03:56 rhyskidd: skeggsb: karol mentioned a scrip 'nvkm-am' that you use to push patches from your oot nouveau to get them mainline ready
03:57 rhyskidd: do you have that available somewhere? i had a super hacky one, but it would be easier to test if more can use the same scripts (like the 'dim' tools danvet created for drm/intel)
04:00 rhyskidd: mupuf: has nvidia_bios moved to gitlab yet? my ssh keys aren't working at the old location
07:01 pmoreau: karolherbst: The vloadn thing was fixed on May 22nd, and will be part of the 2019.4 release, whenever that happens.
07:08 pmoreau: karolherbst: Do you have any changes to !563 that have not pushed to that branch yet? (Since I know you were planning on tweaking the compilation in clover.) Otherwise I’ll go ahead and do a rebase.
08:28 sleepymario: hello, i have 2 questions. i accidentally got hold of an old 8800GTS and was quite surprised to see that it actually works with nouveau, and that without any firmware necessary. First question: Is it in any way possible to make it support a widescreen (2560x1080) instead of a full HD (1920x1080)? I'm on wayland, which kinda worked out of the box. The thing has 2 DVI ports, so maybe not? If anyone has an idea
08:28 sleepymario: about this, please let me know.
08:28 sleepymario: Second question: What's the 'best' videocard that still supports nouveau fine (for normal workloads, simple WM and browser basically), and does not need any firmware to run?
10:06 karolherbst: sleepymario: "best" would be the 780 ti or any kepler or gm10x one
10:06 karolherbst: nouveau provides more or less reliable reclocking on those
10:06 karolherbst: gm10x is 750 and 750 ti I think
10:06 karolherbst: 745 as well as it seems
11:35 sleepymario: karolherbst: okay, thanks a lot
18:08 shellclear: Hi, After upgrade from debian stretch to buster I'm facing issues with nouveau modules. kworker process have cpu nice set to -20 and some cpu cores are stuck in 100% usage. the system is impossible to use because have a cpu load of 12.0 ... to solve this issue I installed nvidia proprietary drivers and the kernel seems ok now... but I would like to use nouveau instead proprietaru drivers... someone know if is a known bug
18:08 shellclear: or if it has a way to fix it?
18:09 shellclear: How I can debug it?
18:09 shellclear: :D
18:10 imirkin: which gpu, and anything interesting in the logs?
18:10 imirkin: generically speaking, this isn't a known issue
18:11 karolherbst: shellclear: is this a laptop by any chance with a fairly new CPU/GPU?
18:11 karolherbst: but yeah.. dmesg would be helpful
18:14 shellclear: imirkin: the gpu is a mx150 nvidia and I dont see nothing interesting in the logs(nothing that I have understodd)
18:14 imirkin: maybe you can provide a dmesg from an affected boot
18:14 imirkin: mx150 is ... which one is that?
18:15 shellclear: karolherbst: its a laptop with hybrid graphic card and +- new
18:15 karolherbst: imirkin: pascal
18:15 karolherbst: gp108
18:16 karolherbst: I think debian stretch simply had an old kernel where nouveau didn't support the GPU
18:16 karolherbst: and now with an updated one you run into our infamous runpm bug
18:16 shellclear: imirkin: https://www.geforce.com/hardware/notebook-gpus/geforce-mx150
18:16 karolherbst: heh, stretch is 4.9
18:16 imirkin: shellclear: well, really just the dmesg. that website won't really offer the info i want either.
18:17 imirkin: if it's a gp108 as karol says, it's fully expected to have lots of weird issues
18:17 karolherbst: yeah...
18:17 karolherbst: 4.9 is GP100 + GP104
18:17 karolherbst: close
18:17 imirkin: karolherbst: he's on buster now -- no clue what kernel that has
18:17 karolherbst: something where nouveau supports the gp108
18:18 karolherbst: 4.19
18:18 shellclear: yes, Im using buster now... a think interesting is that when I booting using kernel 4.9 works very well
18:18 karolherbst: yeah...
18:18 imirkin: shellclear: also booting with nouveau.modeset=0 should work equally well with the newer kernel
18:18 karolherbst: shellclear: I bet if you boot with nouveau.runpm=0 it works
18:18 imirkin: same result -- no nouveau support for your gpu :)
18:18 karolherbst: :D
18:18 karolherbst: I'd prefer disabling runpm though
18:18 karolherbst: although I expect secboot issues....
18:18 karolherbst: it's a mess right now
18:19 karolherbst: especially with pascal
18:19 shellclear: but that strange is that using kernel 4.9 it works
18:19 imirkin: no, it doesn'
18:19 imirkin: there's just no support for your gpu
18:20 karolherbst: new hardware and debian just don't play well as nicely :p except you use debian sid
18:21 shellclear: it possible that with kernel 4.9 only my intel gpu would be used?
18:21 karolherbst: that's the case
18:21 karolherbst: well...
18:21 karolherbst: _if_ i915 supports your gpu
18:21 karolherbst: could be software only
18:22 imirkin: shellclear: no need for your nvidia gpu -- it doesn't really do much
18:22 karolherbst: but intel is usually faster than nouveau, so your intel GPU might be supported where your nvidia gpu isn't
18:22 shellclear: I'll remove nvidia proprietary and try again using that boot parameters
18:23 imirkin: what's your goal?
18:24 shellclear: this laptop is only for work and watch movies, listen music... Im not a gamer if is it what you want to know :D
18:24 imirkin: so then you can just leave the nvidia gpu alone and not worry too much about it
18:24 imirkin: karol's solution will force it to avoid suspending, which means it'll use up battery
18:25 imirkin: not sure if it'll just end up suspended with nouveau not touching it or not ...
18:25 imirkin: but try nouveau.modeset=0 and see if it still eats power
18:27 karolherbst: imirkin: I am sure in terms of battery consumption it's undecided wheather runpm=0 or modeset=0 end up using more power
18:27 imirkin: karolherbst: well, pcie pm and all?
18:27 karolherbst: devinit may do things which end up using less battery... but other stuff might more
18:27 imirkin: [with the newer setups]
18:27 karolherbst: imirkin: by default you have a 8.0 pcie link. devinit puts it down to 2.5
18:27 karolherbst: eg
18:27 imirkin: can't the kernel suspend the gpu now though?
18:28 karolherbst: never checked what saves more power though
18:28 karolherbst: imirkin: with modeset=0? doubtful
18:28 karolherbst: although userspace could enable runpm on the gpu and the bridge
18:28 karolherbst: so that should work
18:28 imirkin: with powertop or whatever
18:28 karolherbst: yes
18:29 karolherbst: I am still not 100% what is causing the issue, but I have a fairly good idea already
18:29 mupuf: Rhy
18:30 mupuf: rhyskidd: sorry, my gitolite is down and we are moving the repos to gitlab!
18:30 karolherbst: just trying to reproduce it on AMD as well
18:31 karolherbst: mupuf: what repo did he ask about
18:31 karolherbst: ?
18:31 karolherbst: rhyskidd: https://gitlab.freedesktop.org/nouveau
18:32 karolherbst: you should be able to see all of those... I think
18:32 karolherbst: yeah, you are a member
18:32 karolherbst: mupuf: we also had a git repository for apitraces, didn't we?
18:32 karolherbst: or was that ftp only?
18:34 shellclear: karolherbst: imirkin: I'm removing nvidia driver and I'll try to boot again but this time using nouveau.modeset=0
18:34 shellclear: thanks