chiark / gitweb /
15 years ago[PATCH] include missing header to udevtest.c [Sun, 12 Dec 2004 01:32:25 +0000 (02:32 +0100)]
[PATCH] include missing header to udevtest.c

15 years ago[PATCH] add -V option to udev to print the version number [Sun, 12 Dec 2004 01:27:50 +0000 (02:27 +0100)]
[PATCH] add -V option to udev to print the version number

On Sat, 2004-12-11 at 18:44 +0200, Martin Schlemmer [c] wrote:
> Any suggestions to determining the version of the installed udev?
> This is now during startup, to see if we can make use of using
> udevsend as hotplug agent.  If the system was up, udevinfo could
> be used, but that is in /usr/bin that might be on a seperate /usr.
> I know we might move udevinfo to /bin, but that might be an issue
> for some, and adding a -V switch to /sbin/udev might be a better
> choice.

15 years ago[PATCH] prevent udev node creatinon for "class" registration [Sun, 12 Dec 2004 01:03:14 +0000 (02:03 +0100)]
[PATCH] prevent udev node creatinon for "class" registration

I've found a /dev/video4linux node and just realized, that libsysfs
searches all subdirs for an attribute name.
So it found /class/video4linux/video0/dev for the videodev class
creation event /class/video4linux and created a node.

Just ignore the SUBSYSTEM="class" events now.

15 years ago[PATCH] udevd: serialization of the event sequence of a chain of devices [Sat, 11 Dec 2004 20:43:08 +0000 (21:43 +0100)]
[PATCH] udevd: serialization of the event sequence of a chain of devices

Currently udevd delays only events for the same DEVPATH.

Example of an "add" event sequence:

With this change, we make sure, that the udev process handling
/block/sda has finished its work (waited for all attributes,
created the node) before we fork the udev event for /block/sda/sda1.
This way the event for sda1 can be sure, that the node for the
main device is already created (may be useful for disk labels).

It will not affect any parallel device handling, only the sequence
of the devices directory chain is serialized. The 10.000 disks
plugged in will still run as parallel events. :)

The main motivation to do this is the program execution of the
dev.d/ and hotplug.d/ directory. If we don't wait for the parent
event to exit, we can't be sure that the executed scripts are
run in the right order.

On Thu, Dec 09, 2004 at 09:18:28AM +0100, Kay Sievers wrote:
> On Wed, 2004-12-08 at 19:07 -0800, David Brownell wrote:
> > Could that argument apply to the underlying hardware, too?

> We now make sure that the sequence of events for a device
> is serialized for every device chain and the class/block
> devices which have a "device" link to a physical device are
> handled after the physical device is fully populated and
> notified to userspace. It will only work this way on kernels
> later than 2.6.10-rc1 cause it depends on the PHYSDEVPATH
> value in the hotplug environment.

15 years ago[PATCH] add a class/fc_host file to the list of what to wait for [Sat, 11 Dec 2004 05:07:31 +0000 (06:07 +0100)]
[PATCH] add a class/fc_host file to the list of what to wait for

15 years ago[PATCH] udev_volume_id: links sysfs.a instead of all objects [Thu, 9 Dec 2004 05:06:17 +0000 (06:06 +0100)]
[PATCH] udev_volume_id: links sysfs.a instead of all objects

cuts down the size from:
  dynamically linked: 48.030 -> 45.187
  statically linked:  149.177 ->  142.362

15 years ago[PATCH] 048 release 048 [Thu, 9 Dec 2004 03:42:47 +0000 (19:42 -0800)]
[PATCH] 048 release

15 years ago[PATCH] fix udev_volume_id so it will now build properly. [Thu, 9 Dec 2004 03:35:29 +0000 (19:35 -0800)]
[PATCH] fix udev_volume_id so it will now build properly.

15 years ago[PATCH] fix scsi_id build errors due to changes in the main udev makefile [Thu, 9 Dec 2004 03:30:41 +0000 (19:30 -0800)]
[PATCH] fix scsi_id build errors due to changes in the main udev makefile

also fixed up the QUIET issue when it was disabled for the extras programs

15 years ago[PATCH] 047 release 047 [Thu, 9 Dec 2004 02:29:04 +0000 (18:29 -0800)]
[PATCH] 047 release

15 years ago[PATCH] add more debian rules files. [Thu, 9 Dec 2004 01:09:40 +0000 (17:09 -0800)]
[PATCH] add more debian rules files.

