01:13 ccr:re-checks his bisect.
05:25 fling: I had not a single hang or crash after switching from xv to x11 in mpv.
05:25 fling: Looks like this solution helps.
08:15 cinatic: hi
08:16 cinatic: i have a problem with nouveau and android emulator, i see nouveau stack trace when starting the emulator and then it crashes
08:23 gnarface: cinatic: usual culprits are your kernel or mesa, if they're old upgrade them then try again
09:09 cinatic: gnarface: mesa is 20.1.8-1 and kernel is 5.8.5
09:11 cinatic: i recently changed to nouveau and wayland is there anything i need to configure additionally in this combination, i had no issues with x11 and the propritary driver before?
09:12 gnarface: cinatic: oh, it could be a wayland thing actually, though it could be a regression, or something specific to just that one card too
09:15 cinatic: ok will dig a bit deeper maybe i will find something, thanks
09:29 RSpliet: cinatic: as a start it could be helpful if you could paste your unfiltered dmesg and the stack trace in question on hastebin or fpaste or whatever, and share the links here
09:40 cinatic: RSpliet: https://hastebin.com/opoxixonul.coffeescript
09:42 RSpliet: cinatic: thanks. That's filtered though, it'd be good to get the log from the start of boot without any grepping or selection
09:43 cinatic: this is the whole dmesg output
09:43 cinatic: https://hastebin.com/ajaguzatux.coffeescript
09:45 RSpliet: Thank you!
09:46 RSpliet: GP107, seems to init fine, that's good
09:50 RSpliet: Your second error gives a reason "PTE", which I believe is a page table error.
09:50 RSpliet: karolherbst: this sounds vaguely familiar?
09:51 RSpliet: First one as well by the way, but I think that's after dying
09:53 cinatic: okay thanks for having a look
11:23 karolherbst: RSpliet: ah yeah, that's just an invalid pointer thing
11:23 karolherbst: ohh, probably multithreading stuff
11:27 karolherbst: https://gitlab.freedesktop.org/karolherbst/mesa/-/commits/nvc0_threading is something people could try out
11:27 karolherbst: still needs some bigger changes, but it should work
13:15 cinatic: ok thanks i am currently building it also noticed that i am using the 20.1.8 probably it is already fixed in 20.2.0?
13:15 karolherbst: I don't think so
13:16 karolherbst: multithreading is just something we have to fix for real and nothing landed yet
13:16 karolherbst: I think the emulator hits those issues.. could be something else though
13:25 cinatic: okay
13:39 cinatic: question, i run the archlinux package builder with the threading branch source files, now i created and installed 6 packages mesa, mesa-vdpau, opencl, vulkan intel / mesa / radeon, is that sufficient or do i need to do more?
13:39 karolherbst: should be fine
13:40 karolherbst: you only really need mesa though
13:40 karolherbst: but you still have to update all which you have installed
13:40 cinatic: ok so far everything runs normally, i get a different message now
13:40 cinatic: [ 91.388225] qemu-system-x86[5043]: segfault at 8 ip 00007f2bbfd5d4bc sp 00007f2b15ff8520 error 4 in nouveau_dri.so[7f2bbf5ba000+e50000]
13:40 cinatic: [ 91.388236] Code: f7 ff 25 4f 8d bd 00 0f 1f 80 00 00 00 00 48 89 f7 ff 25 3f 8d bd 00 0f 1f 80 00 00 00 00 41 54 4c 8d 67 20 55 53 48 8b 6f 28 <48> 8b 5d 08 49 39 ec 74 2e 0f 1f 00 48 8b 7d 18 ff 55 10 48 8b 55
13:40 cinatic: [ 91.388262] audit: type=1701 audit(1601904767.278:326): auid=1000 uid=1000 gid=100 ses=4 pid=4990 comm="MainLoopThread" exe="/opt/android-sdk/emulator/qemu/linux-x86_64/qemu-system-x86_64" sig=11 res=1
13:41 cinatic: [ 91.388267] audit: type=1701 audit(1601904767.278:327): auid=1000 uid=1000 gid=100 ses=4 pid=4990 comm="qemu-system-x86" exe="/opt/android-sdk/emulator/qemu/linux-x86_64/qemu-system-x86_64" sig=11 res=1
13:41 karolherbst: ehh :/
13:42 karolherbst: well..I can look into it myself probably
13:42 karolherbst: you just have the normal android emulator using an x86 image, no?
13:42 karolherbst: doing anything special or just booting it up?
13:43 cinatic: yes nothing special, pixel_xl_android_29.avd just booting up
13:45 cinatic: i created the android vm devices with propritery nvidia driver, maybe google / android is doing some hw checks and uses this to configure its underlying qemu
13:46 karolherbst: probably not
13:46 karolherbst: others were hitting this issue as well afaik
13:46 karolherbst: I do know that android is doing some heavy multithreading stuff
13:46 karolherbst: but it could be something else