5 The udevadm test command has no longer a --force option, nodes and symlinks
6 are always updated with a test run now.
12 We are currently merging the Ubuntu rules in the udev default rules,
13 and get one step closer to provide a common Linux /dev setup, regarding
14 device names, symlinks, and default device permissions. On udev startup,
15 we now expect the following groups to be resolvable to their ids with
17 disk, cdrom, floppy, tape, audio, video, lp, tty, dialout, kmem.
18 LDAP setups need to make sure, that these groups are always resolvable at
19 bootup, with only the rootfs mounted, and without network access available.
21 Some systems may need to add some new, currently not used groups, or need
22 to add some users to new groups, but the cost of this change is minimal,
23 compared to the pain the current, rather random, differences between the
24 various distributions cause for upstream projects and third-party vendors.
26 In general, "normal" users who log into a machine should never be a member
27 of any such group, but the device-access should be managed by dynamic ACLs,
28 which get added and removed for the specific users on login/logout and
29 session activity/inactivity. These groups are only provided for custom setups,
30 and mainly system services, to allow proper privilege separation.
31 A video-streaming daemon uid would be a member of "audio" and "video", to get
32 access to the sound and video devices, but no "normal" user should ever belong
33 to the "audio" group, because he could listen to the built-in microphone with
34 any ssh-session established from the other side of the world.
36 /dev/serial/by-{id,path}/ now contains links for ttyUSB devices,
37 which do not depend on the kernel device name. As usual, unique
38 devices - only a single one per product connected, or a real
39 USB serial number in the device - are always found with the same
40 name in the by-id/ directory.
41 Completely identical devices may overwrite their names in by-id/
42 and can only be found reliably in the by-path/ directory. Devices
43 specified by by-path/ must not change their connection, like the
44 USB port number they are plugged in, to keep their name.
46 To support some advanced features, Linux 2.6.22 is the oldest supported
47 version now. The kernel config with enabled SYSFS_DEPRECATED is no longer
48 supported. Older kernels should still work, and devices nodes should be
49 reliably created, but some rules and libudev will not work correctly because
50 the old kernels do not provide the expected information or interfaces.
56 Fix for a possible segfault while swapping network interface names in udev
63 The group "video" is part of the default rules now.
67 Bugfix for kernels using SYSFS_DEPRECATED* option and finding parent
68 block devices in some cases. No common distro uses this option anymore,
69 and we do not get enough testing for this and recent udev versions. If
70 this option is not needed to run some old distro with a new kernel,
71 it should be disabled in the kernel config.
73 Bugfix for the $links substitution variable, which may crash if no links
74 are created. This should not happen in usual setups because we always
75 create /dev/{block,char}/ links.
77 The strings of the parsed rules, which are kept in memory, no longer
78 contain duplicate entries, or duplicate tails of strings. This, and the
79 new rules parsing/matching code reduces the total in-memory size of
80 a huge distro rule sets to 0.08 MB, compared to the 1.2MB of udev
83 The export of DEVTYPE=disk/partition got removed from the default
84 rules. This value is available from the kernel. The pnp shell script
85 modprobe hack is removed from the default rules. ACPI devices have _proper_
86 modalias support and take care of the same functionality.
87 Installations which support old kernels, but install current default
88 udev rules may want to add that to the compat rules file.
90 Libvolume_id now always probes for all known filesystems, and does not
91 stop at the first match. Some filesystems are marked as "exclusive probe",
92 and if any other filesytem type matches at the same time, libvolume_id
93 will, by default, not return any probing result. This is intended to prevent
94 mis-detection with conflicting left-over signatures found from earlier
95 file system formats. That way, we no longer depend on the probe-order
96 in case of multiple competing signatures. In some setups the kernel allows
97 to mount a volume with just the old filesystem signature still in place.
98 This may damage the new filesystem and cause data-loss, just by mounting
99 it. Because volume_id can not decide which one the correct signature is,
100 the wrong signatures need to be removed manually from the volume, or the
101 volume needs to be reformatted, to enable filesystem detection and possible
106 Fix segfault if compiled without optimization and dbg() does not get
107 compiled out and uses variables which are not available.
111 Bugfixes. (And maybe new bugs. :))
113 The rule matching engine got converted from a rule list to a token
114 array which reduced the in-memory rules representation of a full
115 featured distros with thousends of udev rules from 1.2MB to 0.12 MB.
116 Limits like 5 ENV and ATTR matches, and one single instance for most
117 other keys per rule are gone.
