipv4ll_set_state(ll, IPV4LL_STATE_WAITING_ANNOUNCE, 1); - - ipv4ll_set_next_wakeup(ll, ANNOUNCE_WAIT, 0); - - } else if ((ll->state == IPV4LL_STATE_WAITING_ANNOUNCE && trigger == IPV4LL_TRIGGER_TIMEOUT) || - (ll->state == IPV4LL_STATE_ANNOUNCING && trigger == IPV4LL_TRIGGER_TIMEOUT && ll->iteration < ANNOUNCE_NUM-1)) { - - /* Send announcement packet */ - arp_packet_announcement(&out_packet, ll->address, &ll->mac_addr); - out_packet_ready = 1; - ipv4ll_set_state(ll, IPV4LL_STATE_ANNOUNCING, 0); - - ipv4ll_set_next_wakeup(ll, ANNOUNCE_INTERVAL, 0); - - if (ll->iteration == 0) { - log_ipv4ll(ll, "ANNOUNCE"); - ll->claimed_address = ll->address; - ll = ipv4ll_client_notify(ll, IPV4LL_EVENT_BIND); - if (!ll || ll->state == IPV4LL_STATE_STOPPED) - goto out; - - ll->conflict = 0; - } - - } else if ((ll->state == IPV4LL_STATE_ANNOUNCING && trigger == IPV4LL_TRIGGER_TIMEOUT && - ll->iteration >= ANNOUNCE_NUM-1)) { - - ipv4ll_set_state(ll, IPV4LL_STATE_RUNNING, 0); - ll->next_wakeup_valid = 0; - - } else if (trigger == IPV4LL_TRIGGER_PACKET) { - - int conflicted = 0; - usec_t time_now; - struct ether_arp* in_packet = (struct ether_arp*)trigger_data; - - assert(in_packet); - - if (IN_SET(ll->state, IPV4LL_STATE_ANNOUNCING, IPV4LL_STATE_RUNNING)) { - - if (ipv4ll_arp_conflict(ll, in_packet)) { - - r = sd_event_now(ll->event, clock_boottime_or_monotonic(), &time_now); - if (r < 0) - goto out; - - /* Defend address */ - if (time_now > ll->defend_window) { - ll->defend_window = time_now + DEFEND_INTERVAL * USEC_PER_SEC; - arp_packet_announcement(&out_packet, ll->address, &ll->mac_addr); - out_packet_ready = 1; - } else - conflicted = 1; - } - - } else if (IN_SET(ll->state, IPV4LL_STATE_WAITING_PROBE, - IPV4LL_STATE_PROBING, - IPV4LL_STATE_WAITING_ANNOUNCE)) { - - conflicted = ipv4ll_arp_probe_conflict(ll, in_packet); - } - - if (conflicted) { - log_ipv4ll(ll, "CONFLICT"); - ll = ipv4ll_client_notify(ll, IPV4LL_EVENT_CONFLICT); - if (!ll || ll->state == IPV4LL_STATE_STOPPED) - goto out; +DEFINE_TRIVIAL_CLEANUP_FUNC(sd_ipv4ll*, sd_ipv4ll_unref); +#define _cleanup_ipv4ll_unref_ _cleanup_(sd_ipv4ll_unrefp) - ll->claimed_address = 0; +static void ipv4ll_on_acd(sd_ipv4acd *ll, int event, void *userdata); - /* Pick a new address */ - r = ipv4ll_pick_address(ll, &ll->address); - if (r < 0) - goto out; - ll->conflict++; - ll->defend_window = 0; - ipv4ll_set_state(ll, IPV4LL_STATE_WAITING_PROBE, 1); +int sd_ipv4ll_new(sd_ipv4ll **ret) { + _cleanup_ipv4ll_unref_ sd_ipv4ll *ll = NULL; + int r; - if (ll->conflict >= MAX_CONFLICTS) { - log_ipv4ll(ll, "MAX_CONFLICTS"); - ipv4ll_set_next_wakeup(ll, RATE_LIMIT_INTERVAL, PROBE_WAIT); - } else - ipv4ll_set_next_wakeup(ll, 0, PROBE_WAIT); + assert_return(ret, -EINVAL); - } - } + ll = new0(sd_ipv4ll, 1); + if (!ll) + return -ENOMEM; - if (out_packet_ready) { - r = arp_network_send_raw_socket(ll->fd, &ll->link, &out_packet); - if (r < 0) { - log_ipv4ll(ll, "failed to send arp packet out"); - goto out; - } - } + r = sd_ipv4acd_new(&ll->acd); + if (r < 0) + return r; - if (ll->next_wakeup_valid) { - ll->timer = sd_event_source_unref(ll->timer); - r = sd_event_add_time(ll->event, &ll->timer, clock_boottime_or_monotonic(), - ll->next_wakeup, 0, ipv4ll_timer, ll); - if (r < 0) - goto out; + r = sd_ipv4acd_set_callback(ll->acd, ipv4ll_on_acd, ll); + if (r < 0) + return r; - r = sd_event_source_set_priority(ll->timer, ll->event_priority); - if (r < 0) - goto out; + ll->n_ref = 1; - r = sd_event_source_set_description(ll->timer, "ipv4ll-timer"); - if (r < 0) - goto out; - } + *ret = ll; + ll = NULL; -out: - if (r < 0 && ll) - ipv4ll_stop(ll, r); + return 0; } -static int ipv4ll_receive_message(sd_event_source *s, int fd, - uint32_t revents, void *userdata) { +int sd_ipv4ll_stop(sd_ipv4ll *ll) { int r; - struct ether_arp arp; - sd_ipv4ll *ll = (sd_ipv4ll*)userdata; - assert(ll); - - r = read(fd, &arp, sizeof(struct ether_arp)); - if (r < (int) sizeof(struct ether_arp)) - return 0; + assert_return(ll, -EINVAL); - r = arp_packet_verify_headers(&arp); + r = sd_ipv4acd_stop(ll->acd); if (r < 0) - return 0; - - ipv4ll_run_state_machine(ll, IPV4LL_TRIGGER_PACKET, &arp); + return r; return 0; } int sd_ipv4ll_set_index(sd_ipv4ll *ll, int interface_index) { assert_return(ll, -EINVAL); - assert_return(interface_index > 0, -EINVAL); - assert_return(IN_SET(ll->state, IPV4LL_STATE_INIT, - IPV4LL_STATE_STOPPED), -EBUSY); - ll->index = interface_index; - - return 0; + return sd_ipv4acd_set_index(ll->acd, interface_index); } +#define HASH_KEY SD_ID128_MAKE(df,04,22,98,3f,ad,14,52,f9,87,2e,d1,9c,70,e2,f2) + int sd_ipv4ll_set_mac(sd_ipv4ll *ll, const struct ether_addr *addr) { - bool need_restart = false; + int r; assert_return(ll, -EINVAL); - assert_return(addr, -EINVAL); - - if (memcmp(&ll->mac_addr, addr, ETH_ALEN) == 0) - return 0; - if (!IN_SET(ll->state, IPV4LL_STATE_INIT, IPV4LL_STATE_STOPPED)) { - log_ipv4ll(ll, "Changing MAC address on running IPv4LL " - "client, restarting"); - ll = ipv4ll_stop(ll, IPV4LL_EVENT_STOP); - need_restart = true; - } + if (!ll->random_data) { + uint8_t seed[8]; - if (!ll) - return 0; + /* If no random data is set, generate some from the MAC */ + siphash24(seed, &addr->ether_addr_octet, + ETH_ALEN, HASH_KEY.bytes); - memcpy(&ll->mac_addr, addr, ETH_ALEN); + assert_cc(sizeof(unsigned) <= 8); - if (need_restart) - sd_ipv4ll_start(ll); + r = sd_ipv4ll_set_address_seed(ll, *(unsigned*)seed); + if (r < 0) + return r; + } - return 0; + return sd_ipv4acd_set_mac(ll->acd, addr); } int sd_ipv4ll_detach_event(sd_ipv4ll *ll) { assert_return(ll, -EINVAL); - ll->event = sd_event_unref(ll->event); - - return 0; + return sd_ipv4acd_detach_event(ll->acd); } int sd_ipv4ll_attach_event(sd_ipv4ll *ll, sd_event *event, int priority) { int r; assert_return(ll, -EINVAL); - assert_return(!ll->event, -EBUSY); - - if (event) - ll->event = sd_event_ref(event); - else { - r = sd_event_default(&ll->event); - if (r < 0) { - ipv4ll_stop(ll, IPV4LL_EVENT_STOP); - return r; - } - } - ll->event_priority = priority; + r = sd_ipv4acd_attach_event(ll->acd, event, priority); + if (r < 0) + return r; return 0; } @@ -467,189 +192,147 @@ int sd_ipv4ll_get_address(sd_ipv4ll *ll, struct in_addr *address){ return -ENOENT; address->s_addr = ll->claimed_address; + return 0; } -int sd_ipv4ll_set_address_seed (sd_ipv4ll *ll, uint8_t seed[8]) { - unsigned int entropy; +int sd_ipv4ll_set_address_seed(sd_ipv4ll *ll, unsigned seed) { + _cleanup_free_ struct random_data *random_data = NULL; + _cleanup_free_ char *random_data_state = NULL; int r; assert_return(ll, -EINVAL); assert_return(seed, -EINVAL); - entropy = *seed; + random_data = new0(struct random_data, 1); + if (!random_data) + return -ENOMEM; - free(ll->random_data); - free(ll->random_data_state); + random_data_state = new0(char, 128); + if (!random_data_state) + return -ENOMEM; - ll->random_data = new0(struct random_data, 1); - ll->random_data_state = new0(char, 128); + r = initstate_r(seed, random_data_state, 128, random_data); + if (r < 0) + return r; - if (!ll->random_data || !ll->random_data_state) { - r = -ENOMEM; - goto error; - } + free(ll->random_data); + ll->random_data = random_data; + random_data = NULL; - r = initstate_r((unsigned int)entropy, ll->random_data_state, 128, ll->random_data); - if (r < 0) - goto error; + free(ll->random_data_state); + ll->random_data_state = random_data_state; + random_data_state = NULL; -error: - if (r < 0){ - free(ll->random_data); - free(ll->random_data_state); - ll->random_data = NULL; - ll->random_data_state = NULL; - } - return r; + return 0; } bool sd_ipv4ll_is_running(sd_ipv4ll *ll) { assert_return(ll, false); - return !IN_SET(ll->state, IPV4LL_STATE_INIT, IPV4LL_STATE_STOPPED); + return sd_ipv4acd_is_running(ll->acd); } -#define HASH_KEY SD_ID128_MAKE(df,04,22,98,3f,ad,14,52,f9,87,2e,d1,9c,70,e2,f2) - -int sd_ipv4ll_start (sd_ipv4ll *ll) { +static int ipv4ll_pick_address(sd_ipv4ll *ll) { + struct in_addr in_addr; + be32_t addr; int r; + int32_t random; - assert_return(ll, -EINVAL); - assert_return(ll->event, -EINVAL); - assert_return(ll->index > 0, -EINVAL); - assert_return(IN_SET(ll->state, IPV4LL_STATE_INIT, - IPV4LL_STATE_STOPPED), -EBUSY); + assert(ll); + assert(ll->random_data); - ll->state = IPV4LL_STATE_INIT; + do { + r = random_r(ll->random_data, &random); + if (r < 0) + return r; + addr = htonl((random & 0x0000FFFF) | IPV4LL_NETWORK); + } while (addr == ll->address || + (ntohl(addr) & IPV4LL_NETMASK) != IPV4LL_NETWORK || + (ntohl(addr) & 0x0000FF00) == 0x0000 || + (ntohl(addr) & 0x0000FF00) == 0xFF00); - r = arp_network_bind_raw_socket(ll->index, &ll->link); + in_addr.s_addr = addr; + r = sd_ipv4acd_set_address(ll->acd, &in_addr); if (r < 0) - goto out; + return r; - ll->fd = r; - ll->conflict = 0; - ll->defend_window = 0; - ll->claimed_address = 0; + ll->address = addr; - if (!ll->random_data) { - uint8_t seed[8]; + return 0; +} - /* Fallback to mac */ - siphash24(seed, &ll->mac_addr.ether_addr_octet, - ETH_ALEN, HASH_KEY.bytes); +int sd_ipv4ll_start(sd_ipv4ll *ll) { + int r; - r = sd_ipv4ll_set_address_seed(ll, seed); - if (r < 0) - goto out; - } + assert_return(ll, -EINVAL); + assert_return(ll->random_data, -EINVAL); if (ll->address == 0) { - r = ipv4ll_pick_address(ll, &ll->address); + r = ipv4ll_pick_address(ll); if (r < 0) - goto out; + return r; } - ipv4ll_set_state (ll, IPV4LL_STATE_INIT, 1); - - r = sd_event_add_io(ll->event, &ll->receive_message, ll->fd, - EPOLLIN, ipv4ll_receive_message, ll); + r = sd_ipv4acd_start(ll->acd); if (r < 0) - goto out; - - r = sd_event_source_set_priority(ll->receive_message, ll->event_priority); - if (r < 0) - goto out; - - r = sd_event_source_set_description(ll->receive_message, "ipv4ll-receive-message"); - if (r < 0) - goto out; - - r = sd_event_add_time(ll->event, - &ll->timer, - clock_boottime_or_monotonic(), - now(clock_boottime_or_monotonic()), 0, - ipv4ll_timer, ll); - - if (r < 0) - goto out; - - r = sd_event_source_set_priority(ll->timer, ll->event_priority); - if (r < 0) - goto out; - - r = sd_event_source_set_description(ll->timer, "ipv4ll-timer"); -out: - if (r < 0) - ipv4ll_stop(ll, IPV4LL_EVENT_STOP); + return r; return 0; } -int sd_ipv4ll_stop(sd_ipv4ll *ll) { - ipv4ll_stop(ll, IPV4LL_EVENT_STOP); - if (ll) - ipv4ll_set_state(ll, IPV4LL_STATE_STOPPED, 1); - - return 0; -} - -sd_ipv4ll *sd_ipv4ll_ref(sd_ipv4ll *ll) { - - if (!ll) - return NULL; - - assert(ll->n_ref >= 1); - ll->n_ref++; +static void ipv4ll_client_notify(sd_ipv4ll *ll, int event) { + assert(ll); - return ll; + if (ll->cb) + ll->cb(ll, event, ll->userdata); } -sd_ipv4ll *sd_ipv4ll_unref(sd_ipv4ll *ll) { - - if (!ll) - return NULL; - - assert(ll->n_ref >= 1); - ll->n_ref--; - - if (ll->n_ref > 0) - return ll; - - ll->receive_message = sd_event_source_unref(ll->receive_message); - ll->fd = safe_close(ll->fd); - - ll->timer = sd_event_source_unref(ll->timer); +void ipv4ll_on_acd(sd_ipv4acd *acd, int event, void *userdata) { + sd_ipv4ll *ll = userdata; + IPV4LL_DONT_DESTROY(ll); + int r; - sd_ipv4ll_detach_event(ll); + assert(acd); + assert(ll); - free(ll->random_data); - free(ll->random_data_state); - free(ll); + switch (event) { + case IPV4ACD_EVENT_STOP: + ipv4ll_client_notify(ll, IPV4LL_EVENT_STOP); - return