On 29/11/2019 16:07, Boris Brezillon wrote:
On Fri, 29 Nov 2019 15:48:15 +0000 Steven Price steven.price@arm.com wrote:
On 29/11/2019 13:59, Boris Brezillon wrote:
Userspace might tag a BO purgeable while it's still referenced by GPU jobs. We need to make sure the shrinker does not purge such BOs until all jobs referencing it are finished.
Fixes: 013b65101315 ("drm/panfrost: Add madvise and shrinker support") Cc: stable@vger.kernel.org Signed-off-by: Boris Brezillon boris.brezillon@collabora.com
I'm happy with this, but I would also argue that it was previously user space's job not to mark a BO purgeable while it's in use by the GPU.
I was not aware of this design choice.
When I was maintaining mali_kbase I would have said no to this ;) But thankfully I've moved on! I'm not sure whether anyone actually made a design choice for Panfrost here.
This is in some ways an ABI change so we should be sure this is something that we want to support "forever" more.
Well, in that case, the ABI change is bringing extra robustness, with AFAICT, no downside for those that were taking care of that in userspace.
Yes, there's no downside for user space - this is just giving user space extra freedom.
But if Mesa has 'always' been assuming this behaviour we might as well match.
I think so, and VC4 seems to have the same expectations.
This seems like enough justification to me.
Steve
Reviewed-by: Steven Price steven.price@arm.com
Thanks for the review. _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel