| tjaalton | ffs, intel | 06:34 |
|---|---|---|
| tjaalton | https://bugs.freedesktop.org/show_bug.cgi?id=94161 | 06:34 |
| ubottu | Freedesktop bug 94161 in DRM/Intel "[skl rc6] GPU HANG" [Blocker,New] | 06:34 |
| tjaalton | so, disable rc6 it is then | 06:34 |
| ricotz | that is the one :( | 06:36 |
| tjaalton | is there a bug on lp? | 06:39 |
| tjaalton | or bugs | 06:39 |
| darkxst | I still can't get modesetting to work at boot with nvidia-364 ;( | 06:42 |
| ricotz | tjaalton, not sure, just saying I experienced this bug until disabling rc6 | 06:43 |
| tjaalton | mind filing one? | 06:43 |
| tjaalton | against the kernel | 06:43 |
| darkxst | tjaalton, speaking of intel bugs, have you seen this one elsewhere, bug 1548864 (seen comments #12 + #13) | 06:46 |
| ubottu | bug 1548864 in casper (Ubuntu) "Ubuntu GNOME Xenial live image boots to tty1 and demands password" [Undecided,Incomplete] https://launchpad.net/bugs/1548864 | 06:46 |
| ricotz | tjaalton, https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1547800 | 06:46 |
| ubottu | Launchpad bug 1547800 in xserver-xorg-video-intel (Ubuntu) "gpu stuck on render ring" [Medium,Incomplete] | 06:46 |
| darkxst | (the actual bug though seems to be a race condition unrelated to DRI failure) | 06:47 |
| darkxst | s/bug/bug report/ | 06:47 |
| tjaalton | ricotz: that's haswell | 06:49 |
| tjaalton | darkxst: doesn't look familiar | 06:50 |
| darkxst | not sure how widespread it is, most ISO's work fine, apart from the obvious laggy UI from gnome-shell uner llvmpipe. I can't reproduce it on my same generation (I think) intel laptop | 06:53 |
| darkxst | not to mention there is a bug probably in nvidia_drm preventing plymouth from picking up the current encoder | 09:36 |
| darkxst | I tried to trace it though libdrm, but it seems to just be a shim/wrapper around the kernel interface | 09:37 |
| darkxst | not worth working around at this point, and I have no idea how one would actually file a bug report with nvidia! | 09:38 |
| mamarley | darkxst: Make a topic on the NVIDIA forum or send mail to linux-bugs@nvidia.com. | 09:40 |
| mamarley | darkxst: Also, not sure if you heard, but DRM KMS isn't enabled by default in the Ubuntu NVIDIA packages because it breaks PRIME (and SLI). | 09:41 |
| darkxst | mamarley, ok will do | 09:41 |
| darkxst | and yes aware of that, but I can't get it working with kernel options either | 09:41 |
| darkxst | do the alias' get copied into initramfs? | 09:42 |
| darkxst | it did work with my super hacked, initramfs | 09:42 |
| mamarley | darkxst: Did you change "options nvidia_364_drm modeset=0" to have a 1 in /etc/modprobe.d/nvidia-graphics-drivers.conf and then run "sudo initramfs-update -u"? | 09:43 |
| darkxst | mamarley, I commented that out, and then set nvidia-drm.modeset=1 | 09:43 |
| mamarley | darkxst: On the kernel command line? I don't think that obeys the aliases. | 09:44 |
| darkxst | (on kernel command line) | 09:44 |
| mamarley | I know that "options modulename" and "modinfo" don't obey the aliases. | 09:44 |
| darkxst | mamarley, ok so I probably need to use nvidia-364-drm or so on the kernel? | 09:45 |
| mamarley | darkxst: Or just change the 0 to a 1 in nvidia-graphics-drivers.conf. | 09:46 |
| darkxst | mamarley, I don't want it in the initramfs, its not ready yet, mutter crashes also | 09:46 |
| mamarley | darkxst: That won't put the module in the initramfs, just the configuration for it. (I think.) | 09:47 |
| darkxst | mamarley, the module is in there now, mutter won't run with modeset=1 | 09:48 |
| mamarley | OK, well it probably won't run with that on the kernel command line either. | 09:48 |
| darkxst | although not sure the main nvidia-364.ko needs to be in the initramfs | 09:48 |
| darkxst | drm/modeset should be enough | 09:49 |
| darkxst | I think tseliot did that, but here is not here right now ofc! | 09:49 |
| mamarley | I don't know about that, sorry. | 09:51 |
| tjaalton | he's off this week | 09:51 |
| darkxst | mamarley, I don't expect it to run, which is why I dont want the settings in initramfs | 09:51 |
| mamarley | In that case just leave everything at the defaults and you won't get modeset. | 09:52 |
| darkxst | but I need it for testing!@ | 09:52 |
| * mamarley confoozled. | 09:52 | |
| mamarley | You want to temporarily boot the system with modesetting enabled? | 09:53 |
| darkxst | mamarley, yes exactly | 09:53 |
| mamarley | In that case, passing "nvidia_364_drm.modeset=1" to the kernel should work. | 09:53 |
| mamarley | But I haven't tried it. | 09:53 |
| darkxst | ok | 09:54 |
| darkxst | plymouth should work, but apart from the pre-mentioned bug | 09:55 |
| darkxst | mutter, just spits the dummy, but I don't think it should be failing where it is (i.e. there are no gbm buffers on X11) | 09:56 |
| mamarley | Beats me, sorry. I just use it like normal without Plymouth and with KDE and it works fine. | 09:58 |
| darkxst | erm, unless I was running wayland session at the time | 09:58 |
| darkxst | mamarley, this is my test case for now: https://github.com/dvdhrm/docs/tree/master/drm-howto | 09:59 |
| darkxst | though I should probably try and run some of the libdrm tests | 09:59 |
| darkxst | but if the first don't work, then well, its not working! | 10:01 |
| darkxst | mamarley, anyway don't worry about it, I'll sort it out some time when I have time, right along with my gjs port which is now 2 years in the making,,,, | 10:15 |
| === JanC_ is now known as JanC | ||
| tjaalton | ricotz: wayland rejected, needs ffe | 20:27 |
| tjaalton | which is unsurprising | 20:27 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!