15 years ago[PATCH] move distro specific config files into their own directories [Thu, 9 Dec 2004 01:08:20 +0000 (17:08 -0800)]
[PATCH] move distro specific config files into their own directories

15 years agomore merge fixups
Greg KH [Wed, 27 Apr 2005 06:10:15 +0000 (23:10 -0700)]
more merge fixups

15 years ago[PATCH] update debian rules files. [Thu, 9 Dec 2004 01:05:20 +0000 (17:05 -0800)]
[PATCH] update debian rules files.

15 years ago[PATCH] added asterix rules to the gentoo file. [Thu, 9 Dec 2004 00:55:27 +0000 (16:55 -0800)]
[PATCH] added asterix rules to the gentoo file.

15 years ago[PATCH] fix gentoo scsi cdrom rule [Sat, 4 Dec 2004 02:50:05 +0000 (18:50 -0800)]
[PATCH] fix gentoo scsi cdrom rule

15 years ago[PATCH] Fix the multithreaded build again... [Fri, 3 Dec 2004 09:04:53 +0000 (01:04 -0800)]
[PATCH] Fix the multithreaded build again...

15 years ago[PATCH] use udevstart for udev.init.* files. [Sun, 5 Dec 2004 16:31:37 +0000 (08:31 -0800)]
[PATCH] use udevstart for udev.init.* files.

Thanks to Kay for pointing it out.

15 years ago[PATCH] delete a bunch of files no longer needed. [Sun, 5 Dec 2004 16:31:04 +0000 (08:31 -0800)]
[PATCH] delete a bunch of files no longer needed.

Thanks to Kay for pointing it out.

15 years ago[PATCH] Various typos and other litte errors in [Tue, 7 Dec 2004 22:01:44 +0000 (23:01 +0100)]
[PATCH] Various typos and other litte errors in

Thanks-To: Tobias Klauser <>

While crawling through the udev manpage I noticed some typos and other
grammatical errors. English is not my native language, so I'm not sure
if I fixed everything right. I would be glad if any English-speaking
person could check this patch before applying.

15 years ago[PATCH] update Fedora dev.d/ example and remove unused conf.d/ directory [Tue, 7 Dec 2004 21:55:08 +0000 (22:55 +0100)]
[PATCH] update Fedora dev.d/ example and remove unused conf.d/ directory

15 years ago[PATCH] don't install distribution specific init script on "make install" [Tue, 7 Dec 2004 21:45:36 +0000 (22:45 +0100)]
[PATCH] don't install distribution specific init script on "make install"

Leave that part up to the one who itegrates udev into a system as
there is much more to care of than to call a init stript for a
dynamic dev.

15 years ago[PATCH] restore OWNER/GROUP assignment in rule coming from RESULT [Tue, 7 Dec 2004 21:38:40 +0000 (22:38 +0100)]
[PATCH] restore OWNER/GROUP assignment in rule coming from RESULT

15 years ago[PATCH] make gcov compile scripts working with recent gcc [Sun, 5 Dec 2004 15:31:51 +0000 (16:31 +0100)]
[PATCH] make gcov compile scripts working with recent gcc

15 years ago[PATCH] fix udev-test/ to work with again [Sun, 5 Dec 2004 14:07:39 +0000 (15:07 +0100)]
[PATCH] fix udev-test/ to work with again

15 years ago[PATCH] add net/atml and class/ppdev to the wait_for_sysfs exception list [Sun, 5 Dec 2004 14:03:43 +0000 (15:03 +0100)]
[PATCH] add net/atml and class/ppdev to the wait_for_sysfs exception list

15 years ago[PATCH] add net/nlv* devices to the exception list [Thu, 2 Dec 2004 18:30:23 +0000 (19:30 +0100)]
[PATCH] add net/nlv* devices to the exception list

15 years ago[PATCH] add "pcmcia" and "fc_transport" to the wait_for_sysfs lists [Tue, 30 Nov 2004 10:36:36 +0000 (11:36 +0100)]
[PATCH] add "pcmcia" and "fc_transport" to the wait_for_sysfs lists

15 years ago[PATCH] remove unused timestamp field [Mon, 29 Nov 2004 12:51:20 +0000 (13:51 +0100)]
[PATCH] remove unused timestamp field

15 years ago[PATCH] simplify permission handling [Mon, 29 Nov 2004 12:44:01 +0000 (13:44 +0100)]
[PATCH] simplify permission handling

Initialize the defaults in udev_config.c instead of namedev.c.
Replace macro by expanded code. Switch to mode_t instead
of string value. Add and clarify some comments.