NULL; -} + ll->claimed_address = 0; -DEFINE_TRIVIAL_CLEANUP_FUNC(sd_ipv4ll*, sd_ipv4ll_unref); -#define _cleanup_ipv4ll_free_ _cleanup_(sd_ipv4ll_unrefp) + break; + case IPV4ACD_EVENT_BIND: + ll->claimed_address = ll->address; + ipv4ll_client_notify(ll, IPV4LL_EVENT_BIND); -int sd_ipv4ll_new(sd_ipv4ll **ret) { - _cleanup_ipv4ll_free_ sd_ipv4ll *ll = NULL; + break; + case IPV4ACD_EVENT_CONFLICT: + /* if an address was already bound we must call up to the + user to handle this, otherwise we just try again */ + if (ll->claimed_address != 0) { + ipv4ll_client_notify(ll, IPV4LL_EVENT_CONFLICT); - assert_return(ret, -EINVAL); + ll->claimed_address = 0; + } else { + r = ipv4ll_pick_address(ll); + if (r < 0) + goto error; - ll = new0(sd_ipv4ll, 1); - if (!ll) - return -ENOMEM; + r = sd_ipv4acd_start(ll->acd); + if (r < 0) + goto error; + } - ll->n_ref = 1; - ll->state = IPV4LL_STATE_INIT; - ll->index = -1; - ll->fd = -1; + break; + default: + assert_not_reached("Invalid IPv4ACD event."); + } - *ret = ll; - ll = NULL; + return; - return 0; +error: + ipv4ll_client_notify(ll, IPV4LL_EVENT_STOP); } diff --git a/src/libsystemd-network/test-acd.c b/src/libsystemd-network/test-acd.c new file mode 100644 index 0000000000..f1f14ee119 --- /dev/null +++ b/src/libsystemd-network/test-acd.c @@ -0,0 +1,117 @@ +/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ + +/*** + This file is part of systemd. + + Copyright (C) 2014 Tom Gundersen <teg@jklm.no> + + 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 <stdlib.h> +#include <errno.h> +#include <unistd.h> + +#include <linux/veth.h> +#include <net/if.h> + +#include "sd-event.h" +#include "sd-netlink.h" +#include "sd-ipv4acd.h" + +#include "util.h" +#include "event-util.h" +#include "netlink-util.h" +#include "in-addr-util.h" + +static void acd_handler(sd_ipv4acd *acd, int event, void *userdata) { + assert_se(acd); + + switch (event) { + case IPV4ACD_EVENT_BIND: + log_info("bound"); + break; + case IPV4ACD_EVENT_CONFLICT: + log_info("conflict"); + break; + case IPV4ACD_EVENT_STOP: + log_error("the client was stopped"); + break; + default: + assert_not_reached("invalid ACD event"); + } +} + +static int client_run(int ifindex, const struct in_addr *pa, const struct ether_addr *ha, sd_event *e) { + sd_ipv4acd *acd; + + assert_se(sd_ipv4acd_new(&acd) >= 0); + assert_se(sd_ipv4acd_attach_event(acd, e, 0) >= 0); + + assert_se(sd_ipv4acd_set_index(acd, ifindex) >= 0); + assert_se(sd_ipv4acd_set_mac(acd, ha) >= 0); + assert_se(sd_ipv4acd_set_address(acd, pa) >= 0); + assert_se(sd_ipv4acd_set_callback(acd, acd_handler, NULL) >= 0); + + log_info("starting IPv4ACD client"); + + assert_se(sd_ipv4acd_start(acd) >= 0); + + assert_se(sd_event_loop(e) >= 0); + + assert_se(!sd_ipv4acd_unref(acd)); + + return EXIT_SUCCESS; +} + +static int test_acd(const char *ifname, const char *address) { + _cleanup_event_unref_ sd_event *e = NULL; + _cleanup_netlink_unref_ sd_netlink *rtnl = NULL; + _cleanup_netlink_message_unref_ sd_netlink_message *m = NULL, *reply = NULL; + union in_addr_union pa; + struct ether_addr ha; + int ifindex; + + assert_se(in_addr_from_string(AF_INET, address, &pa) >= 0); + + assert_se(sd_event_new(&e) >= 0); + + assert_se(sd_netlink_open(&rtnl) >= 0); + assert_se(sd_netlink_attach_event(rtnl, e, 0) >= 0); + + assert_se(sd_rtnl_message_new_link(rtnl, &m, RTM_GETLINK, 0) >= 0); + assert_se(sd_netlink_message_append_string(m, IFLA_IFNAME, ifname) >= 0); + assert_se(sd_netlink_call(rtnl, m, 0, &reply) >= 0); + + assert_se(sd_rtnl_message_link_get_ifindex(reply, &ifindex) >= 0); + assert_se(sd_netlink_message_read_ether_addr(reply, IFLA_ADDRESS, &ha) >= 0); + + client_run(ifindex, &pa.in, &ha, e); + + return EXIT_SUCCESS; +} + +int main(int argc, char *argv[]) { + log_set_max_level(LOG_DEBUG); + log_parse_environment(); + log_open(); + + if (argc == 3) + return test_acd(argv[1], argv[2]); + else { + log_error("This program takes two arguments.\n" + "\t %s <ifname> <IPv4 address>", program_invocation_short_name); + return EXIT_FAILURE; + } +} diff --git a/src/libsystemd-network/test-ipv4ll-manual.c b/src/libsystemd-network/test-ipv4ll-manual.c new file mode 100644 index 0000000000..ad664cba51 --- /dev/null +++ b/src/libsystemd-network/test-ipv4ll-manual.c @@ -0,0 +1,129 @@ +/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ + +/*** + This file is part of systemd. + + Copyright (C) 2014 Tom Gundersen <teg@jklm.no> + + 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 <stdlib.h> +#include <errno.h> +#include <unistd.h> + +#include <linux/veth.h> +#include <net/if.h> + +#include "sd-event.h" +#include "sd-netlink.h" +#include "sd-ipv4ll.h" + +#include "util.h" +#include "event-util.h" +#include "netlink-util.h" +#include "in-addr-util.h" + +static void ll_handler(sd_ipv4ll *ll, int event, void *userdata) { + _cleanup_free_ char *address = NULL; + struct in_addr addr = {}; + + assert_se(ll); + + if (sd_ipv4ll_get_address(ll, &addr) >= 0) + assert_se(in_addr_to_string(AF_INET, (const union in_addr_union*) &addr, &address) >= 0); + + switch (event) { + case IPV4LL_EVENT_BIND: + log_info("bound %s", strna(address)); + break; + case IPV4LL_EVENT_CONFLICT: + log_info("conflict on %s", strna(address)); + break; + case IPV4LL_EVENT_STOP: + log_error("the client was stopped with address %s", strna(address)); + break; + default: + assert_not_reached("invalid LL event"); + } +} + +static int client_run(int ifindex, const char *seed_str, const struct ether_addr *ha, sd_event *e) { + sd_ipv4ll *ll; + + assert_se(sd_ipv4ll_new(&ll) >= 0); + assert_se(sd_ipv4ll_attach_event(ll, e, 0) >= 0); + + assert_se(sd_ipv4ll_set_index(ll, ifindex) >= 0); + assert_se(sd_ipv4ll_set_mac(ll, ha) >= 0); + assert_se(sd_ipv4ll_set_callback(ll, ll_handler, NULL) >= 0); + + if (seed_str) { + unsigned seed; + + assert_se(safe_atou(seed_str, &seed) >= 0); + + assert_se(sd_ipv4ll_set_address_seed(ll, seed) >= 0); + } + + log_info("starting IPv4LL client"); + + assert_se(sd_ipv4ll_start(ll) >= 0); + + assert_se(sd_event_loop(e) >= 0); + + assert_se(!sd_ipv4ll_unref(ll)); + + return EXIT_SUCCESS; +} + +static int test_ll(const char *ifname, const char *seed) { + _cleanup_event_unref_ sd_event *e = NULL; + _cleanup_netlink_unref_ sd_netlink *rtnl = NULL; + _cleanup_netlink_message_unref_ sd_netlink_message *m = NULL, *reply = NULL; + struct ether_addr ha; + int ifindex; + + assert_se(sd_event_new(&e) >= 0); + + assert_se(sd_netlink_open(&rtnl) >= 0); + assert_se(sd_netlink_attach_event(rtnl, e, 0) >= 0); + + assert_se(sd_rtnl_message_new_link(rtnl, &m, RTM_GETLINK, 0) >= 0); + assert_se(sd_netlink_message_append_string(m, IFLA_IFNAME, ifname) >= 0); + assert_se(sd_netlink_call(rtnl, m, 0, &reply) >= 0); + + assert_se(sd_rtnl_message_link_get_ifindex(reply, &ifindex) >= 0); + assert_se(sd_netlink_message_read_ether_addr(reply, IFLA_ADDRESS, &ha) >= 0); + + client_run(ifindex, seed, &ha, e); + + return EXIT_SUCCESS; +} + +int main(int argc, char *argv[]) { + log_set_max_level(LOG_DEBUG); + log_parse_environment(); + log_open(); + + if (argc == 2) + return test_ll(argv[1], NULL); + else if (argc == 3) + return test_ll(argv[1], argv[2]); + else { + log_error("This program takes one or two arguments.\n" + "\t %s <ifname> [<seed>]", program_invocation_short_name); + return EXIT_FAILURE; + } +} diff --git a/src/libsystemd-network/test-ipv4ll.c b/src/libsystemd-network/test-ipv4ll.c index d60ee98b25..f0e26bda07 100644 --- a/src/libsystemd-network/test-ipv4ll.c +++ b/src/libsystemd-network/test-ipv4ll.c @@ -31,7 +31,7 @@ #include "event-util.h" #include "sd-ipv4ll.h" -#include "ipv4ll-internal.h" +#include "arp-util.h" static bool verbose = false; static bool extended = false; @@ -56,10 +56,10 @@ static void basic_request_handler(sd_ipv4ll *ll, int event, void *userdata) { } } -int arp_network_send_raw_socket(int fd, const union sockaddr_union *link, - const struct ether_arp *arp) { +static int arp_network_send_raw_socket(int fd, int ifindex, + const struct ether_arp *arp) { assert_se(arp); - assert_se(link); + assert_se(ifindex > 0); assert_se(fd >= 0); if (send(fd, arp, sizeof(struct ether_arp), 0) < 0) @@ -68,51 +68,35 @@ int arp_network_send_raw_socket(int fd, const union sockaddr_union *link, return 0; } -int arp_network_bind_raw_socket(int index, union sockaddr_union *link) { - if (socketpair(AF_UNIX, SOCK_DGRAM | SOCK_NONBLOCK, 0, test_fd) < 0) - return -errno; +int arp_send_probe(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha) { + struct ether_arp ea = {}; - return test_fd[0]; -} + assert(fd >= 0); + assert(ifindex > 0); + assert(pa != 0); + assert(ha); -static void test_arp_header(struct ether_arp *arp) { - assert_se(arp); - assert_se(arp->ea_hdr.ar_hrd == htons(ARPHRD_ETHER)); /* HTYPE */ - assert_se(arp->ea_hdr.ar_pro == htons(ETHERTYPE_IP)); /* PTYPE */ - assert_se(arp->ea_hdr.ar_hln == ETH_ALEN); /* HLEN */ - assert_se(arp->ea_hdr.ar_pln == sizeof arp->arp_spa); /* PLEN */ - assert_se(arp->ea_hdr.ar_op == htons(ARPOP_REQUEST)); /* REQUEST */ + return arp_network_send_raw_socket(fd, ifindex, &ea); } -static void test_arp_probe(void) { - struct ether_arp arp; - struct ether_addr mac_addr = { - .ether_addr_octet = {'A', 'B', 'C', '1', '2', '3'}}; - be32_t pa = 0x3030; +int arp_send_announcement(int fd, int ifindex, + be32_t pa, const struct ether_addr *ha) { + struct ether_arp ea = {}; - if (verbose) - printf("* %s\n", __FUNCTION__); + assert(fd >= 0); + assert(ifindex > 0); + assert(pa != 0); + assert(ha); - arp_packet_probe(&arp, pa, &mac_addr); - test_arp_header(&arp); - assert_se(memcmp(arp.arp_sha, &mac_addr, ETH_ALEN) == 0); - assert_se(memcmp(arp.arp_tpa, &pa, sizeof(pa)) == 0); + return arp_network_send_raw_socket(fd, ifindex, &ea); } -static void test_arp_announce(void) { - struct ether_arp arp; - struct ether_addr mac_addr = { - .ether_addr_octet = {'A', 'B', 'C', '1', '2', '3'}}; - be32_t pa = 0x3131; - - if (verbose) - printf("* %s\n", __FUNCTION__); +int arp_network_bind_raw_socket(int index, be32_t address, const struct ether_addr *eth_mac) { + if (socketpair(AF_UNIX, SOCK_DGRAM | SOCK_NONBLOCK, 0, test_fd) < 0) + return -errno; - arp_packet_announcement(&arp, pa, &mac_addr); - test_arp_header(&arp); - assert_se(memcmp(arp.arp_sha, &mac_addr, ETH_ALEN) == 0); - assert_se(memcmp(arp.arp_tpa, &pa, sizeof(pa)) == 0); - assert_se(memcmp(arp.arp_spa, &pa, sizeof(pa)) == 0); + return test_fd[0]; } static void test_public_api_setters(sd_event *e) { @@ -134,9 +118,8 @@ static void test_public_api_setters(sd_event *e) { assert_se(sd_ipv4ll_set_callback(NULL, NULL, NULL) == -EINVAL); assert_se(sd_ipv4ll_set_callback(ll, NULL, NULL) == 0); - assert_se(sd_ipv4ll_set_address_seed(NULL, NULL) == -EINVAL); - assert_se(sd_ipv4ll_set_address_seed(ll, NULL) == -EINVAL); - assert_se(sd_ipv4ll_set_address_seed(ll, seed) == 0); + assert_se(sd_ipv4ll_set_address_seed(NULL, *(unsigned *) seed) == -EINVAL); + assert_se(sd_ipv4ll_set_address_seed(ll, *(unsigned *) seed) == 0); assert_se(sd_ipv4ll_set_mac(NULL, NULL) == -EINVAL); assert_se(sd_ipv4ll_set_mac(ll, NULL) == -EINVAL); @@ -149,7 +132,7 @@ static void test_public_api_setters(sd_event *e) { assert_se(sd_ipv4ll_set_index(ll, 99) == 0); assert_se(sd_ipv4ll_ref(ll) == ll); - assert_se(sd_ipv4ll_unref(ll) == ll); + assert_se(sd_ipv4ll_unref(ll) == NULL); /* Cleanup */ assert_se(sd_ipv4ll_unref(ll) == NULL); @@ -184,21 +167,20 @@ static void test_basic_request(sd_event *e) { sd_event_run(e, (uint64_t) -1); assert_se(sd_ipv4ll_start(ll) == -EBUSY); + assert_se(sd_ipv4ll_is_running(ll)); + /* PROBE */ sd_event_run(e, (uint64_t) -1); assert_se(read(test_fd[1], &arp, sizeof(struct ether_arp)) == sizeof(struct ether_arp)); - test_arp_header(&arp); if (extended) { /* PROBE */ sd_event_run(e, (uint64_t) -1); assert_se(read(test_fd[1], &arp, sizeof(struct ether_arp)) == sizeof(struct ether_arp)); - test_arp_header(&arp); /* PROBE */ sd_event_run(e, (uint64_t) -1); assert_se(read(test_fd[1], &arp, sizeof(struct ether_arp)) == sizeof(struct ether_arp)); - test_arp_header(&arp); sd_event_run(e, (uint64_t) -1); assert_se(basic_request_handler_bind == 1); @@ -215,11 +197,13 @@ static void test_basic_request(sd_event *e) { int main(int argc, char *argv[]) { _cleanup_event_unref_ sd_event *e = NULL; + log_set_max_level(LOG_DEBUG); + log_parse_environment(); + log_open(); + assert_se(sd_event_new(&e) >= 0); test_public_api_setters(e); - test_arp_probe(); - test_arp_announce(); test_basic_request(e); return 0; diff --git a/src/libsystemd/sd-bus/bus-internal.h b/src/libsystemd/sd-bus/bus-internal.h index 7af61a9433..e399701beb 100644 --- a/src/libsystemd/sd-bus/bus-internal.h +++ b/src/libsystemd/sd-bus/bus-internal.h @@ -396,6 +396,6 @@ int bus_maybe_reply_error(sd_bus_message *m, int r, sd_bus_error *error); #define bus_assert_return(expr, r, error) \ do { \ - if (!assert_log(expr)) \ + if (!assert_log(expr, #expr)) \ return sd_bus_error_set_errno(error, r); \ } while (false) diff --git a/src/libsystemd/sd-bus/bus-socket.c b/src/libsystemd/sd-bus/bus-socket.c index 735a775cb4..d0b1e3d7dc 100644 --- a/src/libsystemd/sd-bus/bus-socket.c +++ b/src/libsystemd/sd-bus/bus-socket.c @@ -985,7 +985,7 @@ int bus_socket_read_message(sd_bus *bus) { return -EIO; } - f = realloc(bus->fds, sizeof(int) + (bus->n_fds + n)); + f = realloc(bus->fds, sizeof(int) * (bus->n_fds + n)); if (!f) { close_many((int*) CMSG_DATA(cmsg), n); return -ENOMEM; diff --git a/src/login/loginctl.c b/src/login/loginctl.c index a7e64071cf..405df49a7c 100644 --- a/src/login/loginctl.c +++ b/src/login/loginctl.c @@ -686,19 +686,165 @@ static int print_seat_status_info(sd_bus *bus, const char *path, bool *new_line) return 0; } +static int print_property(const char *name, sd_bus_message *m, const char *contents) { + int r; + + assert(name); + assert(m); + assert(contents); + + if (arg_property && !strv_find(arg_property, name)) + /* skip what we didn't read */ + return sd_bus_message_skip(m, contents); + + switch (contents[0]) { + + case SD_BUS_TYPE_STRUCT_BEGIN: + + if (contents[1] == SD_BUS_TYPE_STRING && STR_IN_SET(name, "Display", "Seat", "ActiveSession")) { + const char *s; + + r = sd_bus_message_read(m, "(so)", &s, NULL); + if (r < 0) + return bus_log_parse_error(r); + + if (arg_all || !isempty(s)) + printf("%s=%s\n", name, s); + + return 0; + + } else if (contents[1] == SD_BUS_TYPE_UINT32 && streq(name, "User")) { + uint32_t uid; + + r = sd_bus_message_read(m, "(uo)", &uid, NULL); + if (r < 0) + return bus_log_parse_error(r); + + if (UID_IS_INVALID(uid)) { + log_error("Invalid user ID: " UID_FMT, uid); + return -EINVAL; + } + + printf("%s=" UID_FMT "\n", name, uid); + + return 0; + } + + break; + + case SD_BUS_TYPE_ARRAY: + + if (contents[1] == SD_BUS_TYPE_STRUCT_BEGIN && streq(name, "Sessions")) { + const char *s; + bool space = false; + + r = sd_bus_message_enter_container(m, SD_BUS_TYPE_ARRAY, "(so)"); + if (r < 0) + return bus_log_parse_error(r); + + printf("%s=", name); + + while ((r = sd_bus_message_read(m, "(so)", &s, NULL)) > 0) { + printf("%s%s", space ? " " : "", s); + space = true; + } + + printf("\n"); + + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_exit_container(m); + if (r < 0) + return bus_log_parse_error(r); + + return 0; + } + + break; + } + + r = bus_print_property(name, m, arg_all); + if (r < 0) + return bus_log_parse_error(r); + + if (r == 0) { + r = sd_bus_message_skip(m, contents); + if (r < 0) + return bus_log_parse_error(r); + + if (arg_all) + printf("%s=[unprintable]\n", name); + } + + return 0; +} + static int show_properties(sd_bus *bus, const char *path, bool *new_line) { + _cleanup_bus_message_unref_ sd_bus_message *reply = NULL; + _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL; int r; + assert(bus); + assert(path); + assert(new_line); + + r = sd_bus_call_method( + bus, + "org.freedesktop.login1", + path, + "org.freedesktop.DBus.Properties", + "GetAll", + &error, + &reply, + "s", ""); + if (r < 0) + return log_error_errno(r, "Failed to get properties: %s", bus_error_message(&error, r)); + + r = sd_bus_message_enter_container(reply, SD_BUS_TYPE_ARRAY, "{sv}"); + if (r < 0) + return bus_log_parse_error(r); + if (*new_line) printf("\n"); *new_line = true; - r = bus_print_all_properties(bus, "org.freedesktop.login1", path, arg_property, arg_all); + while ((r = sd_bus_message_enter_container(reply, SD_BUS_TYPE_DICT_ENTRY, "sv")) > 0) { + const char *name, *contents; + + r = sd_bus_message_read(reply, "s", &name); + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_peek_type(reply, NULL, &contents); + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_message_enter_container(reply, SD_BUS_TYPE_VARIANT, contents); + if (r < 0) + return bus_log_parse_error(r); + + r = print_property(name, reply, contents); + if (r < 0) + return r; + + r = sd_bus_message_exit_container(reply); + 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); + } if (r < 0) - log_error_errno(r, "Could not get properties: %m"); + return bus_log_parse_error(r); - return r; + r = sd_bus_message_exit_container(reply); + if (r < 0) + return bus_log_parse_error(r); + + return 0; } static int show_session(int argc, char *argv[], void *userdata) { diff --git a/src/network/networkd-ipv4ll.c b/src/network/networkd-ipv4ll.c index 0a27a30278..1c34f55b4b 100644 --- a/src/network/networkd-ipv4ll.c +++ b/src/network/networkd-ipv4ll.c @@ -195,10 +195,7 @@ static void ipv4ll_handler(sd_ipv4ll *ll, int event, void *userdata){ } break; default: - if (event < 0) - log_link_warning(link, "IPv4 link-local error: %s", strerror(-event)); - else - log_link_warning(link, "IPv4 link-local unknown event: %d", event); + log_link_warning(link, "IPv4 link-local unknown event: %d", event); break; } } @@ -218,7 +215,9 @@ int ipv4ll_configure(Link *link) { if (link->udev_device) { r = net_get_unique_predictable_data(link->udev_device, seed); if (r >= 0) { - r = sd_ipv4ll_set_address_seed(link->ipv4ll, seed); + assert_cc(sizeof(unsigned) <= 8); + + r = sd_ipv4ll_set_address_seed(link->ipv4ll, *(unsigned *)seed); if (r < 0) return r; } diff --git a/src/network/networkd-link.c b/src/network/networkd-link.c index 9d4a69b0db..215c47b8af 100644 --- a/src/network/networkd-link.c +++ b/src/network/networkd-link.c @@ -1789,6 +1789,45 @@ static int link_set_ipv6_privacy_extensions(Link *link) { return 0; } +static int link_set_ipv6_accept_ra(Link *link) { + const char *p = NULL, *v = NULL; + bool b; + int r; + + /* Make this a NOP if IPv6 is not available */ + if (!socket_ipv6_is_supported()) + return 0; + + if (link->flags & IFF_LOOPBACK) + return 0; + + /* if unset check the ip forwarding setting maintained for the interface + * and then set it to depending on that. enabled if local forwarding + * is disabled. disabled if local forwarding is enabled. + */ + if (link->network->ipv6_accept_ra < 0) { + if (IN_SET(link->network->ip_forward, ADDRESS_FAMILY_YES, ADDRESS_FAMILY_IPV6)) + b = false; + else + b = true; + } else + b = link->network->ipv6_accept_ra; + + p = strjoina("/proc/sys/net/ipv6/conf/", link->ifname, "/accept_ra"); + v = one_zero(b); + + r = write_string_file(p, v, 0); + if (r < 0) { + /* If the right value is set anyway, don't complain */ + if (verify_one_line_file(p, v) > 0) + return 0; + + log_link_warning_errno(link, r, "Cannot configure IPv6 accept_ra for interface: %m"); + } + + return 0; +} + static int link_configure(Link *link) { int r; @@ -1812,6 +1851,10 @@ static int link_configure(Link *link) { if (r < 0) return r; + r = link_set_ipv6_accept_ra(link); + if (r < 0) + return r; + if (link_ipv4ll_enabled(link)) { r = ipv4ll_configure(link); if (r < 0) diff --git a/src/network/networkd-network-gperf.gperf b/src/network/networkd-network-gperf.gperf index 10ca9dae35..8257ab45da 100644 --- a/src/network/networkd-network-gperf.gperf +++ b/src/network/networkd-network-gperf.gperf @@ -15,88 +15,89 @@ struct ConfigPerfItem; %struct-type %includes %% -Match.MACAddress, config_parse_hwaddr, 0, offsetof(Network, match_mac) -Match.Path, config_parse_strv, 0, offsetof(Network, match_path) -Match.Driver, config_parse_strv, 0, offsetof(Network, match_driver) -Match.Type, config_parse_strv, 0, offsetof(Network, match_type) -Match.Name, config_parse_ifnames, 0, offsetof(Network, match_name) -Match.Host, config_parse_net_condition, CONDITION_HOST, offsetof(Network, match_host) -Match.Virtualization, config_parse_net_condition, CONDITION_VIRTUALIZATION, offsetof(Network, match_virt) -Match.KernelCommandLine, config_parse_net_condition, CONDITION_KERNEL_COMMAND_LINE, offsetof(Network, match_kernel) -Match.Architecture, config_parse_net_condition, CONDITION_ARCHITECTURE, offsetof(Network, match_arch) -Link.MACAddress, config_parse_hwaddr, 0, offsetof(Network, mac) -Link.MTUBytes, config_parse_iec_size, 0, offsetof(Network, mtu) -Network.Description, config_parse_string, 0, offsetof(Network, description) -Network.Bridge, config_parse_netdev, 0, offsetof(Network, bridge) -Network.Bond, config_parse_netdev, 0, offsetof(Network, bond) -Network.VLAN, config_parse_netdev, 0, 0 -Network.MACVLAN, config_parse_netdev, 0, 0 -Network.MACVTAP, config_parse_netdev, 0, 0 -Network.IPVLAN, config_parse_netdev, 0, 0 -Network.VXLAN, config_parse_netdev, 0, 0 -Network.Tunnel, config_parse_tunnel, 0, 0 -Network.DHCP, config_parse_dhcp, 0, offsetof(Network, dhcp) -Network.DHCPServer, config_parse_bool, 0, offsetof(Network, dhcp_server) -Network.LinkLocalAddressing, config_parse_address_family_boolean, 0, offsetof(Network, link_local) -Network.IPv4LLRoute, config_parse_bool, 0, offsetof(Network, ipv4ll_route) -Network.IPv6Token, config_parse_ipv6token, 0, offsetof(Network, ipv6_token) -Network.LLDP, config_parse_bool, 0, offsetof(Network, lldp) -Network.Address, config_parse_address, 0, 0 -Network.Gateway, config_parse_gateway, 0, 0 -Network.Domains, config_parse_domains, 0, offsetof(Network, domains) -Network.DNS, config_parse_strv, 0, offsetof(Network, dns) -Network.LLMNR, config_parse_resolve, 0, offsetof(Network, llmnr) -Network.NTP, config_parse_strv, 0, offsetof(Network, ntp) -Network.IPForward, config_parse_address_family_boolean_with_kernel,0, offsetof(Network, ip_forward) -Network.IPMasquerade, config_parse_bool, 0, offsetof(Network, ip_masquerade) -Network.IPv6PrivacyExtensions, config_parse_ipv6_privacy_extensions, 0, offsetof(Network, ipv6_privacy_extensions) -Network.BindCarrier, config_parse_strv, 0, offsetof(Network, bind_carrier) -Address.Address, config_parse_address, 0, 0 -Address.Peer, config_parse_address, 0, 