drm/i915: Use i915_gem_object_get_dma_address() to populate rotated vmas
Replace the kvmalloc_array() with i915_gem_object_get_dma_address() when populating rotated vmas. One random access mechanism ought to be enough for everyone? To calculate the size of the radix tree I think we can do something like this (assuming 64bit pointers): num_pages = obj_size / 4096 tree_height = ceil(log64(num_pages)) num_nodes = sum(64^n, n, 0, tree_height-1) tree_size = num_nodes * 576 If we compare that with the object size we should get a relative overhead of around .2% to 1% for reasonable sized objects, which framebuffers tend to be. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Suggested-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20181016150413.11577-1-ville.syrjala@linux.intel.com Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
This commit is contained in:
parent
198a2a2f1a
commit
73f522bad1
@ -3637,7 +3637,7 @@ void i915_gem_restore_gtt_mappings(struct drm_i915_private *dev_priv)
|
|||||||
}
|
}
|
||||||
|
|
||||||
static struct scatterlist *
|
static struct scatterlist *
|
||||||
rotate_pages(const dma_addr_t *in, unsigned int offset,
|
rotate_pages(struct drm_i915_gem_object *obj, unsigned int offset,
|
||||||
unsigned int width, unsigned int height,
|
unsigned int width, unsigned int height,
|
||||||
unsigned int stride,
|
unsigned int stride,
|
||||||
struct sg_table *st, struct scatterlist *sg)
|
struct sg_table *st, struct scatterlist *sg)
|
||||||
@ -3646,7 +3646,7 @@ rotate_pages(const dma_addr_t *in, unsigned int offset,
|
|||||||
unsigned int src_idx;
|
unsigned int src_idx;
|
||||||
|
|
||||||
for (column = 0; column < width; column++) {
|
for (column = 0; column < width; column++) {
|
||||||
src_idx = stride * (height - 1) + column;
|
src_idx = stride * (height - 1) + column + offset;
|
||||||
for (row = 0; row < height; row++) {
|
for (row = 0; row < height; row++) {
|
||||||
st->nents++;
|
st->nents++;
|
||||||
/* We don't need the pages, but need to initialize
|
/* We don't need the pages, but need to initialize
|
||||||
@ -3654,7 +3654,8 @@ rotate_pages(const dma_addr_t *in, unsigned int offset,
|
|||||||
* The only thing we need are DMA addresses.
|
* The only thing we need are DMA addresses.
|
||||||
*/
|
*/
|
||||||
sg_set_page(sg, NULL, I915_GTT_PAGE_SIZE, 0);
|
sg_set_page(sg, NULL, I915_GTT_PAGE_SIZE, 0);
|
||||||
sg_dma_address(sg) = in[offset + src_idx];
|
sg_dma_address(sg) =
|
||||||
|
i915_gem_object_get_dma_address(obj, src_idx);
|
||||||
sg_dma_len(sg) = I915_GTT_PAGE_SIZE;
|
sg_dma_len(sg) = I915_GTT_PAGE_SIZE;
|
||||||
sg = sg_next(sg);
|
sg = sg_next(sg);
|
||||||
src_idx -= stride;
|
src_idx -= stride;
|
||||||
@ -3668,22 +3669,11 @@ static noinline struct sg_table *
|
|||||||
intel_rotate_pages(struct intel_rotation_info *rot_info,
|
intel_rotate_pages(struct intel_rotation_info *rot_info,
|
||||||
struct drm_i915_gem_object *obj)
|
struct drm_i915_gem_object *obj)
|
||||||
{
|
{
|
||||||
const unsigned long n_pages = obj->base.size / I915_GTT_PAGE_SIZE;
|
|
||||||
unsigned int size = intel_rotation_info_size(rot_info);
|
unsigned int size = intel_rotation_info_size(rot_info);
|
||||||
struct sgt_iter sgt_iter;
|
|
||||||
dma_addr_t dma_addr;
|
|
||||||
unsigned long i;
|
|
||||||
dma_addr_t *page_addr_list;
|
|
||||||
struct sg_table *st;
|
struct sg_table *st;
|
||||||
struct scatterlist *sg;
|
struct scatterlist *sg;
|
||||||
int ret = -ENOMEM;
|
int ret = -ENOMEM;
|
||||||
|
int i;
|
||||||
/* Allocate a temporary list of source pages for random access. */
|
|
||||||
page_addr_list = kvmalloc_array(n_pages,
|
|
||||||
sizeof(dma_addr_t),
|
|
||||||
GFP_KERNEL);
|
|
||||||
if (!page_addr_list)
|
|
||||||
return ERR_PTR(ret);
|
|
||||||
|
|
||||||
/* Allocate target SG list. */
|
/* Allocate target SG list. */
|
||||||
st = kmalloc(sizeof(*st), GFP_KERNEL);
|
st = kmalloc(sizeof(*st), GFP_KERNEL);
|
||||||
@ -3694,29 +3684,20 @@ intel_rotate_pages(struct intel_rotation_info *rot_info,
|
|||||||
if (ret)
|
if (ret)
|
||||||
goto err_sg_alloc;
|
goto err_sg_alloc;
|
||||||
|
|
||||||
/* Populate source page list from the object. */
|
|
||||||
i = 0;
|
|
||||||
for_each_sgt_dma(dma_addr, sgt_iter, obj->mm.pages)
|
|
||||||
page_addr_list[i++] = dma_addr;
|
|
||||||
|
|
||||||
GEM_BUG_ON(i != n_pages);
|
|
||||||
st->nents = 0;
|
st->nents = 0;
|
||||||
sg = st->sgl;
|
sg = st->sgl;
|
||||||
|
|
||||||
for (i = 0 ; i < ARRAY_SIZE(rot_info->plane); i++) {
|
for (i = 0 ; i < ARRAY_SIZE(rot_info->plane); i++) {
|
||||||
sg = rotate_pages(page_addr_list, rot_info->plane[i].offset,
|
sg = rotate_pages(obj, rot_info->plane[i].offset,
|
||||||
rot_info->plane[i].width, rot_info->plane[i].height,
|
rot_info->plane[i].width, rot_info->plane[i].height,
|
||||||
rot_info->plane[i].stride, st, sg);
|
rot_info->plane[i].stride, st, sg);
|
||||||
}
|
}
|
||||||
|
|
||||||
kvfree(page_addr_list);
|
|
||||||
|
|
||||||
return st;
|
return st;
|
||||||
|
|
||||||
err_sg_alloc:
|
err_sg_alloc:
|
||||||
kfree(st);
|
kfree(st);
|
||||||
err_st_alloc:
|
err_st_alloc:
|
||||||
kvfree(page_addr_list);
|
|
||||||
|
|
||||||
DRM_DEBUG_DRIVER("Failed to create rotated mapping for object size %zu! (%ux%u tiles, %u pages)\n",
|
DRM_DEBUG_DRIVER("Failed to create rotated mapping for object size %zu! (%ux%u tiles, %u pages)\n",
|
||||||
obj->base.size, rot_info->plane[0].width, rot_info->plane[0].height, size);
|
obj->base.size, rot_info->plane[0].width, rot_info->plane[0].height, size);
|
||||||
|
Loading…
Reference in New Issue
Block a user