[Bug 824442] New: 60 s delay when starting plymouth without framebuffer due to udevadm settle.
https://bugzilla.novell.com/show_bug.cgi?id=824442 https://bugzilla.novell.com/show_bug.cgi?id=824442#c0 Summary: 60 s delay when starting plymouth without framebuffer due to udevadm settle. Classification: openSUSE Product: openSUSE 12.3 Version: Final Platform: Other OS/Version: openSUSE 12.3 Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: arvidjaar@gmail.com QAContact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:21.0) Gecko/20100101 Firefox/21.0 On system without framebuffer active starting plymouth results in 60 seconds timeout due to two udevadm settle calls. To reproduce - boot in run level with framebuffer disabled - try to change run level. Even to the same - "init 3" Result: Jun 11 21:22:27 opensuse.site systemd[1]: Trying to enqueue job multi-user.target/start/isolate ... Jun 11 21:22:28 opensuse.site systemd[1]: Installed new job plymouth-quit.service/start as 1127 ... Jun 11 21:22:58 opensuse.site systemd[1]: Got SIGCHLD for process 1755 (udevadm) Jun 11 21:22:58 opensuse.site systemd[1]: Child 1755 died (code=exited, status=1/FAILURE) Jun 11 21:22:58 opensuse.site systemd[1]: Child 1755 belongs to plymouth-start.service ... Jun 11 21:22:58 opensuse.site systemd[1]: plymouth-start.service: control process exited, code=exited status=1 Jun 11 21:22:58 opensuse.site systemd[1]: plymouth-start.service running next control command for state start-post Jun 11 21:22:58 opensuse.site systemd[1]: About to execute /sbin/udevadm settle --timeout=30 --exit-if-exists=/sys/class/graphics/fb0/dev Jun 11 21:22:58 opensuse.site systemd[1]: Forked /sbin/udevadm as 1774 ... Jun 11 21:23:28 opensuse.site systemd[1]: Received SIGCHLD from PID 1774 (udevadm). Jun 11 21:23:28 opensuse.site systemd[1]: Got SIGCHLD for process 1774 (udevadm) Jun 11 21:23:28 opensuse.site systemd[1]: Child 1774 died (code=exited, status=1/FAILURE) Reproducible: Always Steps to Reproduce: 1. 2. 3. -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c1
Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c2
--- Comment #2 from Andrey Borzenkov
Hmm, we should not try to restart plymouth-start on failure, when switching runlevel.
It is not limited to switching run levels; the same problem on boot as well if plymouth was not started in initrd.
How did you disable framebuffer ? nomodeset on kernel command line ?
Simply go to yast bootloader and disable "Use graphical console". I do it by habit as in the past using nVidia driver with framebuffer was asking for troubles. -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c3
--- Comment #3 from Frederic Crozat
(In reply to comment #1)
Hmm, we should not try to restart plymouth-start on failure, when switching runlevel.
It is not limited to switching run levels; the same problem on boot as well if plymouth was not started in initrd.
Starting plymouth in standard boot, if not started in initrd, is expected.
How did you disable framebuffer ? nomodeset on kernel command line ?
Simply go to yast bootloader and disable "Use graphical console". I do it by habit as in the past using nVidia driver with framebuffer was asking for troubles.
What is the end result in your case ? could you paste /proc/cmdline ? -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c4
--- Comment #4 from Andrey Borzenkov
Starting plymouth in standard boot, if not started in initrd, is expected.
Of course, I mean it will have the same artificial 60 seconds delay.
What is the end result in your case ? could you paste /proc/cmdline ?
it does not change kernel cmdline, it changes how bootloader sets up screen. BOOT_IMAGE=/vmlinuz-3.7.10-1.11-desktop root=/dev/mapper/system-root resume=/dev/system/swap splash=silent quiet showopts rd.plymouth=0 3 -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c5
--- Comment #5 from Frederic Crozat
(In reply to comment #3)
Starting plymouth in standard boot, if not started in initrd, is expected.
Of course, I mean it will have the same artificial 60 seconds delay.
Do you still have the issue with yesterday package (when switching between runlevel) ? There was an issue (fixed) which was not ignoring udevadm error code in ExecStartPost.
What is the end result in your case ? could you paste /proc/cmdline ?
it does not change kernel cmdline, it changes how bootloader sets up screen.
BOOT_IMAGE=/vmlinuz-3.7.10-1.11-desktop root=/dev/mapper/system-root resume=/dev/system/swap splash=silent quiet showopts rd.plymouth=0 3
But this shouldn't have an influence on plymouth. Plymouth relies on the kernel framebuffer (or KMS), not on bootloader graphical console. Or it is specific to nvidia proprietary driver.. -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c6
--- Comment #6 from Andrey Borzenkov
Do you still have the issue with yesterday package (when switching between runlevel) ?
Yes.
There was an issue (fixed) which was not ignoring udevadm error code in ExecStartPost.
The problem is not exit code but unconditional waiting for non-existent device.
What is the end result in your case ? could you paste /proc/cmdline ?
it does not change kernel cmdline, it changes how bootloader sets up screen.