0 -Address.Broadcast, config_parse_broadcast, 0, 0 -Address.Label, config_parse_label, 0, 0 -Route.Gateway, config_parse_gateway, 0, 0 -Route.Destination, config_parse_destination, 0, 0 -Route.Source, config_parse_destination, 0, 0 -Route.Metric, config_parse_route_priority, 0, 0 -Route.Scope, config_parse_route_scope, 0, 0 -DHCP.ClientIdentifier, config_parse_dhcp_client_identifier, 0, offsetof(Network, dhcp_client_identifier) -DHCP.UseDNS, config_parse_bool, 0, offsetof(Network, dhcp_dns) -DHCP.UseNTP, config_parse_bool, 0, offsetof(Network, dhcp_ntp) -DHCP.UseMTU, config_parse_bool, 0, offsetof(Network, dhcp_mtu) -DHCP.UseHostname, config_parse_bool, 0, offsetof(Network, dhcp_hostname) -DHCP.UseDomains, config_parse_bool, 0, offsetof(Network, dhcp_domains) -DHCP.UseRoutes, config_parse_bool, 0, offsetof(Network, dhcp_routes) -DHCP.SendHostname, config_parse_bool, 0, offsetof(Network, dhcp_sendhost) -DHCP.Hostname, config_parse_hostname, 0, offsetof(Network, hostname) -DHCP.RequestBroadcast, config_parse_bool, 0, offsetof(Network, dhcp_broadcast) -DHCP.CriticalConnection, config_parse_bool, 0, offsetof(Network, dhcp_critical) -DHCP.VendorClassIdentifier, config_parse_string, 0, offsetof(Network, dhcp_vendor_class_identifier) -DHCP.RouteMetric, config_parse_unsigned, 0, offsetof(Network, dhcp_route_metric) -DHCP.UseTimezone, config_parse_bool, 0, offsetof(Network, dhcp_timezone) -DHCPServer.MaxLeaseTimeSec, config_parse_sec, 0, offsetof(Network, dhcp_server_max_lease_time_usec) -DHCPServer.DefaultLeaseTimeSec,config_parse_sec, 0, offsetof(Network, dhcp_server_default_lease_time_usec) -DHCPServer.EmitDNS, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_dns) -DHCPServer.DNS, config_parse_dhcp_server_dns, 0, 0 -DHCPServer.EmitNTP, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_ntp) -DHCPServer.NTP, config_parse_dhcp_server_ntp, 0, 0 -DHCPServer.EmitTimezone, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_timezone) -DHCPServer.Timezone, config_parse_timezone, 0, offsetof(Network, dhcp_server_timezone) -DHCPServer.PoolOffset, config_parse_uint32, 0, offsetof(Network, dhcp_server_pool_offset) -DHCPServer.PoolSize, config_parse_uint32, 0, offsetof(Network, dhcp_server_pool_size) -Bridge.Cost, config_parse_unsigned, 0, offsetof(Network, cost) -Bridge.UseBPDU, config_parse_bool, 0, offsetof(Network, use_bpdu) -Bridge.HairPin, config_parse_bool, 0, offsetof(Network, hairpin) -Bridge.FastLeave, config_parse_bool, 0, offsetof(Network, fast_leave) -Bridge.AllowPortToBeRoot, config_parse_bool, 0, offsetof(Network, allow_port_to_be_root) -Bridge.UnicastFlood, config_parse_bool, 0, offsetof(Network, unicast_flood) -BridgeFDB.MACAddress, config_parse_fdb_hwaddr, 0, 0 -BridgeFDB.VLANId, config_parse_fdb_vlan_id, 0, 0 +Match.MACAddress, config_parse_hwaddr, 0, offsetof(Network, match_mac) +Match.Path, config_parse_strv, 0, offsetof(Network, match_path) +Match.Driver, config_parse_strv, 0, offsetof(Network, match_driver) +Match.Type, config_parse_strv, 0, offsetof(Network, match_type) +Match.Name, config_parse_ifnames, 0, offsetof(Network, match_name) +Match.Host, config_parse_net_condition, CONDITION_HOST, offsetof(Network, match_host) +Match.Virtualization, config_parse_net_condition, CONDITION_VIRTUALIZATION, offsetof(Network, match_virt) +Match.KernelCommandLine, config_parse_net_condition, CONDITION_KERNEL_COMMAND_LINE, offsetof(Network, match_kernel) +Match.Architecture, config_parse_net_condition, CONDITION_ARCHITECTURE, offsetof(Network, match_arch) +Link.MACAddress, config_parse_hwaddr, 0, offsetof(Network, mac) +Link.MTUBytes, config_parse_iec_size, 0, offsetof(Network, mtu) +Network.Description, config_parse_string, 0, offsetof(Network, description) +Network.Bridge, config_parse_netdev, 0, offsetof(Network, bridge) +Network.Bond, config_parse_netdev, 0, offsetof(Network, bond) +Network.VLAN, config_parse_netdev, 0, 0 +Network.MACVLAN, config_parse_netdev, 0, 0 +Network.MACVTAP, config_parse_netdev, 0, 0 +Network.IPVLAN, config_parse_netdev, 0, 0 +Network.VXLAN, config_parse_netdev, 0, 0 +Network.Tunnel, config_parse_tunnel, 0, 0 +Network.DHCP, config_parse_dhcp, 0, offsetof(Network, dhcp) +Network.DHCPServer, config_parse_bool, 0, offsetof(Network, dhcp_server) +Network.LinkLocalAddressing, config_parse_address_family_boolean, 0, offsetof(Network, link_local) +Network.IPv4LLRoute, config_parse_bool, 0, offsetof(Network, ipv4ll_route) +Network.IPv6Token, config_parse_ipv6token, 0, offsetof(Network, ipv6_token) +Network.LLDP, config_parse_bool, 0, offsetof(Network, lldp) +Network.Address, config_parse_address, 0, 0 +Network.Gateway, config_parse_gateway, 0, 0 +Network.Domains, config_parse_domains, 0, offsetof(Network, domains) +Network.DNS, config_parse_strv, 0, offsetof(Network, dns) +Network.LLMNR, config_parse_resolve, 0, offsetof(Network, llmnr) +Network.NTP, config_parse_strv, 0, offsetof(Network, ntp) +Network.IPForward, config_parse_address_family_boolean_with_kernel,0, offsetof(Network, ip_forward) +Network.IPMasquerade, config_parse_bool, 0, offsetof(Network, ip_masquerade) +Network.IPv6PrivacyExtensions, config_parse_ipv6_privacy_extensions, 0, offsetof(Network, ipv6_privacy_extensions) +Network.IPv6AcceptRouterAdvertisements, config_parse_tristate, 0, offsetof(Network, ipv6_accept_ra) +Network.BindCarrier, config_parse_strv, 0, offsetof(Network, bind_carrier) +Address.Address, config_parse_address, 0, 0 +Address.Peer, config_parse_address, 0, 0 +Address.Broadcast, config_parse_broadcast, 0, 0 +Address.Label, config_parse_label, 0, 0 +Route.Gateway, config_parse_gateway, 0, 0 +Route.Destination, config_parse_destination, 0, 0 +Route.Source, config_parse_destination, 0, 0 +Route.Metric, config_parse_route_priority, 0, 0 +Route.Scope, config_parse_route_scope, 0, 0 +DHCP.ClientIdentifier, config_parse_dhcp_client_identifier, 0, offsetof(Network, dhcp_client_identifier) +DHCP.UseDNS, config_parse_bool, 0, offsetof(Network, dhcp_dns) +DHCP.UseNTP, config_parse_bool, 0, offsetof(Network, dhcp_ntp) +DHCP.UseMTU, config_parse_bool, 0, offsetof(Network, dhcp_mtu) +DHCP.UseHostname, config_parse_bool, 0, offsetof(Network, dhcp_hostname) +DHCP.UseDomains, config_parse_bool, 0, offsetof(Network, dhcp_domains) +DHCP.UseRoutes, config_parse_bool, 0, offsetof(Network, dhcp_routes) +DHCP.SendHostname, config_parse_bool, 0, offsetof(Network, dhcp_sendhost) +DHCP.Hostname, config_parse_hostname, 0, offsetof(Network, hostname) +DHCP.RequestBroadcast, config_parse_bool, 0, offsetof(Network, dhcp_broadcast) +DHCP.CriticalConnection, config_parse_bool, 0, offsetof(Network, dhcp_critical) +DHCP.VendorClassIdentifier, config_parse_string, 0, offsetof(Network, dhcp_vendor_class_identifier) +DHCP.RouteMetric, config_parse_unsigned, 0, offsetof(Network, dhcp_route_metric) +DHCP.UseTimezone, config_parse_bool, 0, offsetof(Network, dhcp_timezone) +DHCPServer.MaxLeaseTimeSec, config_parse_sec, 0, offsetof(Network, dhcp_server_max_lease_time_usec) +DHCPServer.DefaultLeaseTimeSec, config_parse_sec, 0, offsetof(Network, dhcp_server_default_lease_time_usec) +DHCPServer.EmitDNS, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_dns) +DHCPServer.DNS, config_parse_dhcp_server_dns, 0, 0 +DHCPServer.EmitNTP, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_ntp) +DHCPServer.NTP, config_parse_dhcp_server_ntp, 0, 0 +DHCPServer.EmitTimezone, config_parse_bool, 0, offsetof(Network, dhcp_server_emit_timezone) +DHCPServer.Timezone, config_parse_timezone, 0, offsetof(Network, dhcp_server_timezone) +DHCPServer.PoolOffset, config_parse_uint32, 0, offsetof(Network, dhcp_server_pool_offset) +DHCPServer.PoolSize, config_parse_uint32, 0, offsetof(Network, dhcp_server_pool_size) +Bridge.Cost, config_parse_unsigned, 0, offsetof(Network, cost) +Bridge.UseBPDU, config_parse_bool, 0, offsetof(Network, use_bpdu) +Bridge.HairPin, config_parse_bool, 0, offsetof(Network, hairpin) +Bridge.FastLeave, config_parse_bool, 0, offsetof(Network, fast_leave) +Bridge.AllowPortToBeRoot, config_parse_bool, 0, offsetof(Network, allow_port_to_be_root) +Bridge.UnicastFlood, config_parse_bool, 0, offsetof(Network, unicast_flood) +BridgeFDB.MACAddress, config_parse_fdb_hwaddr, 0, 0 +BridgeFDB.VLANId, config_parse_fdb_vlan_id, 0, 0 /* backwards compatibility: do not add new entries to this section */ -Network.IPv4LL, config_parse_ipv4ll, 0, offsetof(Network, link_local) -DHCPv4.UseDNS, config_parse_bool, 0, offsetof(Network, dhcp_dns) -DHCPv4.UseMTU, config_parse_bool, 0, offsetof(Network, dhcp_mtu) -DHCPv4.UseHostname, config_parse_bool, 0, offsetof(Network, dhcp_hostname) -DHCP.UseDomainName, config_parse_bool, 0, offsetof(Network, dhcp_domains) -DHCPv4.UseDomainName, config_parse_bool, 0, offsetof(Network, dhcp_domains) -DHCPv4.CriticalConnection, config_parse_bool, 0, offsetof(Network, dhcp_critical) +Network.IPv4LL, config_parse_ipv4ll, 0, offsetof(Network, link_local) +DHCPv4.UseDNS, config_parse_bool, 0, offsetof(Network, dhcp_dns) +DHCPv4.UseMTU, config_parse_bool, 0, offsetof(Network, dhcp_mtu) +DHCPv4.UseHostname, config_parse_bool, 0, offsetof(Network, dhcp_hostname) +DHCP.UseDomainName, config_parse_bool, 0, offsetof(Network, dhcp_domains) +DHCPv4.UseDomainName, config_parse_bool, 0, offsetof(Network, dhcp_domains) +DHCPv4.CriticalConnection, config_parse_bool, 0, offsetof(Network, dhcp_critical) diff --git a/src/network/networkd-network.c b/src/network/networkd-network.c index ee14401982..57495b58e0 100644 --- a/src/network/networkd-network.c +++ b/src/network/networkd-network.c @@ -120,6 +120,7 @@ static int network_load_one(Manager *manager, const char *filename) { network->link_local = ADDRESS_FAMILY_IPV6; network->ipv6_privacy_extensions = IPV6_PRIVACY_EXTENSIONS_NO; + network->ipv6_accept_ra = -1; r = config_parse(NULL, filename, file, "Match\0" diff --git a/src/network/networkd-network.h b/src/network/networkd-network.h index c3439a70ba..2a43b6b347 100644 --- a/src/network/networkd-network.h +++ b/src/network/networkd-network.h @@ -120,6 +120,8 @@ struct Network { AddressFamilyBoolean ip_forward; bool ip_masquerade; + int ipv6_accept_ra; + union in_addr_union ipv6_token; IPv6PrivacyExtensions ipv6_privacy_extensions; diff --git a/src/notify/notify.c b/src/notify/notify.c index c303bcf718..772e3db69d 100644 --- a/src/notify/notify.c +++ b/src/notify/notify.c @@ -191,7 +191,7 @@ int main(int argc, char* argv[]) { goto finish; } - r = sd_pid_notify(arg_pid, false, n); + r = sd_pid_notify(arg_pid ? arg_pid : getppid(), false, n); if (r < 0) { log_error_errno(r, "Failed to notify init system: %m"); goto finish; diff --git a/src/resolve/resolved-bus.c b/src/resolve/resolved-bus.c index 12c17003e9..bf1b7c8ab4 100644 --- a/src/resolve/resolved-bus.c +++ b/src/resolve/resolved-bus.c @@ -191,7 +191,7 @@ static void bus_method_resolve_hostname_complete(DnsQuery *q) { /* This has a cname? Then update the query with the * new cname. */ - r = dns_query_cname_redirect(q, cname->cname.name); + r = dns_query_cname_redirect(q, cname); if (r < 0) { if (r == -ELOOP) r = sd_bus_reply_method_errorf(q->request, BUS_ERROR_CNAME_LOOP, "CNAME loop on '%s'", q->request_hostname); @@ -220,8 +220,6 @@ static void bus_method_resolve_hostname_complete(DnsQuery *q) { added++; } - // what about the