[PATCH 06/10] drm/ttm: remove no_wait_reserve, v2

All items on the lru list are always reservable, so this is a stupid thing to keep. Not only that, it is used in a way which would guarantee deadlocks if it were ever to be set to block on reserve.

This is a lot of churn, but mostly because of the removal of the argument which can be nested arbitrarily deeply in many places.

Only 1 place changed meaningfully, the rest is just churn that through an arbitrarily complex chain that can end up calling ttm_mem_evict_first. The change is simply that ttm_mem_evict_first always assumes no_wait_reserve is true.

This should work since currently ttm guarantees items on the lru are always reservable, and reserving items blockingly with some bo held are enough to cause you to run into a deadlock.

v2: - Warn if -EBUSY is returned on reservation, all objects on the list should be reservable. Adjusted patch slightly due to conflicts.

Signed-off-by: Maarten Lankhorst maarten.lankhorst@canonical.com --- drivers/gpu/drm/ast/ast_ttm.c | 10 +++--- drivers/gpu/drm/cirrus/cirrus_ttm.c | 10 +++--- drivers/gpu/drm/mgag200/mgag200_ttm.c | 10 +++--- drivers/gpu/drm/nouveau/nouveau_bo.c | 55 ++++++++++++++--------------- drivers/gpu/drm/nouveau/nouveau_bo.h | 2 +- drivers/gpu/drm/nouveau/nouveau_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 8 ++--- drivers/gpu/drm/radeon/radeon_ttm.c | 31 +++++++++-------- drivers/gpu/drm/ttm/ttm_bo.c | 60 ++++++++++++++------------------ drivers/gpu/drm/ttm/ttm_bo_util.c | 6 ++-- drivers/gpu/drm/vmwgfx/vmwgfx_dmabuf.c | 13 ++++--- drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c | 4 +-- drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 5 ++- include/drm/ttm/ttm_bo_api.h | 3 +- include/drm/ttm/ttm_bo_driver.h | 19 ++++------ 15 files changed, 110 insertions(+), 128 deletions(-)

diff --git a/drivers/gpu/drm/ast/ast_ttm.c b/drivers/gpu/drm/ast/ast_ttm.c index 1a026ac..adcac90 100644 --- a/drivers/gpu/drm/ast/ast_ttm.c +++ b/drivers/gpu/drm/ast/ast_ttm.c @@ -186,11 +186,11 @@ static void ast_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *

static int ast_bo_move(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { int r; - r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); return r; }

@@ -383,7 +383,7 @@ int ast_bo_pin(struct ast_bo *bo, u32 pl_flag, u64 *gpu_addr) ast_ttm_placement(bo, pl_flag); for (i = 0; i < bo->placement.num_placement; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -406,7 +406,7 @@ int ast_bo_unpin(struct ast_bo *bo)

for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -431,7 +431,7 @@ int ast_bo_push_sysram(struct ast_bo *bo) for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT;

- ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) { DRM_ERROR("pushing to VRAM failed\n"); return ret; diff --git a/drivers/gpu/drm/cirrus/cirrus_ttm.c b/drivers/gpu/drm/cirrus/cirrus_ttm.c index bc83f83..b0e0365 100644 --- a/drivers/gpu/drm/cirrus/cirrus_ttm.c +++ b/drivers/gpu/drm/cirrus/cirrus_ttm.c @@ -186,11 +186,11 @@ static void cirrus_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_re

static int cirrus_bo_move(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { int r; - r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); return r; }

@@ -388,7 +388,7 @@ int cirrus_bo_pin(struct cirrus_bo *bo, u32 pl_flag, u64 *gpu_addr) cirrus_ttm_placement(bo, pl_flag); for (i = 0; i < bo->placement.num_placement; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -411,7 +411,7 @@ int cirrus_bo_unpin(struct cirrus_bo *bo)

for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -436,7 +436,7 @@ int cirrus_bo_push_sysram(struct cirrus_bo *bo) for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT;

- ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) { DRM_ERROR("pushing to VRAM failed\n"); return ret; diff --git a/drivers/gpu/drm/mgag200/mgag200_ttm.c b/drivers/gpu/drm/mgag200/mgag200_ttm.c index 1504699..44ee1d8 100644 --- a/drivers/gpu/drm/mgag200/mgag200_ttm.c +++ b/drivers/gpu/drm/mgag200/mgag200_ttm.c @@ -186,11 +186,11 @@ static void mgag200_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_r

static int mgag200_bo_move(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { int r; - r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); return r; }

