[Intel-xe] [PATCH v3 1/6] drm/xe/bo: Simplify xe_bo_lock()

Matthew Brost matthew.brost at intel.com
Thu Aug 31 17:01:48 UTC 2023


On Thu, Aug 31, 2023 at 04:43:57PM +0200, Thomas Hellström wrote:
> On Thu, 2023-08-31 at 11:29 +0200, Thomas Hellström wrote:
> > xe_bo_lock() was, although it only grabbed a single lock,
> > unnecessarily
> > using ttm_eu_reserve_buffers(). Simplify and document the interface.
> > 
> > v2:
> > - Update also the xe_display subsysem.
> > 
> > Signed-off-by: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_display.c |  5 +--
> >  drivers/gpu/drm/xe/tests/xe_bo.c             | 24 +++++------
> >  drivers/gpu/drm/xe/xe_bo.c                   | 42 ++++++++++++------
> > --
> >  drivers/gpu/drm/xe/xe_bo.h                   |  5 +--
> >  drivers/gpu/drm/xe/xe_bo_evict.c             | 19 ++++-----
> >  drivers/gpu/drm/xe/xe_gt_pagefault.c         | 41 +++++++-----------
> > -
> >  drivers/gpu/drm/xe/xe_vm.c                   | 22 +++++-----
> >  drivers/gpu/drm/xe/xe_vm_madvise.c           | 30 ++++++--------
> >  8 files changed, 85 insertions(+), 103 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c
> > b/drivers/gpu/drm/i915/display/intel_display.c
> > index d6b481f33b45..9df8081f78d9 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -6938,11 +6938,10 @@ static int
> > i915_gem_object_read_from_page(struct xe_bo *bo,
> >         void *virtual;
> >         bool is_iomem;
> >         int ret;
> > -       struct ww_acquire_ctx ww;
> >  
> >         XE_WARN_ON(size != 8);
> >  
> > -       ret = xe_bo_lock(bo, &ww, 0, true);
> > +       ret = xe_bo_lock(bo, true);
> >         if (ret)
> >                 return ret;
> >  
> > @@ -6959,7 +6958,7 @@ static int
> > i915_gem_object_read_from_page(struct xe_bo *bo,
> >  
> >         ttm_bo_kunmap(&map);
> >  out_unlock:
> > -       xe_bo_unlock(bo, &ww);
> > +       xe_bo_unlock(bo);
> >         return ret;
> >  }
> >  #endif
> > diff --git a/drivers/gpu/drm/xe/tests/xe_bo.c
> > b/drivers/gpu/drm/xe/tests/xe_bo.c
> > index b32a9068d76c..31fd4f9b2d5b 100644
> > --- a/drivers/gpu/drm/xe/tests/xe_bo.c
> > +++ b/drivers/gpu/drm/xe/tests/xe_bo.c
> > @@ -203,9 +203,9 @@ static int evict_test_run_gt(struct xe_device
> > *xe, struct xe_gt *gt, struct kuni
> >                         goto cleanup_bo;
> >                 }
> >  
> > -               xe_bo_lock(external, &ww, 0, false);
> > +               xe_bo_lock(external, false);
> >                 err = xe_bo_pin_external(external);
> > -               xe_bo_unlock(external, &ww);
> > +               xe_bo_unlock(external);
> >                 if (err) {
> >                         KUNIT_FAIL(test, "external bo pin err=%pe\n",
> >                                    ERR_PTR(err));
> > @@ -268,9 +268,9 @@ static int evict_test_run_gt(struct xe_device
> > *xe, struct xe_gt *gt, struct kuni
> >                                            ERR_PTR(err));
> >                                 goto cleanup_all;
> >                         }
> > -                       xe_bo_lock(external, &ww, 0, false);
> > +                       xe_bo_lock(external, false);
> >                         err = xe_bo_validate(external, NULL, false);
> > -                       xe_bo_unlock(external, &ww);
> > +                       xe_bo_unlock(external);
> >                         if (err) {
> >                                 KUNIT_FAIL(test, "external bo valid
> > err=%pe\n",
> >                                            ERR_PTR(err));
