| Message ID | 20190331094620.15185-1-chris@chris-wilson.co.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/i915: Check domains for userptr on release | expand |
On Sun, 31 Mar 2019 at 10:47, Chris Wilson <chris@chris-wilson.co.uk> wrote: > > When we return pages to the system, we release control over them and > should defensively return them to the CPU write domain so that we catch > any external writes on reacquiring them (e.g. to transparently > swapout/swapin). While we did this defensive clflushing for ordinary > shmem pages, it was forgotten for userptr. Fortunately, userptr objects > are normally cache coherent and so oblivious to the forgotten domain > tracking. > > References: a679f58d0510 ("drm/i915: Flush pages on acquisition") > References: 754a25442705 ("drm/i915: Skip object locking around a no-op set-domain ioctl") > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> > Cc: Matthew Auld <matthew.william.auld@gmail.com> > Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index e506e43cfade..c3b4ec52e1b7 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -308,7 +308,7 @@ static void __start_cpu_write(struct drm_i915_gem_object *obj) obj->cache_dirty = true; } -static void +void __i915_gem_object_release_shmem(struct drm_i915_gem_object *obj, struct sg_table *pages, bool needs_clflush) @@ -2202,7 +2202,6 @@ i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj, struct page *page; __i915_gem_object_release_shmem(obj, pages, true); - i915_gem_gtt_finish_pages(obj, pages); if (i915_gem_object_needs_bit17_swizzle(obj)) diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h index 1a24dc97e4fd..ca93a40c0c87 100644 --- a/drivers/gpu/drm/i915/i915_gem_object.h +++ b/drivers/gpu/drm/i915/i915_gem_object.h @@ -502,4 +502,8 @@ void i915_gem_object_set_cache_coherency(struct drm_i915_gem_object *obj, unsigned int cache_level); void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj); +void __i915_gem_object_release_shmem(struct drm_i915_gem_object *obj, + struct sg_table *pages, + bool needs_clflush); + #endif diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c index ad0087127144..215bf3fef10c 100644 --- a/drivers/gpu/drm/i915/i915_gem_userptr.c +++ b/drivers/gpu/drm/i915/i915_gem_userptr.c @@ -673,9 +673,7 @@ i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj, if (!pages) return; - if (obj->mm.madv != I915_MADV_WILLNEED) - obj->mm.dirty = false; - + __i915_gem_object_release_shmem(obj, pages, true); i915_gem_gtt_finish_pages(obj, pages); for_each_sgt_page(page, sgt_iter, pages) {
When we return pages to the system, we release control over them and should defensively return them to the CPU write domain so that we catch any external writes on reacquiring them (e.g. to transparently swapout/swapin). While we did this defensive clflushing for ordinary shmem pages, it was forgotten for userptr. Fortunately, userptr objects are normally cache coherent and so oblivious to the forgotten domain tracking. References: a679f58d0510 ("drm/i915: Flush pages on acquisition") References: 754a25442705 ("drm/i915: Skip object locking around a no-op set-domain ioctl") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Matthew Auld <matthew.william.auld@gmail.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> --- drivers/gpu/drm/i915/i915_gem.c | 3 +-- drivers/gpu/drm/i915/i915_gem_object.h | 4 ++++ drivers/gpu/drm/i915/i915_gem_userptr.c | 4 +--- 3 files changed, 6 insertions(+), 5 deletions(-)