1
0
mirror of https://github.com/systemd/systemd.git synced 2024-10-28 11:55:44 +03:00
The systemd System and Service Manager
Go to file
Kay Sievers 69348b66ff remove special TIMEOUT handling from incoming queue
Moving events directly to the exec queue instead of the reordering
incoming queue, leaves holes in the sequence, that lead to timeouts for
all other events. Remove that part of the special handling.

(With netlink, events can't get out-of-order and the maximum timeout is 5
seconds and should not cause any trouble with the 10 sec timout for the
firmware class anyway. Events with timeouts are still prioritized for
execution, but don't bypass the incoming queue anymore.)

Many thanks to:
  Uberto Barbini <uberto@ubiland.net>

for his endless debugging and sending all the traces, that showed this
failure with his DVB device:
  UEVENT[1124474094] add@/module/stv0299
  UEVENT[1124474094] add@/module/ves1x93
  UEVENT[1124474094] add@/module/ttpci_eeprom
  UEVENT[1124474094] add@/module/saa7146
  UEVENT[1124474094] add@/module/video_buf
  UEVENT[1124474094] add@/module/saa7146_vv
  UEVENT[1124474094] add@/module/dvb_core
  UEVENT[1124474094] add@/module/dvb_ttpci
  UEVENT[1124474094] add@/bus/pci/drivers/dvb
  UEVENT[1124474094] add@/class/firmware/0000:00:14.0
  UDEV  [1124474094] add@/module/dvb_core
  UDEV  [1124474094] add@/module/saa7146_vv
  UDEV  [1124474094] add@/module/dvb_ttpci
  UDEV  [1124474094] add@/module/ves1x93
  UDEV  [1124474094] add@/module/ttpci_eeprom
  UDEV  [1124474094] add@/module/saa7146
  UDEV  [1124474094] add@/module/stv0299
  UDEV  [1124474094] add@/module/video_buf
  UDEV  [1124474094] add@/bus/pci/drivers/dvb
  UEVENT[1124474094] remove@/class/firmware/0000:00:14.0  <- event with TIMEOUT will leave a hole in the incoming
  UDEV  [1124474094] add@/class/firmware/0000:00:14.0        sequence, which will cause a wait for the alarm()
  UEVENT[1124474094] add@/class/i2c-adapter/i2c-1            that flushes the queue
  UEVENT[1124474094] add@/class/i2c-dev/i2c-1
  UDEV  [1124474094] remove@/class/firmware/0000:00:14.0  <- event also has TIMEOUT and is executed immediately
  UEVENT[1124474095] add@/class/dvb/dvb0.demux0
  UEVENT[1124474095] add@/class/dvb/dvb0.dvr0
  UEVENT[1124474095] add@/class/dvb/dvb0.video0
  UEVENT[1124474095] add@/class/dvb/dvb0.audio0
  UEVENT[1124474095] add@/class/dvb/dvb0.ca0
  UEVENT[1124474095] add@/class/dvb/dvb0.osd0
  UEVENT[1124474095] add@/class/dvb/dvb0.net0
  UEVENT[1124474095] add@/class/video4linux/video1
  UEVENT[1124474095] add@/class/dvb/dvb0.frontend0
  UDEV  [1124474099] add@/class/i2c-adapter/i2c-1         <- all others have 5 seconds delay cause of the missing event
  UDEV  [1124474099] add@/class/dvb/dvb0.ca0                 missing events
  UDEV  [1124474099] add@/class/dvb/dvb0.osd0
  UDEV  [1124474099] add@/class/video4linux/video1
  UDEV  [1124474099] add@/class/dvb/dvb0.frontend0
  UDEV  [1124474099] add@/class/dvb/dvb0.video0
  UDEV  [1124474099] add@/class/dvb/dvb0.audio0
  UDEV  [1124474099] add@/class/i2c-dev/i2c-1
  UDEV  [1124474099] add@/class/dvb/dv

My test program that simulates a similar sequence, runs without any delay
now. (With one of the next versions we will make netlink mandatory, then
we can remove the whole input queue crap with the reordering anyway.)

