[Bug 746595] New: KBD_NUMLOCK="yes" in /etc/sysconfig/keyboard does not work under systemd init
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c0 Summary: KBD_NUMLOCK="yes" in /etc/sysconfig/keyboard does not work under systemd init Classification: openSUSE Product: openSUSE 12.2 Version: Factory Platform: PC OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: mrmazda@earthlink.net QAContact: qa@suse.de Found By: --- Blocker: --- y2logs (bug 746594): http://bugzilla.novell.com/attachment.cgi?id=475750 This started in 12.1. see: https://bugzilla.novell.com/show_bug.cgi?id=720486#c11 To reproduce: 1-install 12.2M1/KDE with default init packages 2-set KBD_NUMLOCK="yes" in /etc/sysconfig/keyboard 3-reboot Actual results: 1-NUM state is off on tty[1-6] Expected results: 1-NUM state is on on tty[1-6] -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c zj jia <zjjia@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |zjjia@suse.com AssignedTo|bnc-team-screening@forge.pr |sbrabec@suse.com |ovo.novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c1 --- Comment #1 from Frederic Crozat <fcrozat@suse.com> 2012-02-13 15:36:21 UTC --- as explained in comment 8 from bug 720486 "For KBD_NUMLOCK, it is not supported under systemd." kernel should be fixed to not change numlock state at startup -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c2 --- Comment #2 from Felix Miata <mrmazda@earthlink.net> 2012-02-13 16:56:08 UTC --- As long as systemd is unable to do at least as well as sysvinit did regarding NUM state, and kernel devs continue after close to three decades to disrespect the BIOS state for NUM, systemd won't be usable. If it was up to me, it would be impossible for any keyboard with separate cursor and NUM pads to not produce NUM characters from the NUM pad, and the NUM key if retained at all would be permanently and unalterably on. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Component|Basesystem |Kernel AssignedTo|sbrabec@suse.com |kernel-maintainers@forge.pr | |ovo.novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c3 Wolfgang Rosenauer <wolfgang@rosenauer.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |wolfgang@rosenauer.org Severity|Normal |Major --- Comment #3 from Wolfgang Rosenauer <wolfgang@rosenauer.org> 2012-02-29 09:32:21 UTC --- I don't care where this is going to be fixed! But it needs a fix soon IMHO. openSUSE provides /etc/sysconfig/kbd with KBD_NUMLOCK setting but a default install just ignores it? A statement like "KBD_NUMLOCK, it is not supported under systemd" is not really acceptable. The option is there and does not work. -> BUG And no, this is not a 12.2 bug. This needs a fix in 12.1 as well! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c4 Wolfgang Rosenauer <wolfgang@rosenauer.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sbrabec@suse.com --- Comment #4 from Wolfgang Rosenauer <wolfgang@rosenauer.org> 2012-02-29 09:45:56 UTC --- adding kbd maintainer -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c5 --- Comment #5 from Stanislav Brabec <sbrabec@suse.com> 2012-03-01 18:24:21 CET --- Looking at systemd, it does not use kbd init script at all. It does things on its own in following utilities: /lib/systemd/systemd-localed /lib/systemd/systemd-vconsole-setup and this file /usr/share/systemd/kbd-model-map There exists /lib/systemd/system/kbd.service, which is a symlink to /dev/null. Guessing that removal of this file would fix this bug, but it will probably bring some side effects. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c6 --- Comment #6 from Stanislav Brabec <sbrabec@suse.com> 2012-03-01 20:05:14 CET --- Confirming "rm /lib/systemd/system/kbd.service" as a quick work-around. I do not see any apparent regressions. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c7 Jeff Mahoney <jeffm@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |coolo@suse.com, | |jeffm@suse.com AssignedTo|kernel-maintainers@forge.pr |fcrozat@suse.com |ovo.novell.com | --- Comment #7 from Jeff Mahoney <jeffm@suse.com> 2012-03-14 15:53:03 EDT --- (In reply to comment #6)
Confirming "rm /lib/systemd/system/kbd.service" as a quick work-around. I do not see any apparent regressions.
I believe this means it runs the old kdb script, right? Fine, kill the kdb.service file from the systemd package. This habit of ignoring regressions because they don't fit the systemd model needs to stop. This is a feature openSUSE has had since before there was an openSUSE and the fact that systemd doesn't honor it is a regression. I'd imagine the numlock enable-disable feature works on non-x86 systems as well as x86 systems that don't have a BIOS in which to make that determination. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c8 --- Comment #8 from Frederic Crozat <fcrozat@suse.com> 2012-03-15 08:33:39 UTC --- I don't agree with you, Jeff : kernel should keep the numlock state when it starts and we shouldn't have to read BIOS state with hwinfo to workaround it. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c9 --- Comment #9 from Felix Miata <mrmazda@earthlink.net> 2012-03-15 08:43:46 UTC --- That there may be a better way has nothing to do with whether this is a regression. Having NUM on always without jumping through hoops of irrationality was a significant reason for my choice of (open)SUSE in the first place many moons ago. Until the kernel people get their heads out of the sand, what used to work needs to continue to work, regardless what init system is used. I only use keyboards with separate NUM pads so that I can _rapidly_ touch type (use the keyboard without looking at it) numbers, same as on a calculator/adding machine, just like _every_ competent bookkeeper and accountant who uses a full featured personal computer keyboard. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c10 --- Comment #10 from Jeff Mahoney <jeffm@suse.com> 2012-03-15 10:31:03 EDT --- (In reply to comment #8)
I don't agree with you, Jeff : kernel should keep the numlock state when it starts and we shouldn't have to read BIOS state with hwinfo to workaround it.