BOOT_IMAGE=/vmlinuz-3.7.10-1.11-desktop root=/dev/mapper/system-root resume=/dev/system/swap splash=silent quiet showopts rd.plymouth=0 3
But this shouldn't have an influence on plymouth. Plymouth relies on the kernel framebuffer (or KMS), not on bootloader graphical console.
Of course, but if KMS is disabled kernel defaults to pure text unless either vga= or video= parameter is given or bootloader sets up required video mode before loading kernel. In my case bootloader starts in pure text mode which is then inherited by kernel. And KMS drivers are disabled. It is not directly related to having nvidia. -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c7
Frederic Crozat
(In reply to comment #5)
Do you still have the issue with yesterday package (when switching between runlevel) ?
Yes.
There was an issue (fixed) which was not ignoring udevadm error code in ExecStartPost.
The problem is not exit code but unconditional waiting for non-existent device.
I don't know how we can differentiate between "driver is not loaded, so we wait until it is there, after asking udevadm" and "driver is somehow blacklisted, let's not wait". Robert, any idea ? -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c8
Robert Milasan
From udev point of view, I don't see anything wrong, wait 30 seconds if /sys/class/drm/card0/dev doesn't exist and wait again 30 seconds if /sys/class/graphics/fb0/dev if again it doesn't exist.
Is it possible to move this to systemd somehow to check for /sys/class/graphics/fb0/dev and/or /sys/class/drm/card0/dev and just add: udevadm settle --timeout=30 ? or not use udevadm settle at all in plymouth-start.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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c9
Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c10
--- Comment #10 from Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c
Ye Yuan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c11
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c12
--- Comment #12 from Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c13
--- Comment #13 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c14
--- Comment #14 from Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c15
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c16
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c17
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c18
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c19
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c20
--- Comment #20 from Andrey Borzenkov
Andrey, can you point me to that thread form systemd-devel mailing list?
http://lists.freedesktop.org/archives/systemd-devel/2013-July/011826.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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c21
Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c22
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c23
--- Comment #23 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c24
--- Comment #24 from Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c25
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c26
Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c27
--- Comment #27 from Andrey Borzenkov
I honestly don't know whats the issue and can't reproduce it.
OK, I finally was able to determine what triggers it on my system. It is libvirtd.service (libvirt RPM). Disabling it reliably "fixes" the issue with missing uevent. I tried to compare what happens without and with it loaded. For all I can tell the difference is (apart from obvious reordering of events between boots): +KERNEL add /module/x_tables (module) +KERNEL add /module/ebtables (module) +KERNEL add /module/ebtable_nat (module) +KERNEL add /module/ip_tables (module) +KERNEL add /module/iptable_filter (module) +KERNEL add /module/ip6_tables (module) +KERNEL add /module/ip6table_filter (module) KERNEL change /devices/platform/regulatory.0 (platform) +KERNEL add /devices/virtual/net/lo/queues/rx-0 (queues) +KERNEL add /devices/virtual/net/lo/queues/tx-0 (queues) +KERNEL remove /devices/virtual/net/lo/queues/rx-0 (queues) +KERNEL remove /devices/virtual/net/lo/queues/tx-0 (queues) Here additional events are with libvirtd loaded. Events for +KERNEL add /devices/virtual/net/lo/queues/rx-0 (queues) +KERNEL add /devices/virtual/net/lo/queues/tx-0 (queues) are duplicates - there already are the same events before. But without libvirtd there are no remove events for queues. I do not mean to say libvirtd is at fault here but I'm really curious whether someone else can reproduce it. I attach events without and with libvirtd for reference. -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c28
--- Comment #28 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c29
--- Comment #29 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c30
--- Comment #30 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c31
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c32
--- Comment #32 from Andrey Borzenkov
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c33
James Fehlig
Jim, do you know or heard something about this? Libvirtd issues with kernel and/or udev?
Only what Andrey pointed to in #20. And after re-reading the thread it appears there is no resolution. But that particular issue was regarding events generated in a network namespace. According to #30, that is not the case here right? -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c34
--- Comment #34 from Andrey Borzenkov
But that particular issue was regarding events generated in a network namespace. According to #30, that is not the case here right?
During startup libvirtd creates additional network namespaces: 1817 clone(child_stack=0xb8534884, flags=CLONE_NEWNS|CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWNET|SIGCHLD) = 1862 -- 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=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c35
James Fehlig
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c36
--- Comment #36 from Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c37
Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c38
--- Comment #38 from Frederic Crozat
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c39
--- Comment #39 from Robert Milasan
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c40
Raymond Wooninck
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c41
Mark van Tinteren
https://bugzilla.novell.com/show_bug.cgi?id=824442
https://bugzilla.novell.com/show_bug.cgi?id=824442#c42
--- Comment #42 from Andrey Borzenkov
http://bugzilla.novell.com/show_bug.cgi?id=824442
http://bugzilla.novell.com/show_bug.cgi?id=824442#c43
Robert Milasan
participants (1)
-
bugzilla_noreply@novell.com