aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/i915_vma.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/i915/i915_vma.c')
-rw-r--r--drivers/gpu/drm/i915/i915_vma.c114
1 files changed, 75 insertions, 39 deletions
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index f17c09ead7d7..f51fd9fd4c89 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -26,6 +26,7 @@
#include <linux/dma-fence-array.h>
#include <drm/drm_gem.h>
+#include "display/intel_display.h"
#include "display/intel_frontbuffer.h"
#include "gem/i915_gem_lmem.h"
#include "gem/i915_gem_tiling.h"
@@ -73,14 +74,16 @@ static void vma_print_allocator(struct i915_vma *vma, const char *reason)
char buf[512];
if (!vma->node.stack) {
- DRM_DEBUG_DRIVER("vma.node [%08llx + %08llx] %s: unknown owner\n",
- vma->node.start, vma->node.size, reason);
+ drm_dbg(&to_i915(vma->obj->base.dev)->drm,
+ "vma.node [%08llx + %08llx] %s: unknown owner\n",
+ vma->node.start, vma->node.size, reason);
return;
}
stack_depot_snprint(vma->node.stack, buf, sizeof(buf), 0);
- DRM_DEBUG_DRIVER("vma.node [%08llx + %08llx] %s: inserted at %s\n",
- vma->node.start, vma->node.size, reason, buf);
+ drm_dbg(&to_i915(vma->obj->base.dev)->drm,
+ "vma.node [%08llx + %08llx] %s: inserted at %s\n",
+ vma->node.start, vma->node.size, reason, buf);
}
#else
@@ -416,8 +419,8 @@ i915_vma_resource_init_from_vma(struct i915_vma_resource *vma_res,
i915_vma_resource_init(vma_res, vma->vm, vma->pages, &vma->page_sizes,
obj->mm.rsgt, i915_gem_object_is_readonly(obj),
i915_gem_object_is_lmem(obj), obj->mm.region,
- vma->ops, vma->private, vma->node.start,
- vma->node.size, vma->size);
+ vma->ops, vma->private, __i915_vma_offset(vma),
+ __i915_vma_size(vma), vma->size, vma->guard);
}
/**
@@ -445,7 +448,7 @@ int i915_vma_bind(struct i915_vma *vma,
lockdep_assert_held(&vma->vm->mutex);
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
- GEM_BUG_ON(vma->size > vma->node.size);
+ GEM_BUG_ON(vma->size > i915_vma_size(vma));
if (GEM_DEBUG_WARN_ON(range_overflows(vma->node.start,
vma->node.size,
@@ -567,8 +570,8 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
vma->obj->base.size);
} else if (i915_vma_is_map_and_fenceable(vma)) {
ptr = io_mapping_map_wc(&i915_vm_to_ggtt(vma->vm)->iomap,
- vma->node.start,
- vma->node.size);
+ i915_vma_offset(vma),
+ i915_vma_size(vma));
} else {
ptr = (void __iomem *)
i915_gem_object_pin_map(vma->obj, I915_MAP_WC);
@@ -657,22 +660,26 @@ bool i915_vma_misplaced(const struct i915_vma *vma,
if (test_bit(I915_VMA_ERROR_BIT, __i915_vma_flags(vma)))
return true;
- if (vma->node.size < size)
+ if (i915_vma_size(vma) < size)
return true;
GEM_BUG_ON(alignment && !is_power_of_2(alignment));
- if (alignment && !IS_ALIGNED(vma->node.start, alignment))
+ if (alignment && !IS_ALIGNED(i915_vma_offset(vma), alignment))
return true;
if (flags & PIN_MAPPABLE && !i915_vma_is_map_and_fenceable(vma))
return true;
if (flags & PIN_OFFSET_BIAS &&
- vma->node.start < (flags & PIN_OFFSET_MASK))
+ i915_vma_offset(vma) < (flags & PIN_OFFSET_MASK))
return true;
if (flags & PIN_OFFSET_FIXED &&
- vma->node.start != (flags & PIN_OFFSET_MASK))
+ i915_vma_offset(vma) != (flags & PIN_OFFSET_MASK))
+ return true;
+
+ if (flags & PIN_OFFSET_GUARD &&
+ vma->guard < (flags & PIN_OFFSET_MASK))
return true;
return false;
@@ -685,10 +692,11 @@ void __i915_vma_set_map_and_fenceable(struct i915_vma *vma)
GEM_BUG_ON(!i915_vma_is_ggtt(vma));
GEM_BUG_ON(!vma->fence_size);
- fenceable = (vma->node.size >= vma->fence_size &&
- IS_ALIGNED(vma->node.start, vma->fence_alignment));
+ fenceable = (i915_vma_size(vma) >= vma->fence_size &&