15 years ago[PATCH] handle /etc/hotplug.d/ only if the event comes from udevd [Mon, 29 Nov 2004 03:47:59 +0000 (04:47 +0100)]
[PATCH] handle /etc/hotplug.d/ only if the event comes from udevd

15 years ago[PATCH] trivial cleanups and change some comments [Mon, 29 Nov 2004 01:15:55 +0000 (02:15 +0100)]
[PATCH] trivial cleanups and change some comments

15 years ago[PATCH] remove unused variables [Sun, 28 Nov 2004 13:01:43 +0000 (14:01 +0100)]
[PATCH] remove unused variables

Remove udev.bus, cause it's currently unused and newer kernels will pass
it in the hotplug environment as PHYSDEVBUS.

Remove udev.action, cause it's unused.

Rename udev_set_values() to udev_init_device().

15 years ago[PATCH] udevsend/udevd handle events without a subsystem [Sun, 28 Nov 2004 12:56:22 +0000 (13:56 +0100)]
[PATCH] udevsend/udevd handle events without a subsystem

Accept event without a subsystem and pass it through udevd.
Pass empty environment while starting udevd.

15 years ago[PATCH] use blacklist on device "remove" and remove dev.d/ call code duplication [Sun, 28 Nov 2004 12:47:36 +0000 (13:47 +0100)]
[PATCH] use blacklist on device "remove" and remove dev.d/ call code duplication

15 years ago[PATCH] update the man pages and correct Usage: hints [Sun, 28 Nov 2004 12:41:15 +0000 (13:41 +0100)]
[PATCH] update the man pages and correct Usage: hints

Add UDEV_LOG to the man udev man page. Remove mention of specific
variables from the udevd/udevsend man page as we changed to pass
the whole environment.

Correct printed Usage: of udevtest and udevinfo.

Init the config in udevtest earlier to accept input with and without
the sysfs mount point.

15 years ago[PATCH] don't call the hotplug scripts with a test run [Thu, 25 Nov 2004 09:15:32 +0000 (10:15 +0100)]
[PATCH] don't call the hotplug scripts with a test run

15 years agomore fixups
Greg KH [Wed, 27 Apr 2005 05:54:32 +0000 (22:54 -0700)]
more fixups

15 years agofixups to get back to proper patch order
Greg KH [Wed, 27 Apr 2005 05:52:57 +0000 (22:52 -0700)]
fixups to get back to proper patch order

Damm, it's hard to merge a multi-line tree into one flat line at times...

15 years ago[PATCH] don't call dev.d/ scripts twice, if directory = subsystem [Thu, 25 Nov 2004 08:55:54 +0000 (09:55 +0100)]
[PATCH] don't call dev.d/ scripts twice, if directory = subsystem

The /etc/dev.d/input/ was called twice for /dev/input/mouse.
Skip the execution if we get a directory named after the subsystem.
Move UDEV_NO_DEVD where it belongs.

15 years ago[PATCH] remove archive file if we changed something [Thu, 25 Nov 2004 08:52:19 +0000 (09:52 +0100)]
[PATCH] remove archive file if we changed something

15 years ago[PATCH] link archive insted of objects [Thu, 25 Nov 2004 07:38:06 +0000 (08:38 +0100)]
[PATCH] link archive insted of objects

This cuts the size of the binaries, as only objects that
are actually used are linked into the binary:

 45592 -> 43608 udev
  4380 ->  4380 udevsend
 10380 ->  4652 udevd
 34732 -> 33100 udevinfo
 45432 -> 37208 udevtest

15 years ago[PATCH] rename udev_lib to udev_utils and dev_d to udev_multiplex [Thu, 25 Nov 2004 01:44:38 +0000 (02:44 +0100)]
[PATCH] rename udev_lib to udev_utils and dev_d to udev_multiplex

15 years ago[PATCH] Allow GROUP to have modifiers in it [Fri, 3 Dec 2004 06:23:21 +0000 (22:23 -0800)]
[PATCH] Allow GROUP to have modifiers in it

15 years ago[PATCH] comment out ability to run with valgrind [Fri, 3 Dec 2004 06:16:54 +0000 (22:16 -0800)]
[PATCH] comment out ability to run with valgrind

15 years ago[PATCH] fix spurious valgrind warning in udev. [Fri, 3 Dec 2004 06:14:57 +0000 (22:14 -0800)]
[PATCH] fix spurious valgrind warning in udev.

