#include "amdgpu.h"
#include "amdgpu_trace.h"
#include "amdgpu_amdkfd.h"
+#include "amdgpu_vram_mgr.h"
/**
* DOC: amdgpu_object
if (!amdgpu_bo_support_uswc(bo->flags))
bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
- if (adev->ras_enabled)
- bo->flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE;
+ bo->flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE;
bo->tbo.bdev = &adev->mman.bdev;
if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA |
return r;
if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
- bo->tbo.resource->mem_type == TTM_PL_VRAM &&
- amdgpu_bo_in_cpu_visible_vram(bo))
+ amdgpu_res_cpu_visible(adev, bo->tbo.resource))
amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved,
ctx.bytes_moved);
else
bo->tbo.resource->mem_type == TTM_PL_VRAM) {
struct dma_fence *fence;
- r = amdgpu_fill_buffer(bo, 0, bo->tbo.base.resv, &fence, true);
+ r = amdgpu_ttm_clear_buffer(bo, bo->tbo.base.resv, &fence);
if (unlikely(r))
goto fail_unreserve;
void amdgpu_bo_get_memory(struct amdgpu_bo *bo,
struct amdgpu_mem_stats *stats)
{
+ struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+ struct ttm_resource *res = bo->tbo.resource;
uint64_t size = amdgpu_bo_size(bo);
struct drm_gem_object *obj;
unsigned int domain;
bool shared;
/* Abort if the BO doesn't currently have a backing store */
- if (!bo->tbo.resource)
+ if (!res)
return;
obj = &bo->tbo.base;
shared = drm_gem_object_is_shared_for_memory_stats(obj);
- domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type);
+ domain = amdgpu_mem_type_to_domain(res->mem_type);
switch (domain) {
case AMDGPU_GEM_DOMAIN_VRAM:
stats->vram += size;
- if (amdgpu_bo_in_cpu_visible_vram(bo))
+ if (amdgpu_res_cpu_visible(adev, bo->tbo.resource))
stats->visible_vram += size;
if (shared)
stats->vram_shared += size;
if (WARN_ON_ONCE(!dma_resv_trylock(bo->base.resv)))
return;
- r = amdgpu_fill_buffer(abo, AMDGPU_POISON, bo->base.resv, &fence, true);
+ r = amdgpu_fill_buffer(abo, 0, bo->base.resv, &fence, true);
if (!WARN_ON(r)) {
+ amdgpu_vram_mgr_set_cleared(bo->resource);
amdgpu_bo_fence(abo, fence, false);
dma_fence_put(fence);
}
/* Remember that this BO was accessed by the CPU */
abo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
- if (bo->resource->mem_type != TTM_PL_VRAM)
- return 0;
-
- if (amdgpu_bo_in_cpu_visible_vram(abo))
+ if (amdgpu_res_cpu_visible(adev, bo->resource))
return 0;
/* Can't move a pinned BO to visible VRAM */
/* this should never happen */
if (bo->resource->mem_type == TTM_PL_VRAM &&
- !amdgpu_bo_in_cpu_visible_vram(abo))
+ !amdgpu_res_cpu_visible(adev, bo->resource))
return VM_FAULT_SIGBUS;
ttm_bo_move_to_lru_tail_unlocked(bo);
*/
u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m)
{
+ struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
struct dma_buf_attachment *attachment;
struct dma_buf *dma_buf;
const char *placement;
if (dma_resv_trylock(bo->tbo.base.resv)) {
unsigned int domain;
+
domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type);
switch (domain) {
case AMDGPU_GEM_DOMAIN_VRAM:
- if (amdgpu_bo_in_cpu_visible_vram(bo))
+ if (amdgpu_res_cpu_visible(adev, bo->tbo.resource))
placement = "VRAM VISIBLE";
else
placement = "VRAM";
} else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
!(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
- amdgpu_bo_in_cpu_visible_vram(abo)) {
+ amdgpu_res_cpu_visible(adev, bo->resource)) {
/* Try evicting to the CPU inaccessible part of VRAM
* first, but only set GTT as busy placement, so this
(abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) {
struct dma_fence *wipe_fence = NULL;
- r = amdgpu_fill_buffer(abo, AMDGPU_POISON, NULL, &wipe_fence,
- false);
+ r = amdgpu_fill_buffer(abo, 0, NULL, &wipe_fence,
+ false);
if (r) {
goto error;
} else if (wipe_fence) {
+ amdgpu_vram_mgr_set_cleared(bo->resource);
dma_fence_put(fence);
fence = wipe_fence;
}
return r;
}
- /*
- * amdgpu_mem_visible - Check that memory can be accessed by ttm_bo_move_memcpy
+ /**
+ * amdgpu_res_cpu_visible - Check that resource can be accessed by CPU
+ * @adev: amdgpu device
+ * @res: the resource to check
*
- * Called by amdgpu_bo_move()
+ * Returns: true if the full resource is CPU visible, false otherwise.