119 The NAME assignment is no longer special cased. If later rules assign
120 a NAME value again, the former value will be overwritten. As usual
121 for most other keys, the NAME value can be protected by doing a final
122 assignment with NAME:="<value>".
124 All udev code now uses libudev, which is also exported. The library
125 is still under development, marked as experimental, and its interface
126 may change as long as the DeviceKit integration is not finished.
128 Many thanks to Alan Jenkins for his continuous help, and finding and
129 optimizing some of the computing expensive parts.
135 Kernel devices and device nodes are connected now by reverse indizes in
136 /sys and /dev. A device number retrieved by a stat() or similar, the
137 kernel device directory can be found by looking up:
138 /sys/dev/{block,char}/<maj>:<min>
139 and the device node of the same device by looking up:
140 /dev/{block,char}/<maj>:<min>
144 Fix recently introduced bug, which caused a compilation without large
145 file support, where vol_id does not recognize raid signatures at the end
148 Firewire disks now create both, by-id/scsi-* and by-id/ieee-* links.
149 Seems some kernel versions prevent the creation of the ieee-* links,
150 so people used the scsi-* link which disappeared now.
152 More libudev work. Almost all udevadm functionality comes from libudev
155 udevadm trigger has a new option --type, which allows to trigger events
156 for "devices", for "subsystems", or "failed" devices. The old option
157 --retry-failed" still works, but is no longer mentioned in the man page.
163 The udevadm info --device-id-of-file= output has changed to use
164 the obvious format. Possible current users should use the --export
165 option which is not affected.
167 The old udev commands symlinks to udevadm are not installed, if
168 these symlinks are used, a warning is printed.
174 Optical drive's media is no longer probed for raid signatures,
175 reading the end of the device causes some devices to malfunction.
176 Also the offset of the last session found is used now to probe
179 The volume_id library got a major version number update to 1,
180 some deprecated functions are removed.
182 A shared library "libudev" gets installed now to provide access
183 to udev device information. DeviceKit, the successor of HAL, will
184 need this library to access the udev database and search sysfs for
186 The library is currently in an experimental state, also the API is
187 expected to change, as long as the DeviceKit integration is not
192 We use ./configure now. See INSTALL for details. Current
195 "/usr" - prefix for man pages, include files
197 "" - the root filesystem, prefix for libs and binaries
201 "lib" - directory name for libraries, not a path name
202 multilib 64bit systems may use "lib64" instead of "lib"
204 compile-in verbose debug messages
206 disable all logging and compile-out all log strings
208 link against SELInux libraries, to set the expected context
211 In the default rules, the group "disk" gets permissions 0660 instead
212 of 0640. One small step closer to unify distro rules. Some day, all
213 distros hopefully end up with the same set of rules.
215 No symlinks to udevadm are installed anymore, if they are still needed,
216 they should be provided by the package.
222 Default udev rules, which are not supposed to be edited by the user, should
223 be placed in /lib/udev/rules.d/ now, to make it clear that they are private to
224 the udev package and will be replaced with an update. Udev will pick up rule
226 /lib/udev/rules.d/ - default installed rules
227 /etc/udev/rules.d/ - user rules + on-the-fly generated rules
228 /dev/.udev/rules.d/ - temporary non-persistent rules created after bootup
229 It does not matter in which directory a rule file lives, all files are sorted
232 To help creating /dev/root, we have now:
233 $ udevadm info --export --export-prefix="ROOT_" --device-id-of-file=/
236 In case the current --device-id-of-file is already used, please switch to
237 the --export format version, it saves the output parsing and the old
238 format will be changed to use ':' as a separator, like the format in the
243 Fix cdrom_id to properly recognize blank media.
249 Tape drive id-data is queried from /dev/bsg/* instead of the tape
250 nodes. This avoids rewinding tapes on open().
256 The symlinks udevcontrol and udevtrigger are no longer installed by
259 The scsi_id program does not depend on sysfs anymore. It can speak
260 SGv4 now, so /dev/bsg/* device nodes can be used, to query SCSI device
261 data, which should solve some old problems with tape devices, where
262 we better do not open all tape device nodes to identify the device.
268 The cdrom_id program is replaced by an advanced version, which can
269 detect most common device types, and also properties of the inserted
270 media. This is part of moving some basic functionality from HAL into
271 udev (and the kernel).
277 The last WAIT_FOR_SYSFS rule is removed from the default rules.
279 The symlinks to udevadm for the debugging tools: udevmonitor and
280 udevtest are no longer created.
