01:41 imirkin: mooch3: those docs probably never existed :)
01:42 mooch3: oh?
01:42 imirkin: docs take effort to write
01:43 imirkin: esp for stuff like nv1/nv3, they may not have undertaken that effort
01:43 imirkin: and even if they did, such things are easy to lose in time
01:43 imirkin: it's probably like 5 repository systems ago for them
01:44 imirkin: who knows if everything made every transition
01:44 imirkin: my point is -- even if it's not profitable for them to keep it closed, it's also extra work to open it up
05:35 alkisg: Hi imirkin; sorry for the direct ping; could I try anything to resolve the "segfaults on TNT2 and xfwm/marco/XPresent"? If you could point me to the .c files that are probably responsible, I could put printfs until I pinpoint it?
05:36 alkisg: (it's not the TNT2 per se that I care about, but the same problem causes more rare segfaults in newer hardware, and display artifacts)
05:36 imirkin: alkisg: yeah, either you or someone else filed a bug about it
05:36 imirkin: i was going to look at it this weekend
05:37 imirkin: basically my plan is to run Xorg in gdb and see where it crashes
05:37 imirkin: alkisg: https://bugs.freedesktop.org/show_bug.cgi?id=111868
05:37 alkisg: I didn't know where to file a bug, so it wasn't me (I did file one in marco); do you happen to have a link?
05:37 alkisg: Ty
05:38 imirkin: sounds the same as yours -- crash with mate-session, right?
05:38 imirkin: or is your thing different?
05:38 alkisg: The same
05:38 imirkin: ok, so i need to look into how bad mate's deps are
05:38 alkisg: I can work around it by disabling xpresent, or setting pageflip off, or setting the resolution twice
05:39 alkisg: imirkin: xfwm segfaults too
05:39 imirkin: oh really?
05:39 alkisg: (the most recent versions that have xpresent)
05:39 alkisg: Yes
05:39 alkisg: So you just need to install and run that, nothing more
05:39 imirkin: wait, what's xfwm?
05:39 imirkin: is that xfce's thing?
05:39 alkisg: xfce window manager
05:39 imirkin: right ok
05:39 imirkin: that seems more reasonable
05:39 alkisg: Marco copied its xpresent code from there
05:39 diogenes_: alkisg, xfwm4 has xfwm4 --vblank=off --replace, does marco have anything similar?
05:39 imirkin: link to marco, so i'm just testing the right thing?
05:40 imirkin: diogenes_: xorg shouldn't crash.
05:40 alkisg: imirkin: no; I had to add something similar myself to disable xpresent
05:40 alkisg: imirkin: which distro will you be testing on?
05:40 imirkin: alkisg: gentoo
05:41 alkisg: (08:40:22 AM) alkisg: imirkin: no; I had to add something similar myself to disable xpresent ==> that was for diogenes_, sorry
05:41 alkisg: OK let me get you the upstream link then
05:41 alkisg: https://github.com/mate-desktop/marco
05:41 imirkin: oh, that's part of mate though :)
05:42 imirkin: ok yeah. xfwm may be simpler.
05:42 alkisg: Yes, let me search for xfwm instead
05:42 imirkin: i can track that down myself
05:42 alkisg: Mirror: https://github.com/xfce-mirror/xfwm4
05:42 imirkin: so the idea is that if i start xfwm on one of the affected GPUs, i get a crash?
05:42 alkisg: Right, immediate one
05:42 imirkin: do i need to do anything special, or any sort of configuration?
05:42 imirkin: or just defaults are good?
05:42 alkisg: Original:
05:42 alkisg: https://git.xfce.org/xfce/xfwm4
05:43 alkisg: Defaults should be enough to make it crash
05:43 imirkin: excellent
05:43 imirkin: any particular version, or just anything semi-recent?
05:43 alkisg: Disabling compositing also makes it work, but I don't think anyone defaults to --no-composite
05:43 alkisg: Anything from this year should be fine
05:44 alkisg:thinks he tried with 4.14
05:45 diogenes_: alkisg, only 4.14 has xpresent backend.
