00:14 AndrewR: https://yadi.sk/d/d70TbbE9VPCOpg - wine-preloader.trace.bz2 (Lightsmark2007/8 {under wine} itself tend to die helf-way during benchmarking, even with LLVMpipe, but it was used to work up to crash point .. recently...)
00:42 imirkin: AndrewR: thanks ... there has been some refactoring in st/mesa around vbo's, perhaps some previous expectation changed
01:20 imirkin: AndrewR: ok, assert repro'd. now time to debug.
01:27 imirkin: hm. well i'm pretty sure that limit thing is right, so ... need to bisect to see when it stopped getting set
01:27 imirkin: imho it's a bug
01:27 imirkin: but i guess we'll see when i track down the commit
02:30 imirkin: looks like commit fd6636ebc06d55b59851701c436b8b97f50fd7f4 is the bad one
02:30 imirkin: with a subject like "st/mesa: simplify determination whether a draw needs min/max index", i might have known...
11:48 AndrewR: imirkin, thanks for bisection (I was mostly sleeping all this time)
16:56 imirkin: AndrewR: there's a fix on list
16:56 imirkin: but i need to double-check that i'm not breaking a ton of stuff first
16:56 imirkin: it's a single character change, but those are the trickiest
19:04 lovesegfault: Hey everyone
19:04 lovesegfault: karolherbst: What's the best way to reach you async?
19:09 lovesegfault: D'oh I have to go, karolherbst, please email be at bernardo at meurer dot org when you have time, I have a couple questions about those patches you referred me to
19:57 AndrewR: imirkin, thanks ....
23:28 AndrewR: imirkin, I think your patch works, at least Lightsmark starts again, and few other games (Unreal2, UT, Doom3) and 3Dmark2001SE works ...
23:28 imirkin: yeah
23:52 imirkin: AndrewR: it's pushed
23:53 AndrewR: imirkin, I see. Lets hope nothing will break due to this :}
23:53 imirkin: well, i ran a bunch of tests
23:53 imirkin: on both nv50 and nvc0
23:53 imirkin: seemed happy
23:56 AndrewR: imirkin, good. I was thinking more about radeon/intel, but lets hope .. X reboot