chiark / gitweb /
12 years agoUse LT_INIT to explicit that udev needs libtool series 2.
Diego Elio 'Flameeyes' Pettenò [Sat, 8 Aug 2009 20:51:18 +0000 (22:51 +0200)]
Use LT_INIT to explicit that udev needs libtool series 2.

12 years agoFix “make distcheck” run outside of the source directory.
Diego Elio 'Flameeyes' Pettenò [Sat, 8 Aug 2009 20:12:42 +0000 (22:12 +0200)]
Fix “make distcheck” run outside of the source directory.

Pass the path to keys.txt as second parameter of so that
it can be found in the right path.

12 years agofix spelling
Alan Jenkins [Sat, 8 Aug 2009 13:29:38 +0000 (14:29 +0100)]
fix spelling

Fix spelling in docbook comments, code comments, and a local variable
name.  Thanks to "ispell -h" for docbook HTML and "scspell" for source

Signed-off-by: Alan Jenkins <>
12 years agoFix building of documentation when doing out-of-source builds.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 19:54:35 +0000 (21:54 +0200)]
Fix building of documentation when doing out-of-source builds.

Since gtk-mkhtml is executed in a sub-directory of the build directory, and
make does not know of that, the $(buildir) variable will still be "." and
the $(srcdir) will not properly be found. For this reason, use the absolute
variants for the two functions, which won't be changing.

12 years agoUse the keymap check during “make distcheck” rather than “check”.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 18:05:55 +0000 (20:05 +0200)]
Use the keymap check during “make distcheck” rather than “check”.

Since the script checks for validity the source directory
and the file, instead of running it during user-oriented “make
check”, run it during developed-oriented “make distcheck”.

An invalid keymap will abort the execution which will prevent shipping
an incomplete

To properly support out-of-source builds, pass as single parameter to the
test the path to the source directory.

12 years agoMerge in into
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 17:44:30 +0000 (19:44 +0200)]
Merge in into

This removes another file from the distribution, since we're not using it
anywhere else but the top-level file.

12 years agoAsk gperf to use ANSI-C for generation.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 17:41:17 +0000 (19:41 +0200)]
Ask gperf to use ANSI-C for generation.

This avoids an '80s C prototype which caused a warning during our build.

12 years agoAdd tests to the distribution; this fixes "make distcheck".
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 09:22:49 +0000 (11:22 +0200)]
Add tests to the distribution; this fixes "make distcheck".

12 years agoInclude the correct directory for out-of-source builds.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 09:21:34 +0000 (11:21 +0200)]
Include the correct directory for out-of-source builds.

When building in-source, the source and header files are in the same
directory, but they are not in out-of-source.

12 years agoFix another relative path for the new working directory.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 09:12:48 +0000 (11:12 +0200)]
Fix another relative path for the new working directory.

12 years agoFix building of introspection library on top-level
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 09:04:58 +0000 (11:04 +0200)]
Fix building of introspection library on top-level

Since the library is in a subdirectory, it has to know where to look for

12 years agoAdd silent-rule support for the gudev rules.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:22:21 +0000 (03:22 +0200)]
Add silent-rule support for the gudev rules.

12 years agoMake sure to use dependency/target variables.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:21:00 +0000 (03:21 +0200)]
Make sure to use dependency/target variables.

12 years agoMake sure to clean up all the built sources.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:18:44 +0000 (03:18 +0200)]
Make sure to clean up all the built sources.

12 years agoFinally, also merge gudev into the top-level
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:16:32 +0000 (03:16 +0200)]
Finally, also merge gudev into the top-level

The Introspection rules are not tested yet; more touch-ups have been made
for them.

12 years agoMove pkg-config docs and man pages before conditionals.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:07:51 +0000 (03:07 +0200)]
Move pkg-config docs and man pages before conditionals.

Since gudev conditionally installs further pkg-config data, we have to set
it before the condition might be expanded.

