From: Christian König christian.koenig@amd.com

Better match what it is actually doing.

Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/radeon/evergreen_cs.c | 8 ++++---- drivers/gpu/drm/radeon/r100.c | 8 ++++---- drivers/gpu/drm/radeon/r200.c | 2 +- drivers/gpu/drm/radeon/r300.c | 4 ++-- drivers/gpu/drm/radeon/r600_cs.c | 10 +++++----- drivers/gpu/drm/radeon/radeon.h | 28 ++++++++++++++-------------- drivers/gpu/drm/radeon/radeon_cs.c | 10 +++++----- drivers/gpu/drm/radeon/radeon_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 2 +- drivers/gpu/drm/radeon/radeon_uvd.c | 2 +- drivers/gpu/drm/radeon/radeon_vce.c | 2 +- drivers/gpu/drm/radeon/radeon_vm.c | 6 +++--- 12 files changed, 42 insertions(+), 42 deletions(-)

diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index 5c8b358..a1dece6 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -35,7 +35,7 @@ #define MIN(a,b) (((a)<(b))?(a):(b))

int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc); + struct radeon_bo_list **cs_reloc); struct evergreen_cs_track { u32 group_size; u32 nbanks; @@ -1094,7 +1094,7 @@ static int evergreen_cs_parse_packet0(struct radeon_cs_parser *p, static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) { struct evergreen_cs_track *track = (struct evergreen_cs_track *)p->track; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 last_reg; u32 m, i, tmp, *ib; int r; @@ -1792,7 +1792,7 @@ static bool evergreen_is_safe_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) static int evergreen_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct evergreen_cs_track *track; volatile u32 *ib; unsigned idx; @@ -2685,7 +2685,7 @@ int evergreen_cs_parse(struct radeon_cs_parser *p) int evergreen_dma_cs_parse(struct radeon_cs_parser *p) { struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; - struct radeon_cs_reloc *src_reloc, *dst_reloc, *dst2_reloc; + struct radeon_bo_list *src_reloc, *dst_reloc, *dst2_reloc; u32 header, cmd, count, sub_cmd; volatile u32 *ib = p->ib.ptr; u32 idx; diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index 10f8be0..8174731 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -1254,7 +1254,7 @@ int r100_reloc_pitch_offset(struct radeon_cs_parser *p, int r; u32 tile_flags = 0; u32 tmp; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 value;

r = radeon_cs_packet_next_reloc(p, &reloc, 0); @@ -1293,7 +1293,7 @@ int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, int idx) { unsigned c, i; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; int r = 0; volatile uint32_t *ib; @@ -1542,7 +1542,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp; @@ -1901,7 +1901,7 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, static int r100_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; unsigned idx; volatile uint32_t *ib; diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c index 732d493..c70e6d5 100644 --- a/drivers/gpu/drm/radeon/r200.c +++ b/drivers/gpu/drm/radeon/r200.c @@ -146,7 +146,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp; diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index 1bc4704..91d2442 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -598,7 +598,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp, tile_flags = 0; @@ -1142,7 +1142,7 @@ fail: static int r300_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; unsigned idx; diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index c47537a..b81ba19 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -969,7 +969,7 @@ static int r600_cs_parse_packet0(struct radeon_cs_parser *p, static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) { struct r600_cs_track *track = (struct r600_cs_track *)p->track; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 m, i, tmp, *ib; int r;

@@ -1626,7 +1626,7 @@ static bool r600_is_safe_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) static int r600_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r600_cs_track *track; volatile u32 *ib; unsigned idx; @@ -2354,7 +2354,7 @@ static int r600_cs_parser_relocs_legacy(struct radeon_cs_parser *p) if (p->chunk_relocs_idx == -1) { return 0; } - p->relocs = kzalloc(sizeof(struct radeon_cs_reloc), GFP_KERNEL); + p->relocs = kzalloc(sizeof(struct radeon_bo_list), GFP_KERNEL); if (p->relocs == NULL) { return -ENOMEM; } @@ -2435,7 +2435,7 @@ void r600_cs_legacy_init(void) * GPU offset using the provided start. **/ int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc) + struct radeon_bo_list **cs_reloc) { struct radeon_cs_chunk *relocs_chunk; unsigned idx; @@ -2473,7 +2473,7 @@ int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, int r600_dma_cs_parse(struct radeon_cs_parser *p) { struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; - struct radeon_cs_reloc *src_reloc, *dst_reloc; + struct radeon_bo_list *src_reloc, *dst_reloc; u32 header, cmd, count, tiled; volatile u32 *ib = p->ib.ptr; u32 idx, idx_value; diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 17db846..da1c549 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -450,6 +450,15 @@ struct radeon_mman { #endif };