282 The symlinks to the udevadm man page for the old tool names are
285 Abstract namespace sockets paths in RUN+="socket:@<path>" rules,
286 should be prefixed with '@' to indicate that the path is not a
297 Udevstart is removed from the tree, it did not get installed for
298 a long time now, and is long replaced by trigger and settle.
304 All udev tools are merged into a single binary called udevadm.
305 The old names of the tools are built-in commands in udevadm now.
306 Symlinks to udevadm, with the names of the old tools, provide
307 the same functionality as the standalone tools. There is also
308 only a single udevadm.8 man page left for all tools.
310 Tools like mkinitramfs should be checked, if they need to include
311 udevadm in the list of files.
321 The etc/udev/rules.d/ directory now contains a default set of basic
322 udev rules. This initial version is the result of a rules file merge
323 of Fedora and openSUSE. For these both distros only a few specific
324 rules are left in their own file, named after the distro. Rules which
325 are optionally installed, because they are only valid for a specific
326 architecture, or rules for subsystems which are not always used are
327 in etc/udev/packages/.
333 Dynamic rules can be created in /dev/.udev/rules.d/ to trigger
334 actions by dynamically created rules.
336 SYMLINK=="<value>" matches agains the entries in the list of
337 currently defined symlinks. The links are not created in the
338 filesystem at that point in time, but the values can be matched.
340 RUN{ignore_error}+="<program>" will ignore any exit code from the
341 program and not record as a failed event.
347 Final merge of patches/features from the Ubuntu package.
353 Control characters in filesystem label strings are no longer silenty
354 removed, but hex-encoded, to be able to uniquely identify the device
355 by its symlink in /dev/disk/by-label/.
356 If libvolume_id is used by mount(8), LABEL= will work as expected,
357 if slashes or other characters are used in the label string.
359 To test the existence of a file, TEST=="<file>" and TEST!="<file>"
360 can be specified now. The TEST key accepts an optional mode mask
361 TEST{0100}=="<is executable file>".
363 Scsi_id now supports a mode without expecting scsi-specific sysfs
364 entries to allow the extraction of cciss-device persistent properties.
370 In the future, we may see uuid's which are just simple character
371 strings (see the DDF Raid Specification). For that reason vol_id now
372 exports ID_FS_UUID_SAFE, just like ID_FS_LABEL_SAFE. For things like
373 the creation of symlinks, the *_SAFE values ensure, that no control
374 or whitespace characters are used in the filename.
376 Possible users of libvolume_id, please use the volume_id_get_* functions.
377 The public struct will go away in a future release of the library.
383 Removal of useless extras/eventrecorder.sh.
393 The directory multiplexer for dev.d/ and hotplug.d are finally removed
394 from the udev package.
400 Symlinks can have priorities now, the priority is assigned to the device
401 and specified with OPTIONS="link_priority=100". Devices with higher
402 priorities overwrite the symlinks of devices with lower priorities.
403 If the device that currently owns the link, goes away, the symlink
404 will be removed, and recreated, pointing to the next device with the
405 highest actual priority. This should make /dev/disk/by-{label,uuid,id}
406 more reliable, if multiple devices contain the same metadata and overwrite
409 The dasd_id program is removed from the udev tree, and dasdinfo, with the
410 needed rules, are part of the s390-tools now.
412 Please add KERNEL=="[0-9]*:[0-9]*" to the scsi wait-for-sysfs rule,
413 we may get the scsi sysfs mess fixed some day, and this will only catch
414 the devices we are looking for.
416 USB serial numbers for storage devices have the target:lun now appended,
417 to make it possibble to distinguish broken multi-lun devices with all
418 the same SCSI identifiers.
420 Note: The extra "run_directory" which searches and executes stuff in
421 /etc/hotplug.d/ and /etc/dev.d/ is long deprecated, and will be removed
422 with the next release. Make sure, that you don't use it anymore, or
423 provides your own implementation of that inefficient stuff.
424 We are tired of reports about a "slow udev", because these directories
425 contain stuff, that runs with _every_ event, instead of using rules,
426 that run programs only for the matching events.
436 DRIVER== will match only for devices that actually have a real
437 driver. DRIVERS== must be used, if parent devices should be
438 included in the match.
440 Libvolume_id's "linux_raid" detection needed another fix.
448 Add additional check to volume_id detection of via_raid, cause
449 some company decided to put a matching pattern all over the empty
450 storage area of their music players.