12 years agoMake keymap generation rules be silent (backward-compatible).
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 01:00:23 +0000 (03:00 +0200)]
Make keymap generation rules be silent (backward-compatible).

12 years agoMerge keymap building in the top-level
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 00:58:23 +0000 (02:58 +0200)]
Merge keymap building in the top-level

Slight adjustment around the tests and the rules for the new working

12 years agoChange hook handling to be more portable.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 00:52:49 +0000 (02:52 +0200)]
Change hook handling to be more portable.

12 years agoAlso merge into the top-level the simpler extras.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 00:38:02 +0000 (02:38 +0200)]
Also merge into the top-level the simpler extras.

12 years agoReplace the custom test-run target with the standard make check.
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 00:23:01 +0000 (02:23 +0200)]
Replace the custom test-run target with the standard make check.

A little fix is needed for the script (to be called with the
proper path), but this allows for the test binaries to be only built when
running the tests themselves.

12 years agoMerge libudev, udev, and the unconditional extras in a single
Diego Elio 'Flameeyes' Pettenò [Fri, 7 Aug 2009 00:01:29 +0000 (02:01 +0200)]
Merge libudev, udev, and the unconditional extras in a single

Instead of using multiple recursive files, use a single that sets and builds all the basic suite of libraries and
binaries for udev. This reduces the number of files in the source tree, and
also reduces drastically the build time when using parallel-make.

With this setup, all the compile steps will be executed in parallel, and
just the linking stage will be (partially) serialised on the libraries

12 years agoextras/keymap: Fix Bluetooth key on Acer Aspire 6920
Martin Pitt [Fri, 7 Aug 2009 14:41:54 +0000 (15:41 +0100)]
extras/keymap: Fix Bluetooth key on Acer Aspire 6920


12 years agoman: SYMLINK can be matched as well as assigned
Alan Jenkins [Fri, 7 Aug 2009 12:29:49 +0000 (13:29 +0100)]
man: SYMLINK can be matched as well as assigned

If by-path / by-id links don't quite do what you want, this is a nice
clean way to extend the behaviour.  Real example:

SYMLINK=="serial/by-id/usb-Novatel_Wireless_Inc*CDMA*-if00-port0", \\

Some users ask how to do things like this.

 - create an additional link with a shorter name
 - create a link which matches more loosely
    (omit certain path segments e.g. serial numbers)
 - change permissions on certain USB device nodes

Allow them to realize this without reading the friendly *.c files.

Signed-off-by: Alan Jenkins <>
12 years agoman: fix unused, inaccurate metadata
Alan Jenkins [Fri, 7 Aug 2009 12:28:27 +0000 (13:28 +0100)]
man: fix unused, inaccurate metadata

Dates aren't shown in the manpages.  So they are not really useful,
and no-one is going to remember to update them.

"<refmiscinfo class="version"></refmiscinfo>" sounds even less useful.

I leave the unused "title" and "productname" tags.  They could
theoretically be useful, and aren't hard to maintain.  We just need to
fix the "title" for udevadm.

Signed-off-by: Alan Jenkins <>
12 years agoextras/keymap: Add Samsung NC20
Martin Pitt [Fri, 7 Aug 2009 10:11:48 +0000 (11:11 +0100)]
extras/keymap: Add Samsung NC20

Reported by Dirk Thierbach <> on hal ML.

12 years agohid2hci: install re-trigger for hid device when recovering from S3
Mario Limonciello [Fri, 7 Aug 2009 00:23:44 +0000 (02:23 +0200)]
hid2hci: install re-trigger for hid device when recovering from S3

12 years ago"record_failed" -> "fail_event_on_error"
Kay Sievers [Thu, 6 Aug 2009 17:23:06 +0000 (19:23 +0200)]
"record_failed" -> "fail_event_on_error"

12 years agore-enable failed event tracking
Kay Sievers [Thu, 6 Aug 2009 14:16:26 +0000 (16:16 +0200)]
re-enable failed event tracking

It did not work for the last couple of releases.