+struct radeon_bo_list { + struct radeon_bo *robj; + struct ttm_validate_buffer tv; + uint64_t gpu_offset; + unsigned prefered_domains; + unsigned allowed_domains; + uint32_t tiling_flags; +}; + /* bo virtual address in a specific vm */ struct radeon_bo_va { /* protected by bo being reserved */ @@ -1044,15 +1053,6 @@ void cayman_dma_fini(struct radeon_device *rdev); /* * CS. */ -struct radeon_cs_reloc { - struct radeon_bo *robj; - struct ttm_validate_buffer tv; - uint64_t gpu_offset; - unsigned prefered_domains; - unsigned allowed_domains; - uint32_t tiling_flags; -}; - struct radeon_cs_chunk { uint32_t chunk_id; uint32_t length_dw; @@ -1072,9 +1072,9 @@ struct radeon_cs_parser { unsigned idx; /* relocations */ unsigned nrelocs; - struct radeon_cs_reloc *relocs; - struct radeon_cs_reloc **relocs_ptr; - struct radeon_cs_reloc *vm_bos; + struct radeon_bo_list *relocs; + struct radeon_bo_list **relocs_ptr; + struct radeon_bo_list *vm_bos; struct list_head validated; unsigned dma_reloc_idx; /* indices of various chunks */ @@ -2973,7 +2973,7 @@ int radeon_vm_manager_init(struct radeon_device *rdev); void radeon_vm_manager_fini(struct radeon_device *rdev); int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm); void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm); -struct radeon_cs_reloc *radeon_vm_get_bos(struct radeon_device *rdev, +struct radeon_bo_list *radeon_vm_get_bos(struct radeon_device *rdev, struct radeon_vm *vm, struct list_head *head); struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, @@ -3087,7 +3087,7 @@ bool radeon_cs_packet_next_is_pkt3_nop(struct radeon_cs_parser *p); void radeon_cs_dump_packet(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt); int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc, + struct radeon_bo_list **cs_reloc, int nomm); int r600_cs_common_vline_parse(struct radeon_cs_parser *p, uint32_t *vline_start_end, diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index f1f584a..fb776cb 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -92,7 +92,7 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) if (p->relocs_ptr == NULL) { return -ENOMEM; } - p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_cs_reloc), GFP_KERNEL); + p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_bo_list), GFP_KERNEL); if (p->relocs == NULL) { return -ENOMEM; } @@ -251,7 +251,7 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority

static int radeon_cs_sync_rings(struct radeon_cs_parser *p) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; int r;

list_for_each_entry(reloc, &p->validated, tv.head) { @@ -397,8 +397,8 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) static int cmp_size_smaller_first(void *priv, struct list_head *a, struct list_head *b) { - struct radeon_cs_reloc *la = list_entry(a, struct radeon_cs_reloc, tv.head); - struct radeon_cs_reloc *lb = list_entry(b, struct radeon_cs_reloc, tv.head); + struct radeon_bo_list *la = list_entry(a, struct radeon_bo_list, tv.head); + struct radeon_bo_list *lb = list_entry(b, struct radeon_bo_list, tv.head);