+ IS_ALIGNED(i915_vma_offset(vma), vma->fence_alignment));
- mappable = vma->node.start + vma->fence_size <= i915_vm_to_ggtt(vma->vm)->mappable_end;
+ mappable = i915_ggtt_offset(vma) + vma->fence_size <=
+ i915_vm_to_ggtt(vma->vm)->mappable_end;
if (mappable && fenceable)
set_bit(I915_VMA_CAN_FENCE_BIT, __i915_vma_flags(vma));
@@ -746,15 +754,16 @@ static int
i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
u64 size, u64 alignment, u64 flags)
{
- unsigned long color;
+ unsigned long color, guard;
u64 start, end;
int ret;
GEM_BUG_ON(i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND));
GEM_BUG_ON(drm_mm_node_allocated(&vma->node));
+ GEM_BUG_ON(hweight64(flags & (PIN_OFFSET_GUARD | PIN_OFFSET_FIXED | PIN_OFFSET_BIAS)) > 1);
size = max(size, vma->size);
- alignment = max(alignment, vma->display_alignment);
+ alignment = max_t(typeof(alignment), alignment, vma->display_alignment);
if (flags & PIN_MAPPABLE) {
size = max_t(typeof(size), size, vma->fence_size);
alignment = max_t(typeof(alignment),
@@ -765,6 +774,18 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
GEM_BUG_ON(!IS_ALIGNED(alignment, I915_GTT_MIN_ALIGNMENT));
GEM_BUG_ON(!is_power_of_2(alignment));
+ guard = vma->guard; /* retain guard across rebinds */
+ if (flags & PIN_OFFSET_GUARD) {
+ GEM_BUG_ON(overflows_type(flags & PIN_OFFSET_MASK, u32));
+ guard = max_t(u32, guard, flags & PIN_OFFSET_MASK);
+ }
+ /*
+ * As we align the node upon insertion, but the hardware gets
+ * node.start + guard, the easiest way to make that work is
+ * to make the guard a multiple of the alignment size.
+ */
+ guard = ALIGN(guard, alignment);
+
start = flags & PIN_OFFSET_BIAS ? flags & PIN_OFFSET_MASK : 0;
GEM_BUG_ON(!IS_ALIGNED(start, I915_GTT_PAGE_SIZE));
@@ -776,21 +797,16 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
GEM_BUG_ON(!IS_ALIGNED(end, I915_GTT_PAGE_SIZE));
alignment = max(alignment, i915_vm_obj_min_alignment(vma->vm, vma->obj));
- /*
- * for compact-pt we round up the reservation to prevent
- * any smaller pages being used within the same PDE
- */
- if (NEEDS_COMPACT_PT(vma->vm->i915))
- size = round_up(size, alignment);
- /* If binding the object/GGTT view requires more space than the entire
+ /*
+ * If binding the object/GGTT view requires more space than the entire
* aperture has, reject it early before evicting everything in a vain
* attempt to find space.
*/
- if (size > end) {
- DRM_DEBUG("Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n",
- size, flags & PIN_MAPPABLE ? "mappable" : "total",
- end);
+ if (size > end - 2 * guard) {
+ drm_dbg(&to_i915(vma->obj->base.dev)->drm,
+ "Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n",
+ size, flags & PIN_MAPPABLE ? "mappable" : "total", end);
return -ENOSPC;
}
@@ -804,13 +820,23 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
if (!IS_ALIGNED(offset, alignment) ||
range_overflows(offset, size, end))
return -EINVAL;
+ /*
+ * The caller knows not of the guard added by others and
+ * requests for the offset of the start of its buffer
+ * to be fixed, which may not be the same as the position
+ * of the vma->node due to the guard pages.
+ */
+ if (offset < guard || offset + size > end - guard)
+ return -ENOSPC;
ret = i915_gem_gtt_reserve(vma->vm, ww, &vma->node,
- size, offset, color,
- flags);
+ size + 2 * guard,
+ offset - guard,
+ color, flags);
if (ret)
return ret;
} else {
+ size += 2 * guard;
/*
* We only support huge gtt pages through the 48b PPGTT,
* however we also don't want to force any alignment for
@@ -820,7 +846,8 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
* forseeable future. See also i915_ggtt_offset().