15 years ago[PATCH] DEVNAME on device removal [Tue, 23 Nov 2004 15:49:55 +0000 (07:49 -0800)]
[PATCH] DEVNAME on device removal

  I just noticed that the DEVNAME enviroment variable isn't being set anymore
  in udev 0.046 on device removal, while it was being set in 0.042. We're using
  the property tto do umount -l <devices> when a block device is removed. Afaik
  there is no other way to associate a device with it's DEVNAME on removal ?

  Also are there cases where doing umount -l on the removed devices is wrong?
  I guess the device is gone, so there is no sense in keeping it mounted (it's
  not like the filesystem is gonna come back in a sane state again)..

  Attached (trivial) patch brings back the DEVNAME variable on device removal.

15 years ago[PATCH] fix udevinfo '-q path' option as it was not working [Fri, 3 Dec 2004 05:45:26 +0000 (21:45 -0800)]
[PATCH] fix udevinfo '-q path' option as it was not working

15 years ago[PATCH] handle whole hotplug event with udevd/udev [Tue, 23 Nov 2004 07:01:57 +0000 (08:01 +0100)]
[PATCH] handle whole hotplug event with udevd/udev

If /proc/sys/kernel/hotplug points to /sbin/udevsend we handle the whole
hotplug event with multiplexing /etc/hotplug.d/.

15 years ago[PATCH] integrate wait_for_sysfs in udev [Tue, 23 Nov 2004 06:34:56 +0000 (07:34 +0100)]
[PATCH] integrate wait_for_sysfs in udev

Move the wait_for_sysfs logic into the udev binary. udev is called for
every hotplug event. It also waits for /devices events.

15 years ago[PATCH] remove my test code [Tue, 23 Nov 2004 03:22:16 +0000 (04:22 +0100)]
[PATCH] remove my test code

15 years ago[PATCH] make the searched multiplex directories conditionally [Tue, 23 Nov 2004 05:22:26 +0000 (06:22 +0100)]
[PATCH] make the searched multiplex directories conditionally

If we don't have "name" or a "subsystem" don't look for it.

15 years ago[PATCH] add MANAGED_EVENT to the forked udev environment [Tue, 23 Nov 2004 05:14:21 +0000 (06:14 +0100)]
[PATCH] add MANAGED_EVENT to the forked udev environment

This will prevent a loop, if udev sends events back into the

15 years ago[PATCH] export DEVNAME on remove event [Tue, 23 Nov 2004 04:05:37 +0000 (05:05 +0100)]
[PATCH] export DEVNAME on remove event

Fix from Sjoerd Simons <>, it got lost
on the recent reorganization of the udev processing stages.

15 years ago[PATCH] export udev_log flag to the environment [Tue, 23 Nov 2004 03:47:06 +0000 (04:47 +0100)]
[PATCH] export udev_log flag to the environment

Forked scripts and callouts may want to follow udev's configured behavior
and log only if udev is logging itself.

15 years ago[PATCH] add support for /devices-devices without any file to wait for [Tue, 23 Nov 2004 02:55:24 +0000 (03:55 +0100)]
[PATCH] add support for /devices-devices without any file to wait for

15 years ago[PATCH] fix parallel build error [Tue, 23 Nov 2004 03:23:00 +0000 (19:23 -0800)]
[PATCH] fix parallel build error

15 years ago[PATCH] Patch from Alex Riesen <> [Tue, 23 Nov 2004 02:43:37 +0000 (03:43 +0100)]
[PATCH] Patch from Alex Riesen <>

> I just put const's at some places. It cut down data segments, but
> increased code size.
> Overall still smaller:
> -rwxr-xr-x  1 user users 50420 Nov 19 10:53 ../udev-046/udev
> -rwxr-xr-x  1 user users 49556 Nov 19 10:53 udev
>    text    data     bss     dec     hex filename
>   47245     968   22480   70693   11425 ../udev-046/udev
>   48089     104   22064   70257   11271 udev
> Also, the instance of utsname in udev_lib.c is used only once.

