05:36skeggsb: rhyskidd: oops, my bad, thanks for that! i've squashed it in with a "v2" note saying that you saved the day ;)
05:59rhyskidd: no problem
06:01rhyskidd: i was getting some weird hangs, and had looked at the fecs methods for envytools in the past
06:01rhyskidd: was easy enough to follow
06:15skeggsb: rhyskidd: what were you testing for the weird hangs? should have only effected hmm stuff with invalid addresses
07:15pmoreau: skeggsb: Gustavo asked for this patch (https://github.com/skeggsb/linux/commit/ce9dcf10099d22ad4b45cdbbffd9fe583d2c0627) to be dropped in favour of a more recent one: https://lists.freedesktop.org/archives/nouveau/2019-February/032135.html
07:16pmoreau: Dunno if you want to do that before resending the 5.1 pull request to Dave.
17:12imirkin: so that primitives_generated_instanced thing fails at the draw, as expected, not while doing a query. going to try to repro it without the query stuff.
18:52keegans: can I build nouveau and mesa for the Android ABI, then run it in a chroot in my Linux system?
18:52keegans: would that, in theory, work properly?
19:01imirkin: as long as the chroot contains the various android things
19:33keegans: imirkin: do you know where I can find the instructions for using Android.mk in the source tree? `ndk-build` doesn't work because its a non-standard format
20:50imirkin: keegans: i've never done anything with android. i don't think there's anything nouveau-specific there, so a wider group may be appropriate
20:52keegans: ok, thanks
21:03imirkin: HAH! invocations = 4 works, invocations = 3 fails.
21:04imirkin: and somehow having a diff number of emitted vertices per invocation plays into it as well