nemo | mason: welp. I'm currently on the free drivers, and indeed basically everything is broken | 00:04 |
---|---|---|
nemo | even with latest amdgpu kernel module | 00:04 |
nemo | mason: webgl in firefox unusable slow, most games not rendering | 00:04 |
nemo | soooo back to figuring out how to get amdgpu-pro working again ☹ | 00:05 |
nemo | *sigh* | 00:14 |
nemo | amdgpu-pro for 18.04 no longer builds against latest beowulf kernel | 00:15 |
nemo | amdgpu-pro for 20.04 depends on libc6 2.29 and beowulf is still on 2.28 | 00:15 |
nemo | kinda funny that ubuntu 20.04 using the 5.4 stable kernel and libc6 2.229 | 00:16 |
nemo | er 2.29 | 00:16 |
nemo | but debian beowulf (on backports) uses a 5.8 kernel and libc6 2.28 | 00:16 |
nemo | hmmm | 00:16 |
nemo | is ceres stable enough to use? | 00:16 |
* nemo looks at fsmithred | 00:16 | |
nemo | it's on 2.31 | 00:16 |
fsmithred | nemo, I have chimaera on my old thinkpad and it seems to be working just fine. | 00:17 |
nemo | I guess I could try rolling back latest beowulf kernel update, since I was fine until my last upgrade | 00:17 |
nemo | fsmithred: oh. is it chimæra or ceres? I kinda find the naming confusing | 00:17 |
fsmithred | chimaera and ceres are mostly the same | 00:17 |
nemo | ok | 00:17 |
fsmithred | ceres=sid | 00:17 |
nemo | fsmithred: if I was to try doing this on the family machine what name should I be using? | 00:18 |
fsmithred | chimaera is the next release after beowulf | 00:18 |
fsmithred | alphabetical order except that ceres is always ceres | 00:18 |
fsmithred | just like sid is always sid | 00:18 |
nemo | ok | 00:18 |
nemo | aight. guess I'll give this a shot. fingers crossed | 00:18 |
nemo | fsmithred: so I literally just replace beowulf with chimaera in sources.list... | 00:19 |
nemo | update and pray | 00:19 |
fsmithred | brocashelm has been using ceres and has repeatedly said that it's more stable than most distros | 00:19 |
nemo | interesting | 00:19 |
nemo | but :%s/beowulf/chimaera/g | 00:20 |
fsmithred | yeah, pretty much. apt update, apt upgrade, reboot, apt dist-upgrade (or probably aptitude full-upgrade) | 00:20 |
nemo | ? | 00:20 |
nemo | ok | 00:20 |
fsmithred | I ran into some troubles and needed aptitude to get through it. | 00:20 |
nemo | 404 on "security" and "backports" | 00:20 |
nemo | just disable them? | 00:20 |
fsmithred | also had to remove build-essential because of some conflict with gcc versions | 00:20 |
nemo | E: The repository 'http://deb.devuan.org/merged chimaera-security Release' does not have a Release file. | 00:20 |
nemo | E: The repository 'http://deb.devuan.org/merged chimaera-backports Release' does not have a Release file. | 00:20 |
fsmithred | only one main repo | 00:20 |
nemo | ok | 00:21 |
fsmithred | it's Testing | 00:21 |
fsmithred | but use the codename (chimaera) | 00:21 |
nemo | yeah, I seem to remember that's been hammered home in the channel | 00:21 |
nemo | stick with names | 00:21 |
fsmithred | you got i386 or amd64? | 00:21 |
onefang | Chimaera hasn't actually been released yet, so no backports or security. | 00:21 |
nemo | fsmithred: amd64 | 00:22 |
fsmithred | ok. That upgrade seemed to go easier than the i386 today. | 00:22 |
nemo | welp. here goes nothing. | 00:22 |
nemo | I guess worst case, I reinstall everything | 00:22 |
fsmithred | xfce? | 00:22 |
nemo | MATE | 00:22 |
nemo | plus figure out how to get whatever kernel actually worked with amdgpu-pro | 00:22 |
fsmithred | ok, you're on your own | 00:22 |
fsmithred | lol | 00:22 |
fsmithred | 5.9 is in chimaera now | 00:23 |
nemo | I've tried XFCE in past, but it felt... unpolished. was unintuitive where things were, configs seemed scattered | 00:23 |
nemo | featureset of MATE seemed higher | 00:23 |
mason | nemo: Gah. Sorry to hear it, re: the AMD stuff. | 00:24 |