15 years ago[PATCH] add a bunch of busses to the list of what to wait for [Tue, 23 Nov 2004 02:36:32 +0000 (03:36 +0100)]
[PATCH] add a bunch of busses to the list of what to wait for

15 years ago[PATCH] close connection to syslog in forked udevd child [Tue, 23 Nov 2004 02:30:13 +0000 (03:30 +0100)]
[PATCH] close connection to syslog in forked udevd child

15 years ago[PATCH] udevd exit path cleanup [Tue, 23 Nov 2004 02:28:41 +0000 (03:28 +0100)]
[PATCH] udevd exit path cleanup

15 years ago[PATCH] fix network device naming bug [Sat, 20 Nov 2004 06:42:19 +0000 (22:42 -0800)]
[PATCH] fix network device naming bug

15 years ago[PATCH] 046 release 046 [Fri, 19 Nov 2004 03:39:05 +0000 (19:39 -0800)]
[PATCH] 046 release

15 years ago[PATCH] Don't try to print major/minor for devices without a dev file. [Fri, 19 Nov 2004 03:03:52 +0000 (04:03 +0100)]
[PATCH] Don't try to print major/minor for devices without a dev file.

15 years ago[PATCH] remove get_device_type and merge that into udev_set_values() [Fri, 19 Nov 2004 02:50:05 +0000 (03:50 +0100)]
[PATCH] remove get_device_type and merge that into udev_set_values()

15 years ago[PATCH] prevent udevd crash if DEVPATH is not set [Fri, 19 Nov 2004 02:49:13 +0000 (03:49 +0100)]
[PATCH] prevent udevd crash if DEVPATH is not set

Just move the event straight to the exec list and don't try
to compare a NULL pointer.

15 years ago[PATCH] add ippp and bcrypt to the exception lists of wait_for_sysfs [Fri, 19 Nov 2004 02:43:43 +0000 (03:43 +0100)]
[PATCH] add ippp and bcrypt to the exception lists of wait_for_sysfs

15 years ago[PATCH] let klibc add the trailing newline to syslog conditionally [Fri, 19 Nov 2004 02:41:38 +0000 (03:41 +0100)]
[PATCH] let klibc add the trailing newline to syslog conditionally

We get an empty line after every debug print from libsysfs,
so don't add a second if we already have one.

15 years ago[PATCH] change key names in udevinfo sysfs walk to match the kernel [Sat, 13 Nov 2004 04:33:25 +0000 (05:33 +0100)]
[PATCH] change key names in udevinfo sysfs walk to match the kernel

15 years ago[PATCH] disable logging for udevstart [Sat, 13 Nov 2004 13:51:10 +0000 (14:51 +0100)]
[PATCH] disable logging for udevstart

The slow logging facilites on some systems are a reason for
the reported slowness of udevstart. On one of my boxes udevstart
is down from 9 second to 0.3 seconds.

15 years ago[PATCH] add NAME{ignore_remove} attribute [Sat, 13 Nov 2004 13:43:24 +0000 (14:43 +0100)]
[PATCH] add NAME{ignore_remove} attribute

Some broken ide drivers are generating high event traffic, with
add/remove events. With this attribute, it can be specified,
that the node is always available. It may be used in conjunction
with the new DRIVER= match to catch specific kernel device drivers.

15 years ago[PATCH] remove historical SYSFS_attr="value" format [Sat, 13 Nov 2004 11:50:44 +0000 (12:50 +0100)]
[PATCH] remove historical SYSFS_attr="value" format

This is long ago replaced by the unified key attribute
format SYSFS{attr}="value".

15 years ago[PATCH] don't wait for sysfs if the kernel(2.6.10-rc2) tells us what not to expect [Sat, 13 Nov 2004 11:36:47 +0000 (12:36 +0100)]
[PATCH] don't wait for sysfs if the kernel(2.6.10-rc2) tells us what not to expect

Newer kernels will tell us if we should wait for the "device" or
the "bus" link. So don't waste the time with our lists or wait
for the timeout.

15 years ago[PATCH] support DRIVER as a rule key [Sat, 13 Nov 2004 04:21:12 +0000 (05:21 +0100)]
[PATCH] support DRIVER as a rule key

Match with a rule against a device with a specific kernel driver.