> > @@ -278,28 +278,28 @@ static int evict_test_run_gt(struct xe_device
> > *xe, struct xe_gt *gt, struct kuni
> >                         }
> >                 }
> >  
> > -               xe_bo_lock(external, &ww, 0, false);
> > +               xe_bo_lock(external, false);
> >                 xe_bo_unpin_external(external);
> > -               xe_bo_unlock(external, &ww);
> > +               xe_bo_unlock(external);
> >  
> >                 xe_bo_put(external);
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 __xe_bo_unset_bulk_move(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 xe_bo_put(bo);
> >                 continue;
> >  
> >  cleanup_all:
> > -               xe_bo_lock(external, &ww, 0, false);
> > +               xe_bo_lock(external, false);
> >                 xe_bo_unpin_external(external);
> > -               xe_bo_unlock(external, &ww);
> > +               xe_bo_unlock(external);
> >  cleanup_external:
> >                 xe_bo_put(external);
> >  cleanup_bo:
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 __xe_bo_unset_bulk_move(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 xe_bo_put(bo);
> >                 break;
> >         }
> > diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
> > index 6a8a41dafe88..9d70d12e79b1 100644
> > --- a/drivers/gpu/drm/xe/xe_bo.c
> > +++ b/drivers/gpu/drm/xe/xe_bo.c
> > @@ -1079,13 +1079,11 @@ static void xe_gem_object_close(struct
> > drm_gem_object *obj,
> >         struct xe_bo *bo = gem_to_xe_bo(obj);
> >  
> >         if (bo->vm && !xe_vm_in_fault_mode(bo->vm)) {
> > -               struct ww_acquire_ctx ww;
> > -
> >                 XE_WARN_ON(!xe_bo_is_user(bo));
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 ttm_bo_set_bulk_move(&bo->ttm, NULL);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  }
> >  
> > @@ -1863,26 +1861,36 @@ int xe_gem_mmap_offset_ioctl(struct
> > drm_device *dev, void *data,
> >         return 0;
> >  }
> >  
> > -int xe_bo_lock(struct xe_bo *bo, struct ww_acquire_ctx *ww,
> > -              int num_resv, bool intr)
> > +/**
> > + * xe_bo_lock() - Lock the buffer object's dma_resv object
> > + * @bo: The struct xe_bo whose lock is to be taken
> > + * @intr: Whether to perform any wait interruptible
> > + *
> > + * Locks the buffer object's dma_resv object. If the buffer object
> > is
> > + * pointing to a shared dma_resv object, that shared lock is locked.
> > + *
> > + * Return: 0 on success, -EINTR if @intr is true and the wait for a
> > + * contended lock was interrupted.

Probably should add in kernel doc that it is ok for the call to not
check the return value if intr == false && number fence reservation
slots == 0. The existing code I think follows this pattern but it is not
documented.

> > + */
> > +int xe_bo_lock(struct xe_bo *bo, bool intr)
> >  {
> > -       struct ttm_validate_buffer tv_bo;
> > -       LIST_HEAD(objs);
> > -       LIST_HEAD(dups);
> > +       if (intr)
> > +               return dma_resv_lock_interruptible(bo->ttm.base.resv,
> > NULL);
> >  
> > -       XE_WARN_ON(!ww);
> > +       dma_resv_lock(bo->ttm.base.resv, NULL);
> >  
> > -       tv_bo.num_shared = num_resv;
> > -       tv_bo.bo = &bo->ttm;
> > -       list_add_tail(&tv_bo.head, &objs);
> > -
> > -       return ttm_eu_reserve_buffers(ww, &objs, intr, &dups);
> > +       return 0;
> >  }
> >  
> > -void xe_bo_unlock(struct xe_bo *bo, struct ww_acquire_ctx *ww)
> > +/**
> > + * xe_bo_unlock() - Unlock the buffer object's dma_resv object
> > + * @bo: The struct xe_bo whose lock is to be released.