/* Sort A before B if A is smaller. */ return (int)la->robj->tbo.num_pages - (int)lb->robj->tbo.num_pages; @@ -832,7 +832,7 @@ void radeon_cs_dump_packet(struct radeon_cs_parser *p, * GPU offset using the provided start. **/ int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc, + struct radeon_bo_list **cs_reloc, int nomm) { struct radeon_cs_chunk *relocs_chunk; diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index 12cfaea..6162bd2 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -548,7 +548,7 @@ static void radeon_gem_va_update_vm(struct radeon_device *rdev, struct radeon_bo_va *bo_va) { struct ttm_validate_buffer tv, *entry; - struct radeon_cs_reloc *vm_bos; + struct radeon_bo_list *vm_bos; struct ww_acquire_ctx ticket; struct list_head list; unsigned domain; diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 87b00d9..4ab0747 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -502,7 +502,7 @@ int radeon_bo_list_validate(struct radeon_device *rdev, struct ww_acquire_ctx *ticket, struct list_head *head, int ring) { - struct radeon_cs_reloc *lobj; + struct radeon_bo_list *lobj; struct radeon_bo *bo; int r; u64 bytes_moved = 0, initial_bytes_moved; diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c b/drivers/gpu/drm/radeon/radeon_uvd.c index 11b6624..7c22b93 100644 --- a/drivers/gpu/drm/radeon/radeon_uvd.c +++ b/drivers/gpu/drm/radeon/radeon_uvd.c @@ -488,7 +488,7 @@ static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p, unsigned buf_sizes[], bool *has_msg_cmd) { struct radeon_cs_chunk *relocs_chunk; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; unsigned idx, cmd, offset; uint64_t start, end; int r; diff --git a/drivers/gpu/drm/radeon/radeon_vce.c b/drivers/gpu/drm/radeon/radeon_vce.c index 9e85757..8b4eea4 100644 --- a/drivers/gpu/drm/radeon/radeon_vce.c +++ b/drivers/gpu/drm/radeon/radeon_vce.c @@ -453,7 +453,7 @@ int radeon_vce_cs_reloc(struct radeon_cs_parser *p, int lo, int hi, unsigned size) { struct radeon_cs_chunk *relocs_chunk; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; uint64_t start, end, offset; unsigned idx;

diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c index c4ffe02..0423e29 100644 --- a/drivers/gpu/drm/radeon/radeon_vm.c +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -125,15 +125,15 @@ void radeon_vm_manager_fini(struct radeon_device *rdev) * Add the page directory to the list of BOs to * validate for command submission (cayman+). */ -struct radeon_cs_reloc *radeon_vm_get_bos(struct radeon_device *rdev, +struct radeon_bo_list *radeon_vm_get_bos(struct radeon_device *rdev, struct radeon_vm *vm, struct list_head *head) { - struct radeon_cs_reloc *list; + struct radeon_bo_list *list; unsigned i, idx;

list = drm_malloc_ab(vm->max_pde_used + 2, - sizeof(struct radeon_cs_reloc)); + sizeof(struct radeon_bo_list)); if (!list) return NULL;

From: Christian König christian.koenig@amd.com

Stop using the VM mutex for this

Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/radeon/radeon.h | 3 +++ drivers/gpu/drm/radeon/radeon_vm.c | 36 ++++++++++++++++++++++++++++++------ 2 files changed, 33 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index da1c549..3680cf0 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -929,6 +929,9 @@ struct radeon_vm {

struct rb_root va;

+ /* ptotecting invalidated and freed */ + spinlock_t status_lock; + /* BOs moved, but not yet updated in the PT */ struct list_head invalidated;

diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c index 6bc3821..cde48c4 100644 --- a/drivers/gpu/drm/radeon/radeon_vm.c +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -487,7 +487,9 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, tmp->vm = vm; tmp->addr = bo_va->addr; tmp->bo = radeon_bo_ref(bo_va->bo); + spin_lock(&vm->status_lock); list_add(&tmp->vm_status, &vm->freed); + spin_unlock(&vm->status_lock); }

interval_tree_remove(&bo_va->it, &vm->va); @@ -913,7 +915,9 @@ int radeon_vm_bo_update(struct radeon_device *rdev, return -EINVAL; }

