[Intel-gfx] [PATCH 08/38] drm/i915: Pull VM lists under the VM mutex.
Tvrtko Ursulin
tvrtko.ursulin at linux.intel.com
Fri Jan 18 16:04:43 UTC 2019
On 18/01/2019 14:00, Chris Wilson wrote:
> A starting point to counter the pervasive struct_mutex. For the goal of
> avoiding (or at least blocking under them!) global locks during user
> request submission, a simple but important step is being able to manage
> each clients GTT separately. For which, we want to replace using the
> struct_mutex as the guard for all things GTT/VM and switch instead to a
> specific mutex inside i915_address_space.
>
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
Regards,
Tvrtko
> ---
> drivers/gpu/drm/i915/i915_gem.c | 14 ++++++++------
> drivers/gpu/drm/i915/i915_gem_evict.c | 2 ++
> drivers/gpu/drm/i915/i915_gem_gtt.c | 15 +++++++++++++--
> drivers/gpu/drm/i915/i915_gem_shrinker.c | 4 ++++
> drivers/gpu/drm/i915/i915_gem_stolen.c | 2 ++
> drivers/gpu/drm/i915/i915_vma.c | 11 +++++++++++
> drivers/gpu/drm/i915/selftests/i915_gem_evict.c | 3 +++
> drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 3 +++
> 8 files changed, 46 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index f45186ddb236..538fa5404603 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -245,18 +245,19 @@ int
> i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
> struct drm_file *file)
> {
> - struct drm_i915_private *dev_priv = to_i915(dev);
> - struct i915_ggtt *ggtt = &dev_priv->ggtt;
> + struct i915_ggtt *ggtt = &to_i915(dev)->ggtt;
> struct drm_i915_gem_get_aperture *args = data;
> struct i915_vma *vma;
> u64 pinned;
>
> + mutex_lock(&ggtt->vm.mutex);
> +
> pinned = ggtt->vm.reserved;
> - mutex_lock(&dev->struct_mutex);
> list_for_each_entry(vma, &ggtt->vm.bound_list, vm_link)
> if (i915_vma_is_pinned(vma))
> pinned += vma->node.size;
> - mutex_unlock(&dev->struct_mutex);
> +
> + mutex_unlock(&ggtt->vm.mutex);
>
> args->aper_size = ggtt->vm.total;
> args->aper_available_size = args->aper_size - pinned;
> @@ -1529,20 +1530,21 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
>
> static void i915_gem_object_bump_inactive_ggtt(struct drm_i915_gem_object *obj)
> {
> - struct drm_i915_private *i915;
> + struct drm_i915_private *i915 = to_i915(obj->base.dev);
> struct list_head *list;
> struct i915_vma *vma;
>
> GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));
>
> + mutex_lock(&i915->ggtt.vm.mutex);
> for_each_ggtt_vma(vma, obj) {
> if (!drm_mm_node_allocated(&vma->node))
> continue;
>
> list_move_tail(&vma->vm_link, &vma->vm->bound_list);
> }
> + mutex_unlock(&i915->ggtt.vm.mutex);
>
> - i915 = to_i915(obj->base.dev);
> spin_lock(&i915->mm.obj_lock);
> list = obj->bind_count ? &i915->mm.bound_list : &i915->mm.unbound_list;
> list_move_tail(&obj->mm.link, list);
> diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
> index 5cfe4b75e7d6..dc137701acb8 100644
> --- a/drivers/gpu/drm/i915/i915_gem_evict.c
> +++ b/drivers/gpu/drm/i915/i915_gem_evict.c
> @@ -432,6 +432,7 @@ int i915_gem_evict_vm(struct i915_address_space *vm)
> }
>
> INIT_LIST_HEAD(&eviction_list);
> + mutex_lock(&vm->mutex);
