06:12 <bshah> So a apol is working on EGL_KHR_partial_update support in kwin, but question is how do we test this apart from code paths are being hit?

06:13 <bshah> is there test client or what?

06:17 <anarsoul> make sure your rendering is correct?

06:18 <anarsoul> but anyway scissoring will have the same effect on lima

07:19 <rellla> anarsoul: we don't need to return (size * pres->array_size) because as far as i understand we respect the array_size already within actual_level_size.

07:19 <rellla> so it's just https://pastebin.com/raw/rTZ4Q4b2

07:20 <rellla> where the last part with "size += *" could stay as is imho

07:24 <anarsoul> rellla: OK

07:26 <rellla> i did "#define LIMA_MAX_MIP_LEVELS 12" and changed the cubemap/mipmap/rgb* deqp tests to try a texture size of 2048*2048, which triggers the (really) last level and that worked.

08:01 <anarsoul> rellla: you also need to check all the formats

08:01 <anarsoul> :)

08:02 <rellla> i think i need to try a run with LIMA_MAX_MIP_LEVELS set to 6, so that we reach the last_level in all the other tests, too...

08:05 <rellla> anarsoul: anyway, i pushed an updated version, that doesn't break deqp atm.

08:24 <anarsoul> rellla: I'll check it tomorrow

08:26 <rellla> thanks, i don't succeed the tests with a too small MAX_MIP_LEVELS, because many tests don't work, because they need more miplevels.

08:27 <rellla> here are the results of dEQP-GLES2.functional.texture.filtering.2d.linear_mipmap_linear_linear_clamp_etc1 and dEQP-GLES2.functional.texture.filtering.cube.linear_mipmap_linear_linear_clamp_etc1 anyway, which come to the lastlevel

08:27 <rellla> values seem right to me