+ spin_lock(&vm->status_lock); list_del_init(&bo_va->vm_status); + spin_unlock(&vm->status_lock);

bo_va->flags &= ~RADEON_VM_PAGE_VALID; bo_va->flags &= ~RADEON_VM_PAGE_SYSTEM; @@ -1028,17 +1032,25 @@ int radeon_vm_bo_update(struct radeon_device *rdev, int radeon_vm_clear_freed(struct radeon_device *rdev, struct radeon_vm *vm) { - struct radeon_bo_va *bo_va, *tmp; + struct radeon_bo_va *bo_va; int r;

- list_for_each_entry_safe(bo_va, tmp, &vm->freed, vm_status) { + spin_lock(&vm->status_lock); + while (!list_empty(&vm->freed)) { + bo_va = list_first_entry(&vm->freed, + struct radeon_bo_va, vm_status); + spin_unlock(&vm->status_lock); + r = radeon_vm_bo_update(rdev, bo_va, NULL); radeon_bo_unref(&bo_va->bo); radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); if (r) return r; + + spin_lock(&vm->status_lock); } + spin_unlock(&vm->status_lock); return 0;

} @@ -1057,14 +1069,23 @@ int radeon_vm_clear_freed(struct radeon_device *rdev, int radeon_vm_clear_invalids(struct radeon_device *rdev, struct radeon_vm *vm) { - struct radeon_bo_va *bo_va, *tmp; + struct radeon_bo_va *bo_va; int r;

- list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, vm_status) { + spin_lock(&vm->status_lock); + while (!list_empty(&vm->invalidated)) { + bo_va = list_first_entry(&vm->invalidated, + struct radeon_bo_va, vm_status); + spin_unlock(&vm->status_lock); + r = radeon_vm_bo_update(rdev, bo_va, NULL); if (r) return r; + + spin_lock(&vm->status_lock); } + spin_unlock(&vm->status_lock); + return 0; }

@@ -1087,6 +1108,7 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev,

mutex_lock(&vm->mutex); interval_tree_remove(&bo_va->it, &vm->va); + spin_lock(&vm->status_lock); list_del(&bo_va->vm_status);

if (bo_va->addr) { @@ -1096,6 +1118,7 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev, radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); } + spin_unlock(&vm->status_lock);

mutex_unlock(&vm->mutex); } @@ -1116,10 +1139,10 @@ void radeon_vm_bo_invalidate(struct radeon_device *rdev,

list_for_each_entry(bo_va, &bo->va, bo_list) { if (bo_va->addr) { - mutex_lock(&bo_va->vm->mutex); + spin_lock(&bo_va->vm->status_lock); list_del(&bo_va->vm_status); list_add(&bo_va->vm_status, &bo_va->vm->invalidated); - mutex_unlock(&bo_va->vm->mutex); + spin_unlock(&bo_va->vm->status_lock); } } } @@ -1147,6 +1170,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) } mutex_init(&vm->mutex); vm->va = RB_ROOT; + spin_lock_init(&vm->status_lock); INIT_LIST_HEAD(&vm->invalidated); INIT_LIST_HEAD(&vm->freed);

From: Christian König christian.koenig@amd.com

Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_cs.c | 9 +++++-- drivers/gpu/drm/radeon/radeon_gem.c | 17 ++++++++++---- drivers/gpu/drm/radeon/radeon_kms.c | 5 ++++ drivers/gpu/drm/radeon/radeon_vm.c | 47 +++++++++---------------------------- 5 files changed, 36 insertions(+), 44 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 3680cf0..699446a 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -925,7 +925,7 @@ struct radeon_vm_id { };

struct radeon_vm { - struct mutex mutex; + struct mutex lock;

struct rb_root va;

diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index fb776cb..7a90378 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -195,6 +195,7 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) if (p->cs_flags & RADEON_CS_USE_VM) p->vm_bos = radeon_vm_get_bos(p->rdev, p->ib.vm, &p->validated); + if (need_mmap_lock) down_read(&current->mm->mmap_sem);

@@ -571,7 +572,6 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, if (parser->ring == R600_RING_TYPE_UVD_INDEX) radeon_uvd_note_usage(rdev);