> > + *
> > + * Unlock a buffer object lock that was locked by xe_bo_lock().
> > + */
> > +void xe_bo_unlock(struct xe_bo *bo)
> >  {
> >         dma_resv_unlock(bo->ttm.base.resv);
> > -       ww_acquire_fini(ww);
> >  }
> >  
> >  /**
> > diff --git a/drivers/gpu/drm/xe/xe_bo.h b/drivers/gpu/drm/xe/xe_bo.h
> > index 0823dda0f31b..a7b9e7084225 100644
> > --- a/drivers/gpu/drm/xe/xe_bo.h
> > +++ b/drivers/gpu/drm/xe/xe_bo.h
> > @@ -149,10 +149,9 @@ static inline void xe_bo_assert_held(struct
> > xe_bo *bo)
> >                 dma_resv_assert_held((bo)->ttm.base.resv);
> >  }
> >  
> > -int xe_bo_lock(struct xe_bo *bo, struct ww_acquire_ctx *ww,
> > -              int num_resv, bool intr);
> > +int xe_bo_lock(struct xe_bo *bo, bool intr);
> >  
> > -void xe_bo_unlock(struct xe_bo *bo, struct ww_acquire_ctx *ww);
> > +void xe_bo_unlock(struct xe_bo *bo);
> >  
> >  static inline void xe_bo_unlock_vm_held(struct xe_bo *bo)
> >  {
> > diff --git a/drivers/gpu/drm/xe/xe_bo_evict.c
> > b/drivers/gpu/drm/xe/xe_bo_evict.c
> > index 0d5c3a208ab4..49c05ddea164 100644
> > --- a/drivers/gpu/drm/xe/xe_bo_evict.c
> > +++ b/drivers/gpu/drm/xe/xe_bo_evict.c
> > @@ -27,7 +27,6 @@
> >  int xe_bo_evict_all(struct xe_device *xe)
> >  {
> >         struct ttm_device *bdev = &xe->ttm;
> > -       struct ww_acquire_ctx ww;
> >         struct xe_bo *bo;
> >         struct xe_tile *tile;
> >         struct list_head still_in_list;
> > @@ -62,9 +61,9 @@ int xe_bo_evict_all(struct xe_device *xe)
> >                 list_move_tail(&bo->pinned_link, &still_in_list);
> >                 spin_unlock(&xe->pinned.lock);
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 ret = xe_bo_evict_pinned(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 xe_bo_put(bo);
> >                 if (ret) {
> >                         spin_lock(&xe->pinned.lock);
> > @@ -96,9 +95,9 @@ int xe_bo_evict_all(struct xe_device *xe)
> >                 list_move_tail(&bo->pinned_link, &xe-
> > >pinned.evicted);
> >                 spin_unlock(&xe->pinned.lock);
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 ret = xe_bo_evict_pinned(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 xe_bo_put(bo);
> >                 if (ret)
> >                         return ret;
> > @@ -123,7 +122,6 @@ int xe_bo_evict_all(struct xe_device *xe)
> >   */
> >  int xe_bo_restore_kernel(struct xe_device *xe)
> >  {
> > -       struct ww_acquire_ctx ww;
> >         struct xe_bo *bo;
> >         int ret;
> >  
> > @@ -140,9 +138,9 @@ int xe_bo_restore_kernel(struct xe_device *xe)
> >                 list_move_tail(&bo->pinned_link, &xe-
> > >pinned.kernel_bo_present);
> >                 spin_unlock(&xe->pinned.lock);
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 ret = xe_bo_restore_pinned(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 if (ret) {
> >                         xe_bo_put(bo);
> >                         return ret;
> > @@ -184,7 +182,6 @@ int xe_bo_restore_kernel(struct xe_device *xe)
> >   */
> >  int xe_bo_restore_user(struct xe_device *xe)
> >  {
> > -       struct ww_acquire_ctx ww;
> >         struct xe_bo *bo;
> >         struct xe_tile *tile;
> >         struct list_head still_in_list;
> > @@ -206,9 +203,9 @@ int xe_bo_restore_user(struct xe_device *xe)
> >                 xe_bo_get(bo);
> >                 spin_unlock(&xe->pinned.lock);
> >  
> > -               xe_bo_lock(bo, &ww, 0, false);