15 years ago[PATCH] support SUBSYSTEM as a rule key [Fri, 12 Nov 2004 05:52:55 +0000 (06:52 +0100)]
[PATCH] support SUBSYSTEM as a rule key

This should make it easier to catch e.g all block or net devices with
a single rule.

15 years ago[PATCH] rename udevdb* to udev_db* [Fri, 12 Nov 2004 05:32:19 +0000 (06:32 +0100)]
[PATCH] rename udevdb* to udev_db*

15 years ago[PATCH] Make dev.d/ handling a separate processing stage [Fri, 12 Nov 2004 05:21:16 +0000 (06:21 +0100)]
[PATCH] Make dev.d/ handling a separate processing stage

Move the logic when and how to call the dev.d/ scripts into the
main processing path.

15 years ago[PATCH] make the udev object available to more processing stages [Fri, 12 Nov 2004 05:20:22 +0000 (06:20 +0100)]
[PATCH] make the udev object available to more processing stages

Remove the overwriting of main_argv[] hack and use the values
from the udev object.

Pass the udev object to call_foreach_file().

In the udevstart case, export SUBSYSTEM and UDEVSTART to the

15 years ago[PATCH] remove udev_lib dependency from udevsend, which makes it smaller [Fri, 12 Nov 2004 05:18:28 +0000 (06:18 +0100)]
[PATCH] remove udev_lib dependency from udevsend, which makes it smaller

15 years ago[PATCH] add ACTION to udev object to expose it to the whole process [Fri, 12 Nov 2004 05:17:35 +0000 (06:17 +0100)]
[PATCH] add ACTION to udev object to expose it to the whole process

15 years ago[PATCH] make udevinfo's -r option also workimg for symlink queries [Wed, 10 Nov 2004 09:19:39 +0000 (10:19 +0100)]
[PATCH] make udevinfo's -r option also workimg for symlink queries

[kay@pim udev.kay]$ ./udevinfo -r -q symlink -p /class/video4linux/video0
/dev/camera0 /dev/kamera0 /dev/videocam0 /dev/webcam0

[kay@pim udev.kay]$ ./udevinfo -q symlink -p /class/video4linux/video0
camera0 kamera0 videocam0 webcam0

15 years ago[PATCH] let udev act as udevstart if argv[1] == "udevstart" [Wed, 10 Nov 2004 08:44:58 +0000 (09:44 +0100)]
[PATCH] let udev act as udevstart if argv[1] == "udevstart"

"udev udevstart" will run udev as udevstart. This makes it easier
to run a test in the source tree without the need to create a

15 years ago[PATCH] improve udevinfo sysfs info walk [Wed, 10 Nov 2004 01:15:27 +0000 (02:15 +0100)]
[PATCH] improve udevinfo sysfs info walk

15 years ago[PATCH] add sysfs info walk to udevinfo [Tue, 9 Nov 2004 11:11:46 +0000 (12:11 +0100)]
[PATCH] add sysfs info walk to udevinfo

The option "-s" will get information about the major/minor,
the physical device, the bus value and the driver from sysfs for
all class and block devices:

kay@pim udev.kay]$ ./udevinfo -s

DEVPATH        '/sys/block/sda'
SUBSYSTEM      'block'
NAME           'sda'
MAJORMINOR     '8:0'
PHYSDEVPATH    '/sys/devices/pci0000:00/0000:00:1d.7/usb1/1-3/1-3:1.0/host2/target2:0:0/2:0:0:0'
DRIVER         'sd'

DEVPATH        '/sys/class/input/mice'
SUBSYSTEM      'input'
NAME           'mice'
MAJORMINOR     '13:63'

DEVPATH        '/sys/class/input/mouse0'
SUBSYSTEM      'input'
NAME           'mouse0'
MAJORMINOR     '13:32'


15 years ago[PATCH] pass the whole event environment to udevd [Sat, 6 Nov 2004 13:30:15 +0000 (14:30 +0100)]
[PATCH] pass the whole event environment to udevd

Make _all_ hotplug variables available to the forked udev,
the udev callouts and the udev dev.d/ scripts. We put the
whole environment into a buffer and send it over the udevd
socket. udevd recreates *envp[] and passes it to the exec().

