On Tue, 9 Sep 2025 13:39:39 +0000 Alice Ryhl <alicer...@google.com> wrote:
> On Tue, Sep 09, 2025 at 01:36:22PM +0000, Alice Ryhl wrote: > > When using GPUVM in immediate mode, it is necessary to call > > drm_gpuvm_unlink() from the fence signalling critical path. However, > > unlink may call drm_gpuvm_bo_put(), which causes some challenges: > > > > 1. drm_gpuvm_bo_put() often requires you to take resv locks, which you > > can't do from the fence signalling critical path. > > 2. drm_gpuvm_bo_put() calls drm_gem_object_put(), which is often going > > to be unsafe to call from the fence signalling critical path. > > > > To solve these issues, add a deferred version of drm_gpuvm_unlink() that > > adds the vm_bo to a deferred cleanup list, and then clean it up later. > > > > The new methods take the GEMs GPUVA lock internally rather than letting > > the caller do it because it also needs to perform an operation after > > releasing the mutex again. This is to prevent freeing the GEM while > > holding the mutex (more info as comments in the patch). This means that > > the new methods can only be used with DRM_GPUVM_IMMEDIATE_MODE. > > > > Signed-off-by: Alice Ryhl <alicer...@google.com> > > I'm not sure if we came to a conclusion on the gpuva defer stuff on v1, > but I found a less confusing way to implement the locking. Please check > it out. Yep, I think I prefer this version too.