This is probably true, but is a separate issue. My point is that there are systems that don't have a numlock boot option and that had the ability to set the default are not longer having it honored. Also, it's a pretty unreasonable expectation to require users that want a different default to change their BIOSes. We should honor if it's set and override it if it's set in software. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c11 --- Comment #11 from Stanislav Brabec <sbrabec@suse.com> 2012-03-19 17:09:47 CET --- Bug 750326 seems to be another systemd+kbd related regression. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c12 --- Comment #12 from Stanislav Brabec <sbrabec@suse.com> 2012-03-20 17:10:08 CET --- Bug 744108 seems to be another systemd+kbd related regression. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c13 --- Comment #13 from Stanislav Brabec <sbrabec@suse.com> 2012-03-21 21:48:52 CET --- Created an attachment (id=482490) --> (http://bugzilla.novell.com/attachment.cgi?id=482490) systemd-numlock-suse.patch Here is my attempt to port numlock logic to systemd. Parsing should be OK, but it does not work. It also does not yet contain support for /var/run/numlock-on, which is used by xinitrc (/etc/X11/xinit/xinitrc.common) to set numlock in X. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c14 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #482490|0 |1 is obsolete| | --- Comment #14 from Stanislav Brabec <sbrabec@suse.com> 2012-03-22 20:06:05 CET --- Created an attachment (id=482775) --> (http://bugzilla.novell.com/attachment.cgi?id=482775) systemd-numlock-suse.patch New version of the patch that partially works: - It supports all three values (yes, no, bios) - It works on /dev/tty0. - It works in X. (It sets or deletes /var/run/numlock-on.) To do: - Further research how to do it for all consoles. - Polish the patch. - Run commands asynchronously. - Verify that /var/run already exists and is writable. - Do we want to call hwinfo or check the bios status directly? Testing packages: https://build.opensuse.org/package/show?package=systemd&project=home%3Asbrabec%3Abranches%3AopenSUSE%3A12.1%3AUpdate http://download.opensuse.org/repositories/home:/sbrabec:/branches:/openSUSE:... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c15 --- Comment #15 from Frederic Crozat <fcrozat@suse.com> 2012-03-23 13:29:50 UTC --- thanks, regarding your todo: -/var/run will always exist and is always writable : under systemd, it is a symlink to /run which is a tmpfs - I'm not fond of hwinfo call, but I really prefer we don't clutter systemd with ugly BIOS call. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c16 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #482775|0 |1 is obsolete| | --- Comment #16 from Stanislav Brabec <sbrabec@suse.com> 2012-03-23 20:11:36 CET --- Created an attachment (id=483021) --> (http://bugzilla.novell.com/attachment.cgi?id=483021) systemd-numlock-suse.patch New version of this patch sets /var/run/numlock-on in vconsole-setup.c and then it checks and call setleds in logind-seat.c: vt_allocate(). Calling setleds is postponed to login time. It seems to work as expected. If this approach is acceptable, I'll polish this patch (use exec...() instead of system(), check for errors etc.) and propose to the upstream. The seat section could even be system-wide, letting other distros chance to set /var/run/numlock-on as well. Repositories mentioned upper were updated and you can test the package. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c17 Jeff Mahoney <jeffm@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |howard.bryan.webb@gmail.com --- Comment #17 from Jeff Mahoney <jeffm@suse.com> 2012-03-25 00:26:13 EDT --- *** Bug 726698 has been marked as a duplicate of this bug. *** http://bugzilla.novell.com/show_bug.cgi?id=726698 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c18 --- Comment #18 from Frederic Crozat <fcrozat@suse.com> 2012-03-26 08:41:55 UTC --- some other comments on the patch : - it would be better to use execv rather system so calls aren't affected by PATH and so on. - I'd prefer to have the setleds seat part moved out of logind-seat (we shouldn't do any kind of fork from there) to one of udev rules shipped for seat (or create a separate rule). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c19 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #483021|0 |1 is obsolete| | --- Comment #19 from Stanislav Brabec <sbrabec@suse.com> 2012-03-27 19:35:12 CEST --- Created an attachment (id=483532) --> (http://bugzilla.novell.com/attachment.cgi?id=483532) systemd-numlock-suse.patch New (final?) version of the patch: - Code cleanup. - Move seat specific code to udev rules. - configure checks for paths of hwinfo and setleds. Discussion: Support for /run/numlock-on is now generic, its setup in vconsole-setup.c is SUSE specific. I am not sure, whether other distros would want to implement its own code. When the kernel bug will be fixed once in future, the current SUSE solution (check for /var/run/numlock-on) will not be sufficient. It would require to transfer three values to seat and X: keep, set, unset. It would require either two files: /var/run/numlock-on for force on /var/run/numlock-off to force off, or one file with contents, e. g. /var/run/numlock with contents "on", "off", "keep". First solution would be just an enhancement, the second would break the current "API". The line close(open("/run/numlock-on", O_WRONLY|O_CREAT, 0644)) has no failure check. I guess it is OK. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c20 --- Comment #20 from Frederic Crozat <fcrozat@suse.com> 2012-03-28 13:26:42 UTC --- patch looks good to me. could people who wanted the behaviour back test the package ? thanks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c21 --- Comment #21 from Wolfgang Rosenauer <wolfgang@rosenauer.org> 2012-03-29 14:07:04 UTC --- The test packages seem to work for me. My NumLock is now enabled as expected. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c22 --- Comment #22 from Stanislav Brabec <sbrabec@suse.com> 2012-04-05 16:01:08 CEST --- Frederic, will you take care of upstreaming the patch. Or should I send it somewhere? If there will be any update of systemd for 12.1, it would be nice to add it there. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c23 --- Comment #23 from Frederic Crozat <fcrozat@suse.com> 2012-04-05 14:12:28 UTC --- could you send it to systemd-devel@lists.freedesktop.org for RFC (although I'm not sure it will go in upstream) ? I'll put it for next 12.1 maintenance update, if one is needed. And I'll put it in Factory. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c24 --- Comment #24 from Stanislav Brabec <sbrabec@suse.com> 2012-04-05 17:06:46 CEST --- Well, looking at the current upstream systemd, merging this patch will make only a little sense without 0001-handle-disable_caplock-and-compose_table-and-kbd_rat.patch. Are there any plans to merge this patch upstream? Should I try to upstream it anyway? (It would require to upstream at least parts of 0001-handle-disable_caplock-and-compose_table-and-kbd_rat.patch. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c25 --- Comment #25 from Frederic Crozat <fcrozat@suse.com> 2012-04-05 15:44:41 UTC --- I already tried and it wasn't accepted, see http://lists.freedesktop.org/archives/systemd-devel/2011-August/003174.html and the replies -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c26 --- Comment #26 from Stanislav Brabec <sbrabec@suse.com> 2012-04-05 18:23:45 CEST --- OK. If I understand, upstream accepts only distro-wide features. If the feature will have a distro-wide configuration, then it could be acceptable. SUSE specific section is accepted only for parsing old-style configuration files and only for limited time. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c27 --- Comment #27 from Frederic Crozat <fcrozat@suse.com> 2012-04-05 16:29:17 UTC --- yes, upstream plans to phase out in the future (not right now) distro-specific code (which will need to be moved to distro patches). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c28 Felix Miata <mrmazda@earthlink.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |fcrozat@suse.com --- Comment #28 from Felix Miata <mrmazda@earthlink.net> 2012-04-16 14:20:55 UTC --- Could a package like Mandriva's numlock be used until either systemd can support NUM on its own or kernel devs do what they should have done decades ago? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c29 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|fcrozat@suse.com | --- Comment #29 from Frederic Crozat <fcrozat@suse.com> 2012-04-16 14:52:20 UTC --- there is no point, since Stanislav did all the work. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c30 --- Comment #30 from Stanislav Brabec <sbrabec@suse.com> 2012-04-20 17:36:48 CEST --- I just ported systemd-numlock-suse.patch to systemd-44 in openSUSE:Factory. Created request id 114873 to Base:System. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c31 --- Comment #31 from Bernhard Wiedemann <bwiedemann@suse.com> 2012-04-20 18:00:11 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/114876 Factory / systemd -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c32 --- Comment #32 from Frederic Crozat <fcrozat@suse.com> 2012-04-24 15:43:12 UTC --- I've pushed this patch in my test repository home:fcrozat:systemd for next openSUSE 12.1 maintenance update (not scheduled yet). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c33 --- Comment #33 from Felix Miata <mrmazda@earthlink.net> 2012-04-24 17:26:03 UTC --- I reported this against 12.2. I have no 12.1 systems using systemd. On host big31 reporting 12.2M3 with /etc/sysconfig/keyboard KBD_NUMLOCK="yes", I just installed systemd-44-4.1.i586 from that repo, rebooted, and all tty NUMs are off. # rpmqa systemd systemd-44-4.1.i586 systemd-presets-branding-openSUSE-0.1.0-10.1.noarch systemd-sysvinit-44-3.2.i586 I installed systemd-sysvinit from your home and tried again, but it didn't help. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c34 --- Comment #34 from Stanislav Brabec <sbrabec@suse.com> 2012-04-24 19:36:08 CEST --- Well, I tested it in 12.1, 12.2 fix is straightforward port. I just reproduced strange behavior in Factory as well: Login message in console says, that NumLock is on (you can see hint about NumLock that appears only if NumLock is On). But NumLock acts as it is Off. There is probably something broken or changed outside this patch. My patch does not do any special. If it detects that NumLock should be set to On, it just calls "setleds -D +num" before login on all particular consoles, and setting of /var/run/numlock-on needed by X. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c35 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #35 from Frederic Crozat <fcrozat@suse.com> 2012-04-25 08:39:37 UTC --- closing as fixed, the login message is independant from this fix. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c36 --- Comment #36 from Frederic Crozat <fcrozat@suse.com> 2012-05-25 11:53:16 UTC --- pushed to maintenance team : mr#122366 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c37 --- Comment #37 from Swamp Workflow Management <swamp@suse.de> 2012-05-30 14:09:00 UTC --- openSUSE-RU-2012:0664-1: An update that has four recommended fixes can now be installed. Category: recommended (low) Bug References: 727771,746594,746595,747931 CVE References: Sources used: openSUSE 12.1 (src): systemd-37-3.14.1, systemd-gtk-37-3.14.1 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c38 Felix Miata <mrmazda@earthlink.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #38 from Felix Miata <mrmazda@earthlink.net> 2012-06-11 05:50:34 UTC --- I tested this on 1 12.1 system and 10 12.2B1 systems updated from Factory repo as late as today, but in any event updated no earlier than 12.2B1 announcement. Based upon what's on the mirrors, it's definitely not fixed in Factory, and the reason why 8 of my 10 12.2 installations and all except 1 of my dozen or so 12.1 installations default to sysvinit-init boot. Systemd would get a whole lot more testing here, and a whole lot less attention paid to sysvinit bugs, if this actually got fixed. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c39 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO InfoProvider| |sbrabec@suse.com --- Comment #39 from Frederic Crozat <fcrozat@suse.com> 2012-06-13 17:08:01 UTC --- (In reply to comment #38)