fsmithred | yeah, gnome2 was full-featured | 00:24 |
nemo | fsmithred: yeah, I'm mostly praying that the ubuntu 20.04 amdgpu-pro installs against whatever debian is on | 00:24 |
mason | I'm surprised, and a bit disappointed. | 00:24 |
nemo | mason: I'm not at all surprised | 00:24 |
nemo | mason: been par for course for me w/ FOSS drivers for decades regrettably. with occasional bright spot here and there | 00:24 |
nemo | not counting intel naturally | 00:24 |
fsmithred | 20.04 has 5.4 kernel and same version of libc6 as chimaera/ceres | 00:28 |
rknop | Just updated to chimaera. Looks like wicd is not there. What do people use? (I'd like something lighter-weight than Network-Manatger) | 00:51 |
brocashelm | fsmithred: yep, still using ceres and not having any problems. if i did, i would be complaining here or on the forums | 00:57 |
clort | rknop: there's ceni - curses interface to /etc/networko/interfaces | 00:58 |
brocashelm | i like it because i just have to keep apt update/apt upgrade from here on out without planning for a huge system upgrade and rebooting multiple times (i would say a fresh install is better than lts upgrades) | 00:58 |
fsmithred | nemo, ^^^ | 00:58 |
fsmithred | there's also connman for network. | 00:59 |
fsmithred | network-manager will probably be the default with the desktop install, but that hasn't been decided for sure yet | 00:59 |
brocashelm | i would still recommend sticking with beowulf/current stable if that suits you, because you do get a ton of new updates every hour or so and are better off updating as swiftly as possible in order to resolve known issues | 01:02 |
nemo | uhoh | 01:02 |
nemo | libc6-dev : Breaks: libgcc-8-dev (< 8.4.0-2~) but 8.3.0-6 is to be installed | 01:02 |
nemo | rknop_: well thankfully this machine is wired | 01:03 |
nemo | brocashelm: I'm just trying to get graphics acceleration working again | 01:03 |
nemo | brocashelm: although at this moment in time I've descended to "get functioning desktop" | 01:03 |
brocashelm | nemo: have you tried aptitude full-upgrade? | 01:04 |
brocashelm | it will ask you if you would accept the solutions and should resolve package conflicts | 01:04 |
nemo | brocashelm: can't even install aptitude at the moment :( | 01:05 |
nemo | the breakage above has screwed everything | 01:05 |
brocashelm | what did you do since changing sources.list? | 01:05 |
nemo | apt upgrade | 01:05 |
nemo | reboot | 01:05 |
brocashelm | how about apt dist-upgrade? | 01:05 |
nemo | trying to run it | 01:05 |
brocashelm | do this in tty mode | 01:05 |
nemo | without success | 01:05 |
nemo | trying | 01:06 |
nemo | that's what gives error above | 01:06 |
nemo | apt install gcc-8-base (Which is what I get to after trying to go down from libc6-dev to libgcc-8-dev) actually does give me an install prompt | 01:06 |
nemo | but also wants to remove mate | 01:06 |
nemo | and libreoffice | 01:06 |
nemo | but I guess I could go ahead with it in hopes of reinstalling later | 01:06 |
brocashelm | what does apt say if you want to remove libgcc-8-dev? | 01:07 |
nemo | The following packages will be REMOVED: clang clang-7 g++-8 gcc-8 libgcc-8-dev libgtk2-perl libldb1 libobjc-8-dev libosmesa6 libpcre2-posix0 libreoffice libreoffice-avmedia-backend-gstreamer libreoffice-base libreoffice-calc libreoffice-core libreoffice-draw libreoffice-gtk3 | 01:07 |
nemo | libreoffice-impress libreoffice-librelogo libreoffice-math libreoffice-nlpsolver libreoffice-report-builder libreoffice-report-builder-bin libreoffice-script-provider-python libreoffice-sdbc-firebird libreoffice-sdbc-postgresql | 01:08 |
nemo | libreoffice-style-tango libreoffice-wiki-publisher libreoffice-writer libstdc++-8-dev linux-compiler-gcc-8-x86 linux-headers-4.19.0-11-amd64 linux-headers-4.19.0-12-amd64 linux-headers-5.8.0-0.bpo.2-amd64 mate-control-center | 01:08 |