454 Fix path_id for SAS devices.
458 The udev daemon can be started with --debug-trace now, which will
459 execute all events serialized to get a chance to catch a possible
460 action that crashes the box.
462 A warning is logged, if PHYSDEV* keys, the "device" link, or a parent
463 device attribute like $attr{../file} is used, only WAIT_FOR_SYSFS rules
464 are excluded from the warning. Referencing parent attributes directly
465 may break when something in the kernel driver model changes. Udev will
466 just find the attribute by walking up the parent chain.
468 Udevtrigger now sorts the list of devices depending on the device
469 dependency, so a "usb" device is triggered after the parent "pci"
474 Revert persistent-storage ata-serial '_' '-' replacement.
480 Udevtrigger can now filter the list of devices to be triggered. Matches
481 for subsystems or sysfs attributes can be specified.
483 The entries in /dev/.udev/queue and /dev/.udev/failed have changed to
484 zero-sized files to avoid pointing to /sys and confuse broken tools which
485 scan the /dev directory. To retry failed events, udevtrigger --retry-failed
488 The rules and scripts to create udev rules for persistent network
489 devices and optical drives are in the extras/rules_generator directory
490 now. If you use something similar, please consider replacing your own
491 version with this, to share the support effort. The rule_generator
492 installs its own rules into /etc/udev/rules.d.
494 The cdrom_id tool installs its own rule now in /etc/udev/rules.d, cause
495 the rule_generator depends on cdrom_id to be called in an earlier rule.
501 Renaming of some key names (the old names still work):
502 BUS -> SUBSYSTEMS, ID -> KERNELS, SYSFS -> ATTRS, DRIVER -> DRIVERS.
503 (The behavior of the key DRIVER will change soon in one of the next
504 releases, to match only the event device, please switch to DRIVERS
505 instead. If DRIVER is used, it will behave like DRIVERS, but an error
507 With the new key names, we have a more consistent and simpler scheme.
508 We can match the properties of the event device only, with: KERNEL,
509 SUBSYSTEM, ATTR, DRIVER. Or include all the parent devices in the match,
510 with: KERNELS, SUBSYSTEMS, ATTRS, DRIVERS. ID, BUS, SYSFS, DRIVER are no
511 longer mentioned in the man page and should be switched in the rule
514 ATTR{file}="value" can be used now, to write to a sysfs file of the
515 event device. Instead of:
516 ..., SYSFS{type}=="0|7|14", RUN+="/bin/sh -c 'echo 60 > /sys$$DEVPATH/timeout'"
518 ..., ATTR{type}=="0|7|14", ATTR{timeout}="60"
520 All the PHYSDEV* keys are deprecated and will be removed from a
522 PHYDEVPATH - is the path of a parent device and should not be
524 PHYSDEVBUS - is just a SUBSYSTEM value of a parent, and can be
525 matched with SUBSYSTEMS==
526 PHYSDEVDRIVER - for bus devices it is available as ENV{DRIVER}.
527 Newer kernels will have DRIVER in the environment,
528 for older kernels udev puts in. Class device will
529 no longer carry this property of a parent and
530 DRIVERS== can be used to match such a parent value.
531 Note that ENV{DRIVER} is only available for a few bus devices, where
532 the driver is already bound at device event time. On coldplug, the
533 events for a lot devices are already bound to a driver, and they will have
534 that value set. But on hotplug, at the time the kernel creates the device,
535 it can't know what driver may claim the device after that, therefore
536 in most cases it will be empty.
538 Failed events should now be re-triggered with:
539 udevtrigger --retry-failed.
540 Please switch to this command, so we keep the details of the /dev/.udev/failed/
541 files private to the udev tools. We may need to switch the current symlink
542 target, cause some obviously broken tools try to scan all files in /dev
543 including /dev/.udev/, find the links to /sys and end up stat()'ing sysfs files
544 million times. This takes ages on slow boxes.
546 The udevinfo attribute walk (-a) now works with giving a device node
547 name (-n) instead of a devpath (-p). The query now always works, also when
548 no database file was created by udev.
550 The built-in /etc/passwd /etc/group parser is removed, we always depend on
551 getpwnam() and getgrnam() now. One of the next releases will depend on
552 fnmatch() and may use getopt_long().
556 Bugfixes and small improvements.
560 Fix path_id for recent kernels.
566 Added support for swapping network interface names, by temporarily
567 renaming the device and wait for the target name to become free.
571 The built-in MODALIAS key and substitution is removed.
575 The binary firmware helper is replaced by the usual simple
576 shell script. Udevsend is removed from the tree.