> > +               xe_bo_lock(bo, false);
> >                 ret = xe_bo_restore_pinned(bo);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 xe_bo_put(bo);
> >                 if (ret) {
> >                         spin_lock(&xe->pinned.lock);
> > diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c
> > b/drivers/gpu/drm/xe/xe_gt_pagefault.c
> > index b6f781b3d9d7..73fc9389a663 100644
> > --- a/drivers/gpu/drm/xe/xe_gt_pagefault.c
> > +++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c
> > @@ -171,20 +171,18 @@ static int handle_pagefault(struct xe_gt *gt,
> > struct pagefault *pf)
> >  
> >         /* Lock VM and BOs dma-resv */
> >         bo = xe_vma_bo(vma);
> > -       if (only_needs_bo_lock(bo)) {
> > -               /* This path ensures the BO's LRU is updated */
> > -               ret = xe_bo_lock(bo, &ww, xe->info.tile_count,
> > false);
> > -       } else {
> > +       if (!only_needs_bo_lock(bo)) {
> >                 tv_vm.num_shared = xe->info.tile_count;
> >                 tv_vm.bo = xe_vm_ttm_bo(vm);
> >                 list_add(&tv_vm.head, &objs);
> > -               if (bo) {
> > -                       tv_bo.bo = &bo->ttm;
> > -                       tv_bo.num_shared = xe->info.tile_count;
> > -                       list_add(&tv_bo.head, &objs);
> > -               }
> > -               ret = ttm_eu_reserve_buffers(&ww, &objs, false,
> > &dups);
> >         }
> > +       if (bo) {
> > +               tv_bo.bo = &bo->ttm;
> > +               tv_bo.num_shared = xe->info.tile_count;
> > +               list_add(&tv_bo.head, &objs);
> > +       }
> > +
> > +       ret = ttm_eu_reserve_buffers(&ww, &objs, false, &dups);
> >         if (ret)
> >                 goto unlock_vm;
> >  
> > @@ -227,10 +225,7 @@ static int handle_pagefault(struct xe_gt *gt,
> > struct pagefault *pf)
> >         vma->usm.tile_invalidated &= ~BIT(gt_to_tile(gt)->id);
> >  
> >  unlock_dma_resv:
> > -       if (only_needs_bo_lock(bo))
> > -               xe_bo_unlock(bo, &ww);
> > -       else
> > -               ttm_eu_backoff_reservation(&ww, &objs);
> > +       ttm_eu_backoff_reservation(&ww, &objs);
> >  unlock_vm:
> >         if (!ret)
> >                 vm->usm.last_fault_vma = vma;
> > @@ -534,28 +529,22 @@ static int handle_acc(struct xe_gt *gt, struct
> > acc *acc)
> >  
> >         /* Lock VM and BOs dma-resv */
> >         bo = xe_vma_bo(vma);
> > -       if (only_needs_bo_lock(bo)) {
> > -               /* This path ensures the BO's LRU is updated */
> > -               ret = xe_bo_lock(bo, &ww, xe->info.tile_count,
> > false);
> > -       } else {
> > +       if (!only_needs_bo_lock(bo)) {
> >                 tv_vm.num_shared = xe->info.tile_count;
> >                 tv_vm.bo = xe_vm_ttm_bo(vm);
> >                 list_add(&tv_vm.head, &objs);
> > -               tv_bo.bo = &bo->ttm;
> > -               tv_bo.num_shared = xe->info.tile_count;
> > -               list_add(&tv_bo.head, &objs);
> > -               ret = ttm_eu_reserve_buffers(&ww, &objs, false,
> > &dups);
> >         }
> > +       tv_bo.bo = &bo->ttm;
> > +       tv_bo.num_shared = xe->info.tile_count;
> > +       list_add(&tv_bo.head, &objs);
> > +       ret = ttm_eu_reserve_buffers(&ww, &objs, false, &dups);
> >         if (ret)
> >                 goto unlock_vm;
> >  
> >         /* Migrate to VRAM, move should invalidate the VMA first */