Signed-off-by: Kay Sievers <kay.sievers@suse.de>
2005-08-29 01:19:02 +02:00
docs remove example rules and put the dev.d stuff into the run_directory folder 2005-08-09 20:11:26 +02:00
etc update SUSE rules 2005-08-26 14:27:13 +02:00
extras vol_id: ID_LABEL_SAFE will no longer contain fancy characters 2005-08-28 23:16:56 +02:00
klibc klibc: add missing files 2005-06-14 20:27:08 +02:00
libsysfs Fix libsysfs issue with relying on the detach_state file to be 2005-05-18 23:32:28 -07:00
test start to enforce plain ascii or valid utf8 2005-08-28 15:55:58 +02:00
ccdv.c [PATCH] tweak the ccdv program to handle files in subdirectories being built. 2005-04-26 22:02:44 -07:00
ChangeLog 068 release 2005-08-18 10:04:49 -07:00
COPYING [PATCH] added initial documentation and gpl license 2005-04-26 21:00:32 -07:00
FAQ [PATCH] complete removal of explicit udev permissions config file 2005-04-26 23:17:47 -07:00
HOWTO-udev_for_dev [PATCH] replace tdb database by simple lockless file database 2005-04-26 22:16:40 -07:00
list.h [PATCH] cleanup list.h 2005-04-26 23:51:00 -07:00
logging.h cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
make_gcov.sh [PATCH] rename LOG to USE_LOG in all places 2005-04-26 23:48:48 -07:00
Makefile add uft8 validation for safe volume label exporting 2005-08-28 13:49:32 +02:00
README add EXTRAS documentation to the README file. 2005-08-18 10:02:27 -07:00
README-gcov_for_udev [PATCH] gcov for udev 2005-04-26 21:35:16 -07:00
RELEASE-NOTES fix printed udevmonitor header 2005-08-18 13:43:27 +02:00
run_gcov.sh [PATCH] make gcov compile scripts working with recent gcc 2005-04-26 23:06:10 -07:00
TODO [PATCH] update the TODO list as we already have a devfs config file. 2005-04-26 21:35:07 -07:00
udev_add.c cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udev_config.c rename ...trailing_char() to ...trailing_chars() 2005-08-28 23:15:51 +02:00
udev_db.c don't store devpath in udevdb, we don't need it 2005-08-27 23:27:43 +02:00
udev_db.h don't store devpath in udevdb, we don't need it 2005-08-27 23:27:43 +02:00
udev_device.c rename ...trailing_char() to ...trailing_chars() 2005-08-28 23:15:51 +02:00
udev_event.c store ENV{key}="value" exported keys in the database 2005-08-26 21:38:18 +02:00
udev_libc_wrapper.c switch some strlcpy's to memcpy 2005-08-08 02:21:55 +02:00
udev_libc_wrapper.h cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udev_remove.c cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udev_rules_parse.c cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udev_rules.c rename ...trailing_char() to ...trailing_chars() 2005-08-28 23:15:51 +02:00
udev_rules.h add flag for reading of precompiled rules 2005-08-08 17:43:42 +02:00
udev_selinux.c selinux merge fixups again... 2005-04-26 23:52:14 -07:00
udev_selinux.h selinux merge fixups again... 2005-04-26 23:52:14 -07:00
udev_sysfs.c add pci_express to bus list 2005-08-25 03:37:38 +02:00
udev_sysfs.h fix the fix and change the file to wait for to the "bus" link 2005-06-22 02:11:59 +02:00
udev_utils_file.c split udev_util in several files 2005-08-10 16:54:14 +02:00
udev_utils_run.c use WRITE_END/READ_END for the pipe index 2005-08-28 20:25:37 +02:00
udev_utils_string.c rename ...trailing_char() to ...trailing_chars() 2005-08-28 23:15:51 +02:00
udev_utils.c add uft8 validation for safe volume label exporting 2005-08-28 13:49:32 +02:00
udev_utils.h rename ...trailing_char() to ...trailing_chars() 2005-08-28 23:15:51 +02:00
udev.8.in update README 2005-08-09 19:45:08 +02:00
udev.c bring std(in|out|err) fd's in a sane state 2005-08-18 13:07:15 +02:00
udev.h move code to its own files 2005-07-22 12:35:58 -04:00
udev.spec 068 release 2005-08-18 10:04:49 -07:00
udevcontrol.c add Usage: to udevmonitor and udevcontrol 2005-08-11 20:34:24 +02:00
udevd.8 [PATCH] support log-priority levels in udev.conf 2005-04-26 23:54:59 -07:00
udevd.c remove special TIMEOUT handling from incoming queue 2005-08-29 01:19:02 +02:00
udevd.h add udevmonitor, to debug netlink+udev events at the same time 2005-08-01 20:22:46 +02:00
udeveventrecorder.c cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udevinfo.8 [PATCH] udevinfo: print devpath -> node relationship for all devices 2005-04-26 23:39:48 -07:00
udevinfo.c don't store devpath in udevdb, we don't need it 2005-08-27 23:27:43 +02:00
udevinitsend.c fix exit code of udevinitsend and udevmonitor 2005-08-22 12:01:55 +02:00
udevmonitor.c udevmonitor: cleanup on exit 2005-08-25 00:38:25 +02:00
udevrulescompile.c add flag for reading of precompiled rules 2005-08-08 17:43:42 +02:00
udevsend.c allow RUN to send the environment to a local socket 2005-08-01 05:07:19 +02:00
udevstart.8 [PATCH] update bk ignore list some more. 2005-04-26 21:35:14 -07:00
udevstart.c cleanup some debug output and move to info level + unify select() loops 2005-08-15 11:57:04 +02:00
udevtest.8 [PATCH] support log-priority levels in udev.conf 2005-04-26 23:54:59 -07:00
udevtest.c add flag for reading of precompiled rules 2005-08-08 17:43:42 +02:00