584 Some more keys require the correct use of '==' and '=' depending
585 on the kind of operation beeing an assignment or a match. Rules
586 with invalid operations are skipped and logged to syslog. Please
587 test with udevtest if the parsing of your rules throws errors and
588 fix possibly broken rules.
592 Provide "udevsettle" to wait for all current udev events to finish.
593 It also watches the current kernel netlink queue by comparing the
594 even sequence number to make sure that there are no current pending
595 events that have not already arrived in the daemon.
599 Fix rule to skip persistent rules for removable IDE devices, which
600 also skipped optical IDE drives.
602 All *_id program are installed in /lib/udev/ by default now.
604 No binary is stripped anymore as this should be done in the
605 packaging process and not at build time.
607 libvolume_id is provided as a shared library now and vol_id is
608 linked against it. Also one of the next HAL versions will require
609 this library, and the HAL build process will also require the
610 header file to be installed. The copy of the same code in HAL will
611 be removed to have only a single copy left on the system.
615 Add persistent links for SCSI tapes. The rules file is renamed
616 to 60-persistent-storage.rules.
618 Create persistent path for usb devices. Can be used for all sorts
619 of devices that can't be distinguished by other properties like
620 multiple identical keyboards and mice connected to the same box.
622 Provide "udevtrigger" program to request events on coldplug. The
623 shell script is much too slow with thousends of devices.
627 Fix persistent disk rules to exclude removable IDE drives.
629 Warn if %e, $modalias or MODALIAS is used.
633 Fix queue export, which wasn't correct for subsequent add/remove
634 events for the same device.
638 Fix cramfs detection on big endian.
640 Make WAIT_FOR_SYSFS usable in "normal" rules and silent if the whole
645 If BUS== and SYSFS{}== have been used in the same rule, the sysfs
646 attributes were only checked at the parent device that matched the
647 by BUS requested subsystem. Fix it to also look at the device we
648 received the event for.
650 Build variable CROSS has changed to CROSS_COMPILE to match the kernel
655 Fix a bug where NAME="" would prevent RUN from beeing executed.
657 RUN="/bin/program" does not longer automatically add the subsystem
658 as the first parameter. This is from the days of /sbin/hotplug
659 which is dead now and it's just confusing to need to add a space at
660 the end of the program name to prevent this.
661 If you use rules that need the subsystem as the first parameter,
662 like the old "udev_run_hotlugd" and "udev_run_devd", add the subsystem
663 to the key like RUN+="/bin/program $env{SUBSYSTEM}".
667 The udev man page has moved to udev(7) as it does not describe a command
668 anymore. The programs udev, udevstart and udevsend are no longer installed
669 by default and must be copied manually, if they should be installed or
670 included in a package.
672 Fix a bug where "ignore_device" could run earlier collected RUN keys before
673 the ignore rule was applied.
675 More preparation for future sysfs changes. usb_id and scsi_id no longer
676 depend on a magic order of devices in the /devices chain. Specific devices
677 should be requested by their subsytem.
679 This will always find the scsi parent device without depending on a specific
681 dev = sysfs_device_get(devpath);
682 dev_usb = sysfs_device_get_parent_with_subsystem(dev, "scsi");
684 The "device" link in the current sysfs layout will be automatically
685 _resolved_ as a parent and in the new sysfs layout it will just _be_ the
686 parent in the devpath. If a device is requested by it's symlink, like all
687 class devices in the new sysfs layout will look like, it gets automatically
688 resolved and substituted with the real devpath and not the symlink path.
691 A similar logic must be applied to _all_ sysfs users, including
692 scripts, that search along parent devices in sysfs. The explicit use of
693 the "device" link must be avoided. With the future sysfs layout all
694 DEVPATH's will start with /devices/ and have a "subsystem" symlink poiting
695 back to the "class" or the "bus". The layout of the parent devices in
696 /devices is not necessarily expected to be stable across kernel releases and
697 searching for parents by their subsystem should make sysfs users tolerant
698 for changed parent chains.
702 Prepare udev to work with the experimental kernel patch, that moves
703 /sys/class devices to /sys/devices and /sys/block to /sys/class/block.
705 Clarify BUS, ID, $id usage and fix $id behavior. This prepares for
706 moving the class devices to /sys/devices.
708 Thanks again to Marco for help finding a hopefully nice compromise
709 to make %b simpler and working again.
713 Complete removal of libsysfs, replaced by simple helper functions
714 which are much simpler and a bit faster. The udev daemon operatesentirely
715 on event parameters and does not use sysfs for simple rules anymore.