@@ -382,7 +382,7 @@ int mgag200_bo_pin(struct mgag200_bo *bo, u32 pl_flag, u64 *gpu_addr) mgag200_ttm_placement(bo, pl_flag); for (i = 0; i < bo->placement.num_placement; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -405,7 +405,7 @@ int mgag200_bo_unpin(struct mgag200_bo *bo)

for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT; - ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) return ret;

@@ -430,7 +430,7 @@ int mgag200_bo_push_sysram(struct mgag200_bo *bo) for (i = 0; i < bo->placement.num_placement ; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT;

- ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false, false); + ret = ttm_bo_validate(&bo->bo, &bo->placement, false, false); if (ret) { DRM_ERROR("pushing to VRAM failed\n"); return ret; diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 3ee2295..b6f1c1e 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -315,7 +315,7 @@ nouveau_bo_pin(struct nouveau_bo *nvbo, uint32_t memtype)

nouveau_bo_placement_set(nvbo, memtype, 0);

- ret = nouveau_bo_validate(nvbo, false, false, false); + ret = nouveau_bo_validate(nvbo, false, false); if (ret == 0) { switch (bo->mem.mem_type) { case TTM_PL_VRAM: @@ -351,7 +351,7 @@ nouveau_bo_unpin(struct nouveau_bo *nvbo)

nouveau_bo_placement_set(nvbo, bo->mem.placement, 0);

- ret = nouveau_bo_validate(nvbo, false, false, false); + ret = nouveau_bo_validate(nvbo, false, false); if (ret == 0) { switch (bo->mem.mem_type) { case TTM_PL_VRAM: @@ -392,12 +392,12 @@ nouveau_bo_unmap(struct nouveau_bo *nvbo)

int nouveau_bo_validate(struct nouveau_bo *nvbo, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu) + bool no_wait_gpu) { int ret;

- ret = ttm_bo_validate(&nvbo->bo, &nvbo->placement, interruptible, - no_wait_reserve, no_wait_gpu); + ret = ttm_bo_validate(&nvbo->bo, &nvbo->placement, + interruptible, no_wait_gpu); if (ret) return ret;

@@ -556,8 +556,7 @@ nouveau_bo_evict_flags(struct ttm_buffer_object *bo, struct ttm_placement *pl) static int nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan, struct nouveau_bo *nvbo, bool evict, - bool no_wait_reserve, bool no_wait_gpu, - struct ttm_mem_reg *new_mem) + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct nouveau_fence *fence = NULL; int ret; @@ -567,7 +566,7 @@ nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan, return ret;

ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, evict, - no_wait_reserve, no_wait_gpu, new_mem); + no_wait_gpu, new_mem); nouveau_fence_unref(&fence); return ret; } @@ -965,8 +964,7 @@ nouveau_vma_getmap(struct nouveau_channel *chan, struct nouveau_bo *nvbo,

static int nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, - bool no_wait_reserve, bool no_wait_gpu, - struct ttm_mem_reg *new_mem) + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct nouveau_drm *drm = nouveau_bdev(bo->bdev); struct nouveau_channel *chan = chan = drm->channel; @@ -995,7 +993,6 @@ nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, ret = drm->ttm.move(chan, bo, &bo->mem, new_mem); if (ret == 0) { ret = nouveau_bo_move_accel_cleanup(chan, nvbo, evict, - no_wait_reserve, no_wait_gpu, new_mem); }

@@ -1064,8 +1061,7 @@ nouveau_bo_move_init(struct nouveau_drm *drm)

static int nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr, - bool no_wait_reserve, bool no_wait_gpu, - struct ttm_mem_reg *new_mem) + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { u32 placement_memtype = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING; struct ttm_placement placement; @@ -1078,7 +1074,7 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr,

tmp_mem = *new_mem; tmp_mem.mm_node = NULL; - ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_gpu); if (ret) return ret;

@@ -1086,11 +1082,11 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr, if (ret) goto out;

- ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_reserve, no_wait_gpu, &tmp_mem); + ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_gpu, &tmp_mem); if (ret) goto out;

- ret = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, new_mem); + ret = ttm_bo_move_ttm(bo, true, no_wait_gpu, new_mem); out: ttm_bo_mem_put(bo, &tmp_mem); return ret; @@ -1098,8 +1094,7 @@ out:

static int nouveau_bo_move_flips(struct ttm_buffer_object *bo, bool evict, bool intr, - bool no_wait_reserve, bool no_wait_gpu, - struct ttm_mem_reg *new_mem) + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { u32 placement_memtype = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING; struct ttm_placement placement; @@ -1112,15 +1107,15 @@ nouveau_bo_move_flips(struct ttm_buffer_object *bo, bool evict, bool intr,

tmp_mem = *new_mem; tmp_mem.mm_node = NULL; - ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_gpu); if (ret) return ret;

- ret = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem); + ret = ttm_bo_move_ttm(bo, true, no_wait_gpu, &tmp_mem); if (ret) goto out;

- ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_reserve, no_wait_gpu, new_mem); + ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_gpu, new_mem); if (ret) goto out;

@@ -1195,8 +1190,7 @@ nouveau_bo_vm_cleanup(struct ttm_buffer_object *bo,

static int nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr, - bool no_wait_reserve, bool no_wait_gpu, - struct ttm_mem_reg *new_mem) + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct nouveau_drm *drm = nouveau_bdev(bo->bdev); struct nouveau_bo *nvbo = nouveau_bo(bo); @@ -1220,23 +1214,26 @@ nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr,

/* CPU copy if we have no accelerated method available */ if (!drm->ttm.move) { - ret = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); goto out; }

/* Hardware assisted copy. */ if (new_mem->mem_type == TTM_PL_SYSTEM) - ret = nouveau_bo_move_flipd(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + ret = nouveau_bo_move_flipd(bo, evict, intr, + no_wait_gpu, new_mem); else if (old_mem->mem_type == TTM_PL_SYSTEM) - ret = nouveau_bo_move_flips(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + ret = nouveau_bo_move_flips(bo, evict, intr, + no_wait_gpu, new_mem); else - ret = nouveau_bo_move_m2mf(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + ret = nouveau_bo_move_m2mf(bo, evict, intr, + no_wait_gpu, new_mem);

if (!ret) goto out;

/* Fallback to software copy. */ - ret = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem);

out: if (nv_device(drm->device)->card_type < NV_50) { @@ -1343,7 +1340,7 @@ nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo) nvbo->placement.fpfn = 0; nvbo->placement.lpfn = mappable; nouveau_bo_placement_set(nvbo, TTM_PL_FLAG_VRAM, 0); - return nouveau_bo_validate(nvbo, false, true, false); + return nouveau_bo_validate(nvbo, false, false); }

static int diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h index 61b8980..689b59b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.h +++ b/drivers/gpu/drm/nouveau/nouveau_bo.h @@ -76,7 +76,7 @@ u32 nouveau_bo_rd32(struct nouveau_bo *, unsigned index); void nouveau_bo_wr32(struct nouveau_bo *, unsigned index, u32 val); void nouveau_bo_fence(struct nouveau_bo *, struct nouveau_fence *); int nouveau_bo_validate(struct nouveau_bo *, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu); + bool no_wait_gpu);

struct nouveau_vma * nouveau_bo_vma_find(struct nouveau_bo *, struct nouveau_vm *); diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index 6d8391d..7b9364b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -434,7 +434,7 @@ validate_list(struct nouveau_channel *chan, struct list_head *list, return ret; }

- ret = nouveau_bo_validate(nvbo, true, false, false); + ret = nouveau_bo_validate(nvbo, true, false); if (unlikely(ret)) { if (ret != -ERESTARTSYS) NV_ERROR(drm, "fail ttm_validate\n"); diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 50aa508..2986686 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -240,7 +240,7 @@ int radeon_bo_pin_restricted(struct radeon_bo *bo, u32 domain, u64 max_offset, } for (i = 0; i < bo->placement.num_placement; i++) bo->placements[i] |= TTM_PL_FLAG_NO_EVICT; - r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false, false); + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false); if (likely(r == 0)) { bo->pin_count = 1; if (gpu_addr != NULL) @@ -269,7 +269,7 @@ int radeon_bo_unpin(struct radeon_bo *bo) return 0; for (i = 0; i < bo->placement.num_placement; i++) bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT; - r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false, false); + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false); if (unlikely(r != 0)) dev_err(bo->rdev->dev, "%p validate failed for unpin\n", bo); return r; @@ -355,7 +355,7 @@ int radeon_bo_list_validate(struct list_head *head) retry: radeon_ttm_placement_from_domain(bo, domain); r = ttm_bo_validate(&bo->tbo, &bo->placement, - true, false, false); + true, false); if (unlikely(r)) { if (r != -ERESTARTSYS && domain == RADEON_GEM_DOMAIN_VRAM) { domain |= RADEON_GEM_DOMAIN_GTT; @@ -575,7 +575,7 @@ int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) /* hurrah the memory is not visible ! */ radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_VRAM); rbo->placement.lpfn = rdev->mc.visible_vram_size >> PAGE_SHIFT; - r = ttm_bo_validate(bo, &rbo->placement, false, true, false); + r = ttm_bo_validate(bo, &rbo->placement, false, false); if (unlikely(r != 0)) return r; offset = bo->mem.start << PAGE_SHIFT; diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 563c8ed..1d8ff2f 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -216,7 +216,7 @@ static void radeon_move_null(struct ttm_buffer_object *bo, }

static int radeon_move_blit(struct ttm_buffer_object *bo, - bool evict, int no_wait_reserve, bool no_wait_gpu, + bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem, struct ttm_mem_reg *old_mem) { @@ -266,14 +266,14 @@ static int radeon_move_blit(struct ttm_buffer_object *bo, &fence); /* FIXME: handle copy error */ r = ttm_bo_move_accel_cleanup(bo, (void *)fence, - evict, no_wait_reserve, no_wait_gpu, new_mem); + evict, no_wait_gpu, new_mem); radeon_fence_unref(&fence); return r; }

static int radeon_move_vram_ram(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct radeon_device *rdev; @@ -294,7 +294,7 @@ static int radeon_move_vram_ram(struct ttm_buffer_object *bo, placement.busy_placement = &placements; placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT; r = ttm_bo_mem_space(bo, &placement, &tmp_mem, - interruptible, no_wait_reserve, no_wait_gpu); + interruptible, no_wait_gpu); if (unlikely(r)) { return r; } @@ -308,11 +308,11 @@ static int radeon_move_vram_ram(struct ttm_buffer_object *bo, if (unlikely(r)) { goto out_cleanup; } - r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem, old_mem); + r = radeon_move_blit(bo, true, no_wait_gpu, &tmp_mem, old_mem); if (unlikely(r)) { goto out_cleanup; } - r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, new_mem); + r = ttm_bo_move_ttm(bo, true, no_wait_gpu, new_mem); out_cleanup: ttm_bo_mem_put(bo, &tmp_mem); return r; @@ -320,7 +320,7 @@ out_cleanup:

static int radeon_move_ram_vram(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct radeon_device *rdev; @@ -340,15 +340,16 @@ static int radeon_move_ram_vram(struct ttm_buffer_object *bo, placement.num_busy_placement = 1; placement.busy_placement = &placements; placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT; - r = ttm_bo_mem_space(bo, &placement, &tmp_mem, interruptible, no_wait_reserve, no_wait_gpu); + r = ttm_bo_mem_space(bo, &placement, &tmp_mem, + interruptible, no_wait_gpu); if (unlikely(r)) { return r; } - r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem); + r = ttm_bo_move_ttm(bo, true, no_wait_gpu, &tmp_mem); if (unlikely(r)) { goto out_cleanup; } - r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, new_mem, old_mem); + r = radeon_move_blit(bo, true, no_wait_gpu, new_mem, old_mem); if (unlikely(r)) { goto out_cleanup; } @@ -359,7 +360,7 @@ out_cleanup:

static int radeon_bo_move(struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct radeon_device *rdev; @@ -388,18 +389,18 @@ static int radeon_bo_move(struct ttm_buffer_object *bo, if (old_mem->mem_type == TTM_PL_VRAM && new_mem->mem_type == TTM_PL_SYSTEM) { r = radeon_move_vram_ram(bo, evict, interruptible, - no_wait_reserve, no_wait_gpu, new_mem); + no_wait_gpu, new_mem); } else if (old_mem->mem_type == TTM_PL_SYSTEM && new_mem->mem_type == TTM_PL_VRAM) { r = radeon_move_ram_vram(bo, evict, interruptible, - no_wait_reserve, no_wait_gpu, new_mem); + no_wait_gpu, new_mem); } else { - r = radeon_move_blit(bo, evict, no_wait_reserve, no_wait_gpu, new_mem, old_mem); + r = radeon_move_blit(bo, evict, no_wait_gpu, new_mem, old_mem); }

if (r) { memcpy: - r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); } return r; } diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index e6df086..1d77ad1 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -375,7 +375,7 @@ static int ttm_bo_add_ttm(struct ttm_buffer_object *bo, bool zero_alloc) static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo, struct ttm_mem_reg *mem, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu) + bool no_wait_gpu) { struct ttm_bo_device *bdev = bo->bdev; bool old_is_pci = ttm_mem_reg_is_pci(bdev, &bo->mem); @@ -429,12 +429,12 @@ static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo,

if (!(old_man->flags & TTM_MEMTYPE_FLAG_FIXED) && !(new_man->flags & TTM_MEMTYPE_FLAG_FIXED)) - ret = ttm_bo_move_ttm(bo, evict, no_wait_reserve, no_wait_gpu, mem); + ret = ttm_bo_move_ttm(bo, evict, no_wait_gpu, mem); else if (bdev->driver->move) ret = bdev->driver->move(bo, evict, interruptible, - no_wait_reserve, no_wait_gpu, mem); + no_wait_gpu, mem); else - ret = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, mem); + ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, mem);

if (ret) { if (bdev->driver->move_notify) { @@ -717,7 +717,7 @@ void ttm_bo_unlock_delayed_workqueue(struct ttm_bo_device *bdev, int resched) EXPORT_SYMBOL(ttm_bo_unlock_delayed_workqueue);

static int ttm_bo_evict(struct ttm_buffer_object *bo, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu) + bool no_wait_gpu) { struct ttm_bo_device *bdev = bo->bdev; struct ttm_mem_reg evict_mem; @@ -748,7 +748,7 @@ static int ttm_bo_evict(struct ttm_buffer_object *bo, bool interruptible, placement.num_busy_placement = 0; bdev->driver->evict_flags(bo, &placement); ret = ttm_bo_mem_space(bo, &placement, &evict_mem, interruptible, - no_wait_reserve, no_wait_gpu); + no_wait_gpu); if (ret) { if (ret != -ERESTARTSYS) { pr_err("Failed to find memory space for buffer 0x%p eviction\n", @@ -759,7 +759,7 @@ static int ttm_bo_evict(struct ttm_buffer_object *bo, bool interruptible, }

ret = ttm_bo_handle_move_mem(bo, &evict_mem, true, interruptible, - no_wait_reserve, no_wait_gpu); + no_wait_gpu); if (ret) { if (ret != -ERESTARTSYS) pr_err("Buffer eviction failed\n"); @@ -773,7 +773,7 @@ out:

static int ttm_mem_evict_first(struct ttm_bo_device *bdev, uint32_t mem_type, - bool interruptible, bool no_wait_reserve, + bool interruptible, bool no_wait_gpu) { struct ttm_bo_global *glob = bdev->glob; @@ -781,7 +781,6 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev, struct ttm_buffer_object *bo; int ret, put_count = 0;

-retry: spin_lock(&glob->lru_lock); if (list_empty(&man->lru)) { spin_unlock(&glob->lru_lock); @@ -793,20 +792,11 @@ retry:

ret = ttm_bo_reserve_locked(bo, false, true, false, 0);

- if (unlikely(ret == -EBUSY)) { + if (WARN_ON_ONCE(ret == -EBUSY)) { spin_unlock(&glob->lru_lock); - if (likely(!no_wait_reserve)) - ret = ttm_bo_wait_unreserved(bo, interruptible);

kref_put(&bo->list_kref, ttm_bo_release_list); - - /** - * We *need* to retry after releasing the lru lock. - */ - - if (unlikely(ret != 0)) - return ret; - goto retry; + return ret; }

if (!list_empty(&bo->ddestroy)) { @@ -823,7 +813,7 @@ retry:

ttm_bo_list_ref_sub(bo, put_count, true);

- ret = ttm_bo_evict(bo, interruptible, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_evict(bo, interruptible, no_wait_gpu); ttm_bo_unreserve(bo);

kref_put(&bo->list_kref, ttm_bo_release_list); @@ -848,7 +838,6 @@ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, struct ttm_placement *placement, struct ttm_mem_reg *mem, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu) { struct ttm_bo_device *bdev = bo->bdev; @@ -861,8 +850,8 @@ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, return ret; if (mem->mm_node) break; - ret = ttm_mem_evict_first(bdev, mem_type, interruptible, - no_wait_reserve, no_wait_gpu); + ret = ttm_mem_evict_first(bdev, mem_type, + interruptible, no_wait_gpu); if (unlikely(ret != 0)) return ret; } while (1); @@ -927,7 +916,7 @@ static bool ttm_bo_mt_compatible(struct ttm_mem_type_manager *man, int ttm_bo_mem_space(struct ttm_buffer_object *bo, struct ttm_placement *placement, struct ttm_mem_reg *mem, - bool interruptible, bool no_wait_reserve, + bool interruptible, bool no_wait_gpu) { struct ttm_bo_device *bdev = bo->bdev; @@ -1018,7 +1007,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, }

ret = ttm_bo_mem_force_space(bo, mem_type, placement, mem, - interruptible, no_wait_reserve, no_wait_gpu); + interruptible, no_wait_gpu); if (ret == 0 && mem->mm_node) { mem->placement = cur_flags; return 0; @@ -1033,7 +1022,7 @@ EXPORT_SYMBOL(ttm_bo_mem_space);

int ttm_bo_move_buffer(struct ttm_buffer_object *bo, struct ttm_placement *placement, - bool interruptible, bool no_wait_reserve, + bool interruptible, bool no_wait_gpu) { int ret = 0; @@ -1060,10 +1049,12 @@ int ttm_bo_move_buffer(struct ttm_buffer_object *bo, /* * Determine where to move the buffer. */ - ret = ttm_bo_mem_space(bo, placement, &mem, interruptible, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_mem_space(bo, placement, &mem, + interruptible, no_wait_gpu); if (ret) goto out_unlock; - ret = ttm_bo_handle_move_mem(bo, &mem, false, interruptible, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_handle_move_mem(bo, &mem, false, + interruptible, no_wait_gpu); out_unlock: if (ret && mem.mm_node) ttm_bo_mem_put(bo, &mem); @@ -1092,7 +1083,7 @@ static int ttm_bo_mem_compat(struct ttm_placement *placement,

int ttm_bo_validate(struct ttm_buffer_object *bo, struct ttm_placement *placement, - bool interruptible, bool no_wait_reserve, + bool interruptible, bool no_wait_gpu) { int ret; @@ -1108,7 +1099,8 @@ int ttm_bo_validate(struct ttm_buffer_object *bo, */ ret = ttm_bo_mem_compat(placement, &bo->mem); if (ret < 0) { - ret = ttm_bo_move_buffer(bo, placement, interruptible, no_wait_reserve, no_wait_gpu); + ret = ttm_bo_move_buffer(bo, placement, interruptible, + no_wait_gpu); if (ret) return ret; } else { @@ -1224,7 +1216,7 @@ int ttm_bo_init(struct ttm_bo_device *bdev, goto out_err; }

- ret = ttm_bo_validate(bo, placement, interruptible, false, false); + ret = ttm_bo_validate(bo, placement, interruptible, false); if (ret) goto out_err;

@@ -1303,7 +1295,7 @@ static int ttm_bo_force_list_clean(struct ttm_bo_device *bdev, int ret = 0;

while (!ret) { - ret = ttm_mem_evict_first(bdev, mem_type, false, false, false); + ret = ttm_mem_evict_first(bdev, mem_type, false, false); if (ret == -EBUSY) return 0; else if (ret && !allow_errors) { @@ -1826,7 +1818,7 @@ static int ttm_bo_swapout(struct ttm_mem_shrink *shrink) evict_mem.mem_type = TTM_PL_SYSTEM;

ret = ttm_bo_handle_move_mem(bo, &evict_mem, true, - false, false, false); + false, false); if (unlikely(ret != 0)) goto out; } diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index b9c4e51..9e9c5d2 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -43,7 +43,7 @@ void ttm_bo_free_old_node(struct ttm_buffer_object *bo) }

int ttm_bo_move_ttm(struct ttm_buffer_object *bo, - bool evict, bool no_wait_reserve, + bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct ttm_tt *ttm = bo->ttm; @@ -314,7 +314,7 @@ static int ttm_copy_ttm_io_page(struct ttm_tt *ttm, void *dst, }

int ttm_bo_move_memcpy(struct ttm_buffer_object *bo, - bool evict, bool no_wait_reserve, bool no_wait_gpu, + bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct ttm_bo_device *bdev = bo->bdev; @@ -611,7 +611,7 @@ EXPORT_SYMBOL(ttm_bo_kunmap);

int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, void *sync_obj, - bool evict, bool no_wait_reserve, + bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem) { diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_dmabuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_dmabuf.c index bd78257..2ace48d 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_dmabuf.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_dmabuf.c @@ -66,7 +66,7 @@ int vmw_dmabuf_to_placement(struct vmw_private *dev_priv, if (unlikely(ret != 0)) goto err;

- ret = ttm_bo_validate(bo, placement, interruptible, false, false); + ret = ttm_bo_validate(bo, placement, interruptible, false);

ttm_bo_unreserve(bo);

@@ -123,7 +123,7 @@ int vmw_dmabuf_to_vram_or_gmr(struct vmw_private *dev_priv, else placement = &vmw_vram_gmr_placement;

- ret = ttm_bo_validate(bo, placement, interruptible, false, false); + ret = ttm_bo_validate(bo, placement, interruptible, false); if (likely(ret == 0) || ret == -ERESTARTSYS) goto err_unreserve;

@@ -138,7 +138,7 @@ int vmw_dmabuf_to_vram_or_gmr(struct vmw_private *dev_priv, else placement = &vmw_vram_placement;

- ret = ttm_bo_validate(bo, placement, interruptible, false, false); + ret = ttm_bo_validate(bo, placement, interruptible, false);

err_unreserve: ttm_bo_unreserve(bo); @@ -224,10 +224,9 @@ int vmw_dmabuf_to_start_of_vram(struct vmw_private *dev_priv, if (bo->mem.mem_type == TTM_PL_VRAM && bo->mem.start < bo->num_pages && bo->mem.start > 0) - (void) ttm_bo_validate(bo, &vmw_sys_placement, false, - false, false); + (void) ttm_bo_validate(bo, &vmw_sys_placement, false, false);

- ret = ttm_bo_validate(bo, &placement, interruptible, false, false); + ret = ttm_bo_validate(bo, &placement, interruptible, false);

/* For some reason we didn't up at the start of vram */ WARN_ON(ret == 0 && bo->offset != 0); @@ -316,7 +315,7 @@ void vmw_bo_pin(struct ttm_buffer_object *bo, bool pin) placement.num_placement = 1; placement.placement = &pl_flags;

- ret = ttm_bo_validate(bo, &placement, false, true, true); + ret = ttm_bo_validate(bo, &placement, false, true);

BUG_ON(ret != 0 || bo->mem.mem_type != old_mem_type); } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c index e5775a0..546313f 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c @@ -921,7 +921,7 @@ static int vmw_validate_single_buffer(struct vmw_private *dev_priv, * used as a GMR, this will return -ENOMEM. */

- ret = ttm_bo_validate(bo, &vmw_vram_gmr_placement, true, false, false); + ret = ttm_bo_validate(bo, &vmw_vram_gmr_placement, true, false); if (likely(ret == 0 || ret == -ERESTARTSYS)) return ret;

@@ -931,7 +931,7 @@ static int vmw_validate_single_buffer(struct vmw_private *dev_priv, */

DRM_INFO("Falling through to VRAM.\n"); - ret = ttm_bo_validate(bo, &vmw_vram_placement, true, false, false); + ret = ttm_bo_validate(bo, &vmw_vram_placement, true, false); return ret; }

diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index 0ca3e1c..abe7c72 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -839,7 +839,7 @@ int vmw_surface_do_validate(struct vmw_private *dev_priv, goto out_no_reserve;

ret = ttm_bo_validate(srf->backup, &vmw_srf_placement, - true, false, false); + true, false); if (unlikely(ret != 0)) goto out_no_validate; } @@ -969,8 +969,7 @@ int vmw_surface_evict(struct vmw_private *dev_priv, if (unlikely(ret != 0)) goto out_no_reserve;

- ret = ttm_bo_validate(srf->backup, &vmw_srf_placement, - true, false, false); + ret = ttm_bo_validate(srf->backup, &vmw_srf_placement, true, false); if (unlikely(ret != 0)) goto out_no_validate;

diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 13e33f0..c22a385 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -340,7 +340,6 @@ extern int ttm_bo_wait(struct ttm_buffer_object *bo, bool lazy, * @bo: The buffer object. * @placement: Proposed placement for the buffer object. * @interruptible: Sleep interruptible if sleeping. - * @no_wait_reserve: Return immediately if other buffers are busy. * @no_wait_gpu: Return immediately if the GPU is busy. * * Changes placement and caching policy of the buffer object @@ -353,7 +352,7 @@ extern int ttm_bo_wait(struct ttm_buffer_object *bo, bool lazy, */ extern int ttm_bo_validate(struct ttm_buffer_object *bo, struct ttm_placement *placement, - bool interruptible, bool no_wait_reserve, + bool interruptible, bool no_wait_gpu);

/** diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index 0c8c3b5..887c3c0 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -394,7 +394,7 @@ struct ttm_bo_driver { */ int (*move) (struct ttm_buffer_object *bo, bool evict, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu, + bool no_wait_gpu, struct ttm_mem_reg *new_mem);

/** @@ -706,7 +706,6 @@ extern bool ttm_mem_reg_is_pci(struct ttm_bo_device *bdev, * @proposed_placement: Proposed new placement for the buffer object. * @mem: A struct ttm_mem_reg. * @interruptible: Sleep interruptible when sliping. - * @no_wait_reserve: Return immediately if other buffers are busy. * @no_wait_gpu: Return immediately if the GPU is busy. * * Allocate memory space for the buffer object pointed to by @bo, using @@ -722,7 +721,7 @@ extern int ttm_bo_mem_space(struct ttm_buffer_object *bo, struct ttm_placement *placement, struct ttm_mem_reg *mem, bool interruptible, - bool no_wait_reserve, bool no_wait_gpu); + bool no_wait_gpu);

extern void ttm_bo_mem_put(struct ttm_buffer_object *bo, struct ttm_mem_reg *mem); @@ -904,7 +903,6 @@ extern int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo, * * @bo: A pointer to a struct ttm_buffer_object. * @evict: 1: This is an eviction. Don't try to pipeline. - * @no_wait_reserve: Return immediately if other buffers are busy. * @no_wait_gpu: Return immediately if the GPU is busy. * @new_mem: struct ttm_mem_reg indicating where to move. * @@ -919,15 +917,14 @@ extern int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo, */

extern int ttm_bo_move_ttm(struct ttm_buffer_object *bo, - bool evict, bool no_wait_reserve, - bool no_wait_gpu, struct ttm_mem_reg *new_mem); + bool evict, bool no_wait_gpu, + struct ttm_mem_reg *new_mem);

/** * ttm_bo_move_memcpy * * @bo: A pointer to a struct ttm_buffer_object. * @evict: 1: This is an eviction. Don't try to pipeline. - * @no_wait_reserve: Return immediately if other buffers are busy. * @no_wait_gpu: Return immediately if the GPU is busy. * @new_mem: struct ttm_mem_reg indicating where to move. * @@ -942,8 +939,8 @@ extern int ttm_bo_move_ttm(struct ttm_buffer_object *bo, */

extern int ttm_bo_move_memcpy(struct ttm_buffer_object *bo, - bool evict, bool no_wait_reserve, - bool no_wait_gpu, struct ttm_mem_reg *new_mem); + bool evict, bool no_wait_gpu, + struct ttm_mem_reg *new_mem);

/** * ttm_bo_free_old_node @@ -960,7 +957,6 @@ extern void ttm_bo_free_old_node(struct ttm_buffer_object *bo); * @bo: A pointer to a struct ttm_buffer_object. * @sync_obj: A sync object that signals when moving is complete. * @evict: This is an evict move. Don't return until the buffer is idle. - * @no_wait_reserve: Return immediately if other buffers are busy. * @no_wait_gpu: Return immediately if the GPU is busy. * @new_mem: struct ttm_mem_reg indicating where to move. * @@ -974,8 +970,7 @@ extern void ttm_bo_free_old_node(struct ttm_buffer_object *bo);

extern int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, void *sync_obj, - bool evict, bool no_wait_reserve, - bool no_wait_gpu, + bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem); /** * ttm_io_prot