05:47 alkisg: Hmmm https://git.xfce.org/xfce/xfwm4/commit/?id=e07574d6e7a2dbaa08c3ba4765c6306073d9493e compositor: Revert to GLX as default vblank method (again)
05:47 alkisg: " Although XPresent would be the most sensible choice for a lightweight vblank method, it's still quite fragile and currently broken with modesettings driver on Xorg 1.20.
05:47 alkisg: I don't remember if I had to pass a parameter to select xpresent there
05:50 alkisg: Maybe it needs: xfwm4 --replace --vblank=xpresent
05:50 imirkin: ffs
05:50 alkisg: (if it defaults to glx, can't recall)
05:50 imirkin: installing mate:
05:50 imirkin: Total: 80 packages (76 new, 3 in new slots, 1 reinstall, 1 uninstall), Size of downloads: 134355 KiB
05:51 imirkin: yeah, that ain't happening.
05:51 alkisg: imirkin: why all of that?
05:51 alkisg: Can't you just install marco?
05:51 imirkin: https://hastebin.com/uvepesikoh.rb
05:52 alkisg: LANG=C apt install --no-install-recommends marco ==> The following NEW packages will be installed:
05:52 alkisg: libmarco-private0 marco marco-common mate-desktop-common Need to get 1179 kB of archives.
05:52 imirkin: let's try it without "extras"
05:52 imirkin: yeah, that's coz you already have a ton of that junk installed :)
05:52 alkisg: I don't know gentoo, but it should be possible to just get marco
05:53 imirkin: marco depends on things
05:53 alkisg: No, I tried that on gnome, not mate. But ok, you might miss a lot :)
05:53 imirkin: i have ... none of those.
05:53 imirkin: i never understood what any of that stuff adds, except bloat
05:53 imirkin: ok, cut down a little, it's a bit more reasonable -- Total: 38 packages (35 new, 2 in new slots, 1 reinstall, 1 uninstall), Size of downloads: 72862 KiB
05:54 alkisg: In a chroot without xorg, I get: After this operation, 390 MB of additional disk space will be used.
05:54 alkisg: So yeah you might be missing a lot of these, if you don't have gnome/mate/whatever and you're minimal
05:55 imirkin: xfwm-4.14.0 -- should work ok?
05:55 alkisg: Right
05:55 alkisg: It has all the latest xpresent code, which I tested with, and it's causing the crash
05:55 alkisg: (should work == should crash, in this case :P)
05:55 imirkin: right ok
05:56 imirkin: and all i do is run xinit xfwm4 -- nothing fancy?
05:56 diogenes_: imirkin, xfwm4 --replace
05:56 alkisg: You might need xfwm4 --vblank=xpresent
05:56 alkisg: --replace => only if he's already running xorg...
05:56 alkisg: (and a window manager)
05:57 imirkin: alkisg: which version of xorg, and which version of xf86-video-nouveau?
05:58 imirkin: i'm going to check if i can repro on a nv34 now. if yes - great. otherwise it'll wait until tomorrow
05:58 imirkin: i have a TNT2, so i should def be able to repro
05:59 imirkin: (assuming it hasn't died... iirc a board died, although i think it was a NV18)
06:00 alkisg: imirkin: I tested on both 18.04 and 19.10, so it should happen in all recent xorgs/nouveau
06:02 imirkin: can you get an actual version?
06:02 imirkin: i have no clue what ubuntu ships
06:02 alkisg: Sure, moment
06:02 imirkin: but whatever it is, i assume that "latest" is secretly "2+ years old"
06:03 alkisg: This was one: ii xserver-xorg-core-hwe-18.04 2:1.20.4-1ubuntu3~18.04.1 amd64 Xorg X server - core server
06:03 alkisg: ii xserver-xorg-video-nouveau-hwe-18.04 1:1.0.16-1~18.04.1
06:03 imirkin: cool, thanks!