nemo | mate-desktop-environment mate-desktop-environment-core python-gi python-mate-menu python3-uno python3.7 task-mate-desktop uno-libs3 | 01:08 |
nemo | sorry for multiline paste. my env is a little clunky right now | 01:08 |
brocashelm | hmmm, quite a lot of dependencies | 01:08 |
brocashelm | what i did was apt update, apt upgrade, apt dist-upgrade, reboot, apt update, apt upgrade, and then launch my session | 01:08 |
nemo | The following NEW packages will be installed: cpp-10 g++-10 gcc-10 ... etc | 01:08 |
nemo | oh well, none of those will actually prevent my system from booting | 01:08 |
brocashelm | there were packages being held back, too, but then that got resolved as i adjusted the updates | 01:08 |
nemo | so I'm going to go ahead with it | 01:08 |
nemo | in hopes I can get them back later | 01:09 |
nemo | you do have libreoffice right | 01:09 |
brocashelm | sure, you can always get them back later | 01:09 |
brocashelm | i did; replaced it with abiword + dia + gnumeric | 01:09 |
nemo | hm. interesting. it is fetching some mate stuff | 01:09 |
nemo | brocashelm: haven't used abiword in years | 01:09 |
nemo | regrettably it was not super reliable at generating MSWord docs people could open | 01:09 |
brocashelm | it's probably gotten a bit better since then :) | 01:10 |
nemo | brocashelm: last time I used it, it was on my old laptop w/ 96MiB of RAM :) | 01:10 |
brocashelm | hehe | 01:10 |
nemo | was the only gentoo system I ever used -Os on | 01:10 |
nemo | in hopes I'd save on ram | 01:10 |
brocashelm | ya, just follow what apt tells you with the dependencies and go from there | 01:10 |
nemo | I nfsmounted / to do upgrades | 01:10 |
brocashelm | keep a list of packages removed you want to reinstall if that's easier | 01:10 |
brocashelm | aptitude dist-upgrade will attempt to clean all the errors up | 01:11 |
nemo | if I manage to install aptitude ;) | 01:11 |
nemo | anyway. it's doing a bunch of stuff now | 01:11 |
nemo | thank the deities to screen/tmux + irssi. | 01:11 |
brocashelm | yup, those are quite the trip | 01:11 |
nemo | *for | 01:11 |
fsmithred | like I said, I had to remove build-essential | 01:11 |
nemo | [ 1383.966177] Not activating Mandatory Access Control as /sbin/tomoyo-init does not exist. | 01:12 |
nemo | just appeared on this terminal | 01:12 |
nemo | not sure what 'sactly that means | 01:12 |
fsmithred | never saw that before | 01:12 |
brocashelm | i have; i installed tomoyo and selinux once and regretted them | 01:13 |
brocashelm | try removing tomoyo-tools if you have it installed | 01:14 |
nemo | I've used selinux a little bit, I hear it's a good thing in theory | 01:14 |
nemo | but have found it termendously irritating | 01:15 |
nemo | *tre | 01:15 |
nemo | never sure why 'sactly I can't do something | 01:15 |
brocashelm | tomoyo and apparmor were designed to circumvent its complexities, but only introduce more | 01:15 |
nemo | doesn't really integrate nicely with traditional permissions | 01:15 |
brocashelm | i don't really need them since i don't run a server or anything | 01:15 |
brocashelm | i was just going by what lynis told me to do, lol | 01:16 |
nemo | ah... might've enabled apparmor at some point | 01:16 |
nemo | when trying out stuff | 01:16 |
brocashelm | yeah, aa-status should tell you | 01:16 |
nemo | E: Could not configure 'libc6:amd64'. | 01:16 |
nemo | E: Could not perform immediate configuration on 'libtirpc3:amd64'. Please see man 5 apt.conf under APT::Immediate-Configure for details. (2) | 01:16 |
nemo | but now I can run dist-upgrade | 01:16 |
nemo | so hopefully that will fix things | 01:16 |
brocashelm | yeah, it should | 01:16 |
nemo | it's fetching tons and tons of stuff | 01:16 |
