Hi, On Mon, 14 Apr 2025 at 16:34, Steven Price <steven.pr...@arm.com> wrote: > On 14/04/2025 13:47, Boris Brezillon wrote: > > Hm, I might have been too prompt at claiming this was doable. In > > practice, doing that might regress Lima and Panfrost in situations > > where trying harder than GFP_NOWAIT would free up some memory. Not > > saying this was right to use GFP_KERNEL in the first place, but some > > expectations were set by this original mistake, so I'll probably need > > Lima developers to vouch in for this change after they've done some > > testing on a system under high memory pressure, and I'd need to do the > > same kind of testing for Panfrost and ask Steve if he's okay with that > > too. > > It's a tricky one. The ideal would be to teach user space how to recover > from the memory allocation failure (even on older GPUs it's still in > theory possible to break up the work and do incremental rendering). But > that ship sailed long ago so this would be a regression.
Yeah. Vulkan clients are a little better in this regard, but GL clients are pretty useless at dealing with device loss across the board. Cheers, Daniel