- mutex_lock(&vm->mutex); r = radeon_bo_vm_update_pte(parser, vm); if (r) { goto out; @@ -592,7 +592,6 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, }

out: - mutex_unlock(&vm->mutex); return r; }

@@ -696,6 +695,8 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) }

r = radeon_cs_ib_fill(rdev, &parser); + if (parser.cs_flags & RADEON_CS_USE_VM) + mutex_lock(&parser.ib.vm->lock); if (!r) { r = radeon_cs_parser_relocs(&parser); if (r && r != -ERESTARTSYS) @@ -704,6 +705,8 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)

if (r) { radeon_cs_parser_fini(&parser, r, false); + if (parser.cs_flags & RADEON_CS_USE_VM) + mutex_unlock(&parser.ib.vm->lock); up_read(&rdev->exclusive_lock); r = radeon_cs_handle_lockup(rdev, r); return r; @@ -721,6 +724,8 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) } out: radeon_cs_parser_fini(&parser, r, true); + if (parser.cs_flags & RADEON_CS_USE_VM) + mutex_unlock(&parser.ib.vm->lock); up_read(&rdev->exclusive_lock); r = radeon_cs_handle_lockup(rdev, r); return r; diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index 6162bd2..4eafec6 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -150,8 +150,10 @@ int radeon_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_pri return 0; }

+ mutex_lock(&vm->lock); r = radeon_bo_reserve(rbo, false); if (r) { + mutex_unlock(&vm->lock); return r; }

@@ -162,6 +164,7 @@ int radeon_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_pri ++bo_va->ref_count; } radeon_bo_unreserve(rbo); + mutex_unlock(&vm->lock);

return 0; } @@ -180,8 +183,10 @@ void radeon_gem_object_close(struct drm_gem_object *obj, return; }

+ mutex_lock(&vm->lock); r = radeon_bo_reserve(rbo, true); if (r) { + mutex_unlock(&vm->lock); dev_err(rdev->dev, "leaking bo va because " "we fail to reserve bo (%d)\n", r); return; @@ -193,6 +198,7 @@ void radeon_gem_object_close(struct drm_gem_object *obj, } } radeon_bo_unreserve(rbo); + mutex_unlock(&vm->lock); }

static int radeon_gem_handle_lockup(struct radeon_device *rdev, int r) @@ -576,17 +582,13 @@ static void radeon_gem_va_update_vm(struct radeon_device *rdev, goto error_unreserve; }

- mutex_lock(&bo_va->vm->mutex); r = radeon_vm_clear_freed(rdev, bo_va->vm); if (r) - goto error_unlock; + goto error_unreserve;

if (bo_va->it.start) r = radeon_vm_bo_update(rdev, bo_va, &bo_va->bo->tbo.mem);

-error_unlock: - mutex_unlock(&bo_va->vm->mutex); - error_unreserve: ttm_eu_backoff_reservation(&ticket, &list);

@@ -662,14 +664,18 @@ int radeon_gem_va_ioctl(struct drm_device *dev, void *data, return -ENOENT; } rbo = gem_to_radeon_bo(gobj); + mutex_lock(&fpriv->vm.lock); r = radeon_bo_reserve(rbo, false); if (r) { + mutex_unlock(&fpriv->vm.lock); args->operation = RADEON_VA_RESULT_ERROR; drm_gem_object_unreference_unlocked(gobj); return r; } bo_va = radeon_vm_bo_find(&fpriv->vm, rbo); if (!bo_va) { + radeon_bo_unreserve(rbo); + mutex_unlock(&fpriv->vm.lock); args->operation = RADEON_VA_RESULT_ERROR; drm_gem_object_unreference_unlocked(gobj); return -ENOENT; @@ -698,6 +704,7 @@ int radeon_gem_va_ioctl(struct drm_device *dev, void *data, args->operation = RADEON_VA_RESULT_ERROR; } out: + mutex_unlock(&fpriv->vm.lock); drm_gem_object_unreference_unlocked(gobj); return r; } diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index f4dd26a..d994006 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -613,8 +613,10 @@ int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) }