udev - a userspace device manager

For more information on the design, and structure of this project, see the
files in the docs/ directory.

To use:

- You must be running a 2.6 version of the Linux kernel.

- Your 2.6 kernel must have had CONFIG_HOTPLUG enabled when it was built.

- Make sure sysfs is mounted at /sys. No other location is supported.
  You can mount it by running:
  	mount -t sysfs none /sys

- Make sure you integrate udev with your hotplug setup. There is a copy of
  the rules files for all major distros in the etc/udev folder. You may look
  there how others are doing it.

- Make sure you integrate with the kernel hotplug events. Later versions of
  udev are able to listen directly to a netlink socket, older versions used
  udevsend to feed the udev daemon with the kernel event. The most basic
  setup to run udev is to let the kernel for the udev binary directly:
	echo "/sbin/udev" > /proc/sys/kernel/hotplug

  While this may work in some setups, it is not recommended to do. A recent
  kernel and udev version is able to operate with the event serializing daemon
  udevd, that makes sure, that no "remove" event will beat a "add" event for
  the same device.

- Build the project:
	make

  Note:
      There are a number of different flags that you can use when building
      udev.  They are as follows:
	prefix
		set this to the default root that you want udev to be
		installed into.  This works just like the 'configure --prefix'
		script does.  Default value is ''.  Only override this if you
		really know what you are doing.
	USE_KLIBC
		if set to 'true', udev is built and linked against the
		included version of klibc.  Default value is 'false'.
	USE_LOG
		if set to 'true', udev will emit messages to the syslog when
		it creates or removes device nodes.  This is helpful to see
		what udev is doing.  This is enabled by default.  Note, if you
		are building udev against klibc it is recommended that you
		disable this option (due to klibc's syslog implementation.)
	USE_SELINUX
		if set to 'true', udev will be built with SELinux support
		enabled.  This is disabled by default.
	DEBUG
		if set to 'true', debugging messages will be sent to the syslog
		as udev is run.  Default value is 'false'.
	KERNEL_DIR
		If this is not set it will default to /lib/modules/`uname -r`/build
		This is used if USE_KLIBC=true to find the kernel include
		directory that klibc needs to build against.  This must be set
		if you are not building udev while running a 2.6 kernel.
	EXTRAS
		if set, will build the "extra" helper programs as specified
		as listed (see below for an example.)

      So, if you want to build udev using klibc with debugging messages, you
      would do:
	make USE_KLIBC=true DEBUG=true

      If you want to build the udev helper program cdrom_id and scsi_id you
      would do:
	make EXTRAS="extras/cdrom_id extras/scsi_id"

      udev will follow the setting of the debug level in udev.conf. Adapt this
      value to see the debug in syslog.

- Install the project:
	make install

  This will put the udev binaries in /sbin, create the and /etc/udev
  directories, and place the udev configuration files in /etc/udev/.  You
  will probably want to edit the *.rules files to create custom naming
  rules.  More info on how the config files are set up are contained in
  comments in the files, and is located in the documentation.

- Add and remove devices from the system and marvel as nodes are created
  and removed in /dev based on the device types.

- If you later get sick of it, uninstall it:
	make uninstall

If nothing seems to happen, make sure your build worked properly by
running the udev-test.pl script as root in the test/ subdirectory of the
udev source tree. Running udevstart should populate an empty /dev
directory. You may test, if a node is recreated after running udevstart.

Development and documentation help is very much appreciated, see the TODO
file for a list of things left to be done.

Any comment/questions/concerns please let me and the other udev developers
know by sending a message to the linux-hotplug-devel mailing list at:
	linux-hotplug-devel@lists.sourceforge.net

greg k-h
greg@kroah.com