01:47pmoreau: skeggsb_: I have a *fix* for the interrupt: just add some "if (chipset == 0x96) return some_error_code;" at the beginning of nouveau_gem_new and I can start X without any hang.
01:48pmoreau: Can I use nvif to simulate an ioctl for NOUVEAU_GEM_NEW while running the blob?
02:16Eliasvan: Hi, I've got a question regarding the Gallium API:
02:16Eliasvan: (I had problems connecting to #dri-devel, so I'll ask it here)
02:16Eliasvan: Is there an equivalent of the glFinish() function in the Gallium API?
02:16Eliasvan: I ask this because it could be useful for accelerated screencapturing programs such as SimpleScreenRecorder, to support capturing other APIs than OpenGL in an accelerated way (e.g. d3d9).
02:23Eliasvan: Someone said "device_Present()" was the Gallium equivalent, but it didn't give any matches on Google
02:56Eliasvan: And a last question about Gallium3D: is the API low-level enough to efficiently map new APIs such as Mantle and Direct3D12?
04:17mlankhorst: maybe, it's intending to be
04:53Eliasvan: mlankhorst: cool, thanks! And concerning my first question (about the Gallium equivalent of "glFinish()"), can you also give some insight in that one?
04:53mlankhorst: erm that's glFlush + wait on the resulting fence object
04:59Eliasvan: so when glFlush is processed by the opengl statetracker, does the statetracker call some special Gallium function that typically marks the end of a frame?
05:02mlankhorst: just break on it with a debugger and find out :P
05:03mlankhorst: looks like it would call mesa/state_tracker/st_cb_flush.c
05:03Eliasvan: mlankhorst: ok, you're right, that will definitely give the right answer ;)
05:04Eliasvan: mlankhorst: ok thanks, I'll have a look at it, thanks for the starting point!
05:04mlankhorst: I know the answer, but understanding why is important :-)
05:05Eliasvan: mlankhorst: yep, true :)
10:11tizbac: what is the meaning of RT_WIDTH_OVERRUN exactly?
12:36mlankhorst: probably what it says
12:36mlankhorst: error in the width parameter..
13:38mwk: tizbac: mlankhorst:
13:38mwk: actually it's rendering outside of the rendering area
13:38mwk: which can only happen if you manage to set window dimensions outside of RT dimensions
13:38mwk: since there's always the window clipping step
13:39mwk: or was it called screen scissor instead of window...
13:47tizbac: mwk, no
13:47tizbac: i've found what causes it
13:48tizbac: it's a mixup of render target with and without multisample
13:48tizbac: if color render targets are non-multisample , and zs is multisample , it will cause either these errors , or gpu lockup depending on size and format
16:32pmoreau: Would it be possible - or maybe it already exists - a way to take a snapshot of all registers at some point and dump it into some file?
23:03regbulle: good morning nouveau programmer