and the reason why 8 of my 10 12.2 installations and all except 1 of my dozen or so 12.1 installations default to sysvinit-init boot. Systemd would get a whole lot more testing here, and a whole lot less attention paid to sysvinit bugs, if this actually got fixed.
please refrain from such comments in bugzilla in the future. I make your message more difficult to understand and doesn't give incentive to anybody on fixing bugs. Stay focused on the bug itself and nothing else. If you don't say which package of systemd is installed on your system, it is impossible to guess if you have a version which is supposed to contain the fix or not and how you are testing exactly and what is exactly broken. So far, I've tested on Factory, with systemd-44-5.1 (and on 12.1 maintenance update, they are similar) and numlock is correctly set for X session. For tty, numlock is enabled correctly but the leds itself doesn't seem to be set properly, as stanislav noted in comment34. When starting in runlevel 3 directly, it looks like the numlock is kind of set (login complains about "numlock is off") but neither keyboard nor leds are set properly. Starting X ensures leds is set properly. Stanislav, it looks like something is missing, compared to sysvinit, in the runlevel 3 setup. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c40 --- Comment #40 from Felix Miata <mrmazda@earthlink.net> 2012-06-13 18:05:54 UTC --- (In reply to comment #39)
If you don't say which package of systemd is installed on your system, it is impossible to guess if you have a version which is supposed to contain the fix or not and how you are testing exactly and what is exactly broken.
# rpm --changelog -q systemd | head ... * Fri Apr 20 2012 sbrabec@suse.cz - Implemented SUSE specific sysconfig numlock logic (bnc#746595). ... Don't you think it safe to assume when someone complains of "10 ... systems updated from Factory repo as late as" 6 weeks after a supposed fix is reported that some version containing that "fix" had been installed?
From the host that's booted ATM: # rpm -qa | grep systemd systemd-44-8.1.i586
For the record, these are the 12.2M1 hosts this bug affects (100% of my Factory installations), with their latest (all June) 'zypper dup' dates: gx28b 02 m7ncd 03 big31 06 gx280 07 gx27b 08 kt880 08 s2846 09 gx110 09 t2240 10 gx62b 11 kt400 11 gx260 11 I'm going to pass on booting the other 11 right now just to gather their systemd versions to report here. I test by booting, 39 times out of 40, directly into runlevel 3, where everything is virtually guaranteed to be big enough to be legible, unlike typical X defaults. I try to login, and get login failure because I start using numpad keystrokes for password entry before looking at the screen after the switch from tty1 to tty2 or tty3. Sorry I forgot to include the 3 tidbit in comment 0. I wouldn't know if this had any effect on runlevel 5, as I always set numlock=On in kdmrc before first instance of runlevel 5 is initiated after an installation. What exactly is broken remains as exactly stated in comment 0, as login confirms with "Hint: Num Lock is off". -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c41 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |REOPENED InfoProvider|sbrabec@suse.com | Depends on| |769423 --- Comment #41 from Stanislav Brabec <sbrabec@suse.com> 2012-06-29 17:00:29 CEST --- Regarding Factory: There seems to be a bug (race condition?) in Factory that causes misbehavior of setleds when called from systemd. Reporting this as new bug 769423. The fix above itself seems to be correct: setleds is called whenever it should be called. But setleds itself don't work as expected. X NumLock is handled in the patch above as well, but it does not depend on setleds. That is why NumLock is set properly in X, but not in console. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c42 --- Comment #42 from Stanislav Brabec <sbrabec@suse.com> 2012-07-03 13:39:30 CEST --- Looking deeper into this bug, I see several independent problems: "<$env{DEVNAME}" in the udev rule does not work any more. setleds sometimes sets only the LED (bug 769423). And finally, call of hwinfo --bios causes measurable slowdown of the boot process. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c43 --- Comment #43 from Stanislav Brabec <sbrabec@suse.com> 2012-07-03 17:01:41 CEST --- Well, I just found a fix and going to create a new patch. Problem: "<$env{DEVNAME}" in the udev rule does not work any more. Change: I wrote a simple wrapper that gets $env{DEVNAME} as an argument and passes it to setleds as stdin. I can create patch that would support device name in setleds and send it to the upstream. Problem: TAG "systemd" is no longer set. Change: Removed 'TAG=="systemd",'. After these two changes, I am not able to reproduce bug 769423. This does not mean, that such bug does not exist. It cannot appear, if NumLock is set in all X and text consoles consistently. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c44 --- Comment #44 from Stanislav Brabec <sbrabec@suse.com> 2012-07-03 20:29:21 CEST --- Created request id 127025 with above mentioned fix. It works for me. I do not see bug 769423 any more, however I think, that it still exist, it is just not exposed. 73-seat-numlock.rules is no longer generated. SETLEDS was added to config.h. configure now supports --with-udevhelperdir. Package can be tested from http://download.opensuse.org/repositories/home:/sbrabec:/branches:/Base:/Sys... soon, until the check-in to Base:System. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c45 Felix Miata <mrmazda@earthlink.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO InfoProvider| |sbrabec@suse.com --- Comment #45 from Felix Miata <mrmazda@earthlink.net> 2012-07-04 02:43:42 UTC --- (In reply to comment #44)
http://download.opensuse.org/repositories/home:/sbrabec:/branches:/Base:/Sys...
I installed these on 5 systemd-sysvinit-equipped systems zypper dup'd from Factory repos on or after 28 June. Success rate: 60%: one failure (=NUM off on all ttys from the time kernel turns off BIOS=ON setting) on a 1133MHz PIII (Dell BIOS), the other a 2.8GHz P4 (AMI BIOS). Successes were on 2 P4s and one Athlon. All use the very same IBM brand PS/2-connected keyboard. What info do I need to collect on the "failure" systems? FWIW, one it succeeded on also has KDE3 installed, and in its Konsole, keyboard repeat rate went to virtual infinity, making it nearly impossible to type only one of any character, including BS. In the other two "successes", keyboard rate in Konsole 4 is normal. On the failure, I exited KDE, deleted the home directories containing KDE settings, restarted KDE, reset my preferences, and keyboard was restored to normal behavior, even after KDE session shutdown and restart. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c46 --- Comment #46 from Bernhard Wiedemann <bwiedemann@suse.com> 2012-07-04 12:00:21 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/127074 Factory / systemd -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c47 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |REOPENED InfoProvider|sbrabec@suse.com | --- Comment #47 from Stanislav Brabec <sbrabec@suse.com> 2012-07-04 15:41:12 CEST --- Well, I see two potential problems: 1. There is no dependence between systemd-vconsole-setup.service and sys-devices-virtual-tty-tty*.device. These devices are started to be initialized just after remount-rootfs.service finishes (both remount-rootfs.service and systemd-vconsole-setup.service start at the same time, but time of its completion is not deterministic). There is a potential race condition: Important parts of systemd-vconsole-setup.service don't finish before sys-devices-virtual-tty-tty*.device and: - /run/numlock-on is set after calling setleds udev rule. - setleds is called in parallel with something else in systemd-vconsole-setup.service and 2. There is a race condition in kbd utilities and parallel calling may cause undefined behavior (one of possible explanations of bug 769423). 1. See "systemd-analyze plot >boot-numlock-failed.svg" output. There is no synchronization. virtual-tty-tty*.device should depend on systemd-vconsole-setup.service. 2. Not easy to debug or reproduce. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c48 --- Comment #48 from Stanislav Brabec <sbrabec@suse.com> 2012-07-04 18:12:06 CEST --- Ad 1. I have a (probably ugly) quick hack: Edit /lib/systemd/system/udev-trigger.service and add systemd-vconsole-setup.service to the After= list. It should completely prevent the /run/numlock-on race condition at the cost of longer boot. (Udev triggers devices enumeration much later in the boot process, after finishing of systemd-vconsole-setup.service. Its call to hwinfo --bios is the only running task in that time. Booting needs additional about 600ms on my 8 cores work machine.) I am not sure, how to delay only sys-devices-virtual-tty-tty*.device. And I even don't know, why .device events start to appear after remount-rootfs.service. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c49 --- Comment #49 from Frederic Crozat <fcrozat@suse.com> 2012-07-04 16:36:14 UTC --- you can try by adding something like: After=dev-tty1.device dev-tty2.service to systemd-vconsole-setup.service -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c50 --- Comment #50 from Felix Miata <mrmazda@earthlink.net> 2012-07-04 16:50:57 UTC --- (In reply to comment #47)
1. See "systemd-analyze plot >boot-numlock-failed.svg" output.
???: Traceback (most recent call last): File "/usr/bin/systemd-analyze", line 113, in <module> import cairo, os ImportError: No module named cairo -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c51 --- Comment #51 from Felix Miata <mrmazda@earthlink.net> 2012-07-04 16:54:24 UTC --- (In reply to comment #48)
I have a (probably ugly) quick hack: Edit /lib/systemd/system/udev-trigger.service and add systemd-vconsole-setup.service to the After= list.
