01:25AndrewR: hi all. I have uptime like "04:15:43 up 18 days, 14:19, 2 users, load average: 1.51, 1.47, 1.39" - best uptime this year :} Yet just recently Xorg nearly freezed itself to death on me ..but by somekeyboard action and mouse moving it was ressurected!
01:27AndrewR: https://pastebin.com/031CsGMQ - dmesg tail. (I thin nv50cal errors are due to me just having 384Mb vram). This is on Linux slax 5.1.12-x64 #1 SMP PREEMPT Wed Jun 19 12:31:05 MSK 2019 x86_64 AMD FX(tm)-4300 Quad-Core Processor AuthenticAMD GNU/Linux but with 32-bit mesa and Seamonkey/X ....
01:27AndrewR: I was surprized it survived near-freeze
18:51dustofgoat: Hey folks
18:52joepublic: Hello, greetings, and welcome.
19:02abff: hello my xorg is crashing when I try to startx, and I suspect it's an issue with Nouveau, can someone help me confirm my suspicions? Xorg performs a core dump and the startx process fails. It happens during or after what xorg calls "Initializing extension GLX"
19:02imirkin: which GPU?
19:03abff: it's one of the gpu covered by nouveau, GT218M
19:04imirkin: anything particularly odd about your setup?
19:04imirkin: e.g. things you've set up that might be different from the average setup?
19:04abff: Not that I can think of, it was working properly for about 8 months
19:04imirkin: can you pastebin the xorg log?
19:04abff: and I wasn't just leaving it to run
19:05abff: I don't have much experience diagnosing these errors as xorg usually just works
19:05imirkin: illegal instruction
19:05imirkin: what CPU do you have?
19:05abff: I saw that, but I'm not sure where, do you think it's from nouveau or elsewhere?
19:06abff: it's a first generation i7 the exact model I am not sure one moment
19:06imirkin: simplest guess would be that arch has started requiring an x86 ext that you don't have
19:06imirkin: ok, so you don't have AVX
19:06imirkin: has arch started to require avx?
19:07abff: I checked their update blog usually if there's a breaking change they warn users
19:07abff: but nothing
19:07abff: this machine also does not have an intel integrated gpu that requires the use of optimus software (I can't remember what it's called, bumblebee or something)
19:07abff: have any ideas?
19:08abff: I don't know what avx is, btw
19:08imirkin: remember mmx?
19:08imirkin: and then came sse, sse2, ssse3, sse4, sse4.2
19:08imirkin: and the next one after that is avx :)
19:09abff: I doubt they would make that some sort of requirement
19:09imirkin: i7 has sse4.2
19:09imirkin: (how do i know? i'm running a i7-920 atm)
19:10imirkin: the thing that i would do
19:10imirkin: is ssh in
19:10imirkin: and try to start X inside of gdb
19:10imirkin: and try to poke around to see what's up
19:10imirkin: sometimes glibc does the AVX stuff for stupid shit like memset()
19:10imirkin: (and indeed, there's a call into glibc ~last)
19:10imirkin: normally it detects CPUs correctly though
19:11abff: will do
19:12imirkin: there is some SSE2 code in xf86-video-nouveau, however it's (a) supported by your CPU and (b) not triggered in initialization paths
19:13imirkin: (it's to do YUV planar -> NV12 conversion)
19:13abff: startx is a shell script right?
19:13imirkin: just run "X" should be enough
19:13imirkin: (and if it doesn't trigger the crash, then that's an interesting datapoint)
19:14abff: okay I will putter away at this
19:14abff: thanks for the assistance
19:28abff: imirkin: haha I should have just had more patience
19:29abff: the mesa lib was updated and it fixed my problem
19:29abff: should have checked for updates one last time
19:33abff: imirkin: great success
19:48imirkin: abff: weird, but ok!
19:48abff: lol I was waiting for an update to fix it all day yesterday
19:48abff: woke up this morning, nothing available, went out to do some shopping, came back and was like "right gotta figure this out"
19:48abff: should have checked one more time
19:49imirkin: maybe they accidentally compiled mesa for "native" target, which happened to support avx? dunno
19:50abff: I'm just glad peoplel smarter than me are on the case