nemo | but I had a lot of toys on this machine | 01:17 |
clort | reminds me of the days when my OS consumed my time | 01:17 |
nemo | jitsi, minetest, games and more games | 01:17 |
nemo | this is the living room TV | 01:17 |
nemo | its only job is running devuan | 01:17 |
nemo | and being a functional remote school screen by tomorrow | 01:17 |
clort | remote school is silly | 01:18 |
nemo | clort: I'm not a huge fan. but sure was convenient for taking a family vacation in october | 01:18 |
nemo | clort: at some point we realised we really could be anywhere in the world | 01:19 |
nemo | and would not have any truancy issues | 01:19 |
nemo | classroom on the beach | 01:19 |
nemo | (which is the fallback really - can always break out the laptop if I've hopelessly screwed up this setup) | 01:19 |
clort | perhaps it will break the monopoly racket on schooling | 01:20 |
nemo | clort: that was one of the things I found amusing | 01:20 |
nemo | clort: when I grew up I was homeschooled in canada | 01:21 |
brocashelm | you'll be glad once the upgrade process finishes :) | 01:21 |
nemo | clort: Ontario sent my mom, in a box, everything she needed to teach me and my brother | 01:21 |
nemo | clort: once a year the province tested us to make sure everything was fine. we were 99th percentile, not due to any great brains I think, but rather the 1:2 student teacher ratio and high teacher investment | 01:21 |
nemo | we spent 2-3 hours a day in school, rest of time tearing around the woods | 01:21 |
brocashelm | there's #devuan-offtopic btw | 01:22 |
nemo | ah yes | 01:22 |
clort | he'd have to join it | 01:22 |
golinux | Just go there #devuan-offtopic . . . please | 01:25 |
brocashelm | yup already there | 01:26 |
brocashelm | thanks | 01:26 |
nemo | welp. I got aptitude installed, it recommended a bunch of stuff that did not seem likely to prevent booting so I said yes to it all | 01:35 |
nemo | (the aptitude full-upgrade recommended above) | 01:35 |
fsmithred | cool | 01:35 |
nemo | btw, since I'm here and you mentioned wifi being even more broken in chimaera | 01:35 |
fsmithred | was your beowulf fully udated before you switched to chimaera? | 01:35 |
nemo | is wifi like some systemd dependent thing these days? already beowulf meant giving up on the mate network widget and my SO's wifi constantly having to be reconnected 'cause wicd seemed to not autoreconnect even with it checked | 01:36 |
nemo | fsmithred: yes | 01:36 |
nemo | just surprised that there's not even wicd anymore | 01:36 |
fsmithred | you probably want to install network-manager-gnome | 01:36 |
nemo | hum | 01:37 |
nemo | so long as that doesn't pull in half of gnome I guess | 01:37 |
nemo | fsmithred: for beowulf? | 01:37 |
fsmithred | no, not for beowulf | 01:37 |
fsmithred | and actually there is wicd in ceres | 01:37 |
nemo | looks like aptitude just removed mate. once I'm convinced devuan has run out of things to update in chimaera I will try reinstalling mate | 01:37 |
nemo | fsmithred: ok. just saw a wifi warning earlier | 01:38 |
fsmithred | you could add a line for ceres, pin it to a lower priority, and add things selectively from ceres | 01:38 |
nemo | well. for this particular system no wifi needed | 01:39 |
nemo | have a static ip | 01:39 |
fsmithred | oh, it's the same version of wicd in beowulf and ceres | 01:40 |
fsmithred | I don't know if it works | 01:40 |
nemo | yeah, it just doesn't seem to work as well as the old gnome2 applet | 01:40 |
fsmithred | static ip on wire is better | 01:40 |
nemo | she's still a bit annoyed I upgraded her laptop to beowul | 01:40 |
fsmithred | uh oh | 01:40 |
nemo | been promising to look into wifi thing to see if I can make it disconnect less | 01:40 |
nemo | hm... nginx keeps erroring on update in chimaera | 01:41 |
fsmithred | you could try network-manager in beowulf | 01:41 |
nemo | fsmithred: ok... thought wicd was forced due to some systemd dep | 01:41 |
nemo | but will give it a shot | 01:42 |
fsmithred | no, but there might be gtk issues. Not sure. | 01:42 |
nemo | hmmm. not sure why nginx keeps failing | 01:44 |
nemo | maybe can remove and reinstall it | 01:44 |
nemo | dpkg: error processing package nginx-core (--configure): installed nginx-core package post-installation script subprocess returned error exit status 1 | 01:45 |
fsmithred | you didn't lose an init script, did you? | 01:45 |
nemo | guess I need to find a log | 01:45 |
nemo | hm | 01:45 |
fsmithred | or check that script - /var/lib/dpkg/info/nginz.postinst probably | 01:46 |
fsmithred | with better spelling | 01:46 |
fsmithred | or check for debian bug reports | 01:47 |
nemo | hm. tries to start on port 80. wonder if I somehow installed 2 webservers | 01:47 |
nemo | doesn't look like it | 01:47 |
nemo | /etc/init.d/nginx fails to start | 01:48 |
nemo | which nginx-core calls in postinst | 01:48 |
nemo | /usr/sbin/nginx | 01:49 |
nemo | nginx: [emerg] "load_module" directive is specified too late in /etc/nginx/modules-enabled/70-mod-stream-geoip.conf:1 | 01:49 |
nemo | um okayyyyy | 01:49 |
nemo | removing both geoips allows it to start | 01:52 |
nemo | something to look into it later, jitsi related I suppose | 01:52 |
nemo | and cocnfigure succeeds | 01:52 |
nemo | *configure | 01:52 |
nemo | heh. tried starting a desktop | 01:54 |
nemo | [ 3961.821] (II) AMDGPU(0): Setting screen physical size to 508 x 285 | 01:54 |
nemo | yeaaaaah | 01:54 |
nemo | that's not gonna work | 01:54 |
nemo | oh. a missing lib upgrade | 01:55 |
nemo | and now it starts yay | 01:55 |
nemo | *sigh* synaptic-pkgexec is messed up again. wish I'd noted down what I'd done with it last time | 01:56 |
nemo | I erased something... | 01:56 |
nemo | framerate in glxgears looks promising at least | 01:57 |
nemo | pkexec fails too | 02:02 |
rknop_ | Looks like in chimaera, it's not possible to install the proprietary nvidia drivers at the moment ; they don't compile. | 02:04 |
rknop_ | (That is, the kernel module doesn't compile. Looks like a mismatch between nvidia code and gcc warnings that are now errors.) | 02:05 |
fsmithred | rknop_, are you using the binaries from nvidia? | 02:09 |
fsmithred | CC=/usr/bin/gcc-<right-version> ./nvidia-whatever... | 02:10 |
rknop_ | No, I was just apt-get installing nvidia-driver | 02:11 |
rknop_ | (Well, apt install ) | 02:11 |
Soltis | Problem: php7.4-fpm : Depends: systemd but it is not installable | 07:35 |
golinux | Soltis: Please see this thread: https://dev1galaxy.org/viewtopic.php?id=3331 | 08:10 |
nemo | onefang: btw after upgrading to chimaera, qt now detects broken CPU uuid generation and generates using other methods (I bet urandom) | 16:07 |
nemo | which they probably should've just defaulted to in first place | 16:07 |
nemo | onefang: my test program now generates STDERR warning and a valid uuid | 16:07 |
nemo | WARNING: CPU random generator seem to be failing, disable hardware random number generation WARNING: RDRND generated: 0xffffffff 0xffffffff 0xffffffff 0xffffffff | 16:08 |
nemo | hm... seems like the issue is: | 16:19 |
nemo | https://pkginfo.devuan.org/stage/chimaera/chimaera/amd64-microcode_3.20191218.1.html | 16:19 |
nemo | cimaera is still on the amd64 microcode from december, which I'm guessing does not have the rdrand fix | 16:19 |
nemo | huh. bullseye and sid are too! | 16:20 |
nemo | well, that sucks | 16:20 |
nemo | gonna guess this is AMD's fault then | 16:20 |
nemo | I mean, in some sense this is a feature I guess | 16:21 |
nemo | since rdrand should not be used for cryptographically secure randomness on its own, maybe it failing is good 😃 | 16:21 |
nemo | and it offers no performance benefits for non-cryptographically sensitive like the Qt UUID | 16:21 |
suavedandy | Okay. I'mma test Micro text editor on Regolith to see if it works there. Watch me. | 20:40 |
rknop | I have this weird thing on xfce4 and chimaera. Sometimes the whole screen will lock up. I know this happens running Blender; I've seen it with a couple of different versions. I think I saw it with something else as well. The whole screen sorts of locks up. I can move my mouse, and the pointer will move, but it responds to no keyboard input or clicking... except that one small thing on the Blender interface will change. It acts | 20:41 |
rknop | almost as if every input it gets via mouse click or keyboard is a keyboard left or keyboard right key. the one acception is ctrl-alt-l, which does lock the screen as per xscreensaver... but it doesn't put up the prompt for me to put my password to unlock it. | 20:41 |
rknop | Has anybody seen anything like this? | 20:41 |
rknop | Is there an xdotool or similar command I can tell xfce to flush all input and release all input grabs and redraw everything without having to kill the session and start over? | 20:41 |
rknop | (If I kill the xfce session, it goes back to the slim session manager as normal.)O | 20:42 |
gnarface | rknop: you using a nvidia card? | 20:43 |
gnarface | rknop: with the proprietary drivers? | 20:43 |
rknop | No. The laptop has an nvidia card, but the proprietary drivers won't compile at the moment. (mismatch in chimaera). So, graphics is all via the intel card | 20:44 |
gnarface | rknop: i have seen something like that but i'm not using xfce. restarting the WM does flush it, but i don't know the command for it in xfce | 20:44 |
rknop | Yeah, restarting the wm didn't flush it for me. xfwm --replace | 20:44 |
gnarface | rknop: i've been assuming it's a nvidia driver issue | 20:44 |
gnarface | blender uses opengl, is it only happening with opengl? | 20:44 |
gnarface | or other stuff too? | 20:44 |
rknop | If I do xfwm4 --replace, the whole screen goes black, and then starts showing garbage. | 20:45 |
rknop | (xfwm4 --replace works as expected if the lockup hasn't happened.) | 20:45 |
gnarface | definitely sounds like a driver problem :( | 20:45 |
rknop | I am not sure. I can get it to happen pretty often with blender, though it's random. Haven't seen it with something else. | 20:45 |
gnarface | you're on beowulf? | 20:45 |
rknop | chimaera | 20:45 |
gnarface | oh | 20:45 |
gnarface | right | 20:45 |
gnarface | hmm | 20:45 |
rknop | This is a very new laptop, and I need the testing distribution to have things like drivers for the wireless card, etc. I worked with backports a bit, but ended up needing too much. | 20:46 |
gnarface | does xfce have a compositor you can disable? | 20:46 |
rknop | ...I think so | 20:46 |
rknop | Not sure what it's called. | 20:46 |
rknop | Will look for it. | 20:46 |
gnarface | i'm seeing this in enlightenment on ceres, with nvidia proprietary drivers | 20:47 |
gnarface | don't remember seeing it with blender specifically but haven't tested a lot | 20:47 |
gnarface | mostly happening with wine and thunderbird | 20:47 |
rknop | Disabled the compositor (at least, I clicked a box that claimed I did), going to see if I can get it to lock up again . (Usually only takes a few minutes.) | 20:48 |
suavedandy | Actually, might also download pure Debian and test. | 20:48 |
rknop | I notice that there's a process "xscreensaver-systemd" in addition to xscreensaver -no-splash.... what's up with that? | 20:49 |
rknop | OK, still locked up even with compositor disabled | 20:50 |
rknop | Ah... yeah, something in /var/log/messages. "GPU HANG" | 20:56 |
rknop | Almost certainly related! | 20:57 |