cache? - /* If we didn't find anything, then let's restart the * query, this time with the cname */ if (added <= 0) { diff --git a/src/resolve/resolved-dns-answer.c b/src/resolve/resolved-dns-answer.c index 13ad4ca6bd..89b9b0e1ea 100644 --- a/src/resolve/resolved-dns-answer.c +++ b/src/resolve/resolved-dns-answer.c @@ -149,6 +149,19 @@ int dns_answer_contains(DnsAnswer *a, DnsResourceKey *key) { return 0; } +int dns_answer_match_soa(DnsResourceKey *key, DnsResourceKey *soa) { + if (soa->class != DNS_CLASS_IN) + return 0; + + if (soa->type != DNS_TYPE_SOA) + return 0; + + if (!dns_name_endswith(DNS_RESOURCE_KEY_NAME(key), DNS_RESOURCE_KEY_NAME(soa))) + return 0; + + return 1; +} + int dns_answer_find_soa(DnsAnswer *a, DnsResourceKey *key, DnsResourceRecord **ret) { unsigned i; @@ -164,13 +177,7 @@ int dns_answer_find_soa(DnsAnswer *a, DnsResourceKey *key, DnsResourceRecord **r for (i = 0; i < a->n_rrs; i++) { - if (a->items[i].rr->key->class != DNS_CLASS_IN) - continue; - - if (a->items[i].rr->key->type != DNS_TYPE_SOA) - continue; - - if (dns_name_endswith(DNS_RESOURCE_KEY_NAME(key), DNS_RESOURCE_KEY_NAME(a->items[i].rr->key))) { + if (dns_answer_match_soa(key, a->items[i].rr->key)) { *ret = a->items[i].rr; return 1; } diff --git a/src/resolve/resolved-dns-answer.h b/src/resolve/resolved-dns-answer.h index 0757dd60d0..044d73b19c 100644 --- a/src/resolve/resolved-dns-answer.h +++ b/src/resolve/resolved-dns-answer.h @@ -49,6 +49,7 @@ DnsAnswer *dns_answer_unref(DnsAnswer *a); int dns_answer_add(DnsAnswer *a, DnsResourceRecord *rr, int ifindex); int dns_answer_add_soa(DnsAnswer *a, const char *name, uint32_t ttl); int dns_answer_contains(DnsAnswer *a, DnsResourceKey *key); +int dns_answer_match_soa(DnsResourceKey *key, DnsResourceKey *soa); int dns_answer_find_soa(DnsAnswer *a, DnsResourceKey *key, DnsResourceRecord **ret); DnsAnswer *dns_answer_merge(DnsAnswer *a, DnsAnswer *b); diff --git a/src/resolve/resolved-dns-cache.c b/src/resolve/resolved-dns-cache.c index cbbbed8c8a..ab13636bc1 100644 --- a/src/resolve/resolved-dns-cache.c +++ b/src/resolve/resolved-dns-cache.c @@ -277,13 +277,14 @@ static int dns_cache_put_positive( /* New TTL is 0? Delete the entry... */ if (rr->ttl <= 0) { - if (dns_cache_remove(c, rr->key)) { - r = dns_resource_key_to_string(rr->key, &key_str); - if (r < 0) - return r; + r = dns_resource_key_to_string(rr->key, &key_str); + if (r < 0) + return r; + if (dns_cache_remove(c, rr->key)) log_debug("Removed zero TTL entry from cache: %s", key_str); - } + else + log_debug("Not caching zero TTL cache entry: %s", key_str); return 0; } @@ -361,7 +362,7 @@ static int dns_cache_put_negative( if (r < 0) return r; - log_debug("Ignored negative cache entry with zero SOA TTL: %s", key_str); + log_debug("Not caching negative entry with zero SOA TTL: %s", key_str); return 0; } @@ -402,7 +403,7 @@ static int dns_cache_put_negative( int dns_cache_put( DnsCache *c, - DnsQuestion *q, + DnsResourceKey *key, int rcode, DnsAnswer *answer, unsigned max_rrs, @@ -410,16 +411,16 @@ int dns_cache_put( int owner_family, const union in_addr_union *owner_address) { + DnsResourceRecord *soa = NULL; unsigned cache_keys, i; int r; assert(c); - if (q) { - /* First, if we were passed a question, delete all matching old RRs, + if (key) { + /* First, if we were passed a key, delete all matching old RRs, * so that we only keep complete by_key in place. */ - for (i = 0; i < q->n_keys; i++) - dns_cache_remove(c, q->keys[i]); + dns_cache_remove(c, key); } if (!answer) @@ -437,8 +438,8 @@ int dns_cache_put( cache_keys = answer->n_rrs; - if (q) - cache_keys += q->n_keys; + if (key) + cache_keys ++; /* Make some space for our new entries */ dns_cache_make_space(c, cache_keys); @@ -453,44 +454,63 @@ int dns_cache_put( goto fail; } - if (!q) + if (!key) return 0; - /* Third, add in negative entries for all keys with no RR */ - for (i = 0; i < q->n_keys; i++) { - DnsResourceRecord *soa = NULL; + /* Third, add in negative entries if the key has no RR */ + r = dns_answer_contains(answer, key); + if (r < 0) + goto fail; + if (r > 0) + return 0; - r = dns_answer_contains(answer, q->keys[i]); - if (r < 0) - goto fail; - if (r > 0) - continue; + /* See https://tools.ietf.org/html/rfc2308, which + * say that a matching SOA record in the packet + * is used to to enable negative caching. */ - /* See https://tools.ietf.org/html/rfc2308, which - * say that a matching SOA record in the packet - * is used to to enable negative caching. */ + r = dns_answer_find_soa(answer, key, &soa); + if (r < 0) + goto fail; + if (r == 0) + return 0; - r = dns_answer_find_soa(answer, q->keys[i], &soa); - if (r < 0) - goto fail; - if (r == 0) - continue; + /* Also, if the requested key is an alias, the negative response should + be cached for each name in the redirect chain. Any CNAME record in + the response is from the redirection chain, though only the final one + is guaranteed to be included. This means that we cannot verify the + chain and that we need to cache them all as it may be incomplete. */ + for (i = 0; i < answer->n_rrs; i++) { + DnsResourceRecord *answer_rr = answer->items[i].rr; - r = dns_cache_put_negative(c, q->keys[i], rcode, timestamp, MIN(soa->soa.minimum, soa->ttl), owner_family, owner_address); - if (r < 0) - goto fail; + if (answer_rr->key->type == DNS_TYPE_CNAME) { + _cleanup_(dns_resource_key_unrefp) DnsResourceKey *canonical_key = NULL; + + canonical_key = dns_resource_key_new_redirect(key, answer_rr); + if (!canonical_key) + goto fail; + + /* Let's not add negative cache entries for records outside the current zone. */ + if (!dns_answer_match_soa(canonical_key, soa->key)) + continue; + + r = dns_cache_put_negative(c, canonical_key, rcode, timestamp, MIN(soa->soa.minimum, soa->ttl), owner_family, owner_address); + if (r < 0) + goto fail; + } } + r = dns_cache_put_negative(c, key, rcode, timestamp, MIN(soa->soa.minimum, soa->ttl), owner_family, owner_address); + if (r < 0) + goto fail; + return 0; fail: /* Adding all RRs failed. Let's clean up what we already * added, just in case */ - if (q) { - for (i = 0; i < q->n_keys; i++) - dns_cache_remove(c, q->keys[i]); - } + if (key) + dns_cache_remove(c, key); for (i = 0; i < answer->n_rrs; i++) dns_cache_remove(c, answer->items[i].rr->key); @@ -498,6 +518,29 @@ fail: return r; } +static DnsCacheItem *dns_cache_get_by_key_follow_cname(DnsCache *c, DnsResourceKey *k) { + _cleanup_(dns_resource_key_unrefp) DnsResourceKey *cname_key = NULL; + DnsCacheItem *i, *j; + + assert(c); + assert(k); + + i = hashmap_get(c->by_key, k); + if (i || k->type == DNS_TYPE_CNAME) + return i; + + /* check if we have a CNAME record instead */ + cname_key = dns_resource_key_new_cname(k); + if (!cname_key) + return NULL; + + j = hashmap_get(c->by_key, cname_key); + if (j) + return j; + + return i; +} + int dns_cache_lookup(DnsCache *c, DnsResourceKey *key, int *rcode, DnsAnswer **ret) { _cleanup_(dns_answer_unrefp) DnsAnswer *answer = NULL; unsigned n = 0; @@ -527,7 +570,7 @@ int dns_cache_lookup(DnsCache *c, DnsResourceKey *key, int *rcode, DnsAnswer **r return 0; } - first = hashmap_get(c->by_key, key); + first = dns_cache_get_by_key_follow_cname(c, key); if (!first) { /* If one question cannot be answered we need to refresh */ diff --git a/src/resolve/resolved-dns-cache.h b/src/resolve/resolved-dns-cache.h index 1225e58de4..60cf6a4784 100644 --- a/src/resolve/resolved-dns-cache.h +++ b/src/resolve/resolved-dns-cache.h @@ -39,7 +39,7 @@ typedef struct DnsCache { void dns_cache_flush(DnsCache *c); void dns_cache_prune(DnsCache *c); -int dns_cache_put(DnsCache *c, DnsQuestion *q, int rcode, DnsAnswer *answer, unsigned max_rrs, usec_t timestamp, int owner_family, const union in_addr_union *owner_address); +int dns_cache_put(DnsCache *c, DnsResourceKey *key, int rcode, DnsAnswer *answer, unsigned max_rrs, usec_t timestamp, int owner_family, const union in_addr_union *owner_address); int dns_cache_lookup(DnsCache *c, DnsResourceKey *key, int *rcode, DnsAnswer **answer); int dns_cache_check_conflicts(DnsCache *cache, DnsResourceRecord *rr, int owner_family, const union in_addr_union *owner_address); diff --git a/src/resolve/resolved-dns-query.c b/src/resolve/resolved-dns-query.c index c0b4c8ba81..4b1d18b2ef 100644 --- a/src/resolve/resolved-dns-query.c +++ b/src/resolve/resolved-dns-query.c @@ -831,7 +831,7 @@ void dns_query_ready(DnsQuery *q) { dns_query_complete(q, state); } -int dns_query_cname_redirect(DnsQuery *q, const char *name) { +int dns_query_cname_redirect(DnsQuery *q, const DnsResourceRecord *cname) { _cleanup_(dns_question_unrefp) DnsQuestion *nq = NULL; int r; @@ -840,7 +840,7 @@ int dns_query_cname_redirect(DnsQuery *q, const char *name) { if (q->n_cname_redirects > CNAME_MAX) return -ELOOP; - r = dns_question_cname_redirect(q->question, name, &nq); + r = dns_question_cname_redirect(q->question, cname, &nq); if (r < 0) return r; diff --git a/src/resolve/resolved-dns-query.h b/src/resolve/resolved-dns-query.h index 93d49301fa..e7063d9678 100644 --- a/src/resolve/resolved-dns-query.h +++ b/src/resolve/resolved-dns-query.h @@ -72,7 +72,7 @@ DnsQuery *dns_query_free(DnsQuery *q); int dns_query_go(DnsQuery *q); void dns_query_ready(DnsQuery *q); -int dns_query_cname_redirect(DnsQuery *q, const char *name); +int dns_query_cname_redirect(DnsQuery *q, const DnsResourceRecord *cname); int dns_query_bus_track(DnsQuery *q, sd_bus_message *m); diff --git a/src/resolve/resolved-dns-question.c b/src/resolve/resolved-dns-question.c index c94928d725..1507f22da0 100644 --- a/src/resolve/resolved-dns-question.c +++ b/src/resolve/resolved-dns-question.c @@ -242,13 +242,13 @@ int dns_question_is_equal(DnsQuestion *a, DnsQuestion *b) { return 1; } -int dns_question_cname_redirect(DnsQuestion *q, const char *name, DnsQuestion **ret) { +int dns_question_cname_redirect(DnsQuestion *q, const DnsResourceRecord *cname, DnsQuestion **ret) { _cleanup_(dns_question_unrefp) DnsQuestion *n = NULL; bool same = true; unsigned i; int r; - assert(name); + assert(cname); assert(ret); if (!q) { @@ -262,7 +262,7 @@ int dns_question_cname_redirect(DnsQuestion *q, const char *name, DnsQuestion ** } for (i = 0; i < q->n_keys; i++) { - r = dns_name_equal(DNS_RESOURCE_KEY_NAME(q->keys[i]), name); + r = dns_name_equal(DNS_RESOURCE_KEY_NAME(q->keys[i]), cname->cname.name); if (r < 0) return r; @@ -286,7 +286,7 @@ int dns_question_cname_redirect(DnsQuestion *q, const char *name, DnsQuestion ** for (i = 0; i < q->n_keys; i++) { _cleanup_(dns_resource_key_unrefp) DnsResourceKey *k = NULL; - k = dns_resource_key_new(q->keys[i]->class, q->keys[i]->type, name); + k = dns_resource_key_new_redirect(q->keys[i], cname); if (!k) return -ENOMEM; diff --git a/src/resolve/resolved-dns-question.h b/src/resolve/resolved-dns-question.h index 77de0c7a2c..13cd1f20f3 100644 --- a/src/resolve/resolved-dns-question.h +++ b/src/resolve/resolved-dns-question.h @@ -46,6 +46,6 @@ int dns_question_is_superset(DnsQuestion *q, DnsQuestion *other); int dns_question_contains(DnsQuestion *a, DnsResourceKey *k); int dns_question_is_equal(DnsQuestion *a, DnsQuestion *b); -int dns_question_cname_redirect(DnsQuestion *q, const char *name, DnsQuestion **ret); +int dns_question_cname_redirect(DnsQuestion *q, const