> list_for_each_entry(vma, &vm->bound_list, vm_link) {
> if (i915_vma_is_pinned(vma))
> continue;
> @@ -439,6 +440,7 @@ int i915_gem_evict_vm(struct i915_address_space *vm)
> __i915_vma_pin(vma);
> list_add(&vma->evict_link, &eviction_list);
> }
> + mutex_unlock(&vm->mutex);
>
> ret = 0;
> list_for_each_entry_safe(vma, next, &eviction_list, evict_link) {
> diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
> index 2ad9070a54c1..49b00996a15e 100644
> --- a/drivers/gpu/drm/i915/i915_gem_gtt.c
> +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
> @@ -1931,7 +1931,10 @@ static struct i915_vma *pd_vma_create(struct gen6_hw_ppgtt *ppgtt, int size)
> vma->ggtt_view.type = I915_GGTT_VIEW_ROTATED; /* prevent fencing */
>
> INIT_LIST_HEAD(&vma->obj_link);
> +
> + mutex_lock(&vma->vm->mutex);
> list_add(&vma->vm_link, &vma->vm->unbound_list);
> + mutex_unlock(&vma->vm->mutex);
>
> return vma;
> }
> @@ -3504,9 +3507,10 @@ void i915_gem_restore_gtt_mappings(struct drm_i915_private *dev_priv)
>
> i915_check_and_clear_faults(dev_priv);
>
> + mutex_lock(&ggtt->vm.mutex);
> +
> /* First fill our portion of the GTT with scratch pages */
> ggtt->vm.clear_range(&ggtt->vm, 0, ggtt->vm.total);
> -
> ggtt->vm.closed = true; /* skip rewriting PTE on VMA unbind */
>
> /* clflush objects bound into the GGTT and rebind them. */
> @@ -3516,19 +3520,26 @@ void i915_gem_restore_gtt_mappings(struct drm_i915_private *dev_priv)
> if (!(vma->flags & I915_VMA_GLOBAL_BIND))
> continue;
>
> + mutex_unlock(&ggtt->vm.mutex);
> +
> if (!i915_vma_unbind(vma))
> - continue;
> + goto lock;
>
> WARN_ON(i915_vma_bind(vma,
> obj ? obj->cache_level : 0,
> PIN_UPDATE));
> if (obj)
> WARN_ON(i915_gem_object_set_to_gtt_domain(obj, false));
> +
> +lock:
> + mutex_lock(&ggtt->vm.mutex);
> }
>
> ggtt->vm.closed = false;
> i915_ggtt_invalidate(dev_priv);
>
> + mutex_unlock(&ggtt->vm.mutex);
> +
> if (INTEL_GEN(dev_priv) >= 8) {
> struct intel_ppat *ppat = &dev_priv->ppat;
>
> diff --git a/drivers/gpu/drm/i915/i915_gem_shrinker.c b/drivers/gpu/drm/i915/i915_gem_shrinker.c
> index a76d6c95c824..6da795c7e62e 100644
> --- a/drivers/gpu/drm/i915/i915_gem_shrinker.c
> +++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
> @@ -461,6 +461,7 @@ i915_gem_shrinker_vmap(struct notifier_block *nb, unsigned long event, void *ptr
> I915_SHRINK_VMAPS);
>
> /* We also want to clear any cached iomaps as they wrap vmap */
> + mutex_lock(&i915->ggtt.vm.mutex);
> list_for_each_entry_safe(vma, next,
> &i915->ggtt.vm.bound_list, vm_link) {
> unsigned long count = vma->node.size >> PAGE_SHIFT;
> @@ -468,9 +469,12 @@ i915_gem_shrinker_vmap(struct notifier_block *nb, unsigned long event, void *ptr
> if (!vma->iomap || i915_vma_is_active(vma))
> continue;
>
> + mutex_unlock(&i915->ggtt.vm.mutex);
> if (i915_vma_unbind(vma) == 0)
> freed_pages += count;
> + mutex_lock(&i915->ggtt.vm.mutex);
> }
> + mutex_unlock(&i915->ggtt.vm.mutex);
>
> out:
> shrinker_unlock(i915, unlock);
> diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c
> index a9e365789686..74a9661479ca 100644
> --- a/drivers/gpu/drm/i915/i915_gem_stolen.c
> +++ b/drivers/gpu/drm/i915/i915_gem_stolen.c
