| Age | Commit message (Collapse) | Author |
|---|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Uevents are events of the host, which should not leak into a container.
Containers do not support hotplug at the moment, and devices and uevents
are not namespace aware.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Clang is a bit more strict wrt format-nonliterals:
http://clang.llvm.org/docs/LanguageExtensions.html#format-string-checking
Adding these extra printf attributes also makes gcc able to find more
problems. E.g. this patch uncovers a format issue in udev-builtin-path_id.c
Some parts looked intetional about breaking the format-nonliteral check.
I added some supression for warnings there.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
https://bugs.freedesktop.org/show_bug.cgi?id=72164
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Returning anything else but NULL would suggest the caller's reference
might still be valid, but it isn't, because the caller just invoked
_unref() after all.
This turns the return value into a typesafe shortcut that allows
unreffing and resetting a reference in one line. In contrast to
solutions for this which take a pointer to a pointer to accomplish the
same this solution is just syntactic sugar the developer can make use of
but doesn't have to, and this is particularly useful when immediately
unreffing objects returned by function calls.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
|
|
|
|
This brings it in sync with the default config file.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
This extends 62678ded 'efi: never call qsort on potentially
NULL arrays' to all other places where qsort is used and it
is not obvious that the count is non-zero.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
In the process, rename udev_encode_string which is poorly named
for what it does. It deals specifically with encoding names that
udev creates and has its own rules: utf8 is valid but some ascii
is not (e.g. path separators), and everything else is simply escaped.
Rename it to encode_devnode_name.
Adopted for eudev: Anthony G. Basile <blueness@gentoo.org>
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
There's now some more obvious overlap amongst the two
utf8 validation functions, but no more than there already
was previously. This also adds some menial tests for anyone
who wants to do more merging of these two in the future.
Adopted for eudev: Anthony G. Basile <blueness@gentoo.org>
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
src/udev/Makefile.am: Fix udevadm symlink
|
|
POSIX says:
Note that sysconf(_SC_GETGR_R_SIZE_MAX) may return -1 if there is no
hard limit on the size of the buffer needed to store all the groups
returned.
The example from POSIX uses a default buffer size of 1024 in that case.
Signed-off-by: Michael Forney <mforney@mforney.org>
|
|
Signed-off-by: Michael Forney <mforney@mforney.org>
|
|
The symlink destination should not include DESTDIR.
Signed-off-by: Michael Forney <mforney@mforney.org>
|
|
We remove all code which logs to the journal, preserving only
write_to_console(), write_to_syslog() and write_to_kmsg().
https://github.com/gentoo/eudev/issues/77
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
At the beginning move_later is set to -1, but it is set to different
value only if expression !move_later is true.
|
|
|
|
src/udev/udev-rules.c: In function 'add_rule':
src/udev/udev-rules.c:1078:33: warning: format '%lu' expects argument of type 'long unsigned int', but argument 8 has type 'int' [-Wformat=]
log_error("invalid key/value pair in file %s on line %u,"
^
|
|
systemd-udevd[6260]: invalid key/value pair in file /usr/lib/udev/rules.d/60-ffado.rules on line 46,starting at character 84 ('#')
|
|
A regression introduced when we moved to systemd's logging is that the only
way to adjust the log-level of the udev daemon is via the env var, kernel
commandline or the commandline.
This reintroduces support for specifying this in the configuration file.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
composed one
|
|
|
|
udev_device_get_subsystem() may return NULL if no subsystem could be
figured out by libudev. This might be due to OOM or if the device
disconnected between the udev_device_new() call and
udev_device_get_subsystem(). Therefore, we need to handle subsystem==NULL
safely.
Instead of testing for it in each helper, we treat subsystem==NULL as
empty subsystem in match_subsystem().
