I have an Nvidia GeForce GTX 880M, which nouveau does not play with nicely. I have installed nouveau-fw, so the proprietary firmware is available. When I boot my laptop, I get these errors: nouveau E[ PGRAPH][0000:01:00.0] grctx template channel unload timeout nouveau E[ PGRAPH][0000:01:00.0] failed to construct context nouveau E[ PGRAPH][0000:01:00.0] init failed, -16 Operating system: Chakra OS (64-bit) Relevant packages: - linux-ck 4.1.6 - libdrm 2.4.64 - mesa 10.6.4 - mesa-vdpau 10.6.4 - opencl-mesa 10.6.4 - libva-mesa-driver 10.6.4 - xf86-video-nouveau 1.0.11 - nouveau-fw 340.32 I will also post my dmesg.log, Xorg.0.log and mmiotrace.log.xz. Many thanks in advance. :)
Created attachment 117849 [details] dmesg.log without patches
Created attachment 117850 [details] Xorg.0.log from nvidia driver
(In reply to kyle.devir from comment #0) > I have an Nvidia GeForce GTX 880M, which nouveau does not play with nicely. Please try the following patch: http://cgit.freedesktop.org/~darktama/nouveau/commit/?h=hack-gk106m&id=64b75967fe13499f12876ec10a31d9141e828714 Make sure to use nouveau.runpm=0
My mmiotrace.log.xz is 5.6 MiB... I will have to split it, I guess, because I have already compressed it with "xz -9".
Created attachment 117851 [details] mmiotrace.log.xz-part-aa from nvidia driver
Created attachment 117852 [details] mmiotrace.log.xz-part-ab from nvidia driver
Created attachment 117853 [details] mmiotrace.log.xz-part-ac from nvidia driver
Thanks Ilia. I will test it out, and get back to you.
Created attachment 117854 [details] dmesg.txt with c800 nouveau patch The patch did nothing except turn off the screen, for some reason. With some careful typing, I saved the dmesg output.
Are there any other patches that you could recommend trying?
We have a downstream reporter with what looks like much the same issue: https://bugzilla.redhat.com/show_bug.cgi?id=1264872 Sep 23 20:06:31 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] wait for idle timeout (en: 1, ctxsw: 0, busy: 1) Sep 23 20:06:33 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] wait for idle timeout (en: 1, ctxsw: 0, busy: 1) Sep 23 20:06:35 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] wait for idle timeout (en: 1, ctxsw: 0, busy: 1) Sep 23 20:06:37 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] wait for idle timeout (en: 1, ctxsw: 0, busy: 1) Sep 23 20:06:39 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] grctx template channel unload timeout Sep 23 20:06:39 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] failed to construct context Sep 23 20:06:39 w50.localdomain kernel: nouveau E[ PGRAPH][0000:01:00.0] init failed, -16 Sep 23 20:06:39 w50.localdomain gnome-session[1428]: nvc0_screen_create:716 - Error allocating PGRAPH context for M2MF: -16 Adapter there is: 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK106GLM [Quadro K2100M] [10de:11fc] (rev a1)
Ben pushed an improved version of that c800 patch into 4.3-rc -- http://cgit.freedesktop.org/nouveau/linux-2.6/commit/?h=linux-4.3&id=205877f9156daebb975fb46205488da6fdf5b3f5 and enabled it by default on the Tecra W50: http://cgit.freedesktop.org/nouveau/linux-2.6/commit/?h=linux-4.3&id=778613e583994d947960a7f177226d0e7b5ef124 If you have different hw, you can try it out by booting with nouveau.config=War00C800_0=1
The patch kind of works. A full Plasma session on Xorg did not work. Screen updates froze, updating every full minute, or so. From that observation, the desktop did work. With some luck, I opened an application, which only showed up after a long while. Also, I was able to start Weston, but there were minor visual glitches with applications I started on Xwayland. I was also able to start kwin_wayland with plasmashell running on Xwayland, but plasmashell soon crashed.
afaik plasmashell has its own independent problems with nouveau, so not a great thing to start testing with :)
Created attachment 118419 [details] dmesg.log with 205877f9156daebb975fb46205488da6fdf5b3f5 patch Here is my dmesg.log from use of the updated patch. I used nouveau.config=War00C800_0=1 and nouveau.runpm=0 as my kernel options. My software versions are still the same.
(In reply to Ilia Mirkin from comment #14) > afaik plasmashell has its own independent problems with nouveau, so not a > great thing to start testing with :) That was a shockingly fast reply! Thanks, Ilia. :) So... shall I continue testing with Weston and Xwayland, then? What issues does plasmashell have with Nouveau? Where should I start looking?
(In reply to kyle.devir from comment #16) > (In reply to Ilia Mirkin from comment #14) > > afaik plasmashell has its own independent problems with nouveau, so not a > > great thing to start testing with :) > > That was a shockingly fast reply! Thanks, Ilia. :) > > So... shall I continue testing with Weston and Xwayland, then? What issues > does plasmashell have with Nouveau? Where should I start looking? https://bugs.freedesktop.org/show_bug.cgi?id=92077
Ah, thanks, Ilia! I will keep track of it, then. :)
Created attachment 118437 [details] journal log -b-1 failed boot I am the user Adam Williamson is referring to above with the Toshiba w50 Trying Fedora 23 beta with all installed updates. I tried with kernel options: noveau.config=War00c800_0=1 and: noveau.config=War00c800_0=1 nouveau.runpm=0 It still fails to boot.
Linux version 4.2.1-300.fc23.x86_64 Don't know what fedora patches are in there, but the upstream 4.2.1 doesn't have the nouveau logic in question. You need 4.3-rc1 (or rc2) to get it.
I have tested and can confirm that using rawhide kernel 4.3.0-0.rc2.git1.2.fc24.x86_64 fixes the issues for me and boots using default options. Thanks.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/1081.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.