If RUN{record_failed}+="..." is given, a non-zero execution will mark
the event as failed. Recorded failed events can be re-triggered with:
  udevadm trigger --type=failed

The failed tracking _might_ be useful for things which might not be
ready to be executed at early bootup, but a bit later when the needed
dependencies are available. In many cases though, it indicates that
something is used in a way it should not.

12 years agomake: suppress enter/leaving directory messages
Kay Sievers [Thu, 6 Aug 2009 13:51:47 +0000 (15:51 +0200)]
make: suppress enter/leaving directory messages

12 years agolibudev: silent gcc warning: may be used uninitialized in this function
Kay Sievers [Thu, 6 Aug 2009 13:37:55 +0000 (15:37 +0200)]
libudev: silent gcc warning: may be used uninitialized in this function

12 years agoudev-acl: add joystick devices
Anssi Hannula [Mon, 27 Jul 2009 20:59:44 +0000 (23:59 +0300)]
udev-acl: add joystick devices

User needs read-write access to joystick devices in order to use force
feedback features.

12 years agorules: exclude digitizers from joystick class
Anssi Hannula [Mon, 27 Jul 2009 20:56:55 +0000 (23:56 +0300)]
rules: exclude digitizers from joystick class

Exclude digitizers and similar devices from ID_CLASS joystick by
checking modalias for BTN_DIGI.

This was also done for linux kernel joydev interface in linux commit

12 years agoupdate a few years of copyright
Kay Sievers [Sat, 1 Aug 2009 13:39:18 +0000 (09:39 -0400)]
update a few years of copyright

12 years agofix util_lookup_group to handle large groups
Eric W. Biederman [Sat, 1 Aug 2009 05:26:10 +0000 (22:26 -0700)]
fix util_lookup_group to handle large groups

I have recently been getting the above message on fc11 and
I have traced it down to a bug in util_lookup_group.

As of 145 util_lookup_group reads:

gid_t util_lookup_group(struct udev *udev, const char *group)
char *endptr;
int buflen = sysconf(_SC_GETGR_R_SIZE_MAX);
char buf[buflen];
struct group grbuf;
struct group *gr;
gid_t gid = 0;

if (strcmp(group, "root") == 0)
return 0;
gid = strtoul(group, &endptr, 10);
if (endptr[0] == '\0')
return gid;

errno = 0;
getgrnam_r(group, &grbuf, buf, buflen, &gr);
if (gr != NULL)
return gr->gr_gid;
if (errno == 0 || errno == ENOENT || errno == ESRCH)
err(udev, "specified group '%s' unknown\n", group);
err(udev, "error resolving group '%s': %m\n", group);
return 0;

The errno value from getgrnam_r here is ERANGE which is documented as
"Insufficient buffer space supplied".

When I call get getgrnam_r with a large enough buffer everything
works.  Indicating that the problem is that sysconf is returning
a value too small.

A quick google search tells me that sysconf(_S_GETGR_R_SIZE_MAX)
is documented as:

> sysconf(_S_GETGR_R_SIZE_MAX) returns either -1 or a good
> suggested starting value for buflen.  It does not return the
> worst case possible for buflen.

In my case I have a group with about 50 users in /etc/group
and that is what triggered the problem in udev and caused
all of the udevs group lookups to fail.

The following patch which dynamically allocates the group member buffer
should fix this problem.

Signed-off-by: "Eric W. Biederman" <>
12 years agoenumeration: move ALSA control devices to the end of the enumerated devices of each...
Lennart Poettering [Mon, 27 Jul 2009 21:24:27 +0000 (23:24 +0200)]
enumeration: move ALSA control devices to the end of the enumerated devices of each card

Generally ALSA control devices should be the last ones to be processed
for ACL changes and similar operations because they can then be used as
indicators that ACL management finished for all device nodes of a
specific card.

This patch simple moves each controlC device behind all the pcmC devices
(and similar).

12 years agomake: build internal tools against
Kay Sievers [Sat, 25 Jul 2009 18:37:45 +0000 (20:37 +0200)]
make: build internal tools against

12 years agohid2hci: rewrite (and break) rules and device handling
Kay Sievers [Fri, 24 Jul 2009 16:06:22 +0000 (18:06 +0200)]
hid2hci: rewrite (and break) rules and device handling

We must never access random devices in /dev which do not belong to
the event we are handling. Hard-coding /dev/hidrawX, and looping over all
devices is absolutely not acceptable --> hook into hidraw events.

We can not relay on (rather random) properties merged into the parent
device by earlier rules --> use libudev to find the sibling device
with a matching interface.

Libusb does not fit into udev's use case. We never want want to scan
and open() all usb devices in the system, just to find the device
we are already handling the event for --> put all the stupid scanning
into a single function and prepare for a fixed libusb or drop it later.

12 years agoformat names are not case insensitive
Kay Sievers [Thu, 23 Jul 2009 18:28:16 +0000 (20:28 +0200)]
format names are not case insensitive

12 years agopath_id: fix typo in comment
Kay Sievers [Thu, 23 Jul 2009 18:27:13 +0000 (20:27 +0200)]
path_id: fix typo in comment

12 years agorules: serial - fix path_id call
Kay Sievers [Thu, 23 Jul 2009 18:26:45 +0000 (20:26 +0200)]
rules: serial - fix path_id call

12 years agoextras/keymap: Fix crash for unknown keys
Martin Pitt [Thu, 23 Jul 2009 17:00:38 +0000 (19:00 +0200)]
extras/keymap: Fix crash for unknown keys

The keymap table has some holes in it, which caused the interactive mode to
crash for unknown keys. In these cases, print the numeric key code instead.

12 years agoextras/modem-modeswitch: Add Huawei E1550 GSM modem
Erik Forsberg [Thu, 23 Jul 2009 14:40:25 +0000 (16:40 +0200)]
extras/modem-modeswitch: Add Huawei E1550 GSM modem

What's odd is that this is a huawei modem, not an option modem, so one would
expect it to work better with usb_modeswitch and it's -H (huawei) mode - but
that's not the case, I've tested that as well.

12 years agoudevadm: trigger - add --sysname-match=
Kay Sievers [Wed, 22 Jul 2009 22:02:28 +0000 (00:02 +0200)]
udevadm: trigger - add --sysname-match=

  $ udevadm trigger -n -v --subsystem-match=usb --sysname-match=2-1.1*

12 years agopath_id: make global variable static
Kay Sievers [Wed, 22 Jul 2009 17:09:48 +0000 (19:09 +0200)]
path_id: make global variable static

12 years agohid2hci: remove hid structures and include kernel header
Kay Sievers [Tue, 21 Jul 2009 23:03:33 +0000 (01:03 +0200)]
hid2hci: remove hid structures and include kernel header

12 years agohid2hci: support to hid2hci for recovering Dell BT devices after S3
Mario Limonciello [Tue, 21 Jul 2009 22:33:27 +0000 (00:33 +0200)]
hid2hci: support to hid2hci for recovering Dell BT devices after S3

12 years version bump
Kay Sievers [Mon, 20 Jul 2009 23:40:47 +0000 (01:40 +0200)] version bump

12 years agov4l_id: exit with 0 when --help is given
Kay Sievers [Mon, 20 Jul 2009 23:36:42 +0000 (01:36 +0200)]
v4l_id: exit with 0 when --help is given

12 years agoextras/keymap: Add Samsung SX22S
Martin Pitt [Mon, 20 Jul 2009 15:59:11 +0000 (17:59 +0200)]
extras/keymap: Add Samsung SX22S

Reported on hal ML by Sergey Astanin <>.

12 years agoextras/keymap: teach findkeyboards about USB keyboards
Martin Pitt [Sat, 18 Jul 2009 14:59:46 +0000 (16:59 +0200)]
extras/keymap: teach findkeyboards about USB keyboards

12 years agoextras/keymap: Add Fujitsu Amilo M
Martin Pitt [Sat, 18 Jul 2009 13:30:08 +0000 (15:30 +0200)]
extras/keymap: Add Fujitsu Amilo M

12 years agoextras/keymap: cover more Compaq Evo models
Martin Pitt [Sat, 18 Jul 2009 13:23:55 +0000 (15:23 +0200)]
extras/keymap: cover more Compaq Evo models shows that may Evo N* models share the same
keymap, so generalize the existing rule.

12 years agoextras/keymap: Add HP Presario 2100
Martin Pitt [Sat, 18 Jul 2009 13:17:20 +0000 (15:17 +0200)]
extras/keymap: Add HP Presario 2100

12 years agoextras/keymap: add recently added keymap files to
Martin Pitt [Sat, 18 Jul 2009 13:09:05 +0000 (15:09 +0200)]
extras/keymap: add recently added keymap files to

12 years agoextras/keymap: fix for inline mappings
Martin Pitt [Sat, 18 Jul 2009 13:07:55 +0000 (15:07 +0200)]
extras/keymap: fix for inline mappings

Do not complain about missing keymap files when giving scancode/keyname pairs
on the keymap command line in the rules.

12 years agokeymap: inline one-line key maps
Martin Pitt [Sat, 18 Jul 2009 13:02:25 +0000 (15:02 +0200)]
keymap: inline one-line key maps

Remove key map files which have only one override. Instead, use keymap tools'
new feature of specifying scancode/keyname pairs directly at the command line.

Also add a comment to 95-keymap.rules about how to specify key mappings in the

12 years agokeymap tool: support scancode/keycode pair arguments
Martin Pitt [Sat, 18 Jul 2009 12:52:45 +0000 (14:52 +0200)]
keymap tool: support scancode/keycode pair arguments

This avoids having to create and parse an entire keymap file for cases where
just one or two keys have to be remapped.

12 years agokeymap tool: improve help
Martin Pitt [Sat, 18 Jul 2009 12:27:50 +0000 (14:27 +0200)]
keymap tool: improve help

Do not duplicate help strings, and add missing calling variant for

12 years agoextras/keymap: add Compal Hel80i
Martin Pitt [Sat, 18 Jul 2009 12:13:28 +0000 (14:13 +0200)]
extras/keymap: add Compal Hel80i

12 years agoextras/keymap: add Everex Stepnote XT5000T
Martin Pitt [Sat, 18 Jul 2009 07:27:38 +0000 (09:27 +0200)]
extras/keymap: add Everex Stepnote XT5000T

Reported in

12 years agoudevd: add timestamp to --debug output
Kay Sievers [Fri, 17 Jul 2009 11:24:37 +0000 (13:24 +0200)]
udevd: add timestamp to --debug output

12 years agoextras/keymap: add Zepto ZNote
Martin Pitt [Thu, 16 Jul 2009 15:34:01 +0000 (17:34 +0200)]
extras/keymap: add Zepto ZNote

Reported on

12 years agoadd keymap for Clevo D410J laptop
Martin Pitt [Thu, 16 Jul 2009 09:23:16 +0000 (11:23 +0200)]
add keymap for Clevo D410J laptop

Reported as hal-info patch by Németh Márton <> on hal ML.

12 years agorelease 145
Kay Sievers [Mon, 13 Jul 2009 10:19:01 +0000 (12:19 +0200)]
release 145

12 years agoudevd: use bool
Kay Sievers [Mon, 13 Jul 2009 01:33:15 +0000 (03:33 +0200)]
udevd: use bool

12 years agoudevd: handle SIGCHLD before the worker event message
Kay Sievers [Mon, 13 Jul 2009 01:09:05 +0000 (03:09 +0200)]
udevd: handle SIGCHLD before the worker event message

We may need to handle SIGCHLD before the queued worker message. The last
reference, from the SIGCHLD or the worker message will clean up the worker
context. In case we receive an unexpected SIGCHLD with an error, we let
the event fail and clean up the worker context.