> >         ret = xe_bo_migrate(bo, XE_PL_VRAM0 + tile->id);
> >  
> > -       if (only_needs_bo_lock(bo))
> > -               xe_bo_unlock(bo, &ww);
> > -       else
> > -               ttm_eu_backoff_reservation(&ww, &objs);
> > +       ttm_eu_backoff_reservation(&ww, &objs);
> >  unlock_vm:
> >         up_read(&vm->lock);
> >         xe_vm_put(vm);
> > diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> > index 9cbf2f63d641..5679c47f47ca 100644
> > --- a/drivers/gpu/drm/xe/xe_vm.c
> > +++ b/drivers/gpu/drm/xe/xe_vm.c
> > @@ -267,10 +267,9 @@ static void arm_preempt_fences(struct xe_vm *vm,
> > struct list_head *list)
> >  static int add_preempt_fences(struct xe_vm *vm, struct xe_bo *bo)
> >  {
> >         struct xe_exec_queue *q;
> > -       struct ww_acquire_ctx ww;
> >         int err;
> >  
> > -       err = xe_bo_lock(bo, &ww, vm->preempt.num_exec_queues, true);
> > +       err = xe_bo_lock(bo, true);
> 
> Looks like the fence slot reservation got lost in the rebase here. Will
> fix in next revision.
> 

Yep, please add that back in.

Matt

> /Thomas
> 
> 
> >         if (err)
> >                 return err;
> >  
> > @@ -281,8 +280,8 @@ static int add_preempt_fences(struct xe_vm *vm,
> > struct xe_bo *bo)
> >                                            DMA_RESV_USAGE_BOOKKEEP);
> >                 }
> >  
> > -       xe_bo_unlock(bo, &ww);
> > -       return 0;
> > +       xe_bo_unlock(bo);
> > +       return err;
> >  }
> >  
> >  /**
> > @@ -1021,12 +1020,11 @@ bo_has_vm_references_locked(struct xe_bo *bo,
> > struct xe_vm *vm,
> >  static bool bo_has_vm_references(struct xe_bo *bo, struct xe_vm *vm,
> >                                  struct xe_vma *ignore)
> >  {
> > -       struct ww_acquire_ctx ww;
> >         bool ret;
> >  
> > -       xe_bo_lock(bo, &ww, 0, false);
> > +       xe_bo_lock(bo, false);
> >         ret = !!bo_has_vm_references_locked(bo, vm, ignore);
> > -       xe_bo_unlock(bo, &ww);
> > +       xe_bo_unlock(bo);
> >  
> >         return ret;
> >  }
> > @@ -2267,7 +2265,6 @@ vm_bind_ioctl_ops_create(struct xe_vm *vm,
> > struct xe_bo *bo,
> >                          u32 operation, u8 tile_mask, u32 region)
> >  {
> >         struct drm_gem_object *obj = bo ? &bo->ttm.base : NULL;
> > -       struct ww_acquire_ctx ww;
> >         struct drm_gpuva_ops *ops;
> >         struct drm_gpuva_op *__op;
> >         struct xe_vma_op *op;
> > @@ -2325,11 +2322,11 @@ vm_bind_ioctl_ops_create(struct xe_vm *vm,
> > struct xe_bo *bo,
> >         case XE_VM_BIND_OP_UNMAP_ALL:
> >                 XE_WARN_ON(!bo);
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return ERR_PTR(err);
> >                 ops = drm_gpuva_gem_unmap_ops_create(&vm->mgr, obj);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >                 if (IS_ERR(ops))
> >                         return ops;
> >  
> > @@ -2365,13 +2362,12 @@ static struct xe_vma *new_vma(struct xe_vm
> > *vm, struct drm_gpuva_op_map *op,
> >  {
> >         struct xe_bo *bo = op->gem.obj ? gem_to_xe_bo(op->gem.obj) :
> > NULL;
> >         struct xe_vma *vma;
> > -       struct ww_acquire_ctx ww;
> >         int err;
> >  
> >         lockdep_assert_held_write(&vm->lock);
> >  
> >         if (bo) {
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return ERR_PTR(err);
> >         }
> > @@ -2380,7 +2376,7 @@ static struct xe_vma *new_vma(struct xe_vm *vm,
