18:34karolherbst: ehh.. I think we have a few piglit regressions :/
18:37imirkin: as a result of stuff i did, or your RA change?
18:42karolherbst: imirkin: between my run on the RA branch and end of november
18:43karolherbst: but those regressions are blit related.. so :/
18:43imirkin: heh ok
18:43imirkin: oh, that might be expected
18:43imirkin: whieh ones?
18:43karolherbst: the depthstencil-default_fb-* ones
18:43imirkin: name some names
18:43karolherbst: and indexed drawing..
18:43karolherbst: I will do a master vs master compare anyway and do some git bisects
19:10imirkin: karolherbst: iirc those depthstencil-default_fb ones fail semi-randomly btw
19:25karolherbst: imirkin: apparently piglit changed
19:25imirkin: ah :)
19:25karolherbst: rerun the old master version on piglit master, now there are no changes :)
19:25karolherbst: well.. a few fixes
19:25karolherbst: or... maybe just rerunning them fixed it
19:25karolherbst: or well.. broke it
19:25karolherbst: who knows
19:28karolherbst: imirkin: something is up with spec@ext_image_dma_buf_import@ext_image_dma_buf_import-transcode-nv12-as-r8-gr88 though
19:28karolherbst: well.. all ext_image_dma_buf_import tests
19:28karolherbst: one of the tests seems to randomly crash
19:30imirkin: yeah, i think that's expected
19:31karolherbst: well, it hits an assert: ../src/gallium/drivers/nouveau/nvc0/nvc0_tex.c:144: gm107_create_texture_view: Assertion `!(mt->level.pitch & 0x1f)' failed.
19:31karolherbst: but.. that it happens randomly wonders me a little
19:33imirkin: the egl import one?
19:33imirkin: not surprising
19:33imirkin: we probably don't reject it when we should
19:33imirkin: or we don't make it linear
19:33imirkin: or something. the import / export stuff is completely un-audited
19:33imirkin: there have been tons of changes since it was originally implemented
19:33imirkin: (additional use-cases, etc)
19:35karolherbst: oh well.. anyway, no regressions on pascal with my RA patch