12 years agoudevd: make sure a worker finishes event handling before exiting
Kay Sievers [Sat, 11 Jul 2009 16:21:26 +0000 (18:21 +0200)]
udevd: make sure a worker finishes event handling before exiting

Persistent network rules write out new rules files. When rules change,
we need to kill all workers to update the in-memory copy of the rules.
We need to make sure, that a worker finshes its work for all device
messages it has accepted, before it exits after a SIGTERM from the main

12 years agoudevadm: info - add space after R:, A:, W: on database export
Kay Sievers [Wed, 8 Jul 2009 15:05:55 +0000 (17:05 +0200)]
udevadm: info - add space after R:, A:, W: on database export

<zzam> kay: udevadmin info -e prints all lines with a space after the :, but the W: line
<kay> zzam: yeah, seems so, and the R: and the A:
<zzam> yeah

12 years agoudevd: detach event from worker if we kill a worker
Kay Sievers [Wed, 8 Jul 2009 12:13:20 +0000 (14:13 +0200)]
udevd: detach event from worker if we kill a worker

Jul  8 09:36:41 udevd[663]: worker [5491] did not accept message, kill it
Jul  8 09:36:41 udevd[663]: worker [5491] unexpectedly returned with 0
Jul  8 09:36:41 udevd[663]: worker [5551] unexpectedly returned with 0
Jul  8 09:36:41 kernel: [  156.832086] <6>udevd[663]: segfault at 4 ip 00959fbc sp bfbe7b78 error 6 in udevd[94f000+1c000]

12 years agolibudev: enumerate - sort with qsort()
Kay Sievers [Wed, 8 Jul 2009 00:04:49 +0000 (02:04 +0200)]
libudev: enumerate - sort with qsort()

On machines with many thousands of devices:
  $ time find /sys -name uevent | wc -l

  real      0m33.171s
  user      0m3.329s
  sys       0m29.719s

the current udevtrigger spends minutes sorting the device list:
  $ time /sbin/udevadm trigger --dry-run

  real      4m56.739s
  user      4m45.743s
  sys       0m7.862s

with qsort() it looks better:
  $ time udev/udevadm trigger --dry-run

  real      0m6.495s
  user      0m0.473s
  sys       0m5.923s

12 years agolibudev: device - free values before updating them
Kay Sievers [Wed, 8 Jul 2009 00:02:40 +0000 (02:02 +0200)]
libudev: device - free values before updating them

12 years version bump
Kay Sievers [Wed, 8 Jul 2009 00:02:02 +0000 (02:02 +0200)] version bump

12 years agouse MIN() MAX() from param.h
Kay Sievers [Tue, 7 Jul 2009 17:30:19 +0000 (19:30 +0200)]
use MIN() MAX() from param.h

12 years agoREADME: add CONFIG_BLK_DEV_BSG
Kay Sievers [Tue, 7 Jul 2009 07:48:23 +0000 (09:48 +0200)]

12 years agoscsi_id: correct error handling in prepend_vendor_model
Ian Campbell [Mon, 6 Jul 2009 13:54:52 +0000 (14:54 +0100)]
scsi_id: correct error handling in prepend_vendor_model

The callers of prepend_vendor_model both expect < 0 to be returned on
error and the index to be returned otherwise. However
prepend_vendor_model actually returns 1 on error. Fix this by correctly
returning -1.