Success on PIII host gx150. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c52 --- Comment #52 from Frederic Crozat <fcrozat@suse.com> 2012-07-04 17:11:24 UTC --- (In reply to comment #50)
(In reply to comment #47)
1. See "systemd-analyze plot >boot-numlock-failed.svg" output.
???: Traceback (most recent call last): File "/usr/bin/systemd-analyze", line 113, in <module> import cairo, os ImportError: No module named cairo
install python-cairo -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c53 --- Comment #53 from Felix Miata <mrmazda@earthlink.net> 2012-07-04 17:20:25 UTC --- (In reply to comment #49)
you can try by adding something like: After=dev-tty1.device dev-tty2.service
to systemd-vconsole-setup.service
the following does not succeed on 2.8GHz P4 host hs80e: # systemd-vconsole-setup.service [Unit] Description=Setup Virtual Console DefaultDependencies=no Conflicts=shutdown.target After=systemd-readahead-collect.service systemd-readahead-replay.service dev-tty1.device dev-tty2.service Before=sysinit.target shutdown.target [Service] Type=oneshot RemainAfterExit=yes ExecStart=/lib/systemd/systemd-vconsole-setup <EOF> But, comment 51 does also apply to host hs80e. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c54 --- Comment #54 from Felix Miata <mrmazda@earthlink.net> 2012-07-04 18:17:28 UTC --- Created an attachment (id=497433) --> (http://bugzilla.novell.com/attachment.cgi?id=497433) systemd-analyze plot > boot-numlock-failed.svg output (In reply to comment #47) This done after undo of comment 51 and reboot. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c55 --- Comment #55 from Stanislav Brabec <sbrabec@suse.com> 2012-07-09 15:00:18 CEST --- To comment 49: systemd-vconsole-setup.service needs to be started before dev-tty1.device dev-tty2.device, not after. I'll test to use "Before" and see, whether I need to list all 64 tty devices there. To comment 54: Well, it shows the same race as my boot SVG. On my image, tty devices are initialized in the second half of systemd-vconsole-setup.service, on your SVG they are initialized in about one third of systemd-vconsole-setup.service run. It my explain, why it succeeded on my machine but not on yours. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c56 --- Comment #56 from Stanislav Brabec <sbrabec@suse.com> 2012-07-09 16:48:00 CEST --- Adding dev-tty1.device nor sys-devices-virtual-tty-tty0.device to Before of systemd-vconsole-setup.service does not help. It is ignored, and device rules are started whenever they appear, even if systemd-vconsole-setup.service did not yet finish. (I guess that device nodes tasks are generated by udev and never postponed in respect to systemd rules.) Only comment 48 synchronizes processes correctly. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c57 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO InfoProvider| |kay.sievers@vrfy.org --- Comment #57 from Stanislav Brabec <sbrabec@suse.com> 2012-07-10 14:24:33 CEST --- Kay, is there a better way, how to ensure that systemd-vconsole-setup.service finishes before sys-devices-virtual-tty-tty*.device starts to initialize, than adding After=systemd-vconsole-setup.service to /lib/systemd/system/udev-trigger.service? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c58 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |kay.sievers@vrfy.org InfoProvider|kay.sievers@vrfy.org |fcrozat@suse.com --- Comment #58 from Stanislav Brabec <sbrabec@suse.com> 2012-07-19 15:59:11 CEST --- Frederic, do you think, that adding After=systemd-vconsole-setup.service to /lib/systemd/system/udev-trigger.service is acceptable? Is there a better solution for this race? I see no other way to delay specifically sys-devices-virtual-tty-tty*.device udev events. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c59 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |REOPENED InfoProvider|fcrozat@suse.com | --- Comment #59 from Frederic Crozat <fcrozat@suse.com> 2012-07-20 08:41:33 UTC --- I don't think there is a better way. But instead of using "After=systemd-vconsole-setup.service" in udev-trigger.service, I'd prefer "Before=udev-trigger.service" in systemd-vconsole-setup.service (it has the same meaning) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c60 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |ASSIGNED --- Comment #60 from Stanislav Brabec <sbrabec@suse.com> 2012-07-20 17:30:28 CEST --- OK. I'll submit such fix. If may fix other possible races in the console setup (e. g. setting unicode mode after login prompt appears). By the way, one of my colleagues accepted a HackWeek project that would make possible to replace slow and complex hwinfo --bios | grep "Num Lock:" by a simple utility that only checks NumLock. I guess that with that tool it would be possible to implement the feature in the way acceptable for upstream. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c61 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |tb75252@yahoo.com --- Comment #61 from Stanislav Brabec <sbrabec@suse.com> 2012-07-20 17:33:25 CEST --- *** Bug 740111 has been marked as a duplicate of this bug. *** http://bugzilla.novell.com/show_bug.cgi?id=740111 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c62 --- Comment #62 from Frederic Crozat <fcrozat@suse.com> 2012-07-20 15:38:47 UTC --- (In reply to comment #60)
By the way, one of my colleagues accepted a HackWeek project that would make possible to replace slow and complex hwinfo --bios | grep "Num Lock:" by a simple utility that only checks NumLock. I guess that with that tool it would be possible to implement the feature in the way acceptable for upstream.