15 years ago[PATCH] make spotless for releases. [Sat, 13 Nov 2004 07:43:34 +0000 (23:43 -0800)]
[PATCH] make spotless for releases.

15 years ago[PATCH] replace tdb database by simple lockless file database [Sat, 6 Nov 2004 13:28:01 +0000 (14:28 +0100)]
[PATCH] replace tdb database by simple lockless file database

This makes the udev operation completely lockless by storing a
file for every node in /dev/.udevdb/* This solved the problem
with deadlocking concurrent udev processes waiting for each other
to release the file lock under heavy load.

15 years ago[PATCH] 045 release 045 [Thu, 18 Nov 2004 03:28:03 +0000 (19:28 -0800)]
[PATCH] 045 release

15 years ago[PATCH] Some updates for Gentoo's udev rules [Thu, 18 Nov 2004 03:10:07 +0000 (19:10 -0800)]
[PATCH] Some updates for Gentoo's udev rules


Here is a few updates for the udev.rules.gentoo from udev package.
I will summarise it briefly:

1)  The last change you did to legacy tty's is wrong.

I say this because:
1a)  The original devfs rules had both master and slave in /dev/pty:

nosferatu linux # grep devfs_name drivers/char/pty.c
        pty_driver->devfs_name = "pty/m";
        pty_slave_driver->devfs_name = "pty/s";
nosferatu linux #

1b)  If you refer to ' Tty problems?', you will see that
the /dev/tty/ directory our rules create, replaces this symlink:

nosferatu portage # ls -l /dev/tty
crw-rw-rw-  1 root tty 5, 0 Nov 14 17:06 /dev/tty
nosferatu portage #

which is used to determine the controlling tty.

2)  Somebody added the /dev/cpu/microcode rule, but it was not run
as there was an older rule before that placing it in /dev/misc (which
is wrong).  Just remove the first broken rule

3)  Some form/tab cleanups.  Reorder rules alphabetically according to
device class to make searching/editing easier.

15 years ago[PATCH] 044 release 044 [Thu, 11 Nov 2004 03:38:07 +0000 (19:38 -0800)]
[PATCH] 044 release

15 years ago[PATCH] 043 release 043 [Fri, 5 Nov 2004 14:52:53 +0000 (06:52 -0800)]
[PATCH] 043 release

15 years ago[PATCH] reduce syslog noise of udevsend if multiple instances try to start udevd [Thu, 11 Nov 2004 03:05:20 +0000 (19:05 -0800)]
[PATCH] reduce syslog noise of udevsend if multiple instances try to start udevd

15 years ago[PATCH] add i2c-dev to the list of devices without a bus [Thu, 11 Nov 2004 03:04:38 +0000 (19:04 -0800)]
[PATCH] add i2c-dev to the list of devices without a bus

15 years ago[PATCH] add support to gentoo rules file. [Thu, 11 Nov 2004 02:46:01 +0000 (18:46 -0800)]
[PATCH] add support to gentoo rules file.

15 years ago[PATCH] fix gentoo legacy tty rule [Thu, 11 Nov 2004 02:13:11 +0000 (18:13 -0800)]
[PATCH] fix gentoo legacy tty rule

I thought I had fixed this a long time ago...

15 years ago[PATCH] remove 'sudo' usage from the Makefile [Thu, 11 Nov 2004 02:11:40 +0000 (18:11 -0800)]
[PATCH] remove 'sudo' usage from the Makefile

should fix a lot of gentoo bug reports

15 years ago[PATCH] make test for root permissions before running [Thu, 11 Nov 2004 02:11:00 +0000 (18:11 -0800)]
[PATCH] make test for root permissions before running

Should prevent false error reports from happening.

15 years ago[PATCH] Shut up wait_for_sysfs class/net failure messages, as it's not possible [Fri, 5 Nov 2004 12:17:30 +0000 (13:17 +0100)]
[PATCH] Shut up wait_for_sysfs class/net failure messages, as it's not possible

15 years ago[PATCH] wait() for specific pid to return from fork(). [Fri, 5 Nov 2004 12:16:56 +0000 (13:16 +0100)]
[PATCH] wait() for specific pid to return from fork().

15 years ago[PATCH] Don't use any syslog() in signal handler, cause it may deadlock. [Fri, 5 Nov 2004 12:16:32 +0000 (13:16 +0100)]
[PATCH] Don't use any syslog() in signal handler, cause it may deadlock.