Question inline, probably my lack of understanding of the code.
On 08/29/16 04:20, Christian König wrote:
> From: Christian König <christian.koenig at amd.com>
>
> Add a flag noting that a BO must be created using linear VRAM
> and set this flag on all in kernel users where appropriate.
>
> Hopefully I haven't missed anything.
>
> v2: add it in a few more places, fix CPU mapping.
>
> Signed-off-by: Christian König <christian.koenig at amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 +
> drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 6 ++++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 31 ++++++++++++++++++++++++++++++
> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 3 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c | 3 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 12 ++++++++++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 3 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 10 ++++++++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c | 9 +++++++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 ++++--
> drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 9 ++++++---
> drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 6 ++++--
> include/uapi/drm/amdgpu_drm.h | 2 ++
> 14 files changed, 85 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> index 545789f..b2d95a9 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> @@ -2499,6 +2499,7 @@ static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
> struct amdgpu_bo_va_mapping *
> amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
> uint64_t addr, struct amdgpu_bo **bo);
> +int amdgpu_cs_make_bos_linear(struct amdgpu_cs_parser *parser);
>
> #if defined(CONFIG_DRM_AMD_DAL)
> int amdgpu_dm_display_resume(struct amdgpu_device *adev );
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
> index 040993c..c2dfcd3 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
> @@ -146,7 +146,8 @@ static int amdgpu_cgs_alloc_gpu_mem(struct cgs_device *cgs_device,
> switch(type) {
> case CGS_GPU_MEM_TYPE__VISIBLE_CONTIG_FB:
> case CGS_GPU_MEM_TYPE__VISIBLE_FB:
> - flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
> + flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR;
> domain = AMDGPU_GEM_DOMAIN_VRAM;
> if (max_offset > adev->mc.real_vram_size)
> return -EINVAL;
> @@ -157,7 +158,8 @@ static int amdgpu_cgs_alloc_gpu_mem(struct cgs_device *cgs_device,
> break;
> case CGS_GPU_MEM_TYPE__INVISIBLE_CONTIG_FB:
> case CGS_GPU_MEM_TYPE__INVISIBLE_FB:
> - flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
> + flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR;
> domain = AMDGPU_GEM_DOMAIN_VRAM;
> if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
> place.fpfn =
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index b0e11ca..108408b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@ -1097,3 +1097,34 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
>
> return NULL;
> }
> +
> +/**
> + * amdgpu_cs_make_bos_linear - mark all BOs in this CS as linear
> + *
> + * @parser: command submission parser context
> + *
> + * Helper for UVD/VCE VM emulation, mark all BOs in this CS as linear.
> + */
> +int amdgpu_cs_make_bos_linear(struct amdgpu_cs_parser *parser)
> +{
> + unsigned i;
> + int r;
> +
> + if (!parser->bo_list)
> + return 0;
> +
> + for (i = 0; i < parser->bo_list->num_entries; i++) {
> + struct amdgpu_bo *bo = parser->bo_list->array[i].robj;
> +
> + if (bo->flags |= AMDGPU_GEM_CREATE_VRAM_LINEAR)
> + continue;
> +
> + bo->flags |= AMDGPU_GEM_CREATE_VRAM_LINEAR;
> + amdgpu_ttm_placement_from_domain(bo, bo->allowed_domains);
> + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
> + if (unlikely(r))
> + return r;
> + }
> +
> + return 0;
> +}
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 1b31a7c..38f5315 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -256,7 +256,8 @@ static int amdgpu_vram_scratch_init(struct amdgpu_device *adev)
> if (adev->vram_scratch.robj == NULL) {
> r = amdgpu_bo_create(adev, AMDGPU_GPU_PAGE_SIZE,
> PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &adev->vram_scratch.robj);
> if (r) {
> return r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
> index e694c99..efa9b79 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
> @@ -120,7 +120,8 @@ static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
> aligned_size = ALIGN(size, PAGE_SIZE);
> ret = amdgpu_gem_object_create(adev, aligned_size, 0,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> true, &gobj);
> if (ret) {
> printk(KERN_ERR "failed to allocate framebuffer (%d)\n",
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
> index 0feea34..b5982a9 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
> @@ -126,7 +126,8 @@ int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
> if (adev->gart.robj == NULL) {
> r = amdgpu_bo_create(adev, adev->gart.table_size,
> PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &adev->gart.robj);
> if (r) {
> return r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> index b17734e..6f83909 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> @@ -247,7 +247,8 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
> int r;
>
> r = amdgpu_bo_create(adev, size, align, true, domain,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, bo_ptr);
> if (r) {
> dev_err(adev->dev, "(%d) failed to allocate kernel bo\n", r);
> @@ -617,6 +618,8 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
>
> return 0;
> }
> +
> + bo->flags |= AMDGPU_GEM_CREATE_VRAM_LINEAR;
> amdgpu_ttm_placement_from_domain(bo, domain);
> for (i = 0; i < bo->placement.num_placement; i++) {
> /* force to pin into visible video ram */
> @@ -854,7 +857,9 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
>
> size = bo->mem.num_pages << PAGE_SHIFT;
> offset = bo->mem.start << PAGE_SHIFT;
> - if ((offset + size) <= adev->mc.visible_vram_size)
> + /* TODO: figure out how to map scattered VRAM to the CPU */
> + if ((offset + size) <= adev->mc.visible_vram_size &&
> + (abo->flags & AMDGPU_GEM_CREATE_VRAM_LINEAR))
> return 0;
This is confusing. Why are you adding the offset to the size to compare
against the size? Shouldn't it be size alone?