gnarface | hmmm, not sure about xscreensaver-systemd | 20:57 |
gnarface | this could be a testing issue | 20:57 |
rknop | The "GPU HANG" in messages suggest what you thought: driver issues | 20:58 |
gnarface | rknop: if it's the intel GPU, you have the option of 2 different drivers to try | 20:58 |
rknop | What's the other driver? | 20:59 |
gnarface | well i'm not sure which would auto-load for you but usually these days i expect the modesetting driver to load by default. the other one would be the actual intel drvier | 20:59 |
gnarface | driver* | 20:59 |
rknop | Xorg.0.log suggests that it is indeed the modesetting driver | 21:00 |
gnarface | intel even recommends the modesetting driver on most their hardware these days, but i have found it not to be the better choice for every corner case | 21:00 |
rknop | Eventually, I want to get the fallback rendering to the nvidia card in this thing working, of course, but that will have to wait until I can compile the nvidia drivers again | 21:00 |
rknop | made an xorg conf file to use the "intel" driver; gonna see what happens. The word "modeset" no longer appears on /var/log/Xorg.0.log | 21:05 |
rknop | Nope, still froze up | 21:05 |
rknop | still getting the gpu hang | 21:06 |
gnarface | hmmm, well if changing drivers doesn't help then i don't know | 21:06 |
gnarface | but the intel driver does have some other options that might help | 21:06 |
gnarface | you can force dri2/dri3 or something like that | 21:07 |
rknop | AT some level, I may just have to wait until the driver improves | 21:07 |
gnarface | yea | 21:07 |
gnarface | well i guess it could be a mesa problem too | 21:15 |
rwp | rknop, I didn't read the entire scrollback sorry but I am seeing the same thing, graphics freezes, on ceres on my nvidia based system. | 21:34 |
rwp | It started about a month ago. Initially the nouveau driver would lock up and the kernel watchdog timer would expire and kick it and then it would work again. | 21:35 |
rwp | But then a couple of weeks ago it changed to just freezing for 2m then thawing and nothing in the logs. | 21:35 |
rwp | Been annoying me greatly the last month! | 21:35 |
rwp | I have been contemplating swapping out graphics cards to use the radeon driver to see if that avoids the problem, assuming it is in nouveau somewhere. | 21:36 |
rwp | Tried booting previous kernels but so far no joy on avoiding the graphics freezes. | 21:36 |
rwp | Trips whenever anything fast moves in the graphics. Fast scrolling in an xterm. Opening a new graphics window. Anything that I could guess overflows some double-buffer system or something. | 21:37 |
rknop | Interestingly, so far I haven't had a hang going back to Blender 2.82. Reading around, I've seen a lot of people have gpu hangs with these intel cards, but I also saw a lot of blender people say that issues started with 2.82. Not going to feel too comfortable yet, but it hung faster with later Blender versions. (Still likely a driver issue at the core.) | 22:12 |
mason | rknop: I have no context here, but it's not a GTK2 vs GTK3 thing, is it? | 22:19 |
rknop | Don't think so | 22:20 |
rknop | Almost certainly related to 3d GPU acceleration stuff | 22:20 |
mason | ah | 22:20 |
suavedandy | Testing Micro in Regolith it produces a completely different behavior whether you're in Xorg or in a TTY. | 22:48 |
suavedandy | This is freaking weird. | 22:48 |
rwp | When I get a hang I have sometimes been able to grab the stopwatch and start timing and the thaw from it is often very close to exactly 120 seconds later. Feels like a timeout. | 23:15 |
rwp | I have two older nVidia GeForce 8400 GS cards with nouveau driving a dual display here. Been working fine for years. | 23:16 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!