DnsResourceRecord *cname, DnsQuestion **ret); DEFINE_TRIVIAL_CLEANUP_FUNC(DnsQuestion*, dns_question_unref); diff --git a/src/resolve/resolved-dns-rr.c b/src/resolve/resolved-dns-rr.c index f31644eebc..fd2f53f40b 100644 --- a/src/resolve/resolved-dns-rr.c +++ b/src/resolve/resolved-dns-rr.c @@ -48,6 +48,19 @@ DnsResourceKey* dns_resource_key_new(uint16_t class, uint16_t type, const char * return k; } +DnsResourceKey* dns_resource_key_new_cname(const DnsResourceKey *key) { + assert(key); + + return dns_resource_key_new(key->class, DNS_TYPE_CNAME, DNS_RESOURCE_KEY_NAME(key)); +} + +DnsResourceKey* dns_resource_key_new_redirect(const DnsResourceKey *key, const DnsResourceRecord *cname) { + assert(key); + assert(cname); + + return dns_resource_key_new(key->class, key->type, cname->cname.name); +} + DnsResourceKey* dns_resource_key_new_consume(uint16_t class, uint16_t type, char *name) { DnsResourceKey *k; diff --git a/src/resolve/resolved-dns-rr.h b/src/resolve/resolved-dns-rr.h index 8986a298af..9e2207c0aa 100644 --- a/src/resolve/resolved-dns-rr.h +++ b/src/resolve/resolved-dns-rr.h @@ -177,6 +177,8 @@ static inline const char* DNS_RESOURCE_KEY_NAME(const DnsResourceKey *key) { } DnsResourceKey* dns_resource_key_new(uint16_t class, uint16_t type, const char *name); +DnsResourceKey* dns_resource_key_new_cname(const DnsResourceKey *key); +DnsResourceKey* dns_resource_key_new_redirect(const DnsResourceKey *key, const DnsResourceRecord *cname); DnsResourceKey* dns_resource_key_new_consume(uint16_t class, uint16_t type, char *name); DnsResourceKey* dns_resource_key_ref(DnsResourceKey *key); DnsResourceKey* dns_resource_key_unref(DnsResourceKey *key); diff --git a/src/resolve/resolved-dns-transaction.c b/src/resolve/resolved-dns-transaction.c index 8092bb514d..b30473dd7e 100644 --- a/src/resolve/resolved-dns-transaction.c +++ b/src/resolve/resolved-dns-transaction.c @@ -458,7 +458,7 @@ void dns_transaction_process_reply(DnsTransaction *t, DnsPacket *p) { } /* According to RFC 4795, section 2.9. only the RRs from the answer section shall be cached */ - dns_cache_put(&t->scope->cache, p->question, DNS_PACKET_RCODE(p), p->answer, DNS_PACKET_ANCOUNT(p), 0, p->family, &p->sender); + dns_cache_put(&t->scope->cache, t->key, DNS_PACKET_RCODE(p), p->answer, DNS_PACKET_ANCOUNT(p), 0, p->family, &p->sender); if (DNS_PACKET_RCODE(p) == DNS_RCODE_SUCCESS) dns_transaction_complete(t, DNS_TRANSACTION_SUCCESS); diff --git a/src/shared/bus-util.c b/src/shared/bus-util.c index 4dc4ca581d..fdf41cec19 100644 --- a/src/shared/bus-util.c +++ b/src/shared/bus-util.c @@ -23,22 +23,24 @@ #include "sd-daemon.h" #include "sd-event.h" -#include "util.h" -#include "strv.h" -#include "macro.h" +#include "sd-bus.h" + +#include "bus-error.h" +#include "bus-internal.h" +#include "bus-label.h" +#include "bus-message.h" +#include "cgroup-util.h" #include "def.h" -#include "path-util.h" +#include "macro.h" #include "missing.h" +#include "path-util.h" #include "set.h" #include "signal-util.h" +#include "strv.h" #include "unit-name.h" +#include "util.h" -#include "sd-bus.h" -#include "bus-error.h" -#include "bus-label.h" -#include "bus-message.h" #include "bus-util.h" -#include "bus-internal.h" static int name_owner_change_callback(sd_bus_message *m, void *userdata, sd_bus_error *ret_error) { sd_event *e = userdata; @@ -1423,7 +1425,8 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen if (STR_IN_SET(field, "CPUAccounting", "MemoryAccounting", "BlockIOAccounting", "TasksAccounting", "SendSIGHUP", "SendSIGKILL", "WakeSystem", "DefaultDependencies", - "IgnoreSIGPIPE", "TTYVHangup", "TTYReset", "RemainAfterExit")) { + "IgnoreSIGPIPE", "TTYVHangup", "TTYReset", "RemainAfterExit", + "PrivateTmp", "PrivateDevices", "PrivateNetwork", "NoNewPrivileges")) { r = parse_boolean(eq); if (r < 0) { @@ -1463,10 +1466,21 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen r = sd_bus_message_append(m, "v", "t", n); - } else if (STR_IN_SET(field, "CPUShares", "BlockIOWeight")) { + } else if (STR_IN_SET(field, "CPUShares", "StartupCPUShares")) { + uint64_t u; + + r = cg_cpu_shares_parse(eq, &u); + if (r < 0) { + log_error("Failed to parse %s value %s.", field, eq); + return -EINVAL; + } + + r = sd_bus_message_append(m, "v", "t", u); + + } else if (STR_IN_SET(field, "BlockIOWeight", "StartupBlockIOWeight")) { uint64_t u; - r = safe_atou64(eq, &u); + r = cg_cpu_shares_parse(eq, &u); if (r < 0) { log_error("Failed to parse %s value %s.", field, eq); return -EINVAL; diff --git a/src/shared/pty.c b/src/shared/pty.c deleted file mode 100644 index 35d9ff5f4d..0000000000 --- a/src/shared/pty.c +++ /dev/null @@ -1,633 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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/>. -***/ - -/* - * PTY - * A PTY object represents a single PTY connection between a master and a - * child. The child process is fork()ed so the caller controls what program - * will be run. - * - * Programs like /bin/login tend to perform a vhangup() on their TTY - * before running the login procedure. This also causes the pty master - * to get a EPOLLHUP event as long as no client has the TTY opened. - * This means, we cannot use the TTY connection as reliable way to track - * the client. Instead, we _must_ rely on the PID of the client to track - * them. - * However, this has the side effect that if the client forks and the - * parent exits, we loose them and restart the client. But this seems to - * be the expected behavior so we implement it here. - * - * Unfortunately, epoll always polls for EPOLLHUP so as long as the - * vhangup() is ongoing, we will _always_ get EPOLLHUP and cannot sleep. - * This gets worse if the client closes the TTY but doesn't exit. - * Therefore, the fd must be edge-triggered in the epoll-set so we - * only get the events once they change. - */ - -#include <errno.h> -#include <fcntl.h> -#include <signal.h> -#include <stdbool.h> -#include <stdint.h> -#include <stdlib.h> -#include <sys/epoll.h> -#include <sys/ioctl.h> -#include <sys/uio.h> -#include <sys/wait.h> -#include <termios.h> -#include <unistd.h> - -#include "barrier.h" -#include "macro.h" -#include "ring.h" -#include "util.h" -#include "signal-util.h" -#include "pty.h" - -#define PTY_BUFSIZE 4096 - -enum { - PTY_ROLE_UNKNOWN, - PTY_ROLE_PARENT, - PTY_ROLE_CHILD, -}; - -struct Pty { - unsigned long ref; - Barrier barrier; - int fd; - pid_t child; - sd_event_source *fd_source; - sd_event_source *child_source; - - char in_buf[PTY_BUFSIZE]; - Ring out_buf; - - pty_event_t event_fn; - void *event_fn_userdata; - - bool needs_requeue : 1; - unsigned int role : 2; -}; - -int pty_new(Pty **out) { - _pty_unref_ Pty *pty = NULL; - int r; - - assert_return(out, -EINVAL); - - pty = new0(Pty, 1); - if (!pty) - return -ENOMEM; - - pty->ref = 1; - pty->fd = -1; - pty->barrier = (Barrier) BARRIER_NULL; - - pty->fd = posix_openpt(O_RDWR | O_NOCTTY | O_CLOEXEC | O_NONBLOCK); - if (pty->fd < 0) - return -errno; - - /* - * The slave-node is initialized to uid/gid of the caller of - * posix_openpt(). Only if devpts is mounted with fixed uid/gid this is - * skipped. In that case, grantpt() can overwrite these, but then you - * have to be root to use chown() (or a pt_chown helper has to be - * present). In those cases grantpt() really does something, - * otherwise it's a no-op. We call grantpt() here to try supporting - * those cases, even though no-one uses that, I guess. If you need other - * access-rights, set them yourself after this call returns (no, this is - * not racy, it looks racy, but races regarding your own UID are never - * important as an attacker could ptrace you; and the slave-pty is also - * still locked). - */ - r = grantpt(pty->fd); - if (r < 0) - return -errno; - - r = barrier_create(&pty->barrier); - if (r < 0) - return r; - - *out = pty; - pty = NULL; - return 0; -} - -Pty *pty_ref(Pty *pty) { - if (!pty || pty->ref < 1) - return NULL; - - ++pty->ref; - return pty; -} - -Pty *pty_unref(Pty *pty) { - if (!pty || pty->ref < 1 || --pty->ref > 0) - return NULL; - - pty_close(pty); - pty->child_source = sd_event_source_unref(pty->child_source); - barrier_destroy(&pty->barrier); - ring_clear(&pty->out_buf); - free(pty); - - return NULL; -} - -Barrier *pty_get_barrier(Pty *pty) { - assert(pty); - return &pty->barrier; -} - -bool pty_is_unknown(Pty *pty) { - return pty && pty->role == PTY_ROLE_UNKNOWN; -} - -bool pty_is_parent(Pty *pty) { - return pty && pty->role == PTY_ROLE_PARENT; -} - -bool pty_is_child(Pty *pty) { - return pty && pty->role == PTY_ROLE_CHILD; -} - -bool pty_has_child(Pty *pty) { - return pty_is_parent(pty) && pty->child > 0; -} - -pid_t pty_get_child(Pty *pty) { - return pty_has_child(pty) ? pty->child : -ECHILD; -} - -bool pty_is_open(Pty *pty) { - return pty && pty->fd >= 0; -} - -int pty_get_fd(Pty *pty) { - assert_return(pty, -EINVAL); - - return pty_is_open(pty) ? pty->fd : -EPIPE; -} - -int pty_make_child(Pty *pty) { - _cleanup_free_ char *slave_name = NULL; - int r, fd; - - assert_return(pty, -EINVAL); - assert_return(pty_is_unknown(pty), -EALREADY); - - r = ptsname_malloc(pty->fd, &slave_name); - if (r < 0) - return -errno; - - fd = open(slave_name, O_RDWR | O_CLOEXEC | O_NOCTTY); - if (fd < 0) - return -errno; - - safe_close(pty->fd); - pty->fd = fd; - pty->child = getpid(); - pty->role = PTY_ROLE_CHILD; - barrier_set_role(&pty->barrier, BARRIER_CHILD); - - return 0; -} - -int pty_make_parent(Pty *pty, pid_t child) { - assert_return(pty, -EINVAL); - assert_return(pty_is_unknown(pty), -EALREADY); - - pty->child = child; - pty->role = PTY_ROLE_PARENT; - - return 0; -} - -int pty_unlock(Pty *pty) { - assert_return(pty, -EINVAL); - assert_return(pty_is_unknown(pty) || pty_is_parent(pty), -EINVAL); - assert_return(pty_is_open(pty), -ENODEV); - - return unlockpt(pty->fd) < 0 ? -errno : 0; -} - -int pty_setup_child(Pty *pty) { - struct termios attr; - pid_t pid; - int r; - - assert_return(pty, -EINVAL); - assert_return(pty_is_child(pty), -EINVAL); - assert_return(pty_is_open(pty), -EALREADY); - - r = reset_signal_mask(); - if (r < 0) - return r; - - r = reset_all_signal_handlers(); - if (r < 0) - return r; - - pid = setsid(); - if (pid < 0 && errno != EPERM) - return -errno; - - r = ioctl(pty->fd, TIOCSCTTY, 0); - if (r < 0) - return -errno; - - r = tcgetattr(pty->fd, &attr); - if (r < 0) - return -errno; - - /* erase character should be normal backspace, PLEASEEE! */ - attr.c_cc[VERASE] = 010; - /* always set UTF8 flag */ - attr.c_iflag |= IUTF8; - - r = tcsetattr(pty->fd, TCSANOW, &attr); - if (r < 0) - return -errno; - - if (dup2(pty->fd, STDIN_FILENO) != STDIN_FILENO || - dup2(pty->fd, STDOUT_FILENO) != STDOUT_FILENO || - dup2(pty->fd, STDERR_FILENO) != STDERR_FILENO) - return -errno; - - /* only close FD if it's not a std-fd */ - pty->fd = (pty->fd > 2) ? safe_close(pty->fd) : -1; - - return 0; -} - -void pty_close(Pty *pty) { - if (!pty_is_open(pty)) - return; - - pty->fd_source = sd_event_source_unref(pty->fd_source); - pty->fd = safe_close(pty->fd); -} - -/* - * Drain input-queue and dispatch data via the event-handler. Returns <0 on - * error, 0 if queue is empty and 1 if we couldn't empty the input queue fast - * enough and there's still data left. - */ -static int pty_dispatch_read(Pty *pty) { - unsigned int i; - ssize_t len; - int r; - - /* - * We're edge-triggered, means we need to read the whole queue. This, - * however, might cause us to stall if the writer is faster than we - * are. Therefore, try reading as much as 8 times (32KiB) and only - * bail out then. - */ - - for (i = 0; i < 8; ++i) { - len = read(pty->fd, pty->in_buf, sizeof(pty->in_buf) - 1); - if (len < 0) { - if (errno == EINTR) - continue; - - return (errno == EAGAIN) ? 