From: Jesse B. <jb...@vi...> - 2009-01-09 17:56:48
|
On Friday, January 9, 2009 12:28 am Michel Dänzer wrote: > On Thu, 2009-01-08 at 10:45 -0800, Jesse Barnes wrote: > > Dunno if this is a good idea or not, but it would have helped with some > > of the recent bugs. > > Sure about that? The sequence numbers may be completely different > between CRTCs. Yeah, it depends... In some places we get the count first and then wait on the next one, so as long as the "get count" call gets the count from the fallback crtc subsequent calls should work out ok. But like I said, I don't know if working around broken userspace here is a good idea. It seems like w/DRI1 things are actually in pretty good shape these days, and with the timeout code in libdrm apps should no longer hang, so we should be in good shape going forward as well. (Also the patch is untested; it just came out of a discussion we had yesterday so I thought I'd post it for comment.) -- Jesse Barnes, Intel Open Source Technology Center |