Older kernels (before e5b3cd42: "SCSI: sanitize INQUIRY strings")
truncated the model field in sysfs (or propagated bad results from the
target) to less than the expected/required 16 characters which meant
that the SCSI id was mangled into:
        # /sbin/scsi_id -g -s /block/sdg
when it should have been:
        # /sbin/scsi_id -g -s /block/sdg
        SIET     VIRTUAL-DISK   146cee20

Notice how the serial number has been pasted over the vendor+model at
index 1 instead of being added at the end.

In the former case:
        # cat /sys/devices/platform/host5/session1/target5:0:0/5:0:0:1/model | od -t c -t x1
        0000000   V   I   R   T   U   A   L   -   D   I   S   K  \n
                56 49 52 54 55 41 4c 2d 44 49 53 4b 0a
But it should have been:
        # cat /sys/devices/platform/host5/session1/target5:0:0/5:0:0:1/model | od -t c -t x1
        0000000   V   I   R   T   U   A   L   -   D   I   S   K
                56 49 52 54 55 41 4c 2d 44 49 53 4b 20 20 20 20
        0000020  \n

Signed-off-by: Ian Campbell <>
12 years agorelease 144
Kay Sievers [Sun, 5 Jul 2009 22:52:56 +0000 (00:52 +0200)]
release 144

12 years agobuild: add *exec* to the internal rootlibdir name
Kay Sievers [Sun, 5 Jul 2009 21:56:24 +0000 (23:56 +0200)]
build: add *exec* to the internal rootlibdir name

Automake like variables with *exec* in the directory names,
to decide, that it isn't *data*:

12 years agoupdate NEWS
Kay Sievers [Sat, 4 Jul 2009 14:59:35 +0000 (16:59 +0200)]
update NEWS

12 years agolibudev: monitor - handle kernel supplied DEVNAME properly
Kay Sievers [Fri, 3 Jul 2009 16:40:38 +0000 (18:40 +0200)]
libudev: monitor - handle kernel supplied DEVNAME properly

12 years agobuild: do not delete .la files
Kay Sievers [Fri, 3 Jul 2009 16:38:45 +0000 (18:38 +0200)]
build: do not delete .la files

12 years agoextras/udev-acl: add smartcard readers
Martin Pitt [Fri, 3 Jul 2009 12:57:43 +0000 (14:57 +0200)]
extras/udev-acl: add smartcard readers

gnupg, psc-lite, etc. will ship udev rules for supported smartcard readers
soon. Add the ACL management bit.

Thanks to Michael Bienia!

13 years agoudevadm: test - print list of properties
Kay Sievers [Thu, 2 Jul 2009 20:56:19 +0000 (22:56 +0200)]
udevadm: test - print list of properties

13 years agoMerge branch 'firmware' of git://
Kay Sievers [Thu, 2 Jul 2009 10:13:36 +0000 (12:13 +0200)]
Merge branch 'firmware' of git://git./linux/kernel/git/jcm/udev-jcm

13 years agofirmware: search for third party or sysadmin supplied firmware updates
Jon Masters [Thu, 2 Jul 2009 00:21:03 +0000 (20:21 -0400)]
firmware: search for third party or sysadmin supplied firmware updates

We currently search /lib/firmware and /lib/firmware/`uname -r` for firmware
files for device drivers loaded by the currently running kernel. These are
often packaged by distributions as a subpackage of the kernel or as a
separate package containing firmware. But these files cannot easily be
updated by third parties or sysadmins independently of that package.

This patch causes udev to also look for firmware files in an "updates"
directory, which is almost identical in purpose to the module-init-tools
"updates" directories insomuch as local changes can go in here and will
take preference over firmware supplied by any distribution.

13 years agorules: set group ownership of new firewire driver device files
Stefan Richter [Wed, 1 Jul 2009 19:41:43 +0000 (21:41 +0200)]
rules: set group ownership of new firewire driver device files

The newer firewire-core driver exposes per-device character device files,
called /dev/fw[0-9]*, in contrast to the older raw1394, video1394, dv1394
drivers which created one global file or per-controller files.