Backtrace of udev_enumerate with an input-device disconnecting in exactly
this time-frame:
(gdb) bt
#0 0x00007ffff569dc24 in strnlen () from /usr/lib/libc.so.6
#1 0x00007ffff56d9e04 in fnmatch@@GLIBC_2.2.5 () from /usr/lib/libc.so.6
#2 0x00007ffff5beb83d in match_subsystem (udev_enumerate=0x7a05f0, subsystem=0x0) at src/libudev/libudev-enumerate.c:727
#3 0x00007ffff5bebb30 in parent_add_child (enumerate=enumerate@entry=0x7a05f0, path=<optimized out>) at src/libudev/libudev-enumerate.c:834
#4 0x00007ffff5bebc3f in parent_crawl_children (enumerate=enumerate@entry=0x7a05f0, path=0x7a56b0 "/sys/devices/<shortened>/input/input97", maxdepth=maxdepth@entry=254) at src/libudev/libudev-enumerate.c:866
#5 0x00007ffff5bebc54 in parent_crawl_children (enumerate=enumerate@entry=0x7a05f0, path=0x79e8c0 "/sys/devices/<shortened>/input", maxdepth=maxdepth@entry=255) at src/libudev/libudev-enumerate.c:868
#6 0x00007ffff5bebc54 in parent_crawl_children (enumerate=enumerate@entry=0x7a05f0, path=path@entry=0x753190 "/sys/devices/<shortened>", maxdepth=maxdepth@entry=256) at src/libudev/libudev-enumerate.c:868
#7 0x00007ffff5bec7df in scan_devices_children (enumerate=0x7a05f0) at src/libudev/libudev-enumerate.c:882
#8 udev_enumerate_scan_devices (udev_enumerate=udev_enumerate@entry=0x7a05f0) at src/libudev/libudev-enumerate.c:919
#9 0x00007ffff5df8777 in <random_caller> () at some/file.c:181
|
|
We need to free udev-devices again if they don't match. Funny that no-one
noticed it yet since valgrind is quite verbose about it.
Fix it and free non-matching devices.
|
|
If a realloc() happens in syspath_add(), the move_later pointer could
point to an invalid memory region.
Let move_later store the array index, instead of the pointer to the
entry.
|
|
|
|
|
|
|
|
|
|
while theoretically a nice interface, qsort_r on GLIBC was
implemented with a prototype that shuffles the argument order
around when compared to the traditional BSD implementations,
yielding in 2 separate incompatible implementations.
even worse, the arguments are all of pointer type so one would
not even notice that the order is wrong and so this would yield
in crashes or silent memory corruption.
thus musl does not implement it, because configure scripts would
check for its existance and use it unconditionally, even when
assuming the BSD version.
a more portable solution is to use TLS via __thread, which any
modern GCC should provide. (even 3.4.6 on x86/x86_64 does so).
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
sys/poll.h is a legacy alias used by glibc.
according to POSIX #include <poll.h> is correct.
on GLIBC, the POSIX header includes sys/poll.h, so everything
continues working as it should.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
this is a glibc specific alias for realpath(path, NULL).
to be portable, we need to use the real thing which is even less
verbose.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
strndupa() allocates memory using alloca() which is dangerous and
non-conformant. Some libcs, like musl, deliberately do not implement
it. Instead we allocate a buffer on the stack which is cleaned up
on return.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
If we pass a variable to open()'s flags parameter it really wants a mode
parameter too, otherwise some gcc version whine. Hence, pass 0 in that
case.
|
|
We reorder the functions prototypes and definitions to match
upstream as in commit aa417a4d83999f6d7f092161d5c411b8cbce9977.
The order was lost when that commit was revert and the functions
re-introduced in later commits.
Preserving the order helps to better track upstream changes when
doing a diff between files.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
When a new network iface device is added, scan through the list of rules to see if its
kernel-assigned name is used as a target for another device. If so, and said target device
is not this device, rename it to a temporary interface name. Then rename the device in
accordance with any rename rules that may apply to this device, if applicable.
The temporary name assigned is the basename of the interface with a numeric compoment which is
close to the inverse of the numeric id (127 - id#). This should provide a more user-friendly
output than the old 'rename#' behaviour, when there is no final target name for the iface.
This proactive temporary rename will prevent cases where old-style rule-generator rules are used
and a target NAME= is set for one iface, assigning it to the iface name used by a second iface,
and that second iface has no rename rule to apply. The original rename code would be blocked
due to the conflict and time out when attempting to rename, leaving the interface assigned to the
temporary 'rename[id#]' name and/or failing to rename other ifaces in accordance with the existing
rules. This is a corner case that only occurrs when 75-persistent-net-generator.rules or the
write_net_rules script it 'IMPORTS' fails to generate a new rule and rename the interface and
there is no other interface-renaming rules that apply.
There may also be performance benefits to renaming ifaces early, but no benchmarks have been
run to confirm this.
Signed-off-by: Ian Stakenvicius <axs@gentoo.org>
|
|
Based on systemd commit fb624d04e56018defe297eff1d88888b192caeaa , added
to support commit f8e6d758cd65b28f94b62344bd95e7bab94370d4
Signed-off-by: Ian Stakenvicius <axs@gentoo.org>
|
|
This is a partial combination of two systemd commits:
5926cccae202f1b8869017d4bdaf9e9ce371bba6
4fe88d28a4cfa504c1f2362d4a7030fae4c4af29
...and provides functions needed for the changes that were added
in commit 6284c1ca41dca85e968e8d304cdc154c5a46d6d2
Signed-off-by: Ian Stakenvicius <axs@gentoo.org>
|
