@@ -284,8 +284,11 @@ static void gen8_ggtt_insert_entries(struct i915_address_space *vm,
*/
gte = (gen8_pte_t __iomem *)ggtt->gsm;
- gte += vma_res->start / I915_GTT_PAGE_SIZE;
- end = gte + vma_res->node_size / I915_GTT_PAGE_SIZE;
+ gte += (vma_res->start - vma_res->guard) / I915_GTT_PAGE_SIZE;
+ end = gte + vma_res->guard / I915_GTT_PAGE_SIZE;
+ while (gte < end)
+ gen8_set_pte(gte++, vm->scratch[0]->encode);
+ end += (vma_res->node_size + vma_res->guard) / I915_GTT_PAGE_SIZE;
for_each_sgt_daddr(addr, iter, vma_res->bi.pages)
gen8_set_pte(gte++, pte_encode | addr);
@@ -335,9 +338,12 @@ static void gen6_ggtt_insert_entries(struct i915_address_space *vm,
dma_addr_t addr;
gte = (gen6_pte_t __iomem *)ggtt->gsm;
- gte += vma_res->start / I915_GTT_PAGE_SIZE;
- end = gte + vma_res->node_size / I915_GTT_PAGE_SIZE;
+ gte += (vma_res->start - vma_res->guard) / I915_GTT_PAGE_SIZE;
+ end = gte + vma_res->guard / I915_GTT_PAGE_SIZE;
+ while (gte < end)
+ iowrite32(vm->scratch[0]->encode, gte++);
+ end += (vma_res->node_size + vma_res->guard) / I915_GTT_PAGE_SIZE;
for_each_sgt_daddr(addr, iter, vma_res->bi.pages)
iowrite32(vm->pte_encode(addr, level, flags), gte++);
GEM_BUG_ON(gte > end);
@@ -44,7 +44,8 @@ int i915_gem_gtt_insert(struct i915_address_space *vm,
#define PIN_HIGH BIT_ULL(5)
#define PIN_OFFSET_BIAS BIT_ULL(6)
#define PIN_OFFSET_FIXED BIT_ULL(7)
-#define PIN_VALIDATE BIT_ULL(8) /* validate placement only, no need to call unpin() */
+#define PIN_OFFSET_GUARD BIT_ULL(8)
+#define PIN_VALIDATE BIT_ULL(9) /* validate placement only, no need to call unpin() */
#define PIN_GLOBAL BIT_ULL(10) /* I915_VMA_GLOBAL_BIND */
#define PIN_USER BIT_ULL(11) /* I915_VMA_LOCAL_BIND */
@@ -408,7 +408,7 @@ i915_vma_resource_init_from_vma(struct i915_vma_resource *vma_res,
obj->mm.rsgt, i915_gem_object_is_readonly(obj),
i915_gem_object_is_lmem(obj), obj->mm.region,
vma->ops, vma->private, __i915_vma_offset(vma),
- __i915_vma_size(vma), vma->size);
+ __i915_vma_size(vma), vma->size, vma->guard);
}
/**
@@ -739,7 +739,7 @@ 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;
@@ -747,7 +747,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
GEM_BUG_ON(drm_mm_node_allocated(&vma->node));
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),
@@ -758,6 +758,9 @@ 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 */
+ guard = ALIGN(guard, alignment);
+
start = flags & PIN_OFFSET_BIAS ? flags & PIN_OFFSET_MASK : 0;
GEM_BUG_ON(!IS_ALIGNED(start, I915_GTT_PAGE_SIZE));
@@ -767,6 +770,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
if (flags & PIN_ZONE_4G)
end = min_t(u64, end, (1ULL << 32) - I915_GTT_PAGE_SIZE);
GEM_BUG_ON(!IS_ALIGNED(end, I915_GTT_PAGE_SIZE));
+ GEM_BUG_ON(2 * guard > end);
alignment = max(alignment, i915_vm_obj_min_alignment(vma->vm, vma->obj));
/*
@@ -780,7 +784,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
* aperture has, reject it early before evicting everything in a vain
* attempt to find space.