*/
if (upper_32_bits(end - 1) &&
- vma->page_sizes.sg > I915_GTT_PAGE_SIZE) {
+ vma->page_sizes.sg > I915_GTT_PAGE_SIZE &&
+ !HAS_64K_PAGES(vma->vm->i915)) {
/*
* We can't mix 64K and 4K PTEs in the same page-table
* (2M block), and so to avoid the ugliness and
@@ -857,6 +884,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
GEM_BUG_ON(!i915_gem_valid_gtt_space(vma, color));
list_move_tail(&vma->vm_link, &vma->vm->bound_list);
+ vma->guard = guard;
return 0;
}
@@ -909,7 +937,7 @@ rotate_pages(struct drm_i915_gem_object *obj, unsigned int offset,
struct sg_table *st, struct scatterlist *sg)
{
unsigned int column, row;
- unsigned int src_idx;
+ pgoff_t src_idx;
for (column = 0; column < width; column++) {
unsigned int left;
@@ -1015,7 +1043,7 @@ add_padding_pages(unsigned int count,
static struct scatterlist *
remap_tiled_color_plane_pages(struct drm_i915_gem_object *obj,
- unsigned int offset, unsigned int alignment_pad,
+ unsigned long offset, unsigned int alignment_pad,
unsigned int width, unsigned int height,
unsigned int src_stride, unsigned int dst_stride,
struct sg_table *st, struct scatterlist *sg,
@@ -1074,7 +1102,7 @@ remap_tiled_color_plane_pages(struct drm_i915_gem_object *obj,
static struct scatterlist *
remap_contiguous_pages(struct drm_i915_gem_object *obj,
- unsigned int obj_offset,
+ pgoff_t obj_offset,
unsigned int count,
struct sg_table *st, struct scatterlist *sg)
{
@@ -1107,7 +1135,7 @@ remap_contiguous_pages(struct drm_i915_gem_object *obj,
static struct scatterlist *
remap_linear_color_plane_pages(struct drm_i915_gem_object *obj,
- unsigned int obj_offset, unsigned int alignment_pad,
+ pgoff_t obj_offset, unsigned int alignment_pad,
unsigned int size,
struct sg_table *st, struct scatterlist *sg,
unsigned int *gtt_offset)
@@ -1547,6 +1575,8 @@ static int __i915_ggtt_pin(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
u32 align, unsigned int flags)
{
struct i915_address_space *vm = vma->vm;
+ struct intel_gt *gt;
+ struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
int err;
do {
@@ -1562,14 +1592,15 @@ static int __i915_ggtt_pin(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
}
/* Unlike i915_vma_pin, we don't take no for an answer! */
- flush_idle_contexts(vm->gt);
+ list_for_each_entry(gt, &ggtt->gt_list, ggtt_link)
+ flush_idle_contexts(gt);
if (mutex_lock_interruptible(&vm->mutex) == 0) {
/*
* We pass NULL ww here, as we don't want to unbind
* locked objects when called from execbuf when pinning
* is removed. This would probably regress badly.
*/
- i915_gem_evict_vm(vm, NULL);
+ i915_gem_evict_vm(vm, NULL, NULL);
mutex_unlock(&vm->mutex);
}
} while (1);
@@ -1847,6 +1878,11 @@ int _i915_vma_move_to_active(struct i915_vma *vma,
GEM_BUG_ON(!vma->pages);
+ if (!(flags & __EXEC_OBJECT_NO_REQUEST_AWAIT)) {
+ err = i915_request_await_object(rq, vma->obj, flags & EXEC_OBJECT_WRITE);
+ if (unlikely(err))
+ return err;
+ }
err = __i915_vma_move_to_active(vma, rq);
if (unlikely(err))
return err;
@@ -2114,7 +2150,7 @@ int i915_vma_unbind_async(struct i915_vma *vma, bool trylock_vm)
if (!obj->mm.rsgt)
return -EBUSY;
- err = dma_resv_reserve_fences(obj->base.resv, 1);
+ err = dma_resv_reserve_fences(obj->base.resv, 2);
if (err)
return -EBUSY;
Powered by cgit, Gruvboxed by Blaster4385.