06:04 imirkin: that's what i have, so should be ok
06:04 imirkin: or rather, should be equivalent test
06:04 alkisg: Thank you too; I also have a box with nv4 if you need remote
06:04 imirkin: nv4?!
06:04 imirkin: wow
06:04 alkisg: Sorry, I mean tnt2, I probably got the number wrong,
06:04 imirkin: oh. that's nv5. i have that too :)
06:04 alkisg: although I do have mannnyyyy old cards, so probably all of them :D
06:04 imirkin: nv4 = riva tnt
06:04 imirkin: nv3 = riva 128
06:04 alkisg: I also have some with 2 MB VRAM
06:05 imirkin: nv1 = saturn's gpu :)
06:05 alkisg: OK I have nv3 and nv4
06:05 alkisg: No nv1
06:05 imirkin: (and some pc plug-in versions that were made for development for the saturn, so not exactly a lot of those in circulation)
06:06 imirkin: in case you don't know what a saturn is: https://en.wikipedia.org/wiki/Sega_Saturn
06:06 alkisg:was just about to google that one
06:06 alkisg: #$(*#& my school dns will be back in 10 minutes, can't surf atm
06:07 imirkin: lol
06:11 imirkin: ok. no crash with nv34, it seems :(
06:12 imirkin: very sad.
06:14 alkisg: imirkin: which one is nv34?
06:14 imirkin: FX 5200
06:14 alkisg: Yeah that only crashes once per hour or so
06:15 alkisg: Did you run xfwm4 --vblank=xpresent?
06:15 imirkin: no
06:15 alkisg: It might default to glx, not sure
06:15 alkisg: Give it a try please
06:15 imirkin: sure
06:16 imirkin: well, it's running... going to keep it around
06:17 imirkin: aha, looks like it hung up
06:17 imirkin: no more vblanks?
06:17 imirkin: and it's back...
06:19 alkisg: It'll be so much easier with tnt2, it crashes *immediately* :)
06:22 imirkin: that requires plugging one in
06:25 imirkin: weird, it looks like glxgears froze
06:25 imirkin: but it eventually starts running agian
08:34 imirkin: alkisg: well, hasn't failed yet. heading off to sleep.
08:38 alkisg: imirkin: yeah it needs students and poking for that
08:39 alkisg: try tnt2 whenever you can
08:39 alkisg: it's immediate there
10:58 gst568923: / Hi, I am the owner of an Nvidia Geforce GT 705 (GF119) card and I installed the `nouveau-firmware` package to use video acceleration for playing files with the H264 codec. Going to check the files that the `nouveau-firmware` package goes to install, I noticed that the ones related to my chipset "nvd9" are missing, in the description of the
10:58 gst568923: package is the following statement: "This package is temporary; the nouveau drivers will soon be able togenerate this data on the fly." Can you explain what it means?
12:38 gst568923: ?
13:38 HerrSpliet: gst568923: hmm, well, not sure where that note comes from. Nouveau has not shipped with OSS firmwares for video decoding engines since... ever. And nobody is working on that either.
13:53 gst568923: @HerrSpliet extracting the VP5 engine from the proprietary driver file, with this script it would have been more convenient: https://raw.github.com/imirkin/re-vp2/master/extract_firmware.py pity that only works for some minor versions of VP5
16:43 imirkin: RSpliet: nouveau-firmware is the package containing weird nv4x firmware
16:44 imirkin: i have no idea what he was talking about though -- my script extracts all the relevant firmware.
16:44 imirkin: if one follows instructions
16:44 imirkin: that nouveau-firmware package hasn't been needed in ... a very very long time
19:35 mooch2: Lyude, hey, did you find someone to ask about the riva 128 stuff?
21:37 JacekJagosz: When I do xrandr --listproviders Nvidia card doesn't show up, also I can't use the proprietary driver. Anybody has any idea what could have caused that. I did a fresh install on another drive and it works there, so what config could break it like that?
21:39 imirkin: pastebin xorg log