> > struct drm_gpuva_op_map *op,
> >                             op->va.range - 1, read_only, is_null,
> >                             tile_mask);
> >         if (bo)
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >  
> >         if (xe_vma_is_userptr(vma)) {
> >                 err = xe_vma_userptr_pin_pages(vma);
> > diff --git a/drivers/gpu/drm/xe/xe_vm_madvise.c
> > b/drivers/gpu/drm/xe/xe_vm_madvise.c
> > index c9bc59be5094..05081fa0fbed 100644
> > --- a/drivers/gpu/drm/xe/xe_vm_madvise.c
> > +++ b/drivers/gpu/drm/xe/xe_vm_madvise.c
> > @@ -28,16 +28,15 @@ static int madvise_preferred_mem_class(struct
> > xe_device *xe, struct xe_vm *vm,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->props.preferred_mem_class = value;
> >                 xe_bo_placement_for_flags(xe, bo, bo->flags);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> > @@ -53,16 +52,15 @@ static int madvise_preferred_gt(struct xe_device
> > *xe, struct xe_vm *vm,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->props.preferred_gt = value;
> >                 xe_bo_placement_for_flags(xe, bo, bo->flags);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> > @@ -89,17 +87,16 @@ static int madvise_preferred_mem_class_gt(struct
> > xe_device *xe,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->props.preferred_mem_class = mem_class;
> >                 bo->props.preferred_gt = gt_id;
> >                 xe_bo_placement_for_flags(xe, bo, bo->flags);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> > @@ -112,13 +109,12 @@ static int madvise_cpu_atomic(struct xe_device
> > *xe, struct xe_vm *vm,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >                 if (XE_IOCTL_DBG(xe, !(bo->flags &
> > XE_BO_CREATE_SYSTEM_BIT)))
> >                         return -EINVAL;
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->props.cpu_atomic = !!value;
> > @@ -130,7 +126,7 @@ static int madvise_cpu_atomic(struct xe_device
> > *xe, struct xe_vm *vm,
> >                  */
> >                 if (bo->props.cpu_atomic)
> >                         ttm_bo_unmap_virtual(&bo->ttm);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> > @@ -143,18 +139,17 @@ static int madvise_device_atomic(struct
> > xe_device *xe, struct xe_vm *vm,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >                 if (XE_IOCTL_DBG(xe, !(bo->flags &
> > XE_BO_CREATE_VRAM0_BIT) &&
> >                                  !(bo->flags &
> > XE_BO_CREATE_VRAM1_BIT)))
> >                         return -EINVAL;
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->props.device_atomic = !!value;
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> > @@ -174,16 +169,15 @@ static int madvise_priority(struct xe_device
> > *xe, struct xe_vm *vm,
> >  
> >         for (i = 0; i < num_vmas; ++i) {
> >                 struct xe_bo *bo;
> > -               struct ww_acquire_ctx ww;
> >  
> >                 bo = xe_vma_bo(vmas[i]);
> >  
> > -               err = xe_bo_lock(bo, &ww, 0, true);
> > +               err = xe_bo_lock(bo, true);
> >                 if (err)
> >                         return err;
> >                 bo->ttm.priority = value;
> >                 ttm_bo_move_to_lru_tail(&bo->ttm);
> > -               xe_bo_unlock(bo, &ww);
> > +               xe_bo_unlock(bo);
> >         }
> >  
> >         return 0;
> 


More information about the Intel-xe mailing list