This allows to set ownership, permissions, or/ and access control lists
for each device file based on device type markers obtained from sysfs.
The "units" attribute which is used for this purpose has become available
in Linux 2.6.31(-rc1) by commit 0210b66dd88a2a1e451901b00378a2068b6ccb35.

The added rules match identifiers of

  - IIDC devices:
    industrial cameras and some webcams,

  - AV/C devices:
    camcorders, set-top boxes, TV sets, audio devices, and similar

Signed-off-by: Stefan Richter <>
13 years agoman: udev - add private properties like ENV{.FOO}="bar"
Kay Sievers [Sun, 28 Jun 2009 15:10:04 +0000 (17:10 +0200)]
man: udev - add private properties like ENV{.FOO}="bar"

13 years agoupdate NEWS
Kay Sievers [Sun, 28 Jun 2009 15:09:57 +0000 (17:09 +0200)]
update NEWS

13 years agoscsi_id: delete no longer needed config file
Kay Sievers [Sun, 28 Jun 2009 01:01:31 +0000 (03:01 +0200)]
scsi_id: delete no longer needed config file

13 years agorules: make ata_id properties the default for all ATA block devices
Kay Sievers [Sun, 28 Jun 2009 00:59:48 +0000 (02:59 +0200)]
rules: make ata_id properties the default for all ATA block devices

We need to call ata_id as the default for libata sd* devices. We
want ID_BUS=ata, and the ATA device proeprties, and be independent
of the SCSI emulation with the truncated values. The links
in /dev/disk/by-id/{ata-*,scsi-*} are still the same.

13 years agoata_id: sync ID_SERIAL(_SHORT) with other *_id tools
Kay Sievers [Sun, 28 Jun 2009 00:58:12 +0000 (02:58 +0200)]
ata_id: sync ID_SERIAL(_SHORT) with other *_id tools

ID_SERIAL is the full serial number used for the links, ID_SERIAL_SHORT
is the device serial number.

13 years agoscsi_id: --reformat_serial - use udev_util_replace_whitespace()
Kay Sievers [Sun, 28 Jun 2009 00:27:23 +0000 (02:27 +0200)]
scsi_id: --reformat_serial - use udev_util_replace_whitespace()

13 years agodo not exports properties starting with a '.'
Kay Sievers [Sun, 28 Jun 2009 00:21:58 +0000 (02:21 +0200)]
do not exports properties starting with a '.'

Private variables can be marked like:

13 years agohid2hci: narrow matches to real HCI devices
Martin Pitt [Fri, 26 Jun 2009 06:17:23 +0000 (08:17 +0200)]
hid2hci: narrow matches to real HCI devices

The previous rules just checked bInterfaceProtocol but not the actual device
and interface class. This caused the hci rules to be applied for Dell USB hubs
and attached input devices like keyboards and mouses as well, breaking them

Tighten the match to also check device and interface class/subclass.

13 years version bump
Kay Sievers [Mon, 22 Jun 2009 14:34:55 +0000 (16:34 +0200)] version bump

13 years agoudev/.gitignore: add udev.pc
Kay Sievers [Sun, 21 Jun 2009 22:33:09 +0000 (00:33 +0200)]
udev/.gitignore: add udev.pc

13 years agolibudev: queue - add comments for queue format
Kay Sievers [Sun, 21 Jun 2009 22:32:26 +0000 (00:32 +0200)]
libudev: queue - add comments for queue format

13 years agoTODO: add cleanup of ATA_COMPAT
Kay Sievers [Sun, 21 Jun 2009 22:31:46 +0000 (00:31 +0200)]
TODO: add cleanup of ATA_COMPAT

13 years version bump
Kay Sievers [Sun, 21 Jun 2009 22:30:42 +0000 (00:30 +0200)] version bump

13 years add AM_SILENT_RULES
Kay Sievers [Sat, 20 Jun 2009 01:57:26 +0000 (03:57 +0200)] add AM_SILENT_RULES