716 Please report any new bugs/problems, that may be caused by this big
717 change. They will be fixed immediately.
719 The enumeration format character '%e' is deprecated and will be
720 removed sometimes from a future udev version. It never worked correctly
721 outside of udevstart, so we can't use it with the new parallel
722 coldplug. A simple enumeration is as useless as the devfs naming
723 scheme, just get rid of both if you still use it.
725 MODALIAS and $modalias is not needed and will be removed from one of
726 the next udev versions, replace it in all rules with ENV{MODALIAS} or
727 the sysfs "modalias" value.
729 Thanks a lot to Marco for all his help on finding and fixing bugs.
733 Let scsi_id request libata drive serial numbers from page 0x80.
735 Renamed etc/udev/persistent.rules to persistent-disk.rules and
736 added /dev/disk/by-name/* for device mapper device names.
738 Removed %e from the man page. It never worked reliably outside
739 of udevstart and udevstart is no longer recommended to use.
743 Symlinks are now exported to the event environment. Hopefully it's no
744 longer needed to run udevinfo from an event process, like it was
745 mentioned on the hotplug list:
746 UDEV [1134776873.702967] add@/block/sdb
749 DEVLINKS=/dev/disk/by-id/usb-IBM_Memory_Key_0218B301030027E8 /dev/disk/by-path/usb-0218B301030027E8:0:0:0
753 Fix a problem if udevsend is used as the hotplug handler and tries to use
754 syslog, which causes a "vc" event loop. 2.6.15 will make udevsend obsolete
755 and this kind of problems will hopefully go away soon.
759 All built-in logic to work around bad sysfs timing is removed with this
760 version. The need to wait for sysfs files is almost fixed with a kernel
761 version that doesn't work with this udev version anyway. Until we fix
762 the timing of the "bus" link creation, the former integrated logic should
763 be emulated by a rule placed before all other rules:
764 ACTION=="add", DEVPATH=="/devices/*", ENV{PHYSDEVBUS}=="?*", WAIT_FOR_SYSFS="bus"
766 The option "udev_db" does no longer exist. All udev state will be in
767 /$udev_root/.udev/ now, there is no longer an option to set this
769 If the init script or something else used this value, just depend on
770 this hardcoded path. But remember _all_content_ of this directory is
771 still private to udev and can change at any time.
773 Default location for rule sripts and helper programs is now: /lib/udev/.
774 Everything that is not useful on the commandline should go into this
775 directory. Some of the helpers in the extras folder are installed there
776 now. The rules need to be changed, to find the helpers there.
778 Also /lib/udev/devices is recommended as a directory where packages or
779 the user can place real device nodes, which get copied over to /dev at
780 every boot. This should replace the various solutions with custom config
783 Udevsend does no longer start the udev daemon. This must be done with
784 the init script that prepares /dev on tmpfs and creates the initial nodes,
785 before starting the daemon.
789 Silent a too verbose error logging for the old hotplug.d/ dev.d/
792 The copy of klibc is removed. A systemwide installed version of klibc
793 should be used to build a klibc udev now.
797 NAME="" will not create any nodes, but execute RUN keys. To completely
798 ignore an event the OPTION "ignore_device" should be used.
800 After removal of the reorder queue, events with a TIMEOUT can be executed
801 without any queuing now.
805 Fixed bug in udevd, if inotify is not available. We depend on netlink
806 uevents now, kernels without that event source will not work with that
807 version of udev anymore.
811 The rule parsing happens now in the daemon once at startup, all udev
812 event processes inherit the already parsed rules from the daemon.
813 It is shipped with SUSE10.0 and reduces heavily the system load at
814 startup. The option to save precompiled rules and let the udev process
815 pick the them up is removed, as it's no longer needed.
817 Kernel 2.6.15 will have symlinks at /class/input pointing to the real
818 device. Libsysfs is changed to "translate" the requested link into the
819 real device path, as it would happen with the hotplug event. Otherwise
820 device removal and the udev database will not work.
822 Using 'make STRIPCMD=' will leave the binaries unstripped for debugging
825 A few improvements for vol_id, the filesytem probing code.
829 Fix a stupid typo in extras/run_directory for "make install".
831 scsi_id creates the temporary devnode now in /dev for usage with a
832 non-writable /tmp directory.