*/
- if (size > end) {
+ if (size > end - 2 * guard) {
DRM_DEBUG("Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n",
size, flags & PIN_MAPPABLE ? "mappable" : "total",
end);
@@ -797,13 +801,23 @@ 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
@@ -850,6 +864,7 @@ 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;
}
@@ -127,7 +127,7 @@ static inline bool i915_vma_is_closed(const struct i915_vma *vma)
/* Internal use only. */
static inline u64 __i915_vma_size(const struct i915_vma *vma)
{
- return vma->node.size;
+ return vma->node.size - 2 * vma->guard;
}
/**
@@ -149,7 +149,8 @@ static inline u64 i915_vma_size(const struct i915_vma *vma)
/* Internal use only. */
static inline u64 __i915_vma_offset(const struct i915_vma *vma)
{
- return vma->node.start;
+ /* The actual start of the vma->pages is after the guard pages. */
+ return vma->node.start + vma->guard;
}
/**
@@ -34,8 +34,8 @@ static struct kmem_cache *slab_vma_resources;
* and removal of fences increases as O(ln(pending_unbinds)) instead of
* O(1) for a single fence without interval tree.
*/
-#define VMA_RES_START(_node) ((_node)->start)
-#define VMA_RES_LAST(_node) ((_node)->start + (_node)->node_size - 1)
+#define VMA_RES_START(_node) ((_node)->start - (_node)->guard)
+#define VMA_RES_LAST(_node) ((_node)->start + (_node)->node_size + (_node)->guard - 1)
INTERVAL_TREE_DEFINE(struct i915_vma_resource, rb,
u64, __subtree_last,
VMA_RES_START, VMA_RES_LAST, static, vma_res_itree);
@@ -57,6 +57,7 @@ struct i915_page_sizes {
* @node_size: Size of the allocated range manager node with padding
* subtracted.
* @vma_size: Bind size.
+ * @guard: The size of guard area preceding and trailing the bind.
* @page_sizes_gtt: Resulting page sizes from the bind operation.
* @bound_flags: Flags indicating binding status.
* @allocated: Backend private data. TODO: Should move into @private.
@@ -114,6 +115,7 @@ struct i915_vma_resource {
u64 start;
u64 node_size;
u64 vma_size;
+ u32 guard;
u32 page_sizes_gtt;
u32 bound_flags;
@@ -175,6 +177,7 @@ static inline void i915_vma_resource_put(struct i915_vma_resource *vma_res)
* @start: Offset into the address space of bind range start after padding.
* @node_size: Size of the allocated range manager node minus padding.
* @size: Bind size.
+ * @guard: The size of the guard area preceding and trailing the bind.
*
* Initializes a vma resource allocated using i915_vma_resource_alloc().
* The reason for having separate allocate and initialize function is that
@@ -193,7 +196,8 @@ static inline void i915_vma_resource_init(struct i915_vma_resource *vma_res,
void *private,
u64 start,
u64 node_size,
- u64 size)
+ u64 size,
+ u32 guard)
{
__i915_vma_resource_init(vma_res);
vma_res->vm = vm;
@@ -211,6 +215,7 @@ static inline void i915_vma_resource_init(struct i915_vma_resource *vma_res,
vma_res->start = start;
vma_res->node_size = node_size;
vma_res->vma_size = size;
+ vma_res->guard = guard;
}
static inline void i915_vma_resource_fini(struct i915_vma_resource *vma_res)
@@ -197,14 +197,15 @@ struct i915_vma {
struct i915_fence_reg *fence;
u64 size;
- u64 display_alignment;
struct i915_page_sizes page_sizes;
/* mmap-offset associated with fencing for this vma */
struct i915_mmap_offset *mmo;
+ u32 guard; /* padding allocated around vma->pages within the node */
u32 fence_size;
u32 fence_alignment;
+ u32 display_alignment;
/**
* Count of the number of times this vma has been opened by different