Eric Anholt (![[info]](http://l-stat.livejournal.com/img/userinfo.gif){kind=small} anholt) wrote, |
Re: issues...
#20570 is a nasty one. We unfortunately don't have a way to introspect our hardware (the other side of the wall has some serious infrastructure for capturing all rendering commands, then replaying them step-by-step on a simulator or real hardware to examine what happens. we don't have that). As you saw, I thought I came up with a fix (and it surely fixed some of your hangs), but you've still got more. We're unfortunately working blind in trying to solve these "random" problems, just fixing issues as we get reliable testcases and hoping they help the "random" ones.
#20570 is a nasty one. We unfortunately don't have a way to introspect our hardware (the other side of the wall has some serious infrastructure for capturing all rendering commands, then replaying them step-by-step on a simulator or real hardware to examine what happens. we don't have that). As you saw, I thought I came up with a fix (and it surely fixed some of your hangs), but you've still got more. We're unfortunately working blind in trying to solve these "random" problems, just fixing issues as we get reliable testcases and hoping they help the "random" ones.