834 The uevent kernel socket buffer can carry app. 50.000 events now,
835 let's see who can break this again. :)
837 The upcoming kernel will have a new input driver core integration.
838 Some class devices are now symlinks to the real device. libsysfs
839 needs a fix for this to work correctly. Udevstart of older udev
840 versions will _not_ create these devices!
844 Fix a 'install' target in the Makefile, that prevents EXTRAS from
849 A bunch of mostly trivial bugfixes. From now on no node name or
850 symlink name can contain any character than plain whitelisted ascii
851 characters or validated utf8 byte-streams. This is needed for the
852 /dev/disk/by-label/* links, because we import untrusted data and
853 export it to the filesystem.
857 More bugfixes. If udevd was started from the kernel, we don't
858 have stdin/stdout/stderr, which broke the forked tools in some
863 Bugfix. udevstart event ordering was broken for a long time.
864 The new run_program() uncovered it, because /dev/null was not
865 available while we try to run external programs.
866 Now udevstart should create it before we run anything.
870 Minor bugfixes and some distro rules updates. If you don't have the
871 persistent disk rules in /dev/disk/by-*/* on your distro, just
872 grab it from here. :)
876 We can use socket communication now to pass events from udev to
878 RUN+="socket:/org/freedesktop/hal/udev_event"
879 will pass the whole udev event to the HAL daemon without the need
880 for a forked helper. (See ChangeLog for udevmonitor, as an example)
884 Mostly bugfixes and see ChangeLog.
886 The test for the existence of an environment value should be
888 ENV{KEY}=="*" to ENV{KEY}=="?*"
889 because "*" will not fail anymore, if the key does not exist or
894 Bugfixes and a few tweaks described in the ChangeLog.
898 Mostly a Bugfix release.
900 Added WAIT_FOR_SYSFS="<attribute>" to be able to fight against the sysfs
901 timing with custom rules.
905 We changed the internal rule storage format. Our large rule files took
906 2 MB of RAM, with the change we are down to 99kB.
908 If the device-node has been created with default name and no symlink or
909 options are to remenber, it is not longer stored in the udevdb. HAL will
910 need to be updated to work correctly with that change.
912 To overrride optimization flags, OPTFLAGS may be used now.
920 Major changes happened with this release. The goal is to take over the
921 complete kernel-event handling and provide a more efficient way to dispatch
922 kernel events. Replacing most of the current shell script logic and the
923 kernel forked helper with a netlink-daemon and a rule-based event handling.
925 o udevd listens to netlink events now. The first valid netlink event
926 will make udevd ignore any message from udevsend that contains a
927 SEQNUM, to avoid duplicate events. The forked events can be disabled
929 echo "" > /proc/sys/kernel/hotplug
930 For full support, the broken input-subsytem needs to be fixed, not to
931 bypass the driver core.
933 o /etc/dev.d/ + /etc/hotplug.d/ directory multiplexing is completely
934 removed from udev itself and must be emulated by calling small
935 helper binaries provided in the extras folder:
936 make EXTRAS=extras/run_directory/
937 will build udev_run_devd and udev_run_hotplugd, which can be called
938 from a rule if needed:
939 RUN+="/sbin/udev_run_hotplugd"
940 The recommended way to handle this is to convert all the calls from
941 the directories to explicit udev rules and get completely rid of the
942 multiplexing. (To catch a ttyUSB event, you now no longer need to
943 fork and exit 300 tty script instances you are not interested in, it
944 is just one rule that matches exactly the device.)
946 o udev handles now _all_ events not just events for class and block
947 devices, this way it is possible to control the complete event
948 behavior with udev rules. Especially useful for rules like:
949 ACTION="add", DEVPATH="/devices/*", MODALIAS=="?*", RUN+="/sbin/modprobe $modalias"
951 o As used in the modalias rule, udev supports now textual
952 substitution placeholder along with the usual format chars. This
953 needs to be documented, for now it's only visible in udev_rules_parse.c.
955 o The rule keys support now more operations. This is documented in the
956 man page. It is possible to add values to list-keys like the SYMLINK
957 and RUN list with KEY+="value" and to clear the list by assigning KEY="".
958 Also "final"-assignments are supported by using KEY:="value", which will
959 prevent changing the key by any later rule.
961 o kernel 2.6.12 has the "detached_state" attribute removed from
962 sysfs, which was used to recognize sysfs population. We switched that
963 to wait for the "bus" link, which is only available in kernels after 2.6.11.