0 : -errno; - } else if (len == 0) - continue; - - /* set terminating zero for debugging safety */ - pty->in_buf[len] = 0; - r = pty->event_fn(pty, pty->event_fn_userdata, PTY_DATA, pty->in_buf, len); - if (r < 0) - return r; - } - - /* still data left, make sure we're queued again */ - pty->needs_requeue = true; - - return 1; -} - -/* - * Drain output-queue by writing data to the pty. Returns <0 on error, 0 if the - * output queue is empty now and 1 if we couldn't empty the output queue fast - * enough and there's still data left. - */ -static int pty_dispatch_write(Pty *pty) { - struct iovec vec[2]; - unsigned int i; - ssize_t len; - size_t num; - - /* - * Same as pty_dispatch_read(), we're edge-triggered so we need to call - * write() until either all data is written or it returns EAGAIN. We - * call it twice and if it still writes successfully, we reschedule. - */ - - for (i = 0; i < 2; ++i) { - num = ring_peek(&pty->out_buf, vec); - if (num < 1) - return 0; - - len = writev(pty->fd, vec, (int)num); - if (len < 0) { - if (errno == EINTR) - continue; - - return (errno == EAGAIN) ? 1 : -errno; - } else if (len == 0) - continue; - - ring_pull(&pty->out_buf, (size_t)len); - } - - /* still data left, make sure we're queued again */ - if (ring_get_size(&pty->out_buf) > 0) { - pty->needs_requeue = true; - return 1; - } - - return 0; -} - -static int pty_fd_fn(sd_event_source *source, int fd, uint32_t revents, void *userdata) { - Pty *pty = userdata; - int r_hup = 0, r_write = 0, r_read = 0, r; - - /* - * Whenever we encounter I/O errors, we have to make sure to drain the - * input queue first, before we handle any HUP. A child might send us - * a message and immediately close the queue. We must not handle the - * HUP first or we loose data. - * Therefore, if we read a message successfully, we always return - * success and wait for the next event-loop iteration. Furthermore, - * whenever there is a write-error, we must try reading from the input - * queue even if EPOLLIN is not set. The input might have arrived in - * between epoll_wait() and write(). Therefore, write-errors are only - * ever handled if the input-queue is empty. In all other cases they - * are ignored until either reading fails or the input queue is empty. - */ - - if (revents & (EPOLLHUP | EPOLLERR)) - r_hup = -EPIPE; - - if (revents & EPOLLOUT) - r_write = pty_dispatch_write(pty); - - /* Awesome! Kernel signals HUP without IN but queues are not empty.. */ - if ((revents & EPOLLIN) || r_hup < 0 || r_write < 0) { - r_read = pty_dispatch_read(pty); - if (r_read > 0) - return 0; /* still data left to fetch next round */ - } - - if (r_hup < 0 || r_write < 0 || r_read < 0) { - /* PTY closed and input-queue drained */ - pty_close(pty); - r = pty->event_fn(pty, pty->event_fn_userdata, PTY_HUP, NULL, 0); - if (r < 0) - return r; - } - - return 0; -} - -static int pty_fd_prepare_fn(sd_event_source *source, void *userdata) { - Pty *pty = userdata; - int r; - - if (pty->needs_requeue) { - /* - * We're edge-triggered. In case we couldn't handle all events - * or in case new write-data is queued, we set needs_requeue. - * Before going asleep, we set the io-events *again*. sd-event - * notices that we're edge-triggered and forwards the call to - * the kernel even if the events didn't change. The kernel will - * check the events and re-queue us on the ready queue in case - * an event is pending. - */ - r = sd_event_source_set_io_events(source, EPOLLHUP | EPOLLERR | EPOLLIN | EPOLLOUT | EPOLLET); - if (r >= 0) - pty->needs_requeue = false; - } - - return 0; -} - -static int pty_child_fn(sd_event_source *source, const siginfo_t *si, void *userdata) { - Pty *pty = userdata; - int r; - - pty->child = 0; - - r = pty->event_fn(pty, pty->event_fn_userdata, PTY_CHILD, si, sizeof(*si)); - if (r < 0) - return r; - - return 0; -} - -int pty_attach_event(Pty *pty, sd_event *event, pty_event_t event_fn, void *event_fn_userdata) { - int r; - - assert_return(pty, -EINVAL); - assert_return(event, -EINVAL); - assert_return(event_fn, -EINVAL); - assert_return(pty_is_parent(pty), -EINVAL); - - pty_detach_event(pty); - - if (pty_is_open(pty)) { - r = sd_event_add_io(event, - &pty->fd_source, - pty->fd, - EPOLLHUP | EPOLLERR | EPOLLIN | EPOLLOUT | EPOLLET, - pty_fd_fn, - pty); - if (r < 0) - goto error; - - r = sd_event_source_set_prepare(pty->fd_source, pty_fd_prepare_fn); - if (r < 0) - goto error; - } - - if (pty_has_child(pty)) { - r = sd_event_add_child(event, - &pty->child_source, - pty->child, - WEXITED, - pty_child_fn, - pty); - if (r < 0) - goto error; - } - - pty->event_fn = event_fn; - pty->event_fn_userdata = event_fn_userdata; - - return 0; - -error: - pty_detach_event(pty); - return r; -} - -void pty_detach_event(Pty *pty) { - if (!pty) - return; - - pty->child_source = sd_event_source_unref(pty->child_source); - pty->fd_source = sd_event_source_unref(pty->fd_source); - pty->event_fn = NULL; - pty->event_fn_userdata = NULL; -} - -int pty_write(Pty *pty, const void *buf, size_t size) { - bool was_empty; - int r; - - assert_return(pty, -EINVAL); - assert_return(pty_is_open(pty), -ENODEV); - assert_return(pty_is_parent(pty), -ENODEV); - - if (size < 1) - return 0; - - /* - * Push @buf[0..@size] into the output ring-buffer. In case the - * ring-buffer wasn't empty beforehand, we're already waiting for - * EPOLLOUT and we're done. If it was empty, we have to re-queue the - * FD for EPOLLOUT as we're edge-triggered and wouldn't get any new - * EPOLLOUT event. - */ - - was_empty = ring_get_size(&pty->out_buf) < 1; - - r = ring_push(&pty->out_buf, buf, size); - if (r < 0) - return r; - - if (was_empty) - pty->needs_requeue = true; - - return 0; -} - -int pty_signal(Pty *pty, int sig) { - assert_return(pty, -EINVAL); - assert_return(pty_is_open(pty), -ENODEV); - assert_return(pty_is_parent(pty), -ENODEV); - - return ioctl(pty->fd, TIOCSIG, sig) < 0 ? -errno : 0; -} - -int pty_resize(Pty *pty, unsigned short term_width, unsigned short term_height) { - struct winsize ws = { - .ws_col = term_width, - .ws_row = term_height, - }; - - assert_return(pty, -EINVAL); - assert_return(pty_is_open(pty), -ENODEV); - assert_return(pty_is_parent(pty), -ENODEV); - - /* - * This will send SIGWINCH to the pty slave foreground process group. - * We will also get one, but we don't need it. - */ - return ioctl(pty->fd, TIOCSWINSZ, &ws) < 0 ? -errno : 0; -} - -pid_t pty_fork(Pty **out, sd_event *event, pty_event_t event_fn, void *event_fn_userdata, unsigned short initial_term_width, unsigned short initial_term_height) { - _pty_unref_ Pty *pty = NULL; - int r; - pid_t pid; - - assert_return(out, -EINVAL); - assert_return((event && event_fn) || (!event && !event_fn), -EINVAL); - - r = pty_new(&pty); - if (r < 0) - return r; - - r = pty_unlock(pty); - if (r < 0) - return r; - - pid = fork(); - if (pid < 0) - return -errno; - - if (pid == 0) { - /* child */ - - r = pty_make_child(pty); - if (r < 0) - _exit(-r); - - r = pty_setup_child(pty); - if (r < 0) - _exit(-r); - - /* sync with parent */ - if (!barrier_place_and_sync(&pty->barrier)) - _exit(1); - - /* fallthrough and return the child's PTY object */ - } else { - /* parent */ - - r = pty_make_parent(pty, pid); - if (r < 0) - goto parent_error; - - r = pty_resize(pty, initial_term_width, initial_term_height); - if (r < 0) - goto parent_error; - - if (event) { - r = pty_attach_event(pty, event, event_fn, event_fn_userdata); - if (r < 0) - goto parent_error; - } - - /* sync with child */ - if (!barrier_place_and_sync(&pty->barrier)) { - r = -ECHILD; - goto parent_error; - } - - /* fallthrough and return the parent's PTY object */ - } - - *out = pty; - pty = NULL; - return pid; - -parent_error: - barrier_abort(&pty->barrier); - waitpid(pty->child, NULL, 0); - pty->child = 0; - return r; -} diff --git a/src/shared/pty.h b/src/shared/pty.h deleted file mode 100644 index 63c7db2833..0000000000 --- a/src/shared/pty.h +++ /dev/null @@ -1,72 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -#pragma once - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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 <unistd.h> - -#include "barrier.h" -#include "macro.h" -#include "sd-event.h" - -typedef struct Pty Pty; - -enum { - PTY_CHILD, - PTY_HUP, - PTY_DATA, -}; - -typedef int (*pty_event_t) (Pty *pty, void *userdata, unsigned int event, const void *ptr, size_t size); - -int pty_new(Pty **out); -Pty *pty_ref(Pty *pty); -Pty *pty_unref(Pty *pty); - -#define _pty_unref_ _cleanup_(pty_unrefp) -DEFINE_TRIVIAL_CLEANUP_FUNC(Pty*, pty_unref); - -Barrier *pty_get_barrier(Pty *pty); - -bool pty_is_unknown(Pty *pty); -bool pty_is_parent(Pty *pty); -bool pty_is_child(Pty *pty); -bool pty_has_child(Pty *pty); -pid_t pty_get_child(Pty *pty); - -bool pty_is_open(Pty *pty); -int pty_get_fd(Pty *pty); - -int pty_make_child(Pty *pty); -int pty_make_parent(Pty *pty, pid_t child); -int pty_unlock(Pty *pty); -int pty_setup_child(Pty *pty); -void pty_close(Pty *pty); - -int pty_attach_event(Pty *pty, sd_event *event, pty_event_t event_fn, void *event_fn_userdata); -void pty_detach_event(Pty *pty); - -int pty_write(Pty *pty, const void *buf, size_t size); -int pty_signal(Pty *pty, int sig); -int pty_resize(Pty *pty, unsigned short term_width, unsigned short term_height); - -pid_t pty_fork(Pty **out, sd_event *event, pty_event_t event_fn, void *event_fn_userdata, unsigned short initial_term_width, unsigned short initial_term_height); diff --git a/src/systemctl/systemctl.c b/src/systemctl/systemctl.c index d21ba9a566..e20fc1bbbb 100644 --- a/src/systemctl/systemctl.c +++ b/src/systemctl/systemctl.c @@ -3005,6 +3005,7 @@ static int prepare_firmware_setup(sd_bus *bus) { } static int start_special(sd_bus *bus, char **args) { + _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL; enum action a; int r; @@ -3029,6 +3030,31 @@ static int start_special(sd_bus *bus, char **args) { r = update_reboot_param_file(args[1]); if (r < 0) return r; + } else if (a == ACTION_EXIT && strv_length(args) > 1) { + /* If the exit code is not given on the command line, don't + * reset it to zero: just keep it as it might have been set + * previously. */ + uint8_t code = 0; + + r = safe_atou8(args[1], &code); + if (r < 0) { + log_error("Invalid exit code."); + return -EINVAL; + } + + r = sd_bus_call_method( + bus, + "org.freedesktop.systemd1", + "/org/freedesktop/systemd1", + "org.freedesktop.systemd1.Manager", + "SetExitCode", + &error, + NULL, + "y", code); + if (r < 0) { + log_error("Failed to execute operation: %s", bus_error_message(&error, r)); + return r; + } } if (arg_force >= 2 && @@ -3107,7 +3133,7 @@ static int check_unit_failed(sd_bus *bus, char **args) { static int kill_unit(sd_bus *bus, char **args) { _cleanup_strv_free_ char **names = NULL; - char **name; + char *kill_who = NULL, **name; int r, q; assert(bus); @@ -3118,6 +3144,10 @@ static int kill_unit(sd_bus *bus, char **args) { if (!arg_kill_who) arg_kill_who = "all"; + /* --fail was specified */ + if (streq(arg_job_mode, "fail")) + kill_who = strjoina(arg_kill_who, "-fail", NULL); + r = expand_names(bus, args + 1, NULL, &names); if (r < 0) log_error_errno(r, "Failed to expand names: %m"); @@ -3133,7 +3163,7 @@ static int kill_unit(sd_bus *bus, char **args) { "KillUnit", &error, NULL, - "ssi", *names, arg_kill_who, arg_signal); + "ssi", *names, kill_who ? kill_who : arg_kill_who, arg_signal); if (q < 0) { log_error("Failed to kill unit %s: %s", *names, bus_error_message(&error, q)); if (r == 0) @@ -6220,7 +6250,7 @@ static void systemctl_help(void) { " poweroff Shut down and power-off the system\n" " reboot [ARG] Shut down and reboot the system\n" " kexec Shut down and reboot the system with kexec\n" - " exit Request user instance exit\n" + " exit [EXIT_CODE] Request user instance or container exit\n" " switch-root ROOT [INIT] Change to a different root file system\n" " suspend Suspend the system\n" " hibernate Hibernate the system\n" @@ -7207,7 +7237,7 @@ static int systemctl_main(sd_bus *bus, int argc, char *argv[], int bus_error) { { "default", EQUAL, 1, start_special }, { "rescue", EQUAL, 1, start_special }, { "emergency", EQUAL, 1, start_special }, - { "exit", EQUAL, 1, start_special }, + { "exit", LESS, 2, start_special }, { "reset-failed", MORE, 1, reset_failed }, { "enable", MORE, 2, enable_unit, NOBUS }, { "disable", MORE, 2, enable_unit, NOBUS }, diff --git a/src/systemd/sd-ipv4acd.h b/src/systemd/sd-ipv4acd.h new file mode 100644 index 0000000000..8844ae848d --- /dev/null +++ b/src/systemd/sd-ipv4acd.h @@ -0,0 +1,55 @@ +/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ + +#ifndef foosdipv4acdfoo +#define foosdipv4acdfoo + +/*** + This file is part of systemd. + + Copyright (C) 2014 Axis Communications AB. All rights reserved. + Copyright (C) 2015 Tom Gundersen + + 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 <netinet/in.h> +#include <net/ethernet.h> + +#include "sd-event.h" + +enum { + IPV4ACD_EVENT_STOP = 0, + IPV4ACD_EVENT_BIND = 1, + IPV4ACD_EVENT_CONFLICT = 2, +}; + +typedef struct sd_ipv4acd sd_ipv4acd; +typedef void (*sd_ipv4acd_cb_t)(sd_ipv4acd *ll, int event, void *userdata); + +int sd_ipv4acd_detach_event(sd_ipv4acd *ll); +int sd_ipv4acd_attach_event(sd_ipv4acd *ll, sd_event *event, int priority); +int sd_ipv4acd_get_address(sd_ipv4acd *ll, struct in_addr *address); +int sd_ipv4acd_set_callback(sd_ipv4acd *ll, sd_ipv4acd_cb_t cb, void *userdata); +int sd_ipv4acd_set_mac(sd_ipv4acd *ll, const struct ether_addr *addr); +int sd_ipv4acd_set_index(sd_ipv4acd *ll, int interface_index); +int sd_ipv4acd_set_address(sd_ipv4acd *ll, const struct in_addr *address); +bool sd_ipv4acd_is_running(sd_ipv4acd *ll); +int sd_ipv4acd_start(sd_ipv4acd *ll); +int sd_ipv4acd_stop(sd_ipv4acd *ll); +sd_ipv4acd *sd_ipv4acd_ref(sd_ipv4acd *ll); +sd_ipv4acd *sd_ipv4acd_unref(sd_ipv4acd *ll); +int sd_ipv4acd_new (sd_ipv4acd **ret); + +#endif diff --git a/src/systemd/sd-ipv4ll.h b/src/systemd/sd-ipv4ll.h index d017158154..9581e99d31 100644 --- a/src/systemd/sd-ipv4ll.h +++ b/src/systemd/sd-ipv4ll.h @@ -43,7 +43,7 @@ int sd_ipv4ll_get_address(sd_ipv4ll *ll, struct in_addr *address); int sd_ipv4ll_set_callback(sd_ipv4ll *ll, sd_ipv4ll_cb_t cb, void *userdata); int sd_ipv4ll_set_mac(sd_ipv4ll *ll, const struct ether_addr *addr); int sd_ipv4ll_set_index(sd_ipv4ll *ll, int interface_index); -int sd_ipv4ll_set_address_seed(sd_ipv4ll *ll, uint8_t seed[8]); +int sd_ipv4ll_set_address_seed(sd_ipv4ll *ll, unsigned seed); bool sd_ipv4ll_is_running(sd_ipv4ll *ll); int sd_ipv4ll_start(sd_ipv4ll *ll); int sd_ipv4ll_stop(sd_ipv4ll *ll); diff --git a/src/test/test-execute.c b/src/test/test-execute.c index 0f4172e722..dd8ab7dcb8 100644 --- a/src/test/test-execute.c +++ b/src/test/test-execute.c @@ -137,6 +137,12 @@ static void test_exec_umask(Manager *m) { test(m, "exec-umask-0177.service", 0, CLD_EXITED); } +static void test_exec_runtimedirectory(Manager *m) { + test(m, "exec-runtimedirectory.service", 0, CLD_EXITED); + test(m, "exec-runtimedirectory-mode.service", 0, CLD_EXITED); + test(m, "exec-runtimedirectory-owner.service", 0, CLD_EXITED); +} + int main(int argc, char *argv[]) { test_function_t tests[] = { test_exec_workingdirectory, @@ -150,6 +156,7 @@ int main(int argc, char *argv[]) { test_exec_group, test_exec_environment, test_exec_umask, + test_exec_runtimedirectory, NULL, }; test_function_t *test = NULL; @@ -165,6 +172,7 @@ int main(int argc, char *argv[]) { return EXIT_TEST_SKIP; } + assert_se(setenv("XDG_RUNTIME_DIR", "/tmp/", 1) == 0); assert_se(set_unit_path(TEST_DIR) >= 0); r = manager_new(MANAGER_USER, true, &m); diff --git a/src/test/test-pty.c b/src/test/test-pty.c deleted file mode 100644 index fbab3d4ebe..0000000000 --- a/src/test/test-pty.c +++ /dev/null @@ -1,142 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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 <locale.h> -#include <string.h> -#include <sys/wait.h> -#include <unistd.h> - -#include "pty.h" -#include "util.h" -#include "signal-util.h" - -static const char sndmsg[] = "message\n"; -static const char rcvmsg[] = "message\r\n"; -static char rcvbuf[128]; -static size_t rcvsiz = 0; -static sd_event *event; - -static void run_child(Pty *pty) { - ssize_t r, l; - char buf[512]; - - r = read(0, buf, sizeof(buf)); - assert_se((size_t)r == strlen(sndmsg)); - assert_se(!strncmp(buf, sndmsg, r)); - - l = write(1, buf, r); - assert_se(l == r); -} - -static int pty_fn(Pty *pty, void *userdata, unsigned int ev, const void *ptr, size_t size) { - switch (ev) { - case PTY_DATA: - assert_se(rcvsiz < strlen(rcvmsg) * 2); - assert_se(rcvsiz + size < sizeof(rcvbuf)); - - memcpy(&rcvbuf[rcvsiz], ptr, size); - rcvsiz += size; - - if (rcvsiz >= strlen(rcvmsg) * 2) { - assert_se(rcvsiz == strlen(rcvmsg) * 2); - assert_se(!memcmp(rcvbuf, rcvmsg, strlen(rcvmsg))); - assert_se(!memcmp(&rcvbuf[strlen(rcvmsg)], rcvmsg, strlen(rcvmsg))); - } - - break; - case PTY_HUP: - /* This is guaranteed to appear _after_ the input queues are - * drained! */ - assert_se(rcvsiz == strlen(rcvmsg) * 2); - break; - case PTY_CHILD: - /* this may appear at any time */ - break; - default: - assert_se(0); - break; - } - - /* if we got HUP _and_ CHILD, exit */ - if (pty_get_fd(pty) < 0 && pty_get_child(pty) < 0) - sd_event_exit(event, 0); - - return 0; -} - -static void run_parent(Pty *pty) { - int r; - - /* write message to pty, ECHO mode guarantees that we get it back - * twice: once via ECHO, once from the run_child() fn */ - assert_se(pty_write(pty, sndmsg, strlen(sndmsg)) >= 0); - - r = sd_event_loop(event); - assert_se(r >= 0); -} - -static void test_pty(void) { - pid_t pid; - Pty *pty = NULL; - - rcvsiz = 0; - zero(rcvbuf); - - assert_se(sd_event_default(&event) >= 0); - - pid = pty_fork(&pty, event, pty_fn, NULL, 80, 25); - assert_se(pid >= 0); - - if (pid == 0) { - /* child */ - run_child(pty); - exit(0); - } - - /* parent */ - run_parent(pty); - - /* Make sure the PTY recycled the child; yeah, this is racy if the - * PID was already reused; but that seems fine for a test. */ - assert_se(waitpid(pid, NULL, WNOHANG) < 0 && errno == ECHILD); - - pty_unref(pty); - sd_event_unref(event); -} - -int main(int argc, char *argv[]) { - unsigned int i; - - log_parse_environment(); - log_open(); - - assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGCHLD, -1) >= 0); - - /* Oh, there're ugly races in the TTY layer regarding HUP vs IN. Turns - * out they appear only 10% of the time. I fixed all of them and - * don't see them, anymore. But let's be safe and run this 1000 times - * so we catch any new ones, in case they appear again. */ - for (i = 0; i < 1000; ++i) - test_pty(); - - return 0; -} diff --git a/src/test/test-ring.c b/src/test/test-ring.c deleted file mode 100644 index cb8a5d4e9e..0000000000 --- a/src/test/test-ring.c +++ /dev/null @@ -1,130 +0,0 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - -/*** - This file is part of systemd. - - Copyright 2014 David Herrmann <dh.herrmann@gmail.com> - - 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 <string.h> - -#include "def.h" -#include "ring.h" - -static void test_ring(void) { - static const char buf[8192]; - Ring r; - size_t l; - struct iovec vec[2]; - int s; - - zero(r); - - l = ring_peek(&r, vec); - assert_se(l == 0); - - s = ring_push(&r, buf, 2048); - assert_se(!s); - assert_se(ring_get_size(&r) == 2048); - - l = ring_peek(&r, vec); - assert_se(l == 1); - assert_se(vec[0].iov_len == 2048); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(ring_get_size(&r) == 2048); - - ring_pull(&r, 2048); - assert_se(ring_get_size(&r) == 0); - - l = ring_peek(&r, vec); - assert_se(l == 0); - assert_se(ring_get_size(&r) == 0); - - s = ring_push(&r, buf, 2048); - assert_se(!s); - assert_se(ring_get_size(&r) == 2048); - - l = ring_peek(&r, vec); - assert_se(l == 1); - assert_se(vec[0].iov_len == 2048); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(ring_get_size(&r) == 2048); - - s = ring_push(&r, buf, 1); - assert_se(!s); - assert_se(ring_get_size(&r) == 2049); - - l = ring_peek(&r, vec); - assert_se(l == 2); - assert_se(vec[0].iov_len == 2048); - assert_se(vec[1].iov_len == 1); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(!memcmp(vec[1].iov_base, buf, vec[1].iov_len)); - assert_se(ring_get_size(&r) == 2049); - - ring_pull(&r, 2048); - assert_se(ring_get_size(&r) == 1); - - l = ring_peek(&r, vec); - assert_se(l == 1); - assert_se(vec[0].iov_len == 1); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(ring_get_size(&r) == 1); - - ring_pull(&r, 1); - assert_se(ring_get_size(&r) == 0); - - s = ring_push(&r, buf, 2048); - assert_se(!s); - assert_se(ring_get_size(&r) == 2048); - - s = ring_push(&r, buf, 2049); - assert_se(!s); - assert_se(ring_get_size(&r) == 4097); - - l = ring_peek(&r, vec); - assert_se(l == 1); - assert_se(vec[0].iov_len == 4097); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(ring_get_size(&r) == 4097); - - ring_pull(&r, 1); - assert_se(ring_get_size(&r) == 4096); - - s = ring_push(&r, buf, 4096); - assert_se(!s); - assert_se(ring_get_size(&r) == 8192); - - l = ring_peek(&r, vec); - assert_se(l == 2); - assert_se(vec[0].iov_len == 8191); - assert_se(vec[1].iov_len == 1); - assert_se(!memcmp(vec[0].iov_base, buf, vec[0].iov_len)); - assert_se(!memcmp(vec[1].iov_base, buf, vec[1].iov_len)); - assert_se(ring_get_size(&r) == 8192); - - ring_clear(&r); - assert_se(ring_get_size(&r) == 0); -} - -int main(int argc, char *argv[]) { - log_parse_environment(); - log_open(); - - test_ring(); - - return 0; -} diff --git a/src/timesync/timesyncd.c b/src/timesync/timesyncd.c index 7b4178c993..3cb7d435cd 100644 --- a/src/timesync/timesyncd.c +++ b/src/timesync/timesyncd.c @@ -131,7 +131,7 @@ int main(int argc, char *argv[]) { if (r < 0) log_warning_errno(r, "Failed to parse configuration file: %m"); - log_debug("systemd-timesyncd running as pid %lu", (unsigned long) getpid()); + log_debug("systemd-timesyncd running as pid " PID_FMT, getpid()); sd_notify(false, "READY=1\n" "STATUS=Daemon is running"); |