if (rdev->accel_working) { + mutex_lock(&vm->lock); r = radeon_bo_reserve(rdev->ring_tmp_bo.bo, false); if (r) { + mutex_unlock(&vm->lock); radeon_vm_fini(rdev, vm); kfree(fpriv); return r; @@ -628,6 +630,7 @@ int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) RADEON_VA_IB_OFFSET, RADEON_VM_PAGE_READABLE | RADEON_VM_PAGE_SNOOPED); + mutex_unlock(&vm->lock); if (r) { radeon_vm_fini(rdev, vm); kfree(fpriv); @@ -662,12 +665,14 @@ void radeon_driver_postclose_kms(struct drm_device *dev, int r;

if (rdev->accel_working) { + mutex_lock(&vm->lock); r = radeon_bo_reserve(rdev->ring_tmp_bo.bo, false); if (!r) { if (vm->ib_bo_va) radeon_vm_bo_rmv(rdev, vm->ib_bo_va); radeon_bo_unreserve(rdev->ring_tmp_bo.bo); } + mutex_unlock(&vm->lock); }

radeon_vm_fini(rdev, vm); diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c index cde48c4..658183f 100644 --- a/drivers/gpu/drm/radeon/radeon_vm.c +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -172,7 +172,7 @@ struct radeon_bo_list *radeon_vm_get_bos(struct radeon_device *rdev, * Allocate an id for the vm (cayman+). * Returns the fence we need to sync to (if any). * - * Global and local mutex must be locked! + * Mutex must be locked! */ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, struct radeon_vm *vm, int ring) @@ -231,7 +231,7 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, * * Flush the vm (cayman+). * - * Global and local mutex must be locked! + * Mutex must be locked! */ void radeon_vm_flush(struct radeon_device *rdev, struct radeon_vm *vm, @@ -263,7 +263,7 @@ void radeon_vm_flush(struct radeon_device *rdev, * Fence the vm (cayman+). * Set the fence used to protect page table and id. * - * Global and local mutex must be locked! + * Mutex must be locked! */ void radeon_vm_fence(struct radeon_device *rdev, struct radeon_vm *vm, @@ -314,7 +314,7 @@ struct radeon_bo_va *radeon_vm_bo_find(struct radeon_vm *vm, * Add @bo to the list of bos associated with the vm * Returns newly added bo_va or NULL for failure * - * Object has to be reserved! + * Mutex must be locked and object has to be reserved! */ struct radeon_bo_va *radeon_vm_bo_add(struct radeon_device *rdev, struct radeon_vm *vm, @@ -336,9 +336,7 @@ struct radeon_bo_va *radeon_vm_bo_add(struct radeon_device *rdev, INIT_LIST_HEAD(&bo_va->bo_list); INIT_LIST_HEAD(&bo_va->vm_status);

- mutex_lock(&vm->mutex); list_add_tail(&bo_va->bo_list, &bo->va); - mutex_unlock(&vm->mutex);

return bo_va; } @@ -441,7 +439,8 @@ error_unreserve: * Validate and set the offset requested within the vm address space. * Returns 0 for success, error for failure. * - * Object has to be reserved and gets unreserved by this function! + * Mutex must be locked and object has to be reserved and gets + * unreserved by this function! */ int radeon_vm_bo_set_addr(struct radeon_device *rdev, struct radeon_bo_va *bo_va, @@ -472,14 +471,12 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, eoffset = last_pfn = 0; }

- mutex_lock(&vm->mutex); if (bo_va->it.start || bo_va->it.last) { if (bo_va->addr) { /* add a clone of the bo_va to clear the old address */ struct radeon_bo_va *tmp; tmp = kzalloc(sizeof(struct radeon_bo_va), GFP_KERNEL); if (!tmp) { - mutex_unlock(&vm->mutex); return -ENOMEM; } tmp->it.start = bo_va->it.start; @@ -509,7 +506,6 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, dev_err(rdev->dev, "bo %p va 0x%010Lx conflict with " "(bo %p 0x%010lx 0x%010lx)\n", bo_va->bo, soffset, tmp->bo, tmp->it.start, tmp->it.last); - mutex_unlock(&vm->mutex); return -EINVAL; } bo_va->it.start = soffset; @@ -537,9 +533,6 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, if (vm->page_tables[pt_idx].bo) continue;