*/
- static bool amdgpu_mem_visible(struct amdgpu_device *adev,
- struct ttm_resource *mem)
+ bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
+ struct ttm_resource *res)
{
- u64 mem_size = (u64)mem->size;
struct amdgpu_res_cursor cursor;
- u64 end;
- if (mem->mem_type == TTM_PL_SYSTEM ||
- mem->mem_type == TTM_PL_TT)
+ if (!res)
+ return false;
+
+ if (res->mem_type == TTM_PL_SYSTEM || res->mem_type == TTM_PL_TT ||
+ res->mem_type == AMDGPU_PL_PREEMPT)
return true;
- if (mem->mem_type != TTM_PL_VRAM)
+
+ if (res->mem_type != TTM_PL_VRAM)
return false;
- amdgpu_res_first(mem, 0, mem_size, &cursor);
- end = cursor.start + cursor.size;
+ amdgpu_res_first(res, 0, res->size, &cursor);
while (cursor.remaining) {
+ if ((cursor.start + cursor.size) >= adev->gmc.visible_vram_size)
+ return false;
amdgpu_res_next(&cursor, cursor.size);
+ }
- if (!cursor.remaining)
- break;
+ return true;
+ }
- /* ttm_resource_ioremap only supports contiguous memory */
- if (end != cursor.start)
- return false;
+ /*
+ * amdgpu_res_copyable - Check that memory can be accessed by ttm_bo_move_memcpy
+ *
+ * Called by amdgpu_bo_move()
+ */
+ static bool amdgpu_res_copyable(struct amdgpu_device *adev,
+ struct ttm_resource *mem)
+ {
+ if (!amdgpu_res_cpu_visible(adev, mem))
+ return false;
- end = cursor.start + cursor.size;
- }
+ /* ttm_resource_ioremap only supports contiguous memory */
+ if (mem->mem_type == TTM_PL_VRAM &&
+ !(mem->placement & TTM_PL_FLAG_CONTIGUOUS))
+ return false;
- return end <= adev->gmc.visible_vram_size;
+ return true;
}
/*
if (r) {
/* Check that all memory is CPU accessible */
- if (!amdgpu_mem_visible(adev, old_mem) ||
- !amdgpu_mem_visible(adev, new_mem)) {
+ if (!amdgpu_res_copyable(adev, old_mem) ||
+ !amdgpu_res_copyable(adev, new_mem)) {
pr_err("Move buffer fallback to memcpy unavailable\n");
return r;
}
struct ttm_resource *mem)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
- size_t bus_size = (size_t)mem->size;
switch (mem->mem_type) {
case TTM_PL_SYSTEM:
break;
case TTM_PL_VRAM:
mem->bus.offset = mem->start << PAGE_SHIFT;
- /* check if it's visible */
- if ((mem->bus.offset + bus_size) > adev->gmc.visible_vram_size)
- return -EINVAL;
if (adev->mman.aper_base_kaddr &&
mem->placement & TTM_PL_FLAG_CONTIGUOUS)
return 0;
}
+/**
+ * amdgpu_ttm_clear_buffer - clear memory buffers
+ * @bo: amdgpu buffer object
+ * @resv: reservation object
+ * @fence: dma_fence associated with the operation
+ *
+ * Clear the memory buffer resource.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
+ struct dma_resv *resv,
+ struct dma_fence **fence)
+{
+ struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+ struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+ struct amdgpu_res_cursor cursor;
+ u64 addr;
+ int r;
+
+ if (!adev->mman.buffer_funcs_enabled)
+ return -EINVAL;
+
+ if (!fence)
+ return -EINVAL;
+
+ *fence = dma_fence_get_stub();
+
+ amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &cursor);
+
+ mutex_lock(&adev->mman.gtt_window_lock);
+ while (cursor.remaining) {
+ struct dma_fence *next = NULL;
+ u64 size;
+
+ if (amdgpu_res_cleared(&cursor)) {
+ amdgpu_res_next(&cursor, cursor.size);
+ continue;
+ }
+
+ /* Never clear more than 256MiB at once to avoid timeouts */
+ size = min(cursor.size, 256ULL << 20);
+
+ r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &cursor,
+ 1, ring, false, &size, &addr);
+ if (r)
+ goto err;
+
+ r = amdgpu_ttm_fill_mem(ring, 0, addr, size, resv,
+ &next, true, true);
+ if (r)
+ goto err;
+
+ dma_fence_put(*fence);
+ *fence = next;
+
+ amdgpu_res_next(&cursor, size);
+ }
+err:
+ mutex_unlock(&adev->mman.gtt_window_lock);
+
+ return r;
+}
+
int amdgpu_fill_buffer(struct amdgpu_bo *bo,
uint32_t src_data,
struct dma_resv *resv,
projects / J-linux.git / commitdiff