> @@ -702,7 +702,9 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_i915_private *dev_priv
> vma->flags |= I915_VMA_GLOBAL_BIND;
> __i915_vma_set_map_and_fenceable(vma);
>
> + mutex_lock(&ggtt->vm.mutex);
> list_move_tail(&vma->vm_link, &ggtt->vm.bound_list);
> + mutex_unlock(&ggtt->vm.mutex);
>
> spin_lock(&dev_priv->mm.obj_lock);
> list_move_tail(&obj->mm.link, &dev_priv->mm.bound_list);
> diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
> index 7de28baffb8f..dcbd0d345c72 100644
> --- a/drivers/gpu/drm/i915/i915_vma.c
> +++ b/drivers/gpu/drm/i915/i915_vma.c
> @@ -213,7 +213,10 @@ vma_create(struct drm_i915_gem_object *obj,
> }
> rb_link_node(&vma->obj_node, rb, p);
> rb_insert_color(&vma->obj_node, &obj->vma_tree);
> +
> + mutex_lock(&vm->mutex);
> list_add(&vma->vm_link, &vm->unbound_list);
> + mutex_unlock(&vm->mutex);
>
> return vma;
>
> @@ -656,7 +659,9 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
> GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
> GEM_BUG_ON(!i915_gem_valid_gtt_space(vma, cache_level));
>
> + mutex_lock(&vma->vm->mutex);
> list_move_tail(&vma->vm_link, &vma->vm->bound_list);
> + mutex_unlock(&vma->vm->mutex);
>
> if (vma->obj) {
> struct drm_i915_gem_object *obj = vma->obj;
> @@ -689,8 +694,10 @@ i915_vma_remove(struct i915_vma *vma)
>
> vma->ops->clear_pages(vma);
>
> + mutex_lock(&vma->vm->mutex);
> drm_mm_remove_node(&vma->node);
> list_move_tail(&vma->vm_link, &vma->vm->unbound_list);
> + mutex_unlock(&vma->vm->mutex);
>
> /*
> * Since the unbound list is global, only move to that list if
> @@ -802,7 +809,11 @@ static void __i915_vma_destroy(struct i915_vma *vma)
> GEM_BUG_ON(i915_gem_active_isset(&vma->last_fence));
>
> list_del(&vma->obj_link);
> +
> + mutex_lock(&vma->vm->mutex);
> list_del(&vma->vm_link);
> + mutex_unlock(&vma->vm->mutex);
> +
> if (vma->obj)
> rb_erase(&vma->obj_node, &vma->obj->vma_tree);
>
> diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_evict.c b/drivers/gpu/drm/i915/selftests/i915_gem_evict.c
> index 9e6b9304f2bd..c8deb961a020 100644
> --- a/drivers/gpu/drm/i915/selftests/i915_gem_evict.c
> +++ b/drivers/gpu/drm/i915/selftests/i915_gem_evict.c
> @@ -109,10 +109,13 @@ static int populate_ggtt(struct drm_i915_private *i915)
>
> static void unpin_ggtt(struct drm_i915_private *i915)
> {
> + struct i915_ggtt *ggtt = &i915->ggtt;
> struct i915_vma *vma;
>
> + mutex_lock(&ggtt->vm.mutex);
> list_for_each_entry(vma, &i915->ggtt.vm.bound_list, vm_link)
> i915_vma_unpin(vma);
> + mutex_unlock(&ggtt->vm.mutex);
> }
>
> static void cleanup_objects(struct drm_i915_private *i915)
> diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> index 852b06cb50a0..35eb40e5de91 100644
> --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> @@ -1237,7 +1237,10 @@ static void track_vma_bind(struct i915_vma *vma)
> __i915_gem_object_pin_pages(obj);
>
> vma->pages = obj->mm.pages;
> +
> + mutex_lock(&vma->vm->mutex);
> list_move_tail(&vma->vm_link, &vma->vm->bound_list);
> + mutex_unlock(&vma->vm->mutex);
> }
>
> static int exercise_mock(struct drm_i915_private *i915,
>
More information about the Intel-gfx
mailing list