- /* drop mutex to allocate and clear page table */ - mutex_unlock(&vm->mutex); - r = radeon_bo_create(rdev, RADEON_VM_PTE_COUNT * 8, RADEON_GPU_PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, 0, @@ -554,21 +547,10 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, return r; }

- /* aquire mutex again */ - mutex_lock(&vm->mutex); - if (vm->page_tables[pt_idx].bo) { - /* someone else allocated the pt in the meantime */ - mutex_unlock(&vm->mutex); - radeon_bo_unref(&pt); - mutex_lock(&vm->mutex); - continue; - } - vm->page_tables[pt_idx].addr = 0; vm->page_tables[pt_idx].bo = pt; }

- mutex_unlock(&vm->mutex); return 0; }

@@ -627,7 +609,7 @@ static uint32_t radeon_vm_page_flags(uint32_t flags) * and updates the page directory (cayman+). * Returns 0 for success, error for failure. * - * Global and local mutex must be locked! + * Mutex must be locked! */ int radeon_vm_update_page_directory(struct radeon_device *rdev, struct radeon_vm *vm) @@ -717,8 +699,6 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev, * @pe_end: last PTE to handle * @addr: addr those PTEs should point to * @flags: hw mapping flags - * - * Global and local mutex must be locked! */ static void radeon_vm_frag_ptes(struct radeon_device *rdev, struct radeon_ib *ib, @@ -798,7 +778,6 @@ static void radeon_vm_frag_ptes(struct radeon_device *rdev, * * Update the page tables in the range @start - @end (cayman+). * - * Global and local mutex must be locked! */ static int radeon_vm_update_ptes(struct radeon_device *rdev, struct radeon_vm *vm, @@ -870,7 +849,6 @@ static int radeon_vm_update_ptes(struct radeon_device *rdev, * * Fence the page tables in the range @start - @end (cayman+). * - * Global and local mutex must be locked! */ static void radeon_vm_fence_pts(struct radeon_vm *vm, uint64_t start, uint64_t end, @@ -896,7 +874,7 @@ static void radeon_vm_fence_pts(struct radeon_vm *vm, * Fill in the page table entries for @bo (cayman+). * Returns 0 for success, -EINVAL for failure. * - * Object have to be reserved and mutex must be locked! + * Mutex must be locked and BOs have to be reserved! */ int radeon_vm_bo_update(struct radeon_device *rdev, struct radeon_bo_va *bo_va, @@ -1097,7 +1075,7 @@ int radeon_vm_clear_invalids(struct radeon_device *rdev, * * Remove @bo_va->bo from the requested vm (cayman+). * - * Object have to be reserved! + * Mutex must be locked and object has to be reserved! */ void radeon_vm_bo_rmv(struct radeon_device *rdev, struct radeon_bo_va *bo_va) @@ -1106,7 +1084,6 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev,

list_del(&bo_va->bo_list);

- mutex_lock(&vm->mutex); interval_tree_remove(&bo_va->it, &vm->va); spin_lock(&vm->status_lock); list_del(&bo_va->vm_status); @@ -1119,8 +1096,6 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev, kfree(bo_va); } spin_unlock(&vm->status_lock); - - mutex_unlock(&vm->mutex); }

/** @@ -1168,7 +1143,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) vm->ids[i].flushed_updates = NULL; vm->ids[i].last_id_use = NULL; } - mutex_init(&vm->mutex); + mutex_init(&vm->lock); vm->va = RB_ROOT; spin_lock_init(&vm->status_lock); INIT_LIST_HEAD(&vm->invalidated); @@ -1245,5 +1220,5 @@ void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) radeon_fence_unref(&vm->ids[i].last_id_use); }

- mutex_destroy(&vm->mutex); + mutex_destroy(&vm->lock); }