03:35pavolzetor: What would be a useful information to identify cause of GPU lockup (dpm)?
06:05pavolzetor: I am getting frequent lockups. Any tips how to get the dmesg and debug info?
06:05pavolzetor: frequent -> once a day
06:54pinchartl: tagr_: any news regarding your connector name support patch for modetest ? I'm getting so tired of having to edid my scripts that I will implement that myself if you don't post patches :-)
11:02Droste: is mesa using libgcrypt for something different than shader cache? I got "libGL: dlopen /usr/local/lib/dri/r600_dri.so failed (/usr/lib/libgcrypt.so.20: symbol gpgrt_lock_lock, version GPG_ERROR_1.0 not defined in file libgpg-error.so.0 with link time reference)" when I tried running steam with latest mesa, but this error still shows up if I compile with "--disable-shader-cache"
11:20mlankhorst: mismatched libraries probably
11:22mlankhorst: does steam have libgpg-error?
11:41kisak: steam runtime contains libgpg-error.so.0.8.0
11:42kisak: and a .so.0 symlink
11:42kisak: 32bit and 64bit
12:24mlankhorst: zap it
13:05Droste: mlankhorst: I know compiled with nettle which now works, but I was wondering why it would still need that library with shader cache disabled
13:55jekstrand: mattst88: I haven't looked at your cmod branch yet (I'll try to on Tuesday) but one thing we may want to do is effectivly register-coalesce the flag. NIR generates piles of mov-to-flag when we could just dump the result in the flage register.
13:55jekstrand: mattst88: Maybe that's what your branch does?
14:48mattst88: jekstrand: register coalescing isn't the right way to describe it
14:49mattst88: it's like the saturate propagation pass. it moves the conditional mod from something like a compare instruction onto the instruction that actually generated the result you're testing
15:22kamil7: hi guys, anyone here? have a little question
20:01chrisf: just pushed micah's ARB_shader_precision tests and mesa infrastructure
20:02chrisf: a bunch of them are unhappy on my haswell (mod, plus plenty of others)