>
> /* Can't move a pinned BO to visible VRAM */
> @@ -862,6 +867,7 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
> return -EINVAL;
>
> /* hurrah the memory is not visible ! */
> + abo->flags |= AMDGPU_GEM_CREATE_VRAM_LINEAR;
> amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM);
> lpfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
> for (i = 0; i < abo->placement.num_placement; i++) {
> @@ -920,6 +926,8 @@ u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
> WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM);
> WARN_ON_ONCE(!ww_mutex_is_locked(&bo->tbo.resv->lock) &&
> !bo->pin_count);
> + WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_VRAM &&
> + !(bo->flags & AMDGPU_GEM_CREATE_VRAM_LINEAR));
>
> return bo->tbo.offset;
> }
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index e2f6a46..ea480bb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -1024,7 +1024,8 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
>
> r = amdgpu_bo_create(adev, 256 * 1024, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &adev->stollen_vga_memory);
> if (r) {
> return r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> index 5888e8a..5f5380c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> @@ -891,6 +891,10 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
> return -EINVAL;
> }
>
> + r = amdgpu_cs_make_bos_linear(parser);
> + if (r)
> + return r;
> +
> ctx.parser = parser;
> ctx.buf_sizes = buf_sizes;
> ctx.ib_idx = ib_idx;
> @@ -1007,7 +1011,8 @@ int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
>
> r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &bo);
> if (r)
> return r;
> @@ -1056,7 +1061,8 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
>
> r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &bo);
> if (r)
> return r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
> index 9b71d6c..bb97a8e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
> @@ -157,7 +157,8 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
>
> r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &adev->vce.vcpu_bo);
> if (r) {
> dev_err(adev->dev, "(%d) failed to allocate VCE bo\n", r);
> @@ -634,7 +635,11 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
> uint32_t allocated = 0;
> uint32_t tmp, handle = 0;
> uint32_t *size = &tmp;
> - int i, r = 0, idx = 0;
> + int i, r, idx = 0;
> +
> + r = amdgpu_cs_make_bos_linear(p);
> + if (r)
> + return r;
>
> while (idx < ib->length_dw) {
> uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 7660f82..ea1bd67 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -1398,7 +1398,8 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
> AMDGPU_GPU_PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
> - AMDGPU_GEM_CREATE_SHADOW,
> + AMDGPU_GEM_CREATE_SHADOW |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, resv, &pt);
> if (r)
> goto error_free;
> @@ -1598,7 +1599,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
> r = amdgpu_bo_create(adev, pd_size, align, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
> - AMDGPU_GEM_CREATE_SHADOW,
> + AMDGPU_GEM_CREATE_SHADOW |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL, &vm->page_directory);
> if (r)
> goto error_free_sched_entity;
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
> index f055d49..1517a8c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
> @@ -3213,7 +3213,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
> if (adev->gfx.rlc.save_restore_obj == NULL) {
> r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL,
> &adev->gfx.rlc.save_restore_obj);
> if (r) {
> @@ -3257,7 +3258,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
> if (adev->gfx.rlc.clear_state_obj == NULL) {
> r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL,
> &adev->gfx.rlc.clear_state_obj);
> if (r) {
> @@ -3297,7 +3299,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
> if (adev->gfx.rlc.cp_table_obj == NULL) {
> r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL,
> &adev->gfx.rlc.cp_table_obj);
> if (r) {
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> index 6e01392..26e614e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> @@ -1273,7 +1273,8 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
> if (adev->gfx.rlc.clear_state_obj == NULL) {
> r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL,
> &adev->gfx.rlc.clear_state_obj);
> if (r) {
> @@ -1315,7 +1316,8 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
> if (adev->gfx.rlc.cp_table_obj == NULL) {
> r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
> AMDGPU_GEM_DOMAIN_VRAM,
> - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
> + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
> + AMDGPU_GEM_CREATE_VRAM_LINEAR,
> NULL, NULL,
> &adev->gfx.rlc.cp_table_obj);
> if (r) {
> diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
> index 403b676..2aede6a 100644
> --- a/include/uapi/drm/amdgpu_drm.h
> +++ b/include/uapi/drm/amdgpu_drm.h
> @@ -79,6 +79,8 @@
> #define AMDGPU_GEM_CREATE_VRAM_CLEARED (1 << 3)
> /* Flag that create shadow bo(GTT) while allocating vram bo */
> #define AMDGPU_GEM_CREATE_SHADOW (1 << 4)
> +/* Flag that allocating the BO should use linear VRAM */
> +#define AMDGPU_GEM_CREATE_VRAM_LINEAR (1 << 5)
>
> struct drm_amdgpu_gem_create_in {
> /** the requested memory size */