Well, if he can get this code to be in the kernel directly, it could be a nice project too ;) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c63 --- Comment #63 from Felix Miata <mrmazda@earthlink.net> 2012-07-20 16:45:05 UTC --- (In reply to comment #62)
Well, if he can get this code to be in the kernel directly, it could be a nice project too ;)
Don't the kernel devs have an unofficial "thou shalt pay no mind to the BIOS NUM setting" policy? Why else hasn't the the kernel been handling this since the beginning of time? What are Mandriva & Mageia doing differently from openSUSE WRT NUM state on ttys? On those, NUM seems handled automatically, with no string "NUM" anywhere in their /etc/sysconfig/keyboards, and nothing that comes to my attention in their kernel configs. Once I've turned it on in any tty in either of those, that's where it stays across boots for that tty. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c64 --- Comment #64 from Stanislav Brabec <sbrabec@suse.com> 2012-07-20 19:48:30 CEST --- Fix submitted to: openSUSE:12.2: created maintenance request id 128495 openSUSE:Factory: created request id 128498 to Base:System. Please review. I guess you can close this bug after pushing the fix to Factory. I cannot reproduce, so I cannot verify the fix. But "systemd-analyze plot" now shows no race, so I am considering it as fixed. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c65 --- Comment #65 from Bernhard Wiedemann <bwiedemann@suse.com> 2012-07-21 23:00:09 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/128596 Factory / systemd -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c66 --- Comment #66 from Felix Miata <mrmazda@earthlink.net> 2012-08-30 04:21:51 UTC --- Current NUM pass/fail status booting with 3 on cmdline on my five 12.2 hosts that do not have sysvinit-init installed: d5000 pass 2012/08/29 600MHz Piii gx62b fail 2012/08/29 3.0GHz P4 HT hs80e pass 2012/08/29 2.93GHz C2D kt400 pass 2012/08/28 1667MHz Athlon XP+ gx150 1133MHz Piii fail right before 2012/08/29 dup, previous & subsequent passes -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c67 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO InfoProvider| |mrmazda@earthlink.net --- Comment #67 from Stanislav Brabec <sbrabec@suse.com> 2012-09-03 15:11:04 CEST --- Could you check systemd-analyze plot graph. Does systemd-vconsole-setup.service complete start before udev-trigger.service (and consequent sys-devices-virtual-tty-tty*.device) starts? Does your systemd changelog include: * Fri Jul 20 2012 sbrabec@suse.cz - Fix NumLock detection/set race condition (bnc#746595#c47). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c68 Felix Miata <mrmazda@earthlink.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED InfoProvider|mrmazda@earthlink.net | --- Comment #68 from Felix Miata <mrmazda@earthlink.net> 2012-09-03 20:57:54 UTC --- (In reply to comment #67)
Could you check systemd-analyze plot graph.
Host gx62b: ???: Traceback (most recent call last): File "/usr/bin/systemd-analyze", line 3, in <module> import dbus, sys ImportError: No module named dbus Host gx150: Does by a very small amount, less than the width of red for udev-trigger.service.
Does your systemd changelog include: * Fri Jul 20 2012 sbrabec@suse.cz - Fix NumLock detection/set race condition (bnc#746595#c47).
Host gx62b: It didn't. After installing the comment 44 package I locked it pending BS backlog reduction and hadn't unset the lock. Now with systemd-44-10.1.1 NUM is working. Host gx150: NUM is working now. Is http://lists.opensuse.org/opensuse-factory/2012-09/msg00108.html related to this racing? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c Bug 746595 depends on bug 769423, which changed state. Bug 769423 Summary: setleds +num call in the systemd only sets leds, not NumLock http://bugzilla.novell.com/show_bug.cgi?id=769423 What |Old Value |New Value ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c69 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #69 from Frederic Crozat <fcrozat@suse.com> 2013-03-11 15:39:31 UTC --- closing as fixed, based on latest comments -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c70 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Component|Kernel |Kernel Version|Factory |Final Resolution|FIXED | AssignedTo|fcrozat@suse.com |kernel-maintainers@forge.pr | |ovo.novell.com Product|openSUSE 12.2 |openSUSE 12.3 --- Comment #70 from Stanislav Brabec <sbrabec@suse.com> 2013-03-11 17:05:31 CET --- Well, it is again an issue in 12.3. The systemd-numlock-suse.patch is incorrectly ported, and it first checks for vc_kbd_numlock and then (never) parses /etc/sysconfig/keyboard. I'll prepare a fixed patch. How to reproduce: Edit your boot parameters, add 3. NumLock status is always off. Boot to X, Ctrl+Alt+F1. NumLock lights, but it behaves like NumLock off. Surprisingly NumLock is on in X. I still don't know why. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c71 --- Comment #71 from Stanislav Brabec <sbrabec@suse.com> 2013-03-28 19:31:45 CET --- I just found, that it was incorrectly ported 0001-handle-disable_caplock-and-compose_table-and-kbd_rat.patch that causes errors in systemd-numlock-suse.patch. So nor disabling CapsLock nor enabling NumLock works. I'll test and submit a fix to openSUSE:Factory and openSUSE:12.3:Update. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c72 --- Comment #72 from Stanislav Brabec <sbrabec@suse.com> 2013-03-28 20:20:22 CET --- Created OBS request id 161657 to Base:System. Created OBS maintenance request id 161656 to openSUSE:12.3:Update. I still did not research, why NumLock is on in X. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c73 --- Comment #73 from Bernhard Wiedemann <bwiedemann@suse.com> 2013-04-03 10:00:26 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/162336 Factory / systemd https://build.opensuse.org/request/show/162339 Maintenance / -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c74 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #74 from Stanislav Brabec <sbrabec@suse.com> 2013-04-03 19:06:28 CEST --- I just looked at the NumLock behavior in X (GNOME). The setting in /etc/sysconfig/keyboard interferes there with /desktop/gnome/peripherals/keyboard/remember_numlock_state GConf key, which takes precedence. There is nothing to fix, i just needs to be documented. Updating the key description to: NumLock on? ("yes" or "no" or empty or "bios" for BIOS setting) This setting may interfere with GNOME /desktop/gnome/peripherals/keyboard/remember_numlock_state GConf key. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c75 --- Comment #75 from Bernhard Wiedemann <bwiedemann@suse.com> 2013-04-03 20:00:16 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/162495 Factory / kbd -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c76 --- Comment #76 from Bernhard Wiedemann <bwiedemann@suse.com> 2013-04-04 19:00:19 CEST --- This is an autogenerated message for OBS integration: This bug (746595) was mentioned in https://build.opensuse.org/request/show/162636 Factory / kbd -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c77 --- Comment #77 from Swamp Workflow Management <swamp@suse.de> 2013-04-08 15:04:47 UTC --- openSUSE-RU-2013:0642-1: An update that has two recommended fixes can now be installed. Category: recommended (low) Bug References: 746595,809843 CVE References: Sources used: openSUSE 12.3 (src): systemd-195-13.18.1, systemd-mini-195-13.18.1 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c78 --- Comment #78 from Stanislav Brabec <sbrabec@suse.com> 2013-04-10 19:27:56 CEST --- I just created much simpler patch that does exactly the same in a many times faster way. Instead of time consuming call of hwinfo it just looks into the BIOS Data Area. According to Michal Vyskočil measurements, it may save about 10% of boot time. https://build.opensuse.org/package/show?package=systemd.openSUSE_12.3_Update&project=home%3Asbrabec%3Abranches%3AopenSUSE%3A12.3%3AUpdate If you plan to create an update for 12.3, you can pick it. (Maybe it could make sense even to release it stand-alone.) Now I will think about an upstream-able form of this patch. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c79 --- Comment #79 from Stanislav Brabec <sbrabec@suse.com> 2013-04-12 19:09:15 CEST --- And here is an upstream friendly version of this patch: http://lists.freedesktop.org/archives/systemd-devel/2013-April/010368.html (I also have a sysconfig support patch build on top of this one.) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c80 Stanislav Brabec <sbrabec@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |crrodriguez@opensuse.org --- Comment #80 from Stanislav Brabec <sbrabec@suse.com> 2013-04-15 17:28:44 CEST --- Upstream rejected this type of features, but Cristian Rodríguez provided a useful improvement of my code. I'll merge Cristian's fixes with the downstream addition (sysconfig support) and create a new SUSE specific patch. It will exist until we switch to the new implementation of consoles. Not calling external applications would make this working much faster. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c81 --- Comment #81 from Stanislav Brabec <sbrabec@suse.com> 2013-04-15 17:50:55 CEST --- Regarding the rejection: Lennart Poettering: http://lists.freedesktop.org/archives/systemd-devel/2013-April/010390.html Kay Sievers: http://lists.freedesktop.org/archives/systemd-devel/2013-April/010384.html Regarding the fix by Cristian Rodríguez: objections: http://lists.freedesktop.org/archives/systemd-devel/2013-April/010372.html patch: http://lists.freedesktop.org/archives/systemd-devel/2013-April/010374.html -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c82 --- Comment #82 from Swamp Workflow Management <swamp@suse.de> 2013-06-10 09:06:37 UTC --- openSUSE-RU-2013:0877-1: An update that has 8 recommended fixes can now be installed. Category: recommended (moderate) Bug References: 746595,789589,801878,808319,809646,809695,809820,812874 CVE References: Sources used: openSUSE 12.3 (src): systemd-195-13.25.1, systemd-mini-195-13.25.1 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=746595 https://bugzilla.novell.com/show_bug.cgi?id=746595#c83 --- Comment #83 from Stanislav Brabec <sbrabec@suse.com> 2013-06-13 11:36:03 CEST --- I just finished the new version of this patch: - It does not call hwinfo, and uses direct BIOS memory lookup (It can shorten boot time by 10% on some systems.) - It does not call setleds, the udev helper itself sets the state. Created request id 178889 to Base:System. I have also prepared backport for 12.3: https://build.opensuse.org/package/show?package=systemd.openSUSE_12.3_Update&project=home%3Asbrabec%3Abranches%3AopenSUSE%3A12.3%3AUpdate If you have any planned systemd maintenance updates, feel free to pick-up my package. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com