00:32 karolherbst: uis: I fixed one, not sure if it was yours though
00:32 karolherbst: but maybe you hit the unfixable one
00:36 lroop: RSpliet: I guess I Forgot to mention it, NvForcePost did work.
00:37 lroop: since this is a machine with both Intel and NVIDIA graphics and IIRC it defaults to the Intel, it may expect the driver to POST the GPU?
00:38 lroop: IIRC the BIOS setup menu doesn't even have an option to exclusively use one or the other, I guess they assume that Optimus will do the right thing on Windows and probably don't care that much about Linux
08:22 RSpliet: lroop: that's interesting! Yeah, there might be cases in which nouveau should decide to POST the card. I don't know what the criteria are though
08:22 RSpliet: karolherbst: isn't there a "this card has POSTed already" bit that, if cleared, causes nouveau to POST?
09:30 karolherbst: RSpliet: we even have a workaround for some lenovo laptops where something in this area went wrong
09:31 karolherbst: quirk_reset_lenovo_thinkpad_p50_nvgpu
12:32 rowbee: does anyone know if ben skeggs is reachable via e-mail? i had a couple of questions about his devel-clk tree, and he is apparently not on irc
12:32 rowbee: i will try emailing him via the email on the commits, and see what happens
13:37 uis: nouveau: kernel rejected pushbuf: Invalid argument
13:37 uis: nouveau: ch12: krec 0 pushes 1 bufs 0 relocs 0
13:37 uis: nouveau: ch12: psh 00000000 00000537f0 0000053804
13:38 uis: Kernel: 5.4.60 with gentoo patches
13:38 uis: Mesa: 20.1.7
13:38 uis: nouveau: kernel rejected pushbuf: No such file or directory
13:38 uis: nouveau: ch11: krec 0 pushes 0 bufs 1 relocs 0
13:38 uis: nouveau: ch11: buf 00000000 00000002 00000004 00000004 00000000
13:39 uis: aces: ../libdrm-2.4.102/nouveau/pushbuf.c:723: nouveau_pushbuf_data: Assertion `kref' failed.
13:39 uis: Mesa: 20.0.8
13:39 AndrewR: karolherbst, it seems internals of OpenCL just as complicated (for driver writers) as those written during implementation of GLSL in mesa (15 years ago). A lot of functions , and interactions ....
13:43 uis: Ouch
13:43 uis: It can be broken instllation
14:10 uis: Nope
14:11 uis: Maybe is's problem of mesa/kernel version combination?
14:24 uis: nouveau: kernel rejected pushbuf: No such file or directory
14:24 uis: nouveau: ch12: krec 0 pushes 2 bufs 13 relocs 0
14:24 uis: nouveau: ch12: buf 00000000 00000002 00000004 00000004 00000000
14:24 uis: nouveau: ch12: buf 00000001 00000006 00000004 00000000 00000004
14:24 uis: nouveau: ch12: buf 00000002 0000002a 00000002 00000000 00000002
14:24 uis: nouveau: ch12: buf 00000003 00000039 00000002 00000002 00000000
14:24 uis: nouveau: ch12: buf 00000004 00000026 00000002 00000002 00000000
14:24 uis: nouveau: ch12: buf 00000005 00000028 00000002 00000002 00000000
14:24 uis: nouveau: ch12: buf 00000006 00000012 00000002 00000000 00000002
14:24 uis: nouveau: ch12: buf 00000007 00000014 00000004 00000004 00000000
14:24 uis: nouveau: ch12: buf 00000008 00000008 00000002 00000002 00000000
14:24 uis: nouveau: kernel rejected pushbuf: No such file or directory
14:24 uis: nouveau: ch12: krec 0 pushes 2 bufs 34 relocs 0
14:24 uis: nouveau: ch12: buf 00000000 00000002 00000004 00000004 00000000
14:24 uis: aces: ../libdrm-2.4.102/nouveau/pushbuf.c:723: nouveau_pushbuf_data: Assertion `kref' failed.
14:26 kwizart: uis, you are using a rather recent libdrm with a older kernel, best is to use stable kernel instead
14:26 kwizart: uis, technically, the issue in on the libdrm side
14:27 kwizart: but it can be a interaction of kernel/libdrm/mesa, indeed
14:37 uis: 5.7.19 or 5.8.10?
14:44 kwizart: 5.7 is EOL branch, so 5.8.x
14:44 kwizart: (aka mainline)
15:33 uis: Need I recompile libdrm after kernel update?
15:37 kwizart: uis, usually no, because libdrm bundle it's own kernel headers
16:09 karolherbst: uis: that's a mesa bug anyway
16:09 karolherbst: essentially the GPU context crashed
16:10 karolherbst: so instead of allowing userspace to free we essentially just crash the applicaiton
16:10 karolherbst: but I guess we should handle that a bit better overall
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: ERROR: no viable spill candidates left
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: nvc0_program_translate:638 - shader translation failed: -4
16:14 uis: double free or corruption (fasttop)
16:14 uis: double free or corruption (out)
16:14 uis: It's started
16:14 uis: With 20.1.7
16:15 uis: But not with 20.2.0_rc4
16:22 karolherbst: uis: if it's a regression, mind doing a git bisect to figure out what commit breaks it?
16:22 karolherbst: I mean.. I know I fixed a memory corruption bug
16:47 uis: What difference between 20.1.7 and 20.2.0_rc4?
16:47 uis: And 20.1.8