[Bug 889660] New: nVidia proprietary driver does not consistently resume from sleep/suspend.
https://bugzilla.novell.com/show_bug.cgi?id=889660 https://bugzilla.novell.com/show_bug.cgi?id=889660#c0 Summary: nVidia proprietary driver does not consistently resume from sleep/suspend. Classification: openSUSE Product: openSUSE 13.1 Version: Final Platform: All OS/Version: openSUSE 13.1 Status: NEW Severity: Critical Priority: P5 - None Component: Kernel AssignedTo: kernel-maintainers@forge.provo.novell.com ReportedBy: gordon@dickens.com QAContact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Firefox/31.0 Resume from sleep/suspend does not work consistently with the nVidia proprietary device drivers. I have experienced this problem with the OpenSUSE packaged G02 and G03 drivers up through G02 version 304.121 and G03 version 331.89 and with several of the stock OpenSUSE kernels. With the OpenSUSE packaged drivers and stock kernels, resume from sleep/suspend will fail 25% to 50% of the time. Most recently, I have installed the latest nVidia driver version 340.24 from the nVidia website and the latest kernel version 3.15.6-2.gedc5ddf-desktop from the kernel-stable repo. With this more recent driver/kernel combo, resume from sleep/suspend fails approximately 10% of the time. So, the problem is not gone but it occurs less often with the latest driver and kernel. Also, the failure rate appears higher if I am running the kcometen4 screen saver and the failure rate appears lower if I am not running a screen saver but am using the simple locker instead. I am running OpenSUSE 13.1, KDE 4.13.3 and a nVidia Quadro FX 3600M graphics card on a Dell Precision 6300 laptop. This problem has been reported in the OpenSUSE and KDE forums here: https://forums.opensuse.org/showthread.php/494706-openSUSE-13-1-KDE-inconsis... https://forum.kde.org/viewtopic.php?f=111&t=121590 A very similar problem was also reported as a Ubuntu bug which has purportedly been fixed here: https://bugs.launchpad.net/ubuntu/+source/nvidia-graphics-drivers-331/+bug/1... Reproducible: Sometimes Steps to Reproduce: 1. Put the computer in the sleep/suspend state. 2. Bring the computer out of the sleep/suspend state (Resume). Actual Results: 10% to 50% of the time resume from sleep/suspend will fail and the screen will be garbled and frozen. Expected Results: The computer should resume from sleep/suspend without a garbled and frozen display. -- 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=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c1
Takashi Iwai
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c2
Stefan Dirsch
The openSUSE 13.1 kernel already contains the framework for VT-switchless resume. The rest is to patch Nvidia binary-only stuff.
Not sure which patches you mean, Takashi. Adding NVIDIA here. -- 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=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c
Stefan Dirsch
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c3
Takashi Iwai
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c4
Stefan Dirsch
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c5
--- Comment #5 from Daniel Dadap
Daniel, is this something you're planning to add to the driver?
In fact, I provided that very patch to Canonical when I made that change to the NVIDIA driver, so that they could patch their packages for existing releases while waiting for the next release which contained the change that I made. :) 331.67, and all subsequent versions of the NVIDIA Linux driver, including the 340.24 driver mentioned in the bug description, already include this change. To clarify the issue tracked in that launchpad bug: the problem was that the kernel was skipping VT switches on PRIME configurations; the NVIDIA driver relies on VT switches for proper restoration of X after suspend/resume, which was causing corrupted/frozen displays on PRIME configurations. The reason this was happening on PRIME configurations was because the new code in the kernel to optionally skip VT switches will preserve the old behavior of always VT switching if no driver explicitly indicates to the kernel, via pm_vt_switch_required(), that either does or does not require VT switches. If *any* driver calls pm_vt_switch_required(), and all drivers which do this indicate that no VT switch is required, the kernel will skip VT switches. This was the case under PRIME with older NVIDIA drivers, since the NVIDIA driver did not register with pm_vt_switch_required(), and the Intel driver registered that VT switches were not required. Anyway, based on a skim of the openSUSE forum thread, I doubt this is related to the launchpad bug, since it doesn't seem like those users are using PRIME, and they're still reporting problems with drivers that contain the fix for the pm_vt_switch_required() issue. Unfortunately, the class of "failure to properly resume from suspend" issues is a somewhat broad one, so although the symptom reported on the openSUSE forums is quite similar to the one in the launchpad bug, their root causes appear to be disjoint. Based on the description of the symptoms in the forum threads, it is possible that this is due to improper FBO reuse after a modeswitch: the contents of FBOs are not guaranteed to be preserved across modeswitch events, which are incurred on suspend/resume, among other situations. This has resulted in similar bugs in other composite managers and GL applications, where screen contents were corrupted after suspend/resume. However, it's also possible that it's due to a bug in the NVIDIA driver. I've filed NVIDIA bug number 1540794 to track this issue: we'll attempt to reproduce the problem internally, and determine whether or not it appears to be a driver bug. There is a recently fixed NVIDIA driver bug (the fix is not yet in any released driver, but some soon-to-be-released drivers will include it) involving FBO refreshes in some situations, but I do not believe it is related to this issue, since that bug was a regression in the 337 driver series: this bug is being reported on 304 and 331. -- 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=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c6
Stefan Dirsch
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c7
--- Comment #7 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c8
--- Comment #8 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c9
--- Comment #9 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c10
Bazon Bloch
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c11
--- Comment #11 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c12
--- Comment #12 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c13
Oliver Neukum
Gordon Dickens said:
I had this issue, too (on arch linux with 3.15.8 kernel and Nvidia binary driver), and moving to the older lts-kernel 3.14.15 worked: No more screen errors after resume or the other method (see below). So for me, it seems really to be an kernel issue.
In this case please bisect the kernel. Somebody who can reproduce the issue needs to do that. -- 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=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c14
--- Comment #14 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c15
--- Comment #15 from Gordon Dickens
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c16
--- Comment #16 from Bazon Bloch
https://bugzilla.novell.com/show_bug.cgi?id=889660
https://bugzilla.novell.com/show_bug.cgi?id=889660#c17
Stefan Dirsch
participants (1)
-
bugzilla_noreply@novell.com