Description of problem: I installed 4.4.32 recently (automatic update). After reboot, all input devices of my Dell Precision M4800 failed. No trackpad, no trackpoint, no keyboard. Tried USB devices (mouse and keyboard) without success. No way to login. Rebooting with 4.4.30 solved the problem. Systemd logs do not contain related info. Version-Release number of selected component (if applicable): 4.4.32-desktop-1.mga5 How reproducible: Every time I select 4.4.32 Steps to Reproduce: 1. Boot with default GRUB entry 2. 3.
Hm, on a quick look nothing stands out between 4.4.30 and 4.4.32 Can you boot with 4.4.32 and wait some 30 sec after its fully booted to login screen. Then rebooot into 4.4.30, open a terminal and as root do: lspcidrake -v >lspcidrake.log journalctl -b >boot-4.4.30.log journalctl -b -1 >boot-4.4.32.log and _attach_ the 3 logfiles created to this bugreport.
CC: (none) => tmb
Keywords: (none) => NEEDINFOCC: (none) => marja11Assignee: bugsquad => kernel
Created attachment 8726 [details] Required logfiles I don't know if it's related but 4.4.32 boots quicker than 4.4.30
Interestingly 4.4.32 seems to detect your hw better as on 4.4.30 you have a lot of lines like: local4.local kernel: pci 0000:04:03.0: BAR 14: no space for [mem size 0x00200000] local4.local kernel: pci 0000:04:03.0: BAR 14: failed to assign [mem size 0x00200000] that are properly detected as resources in 4.4.32 Can you also try with 4.4.36-1 that is available in core updates_testing ? also please provide output of lsusb lsusb -t
Created attachment 8728 [details] Required lsusb files
Created attachment 8730 [details] 2 logfiles for 4.4.35. One good and 1 bad I installed 4.4.35-2. It's the latest I found in "testing". The first time I booted 4.4.35, it worked fine. A log is attached. All subsequent boots failed after login. System freezed between login and desktop appearance. No disk activity. Through SSH, I found that CPU was quiet. I've attached a log of a failed boot. I'm back to 4.4.30 I need guidance to diagnose the problem.
CC: (none) => gallard
(In reply to Gilles Allard from comment #5) > Created attachment 8730 [details] > 2 logfiles for 4.4.35. One good and 1 bad > Something went wrong here... ther OK log is from a 4.1.15 kernel. Anyway, can you try with the new 4.4.36-2 in testing ? If you dont find it on your local mirror, grab it from here: http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/core/updates_testing/kernel-desktop-4.4.36-2.mga5-1-1.mga5.x86_64.rpm and if you use any dkms packages you need the -devel package too: http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/core/updates_testing/kernel-desktop-devel-4.4.36-2.mga5-1-1.mga5.x86_64.rpm
Also does it make any difference between cold boot and warm boot ? (cold boot = start from computer completely off) (warm boot = reboot from working kernel)
Created attachment 8739 [details] Logfile of 4.4.36 boot 4.4.36-2 is similar to 4.4.35-2: the display freeze after login. Access through SSH shows that X is running and systemd has completed the boot. Logfile contain some stack traces at the end but I think they are unrelated to my problem. Cold_boot and warm_boot are identical. Next step?
Can you try kernel-linus-4.4.32-1.mga5 available in updates to confirm its an upstream issue (or not)
Oh, and the stack straces is an aftereffect of this: déc 07 09:11:39 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 10020msec déc 07 09:11:39 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:39 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 10521msec déc 07 09:11:39 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:40 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 11022msec déc 07 09:11:40 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:40 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 11523msec déc 07 09:11:40 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:41 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 12024msec déc 07 09:11:41 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:41 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 12525msec déc 07 09:11:41 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3) déc 07 09:11:42 local4.local kernel: radeon 0000:01:00.0: ring 3 stalled for more than 13026msec déc 07 09:11:42 local4.local kernel: radeon 0000:01:00.0: GPU lockup (current fence id 0x000000000000005e last fence id 0x000000000000005f on ring 3)
And I missed those in the 4.4.32 logs on first check... Now there is only a few gpu patches between 4.4.30 and 4.4.32 so it should hopefully not be hard to check wich one triggers
Can you try 2.1, 2.2 and 2.3 from: ftp://ftp.free.fr/mirrors/mageia.org/people/tmb/5/bugs/19892/ and see if any of them locks up ?
also, on 4.4.36-2 try to add radeon.runpm=0 to kernel command line, does that change anything?
Created attachment 8752 [details] log of 4.4.36 boot with kernel parm: radeon.runpm=0 2.1 seems OK (similar to 4.4.30-2) 2.2 freeze after login (tried twice) 2.3 freeze after login (tried once) with radeon kernel parm, we have something different: - after login, the desktop is displayed - no mouse (neither touchpad, nor trackpoint, nor external USB - no keyboard (internal) - no sshd A logfile is attached. For your knowledge, a Dell Precision M4800 is a weird machine. It has 2 mice (touchpad and trackpoint) and 2 graphic chipset (Intel G4 and Radeon 8870M).
ok, so here is a fully updated 4.4.38 with a "fix" for this, try: ftp://ftp.free.fr/mirrors/mageia.org/people/tmb/5/bugs/19892/kernel-desktop-4.4.38-1.1.mga5-1-1.mga5.x86_64.rpm
4.4.38-1 seem to work fine. Thank you very much Thomas. May I know what was the problem (with some details)? Will the fix be included in mainstream kernel? Regards Gilles Allard
The bug was introduced by another "possible stability fix" in 4.4.31. But as that caused regression on your hw, and fixing it would need backporting of several upstream changes, some of wich could cause other regressions in the 4.4 branch. So the "fix" I did was to revert the commit introducing this problem, as it's better to stay on "known good and working" code and keep everyone happy. The fix is now in the 4.4.39-1 kernel released as: http://advisories.mageia.org/MGASA-2016-0429.html
Status: NEW => RESOLVEDResolution: (none) => FIXED
Upstream has now fixed the regression differently... Can you test 4.4.45-1.mga5 currently in updates_testing that it still works for you
Hi Thomas! Where should be 4.4.45? I can't find it. "Core Update Testing" is enabled
Can't find 4.4.45 but I tried 4.4.50 and it works well on my Dell Precision M4800. Fixed and resolved.
Great, I just wanted to verify that it still works for you...