964 Running this udev version on older kernels may cause a short delay for
967 o To provide infrastructure for persistent device naming, the id programs:
968 scsi_id, vol_id (former udev_volume_id), and ata_id (new) are able now
969 to export the probed data in environment key format:
970 pim:~ # /sbin/ata_id --export /dev/hda
971 ID_MODEL=HTS726060M9AT00
972 ID_SERIAL=MRH401M4G6UM9B
976 KERNEL="hd*[!0-9]", IMPORT="/sbin/ata_id --export $tempnode"
977 KERNEL="hd*[!0-9]", ENV{ID_SERIAL}=="?*", SYMLINK+="disk/by-id/$env{ID_MODEL}_$env{ID_SERIAL}"
980 kay@pim:~> tree /dev/disk
983 | |-- HTS726060M9AT00_MRH401M4G6UM9B -> ../../hda
984 | `-- IBM-Memory_Key -> ../../sda
986 | |-- swap -> ../../hda1
987 | |-- date -> ../../sda1
988 | `-- home -> ../../hda3
990 |-- 2E08712B0870F2E7 -> ../../hda3
991 |-- 9352cfef-7687-47bc-a2a3-34cf136f72e1 -> ../../hda1
992 |-- E845-7A89 -> ../../sda1
993 `-- b2a61681-3812-4f13-a4ff-920d70604299 -> ../../hda2
995 The IMPORT= operation will import these keys in the environment and make
996 it available for later PROGRAM= and RUN= executed programs. The keys are
997 also stored in the udevdb and can be queried from there with one of the
1000 o A few binaries are silently added to the repository, which can be used
1001 to replay kernel events from initramfs instead of using coldplug. udevd
1002 can be instructed now to queue-up events while the stored events from
1003 initramfs are filled into the udevd-queue. This code is still under
1004 development and there is no documentation now besides the code itself.
1005 The additional binaries get compiled, but are not installed by default.
1007 o There is also a temporary fix for a performance problem where too many
1008 events happen in parallel and every event needs to parse the rules.
1009 udev can now read precompiled rules stored on disk. This is likely to be
1010 replaced by a more elegant solution in a future udev version.
1014 With kernel version 2.6.12, the sysfs file "detached_state" was removed.
1015 Fix for libsysfs not to expect this file was added.
1019 All rules are applied now, but only the first matching rule with a NAME-key
1020 will be applied. All later rules with NAME-key are completely ignored. This
1021 way system supplied symlinks or permissions gets applied to user-defined
1025 Please check your rules setup, if you may need to add OPTIONS="last_rule"
1026 to some rules, to keep the old behavior.
1028 The rules are read on "remove"-events too. That makes is possible to match
1029 with keys that are available on remove (KERNEL, SUBSYSTEM, ID, ENV, ...) to
1030 instruct udev to ignore an event (OPTIONS="ignore_device").
1031 The new ACTION-key may be used to let a rule act only at a "remove"-event.
1033 The new RUN-key supports rule-based execution of programs after device-node
1034 handling. This is meant as a general replacement for the dev.d/-directories
1035 to give fine grained control over the execution of programs.
1037 The %s{}-sysfs format char replacement values are searched at any of the
1038 devices in the device chain now, not only at the class-device.
1040 We support log priority levels now. The value udev_log in udev.conf is used
1041 to determine what is printed to syslog. This makes it possible to
1042 run a version with compiled-in debug messages in a production environment
1043 which is sometimes needed to find a bug.
1044 It is still possible to supress the inclusion of _any_ syslog usage with
1045 USE_LOG=false to create the smallest possible binaries if needed.
1046 The configured udev_log value can be overridden with the environment variable
1051 Possible use of a system-wide klibc:
1052 make USE_KLIBC=true KLCC=/usr/bin/klcc all
1053 will link against an external klibc and our own version will be ignored.
1057 We support an unlimited count of symlinks now.
1059 If USE_STATIC=true is passed to a glibc build, we link statically and use
1060 a built-in userdb parser to resolve user and group names.
1062 The PLACE= key is gone. It can be replaced by an ID= for a long time, because
1063 we walk up the chain of physical devices to find a match.
1065 The KEY="<value>" format supports '=', '==', '!=,' , '+=' now. This makes it
1066 easy to skip certain attribute matches without composing rules with weird
1067 character class negations like:
1068 KERNEL="[!s][!c][!d]*"
1069 this can now be replaced with:
1071 The current simple '=' is still supported, and should work as it does today,
1072 but existing rules should be converted if possible, to be better readable.
1074 We have new ENV{}== key now, to match against a maximum of 5 environment
1077 udevstart is its own binary again, because we don't need co carry this araound
1078 with every forked event.