drm/amdgpu: move sync into job object

No need to keep that for every IB.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

13 files changed, 46 insertions, 38 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 6cefde4aab49..99e660fec190 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -776,7 +776,6 @@ struct amdgpu_ib {
         bool                            grabbed_vmid;
         struct amdgpu_vm                *vm;
         struct amdgpu_ctx               *ctx;
-        struct amdgpu_sync              sync;
         uint32_t                        gds_base, gds_size;
         uint32_t                        gws_base, gws_size;
         uint32_t                        oa_base, oa_size;
@@ -1178,6 +1177,7 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib);
 int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
                        struct amdgpu_ib *ib, void *owner,
+                       struct fence *last_vm_update,
                        struct fence **f);
 int amdgpu_ib_pool_init(struct amdgpu_device *adev);
 void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
@@ -1236,6 +1236,7 @@ struct amdgpu_job {
         struct amd_sched_job    base;
         struct amdgpu_device    *adev;
         struct amdgpu_ring      *ring;
+        struct amdgpu_sync      sync;
         struct amdgpu_ib        *ibs;
         uint32_t                num_ibs;
         void                    *owner;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index a5311623a489..52c3eb96b199 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -411,7 +411,7 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 
         list_for_each_entry(e, &p->validated, tv.head) {
                 struct reservation_object *resv = e->robj->tbo.resv;
-                r = amdgpu_sync_resv(p->adev, &p->job->ibs[0].sync, resv, p->filp);
+                r = amdgpu_sync_resv(p->adev, &p->job->sync, resv, p->filp);
 
                 if (r)
                         return r;
@@ -491,7 +491,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p,
         if (r)
                 return r;
 
-        r = amdgpu_sync_fence(adev, &p->job->ibs[0].sync, vm->page_directory_fence);
+        r = amdgpu_sync_fence(adev, &p->job->sync, vm->page_directory_fence);
         if (r)
                 return r;
 
@@ -517,14 +517,14 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p,
                                 return r;
 
                         f = bo_va->last_pt_update;
-                        r = amdgpu_sync_fence(adev, &p->job->ibs[0].sync, f);
+                        r = amdgpu_sync_fence(adev, &p->job->sync, f);
                         if (r)
                                 return r;
                 }
 
         }
 
-        r = amdgpu_vm_clear_invalids(adev, vm, &p->job->ibs[0].sync);
+        r = amdgpu_vm_clear_invalids(adev, vm, &p->job->sync);
 
         if (amdgpu_vm_debug && p->bo_list) {
                 /* Invalidate all BOs to test for userspace bugs */
@@ -698,11 +698,8 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
                                   struct amdgpu_cs_parser *p)
 {
         struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-        struct amdgpu_ib *ib;
         int i, j, r;
 
-        /* Add dependencies to first IB */
-        ib = &p->job->ibs[0];
         for (i = 0; i < p->nchunks; ++i) {
                 struct drm_amdgpu_cs_chunk_dep *deps;
                 struct amdgpu_cs_chunk *chunk;
@@ -740,7 +737,8 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
                                 return r;
 
                         } else if (fence) {
-                                r = amdgpu_sync_fence(adev, &ib->sync, fence);
+                                r = amdgpu_sync_fence(adev, &p->job->sync,
+                                                      fence);
                                 fence_put(fence);
                                 amdgpu_ctx_put(ctx);
                                 if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 4b2c176b83f2..b5bdd5d59b58 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -74,8 +74,6 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
                         ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
         }
 
-        amdgpu_sync_create(&ib->sync);
-
         ib->vm = vm;
 
         return 0;
@@ -91,7 +89,6 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
  */
 void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib)
 {
-        amdgpu_sync_free(&ib->sync);
         amdgpu_sa_bo_free(adev, &ib->sa_bo, &ib->fence->base);
         if (ib->fence)
                 fence_put(&ib->fence->base);
@@ -121,6 +118,7 @@ void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib)
  */
 int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
                        struct amdgpu_ib *ibs, void *owner,
+                       struct fence *last_vm_update,
                        struct fence **f)
 {
         struct amdgpu_device *adev = ring->adev;
@@ -152,16 +150,9 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
                 return r;
         }
 
-        r = amdgpu_sync_wait(&ibs->sync);
-        if (r) {
-                amdgpu_ring_undo(ring);
-                dev_err(adev->dev, "failed to sync wait (%d)\n", r);
-                return r;
-        }
-
         if (vm) {
                 /* do context switch */
-                amdgpu_vm_flush(ring, vm, ib->sync.last_vm_update);
+                amdgpu_vm_flush(ring, vm, last_vm_update);
 
                 if (ring->funcs->emit_gds_switch)
                         amdgpu_ring_emit_gds_switch(ring, ib->vm->ids[ring->idx].id,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 6f3e757e056e..0f6719e0ace0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -46,6 +46,8 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
         (*job)->ibs = (void *)&(*job)[1];
         (*job)->num_ibs = num_ibs;
 
+        amdgpu_sync_create(&(*job)->sync);
+
         return 0;
 }
 
@@ -73,6 +75,7 @@ void amdgpu_job_free(struct amdgpu_job *job)
                 amdgpu_ib_free(job->adev, &job->ibs[i]);
 
         amdgpu_bo_unref(&job->uf.bo);
+        amdgpu_sync_free(&job->sync);
         kfree(job);
 }
 
@@ -99,23 +102,22 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
 static struct fence *amdgpu_job_dependency(struct amd_sched_job *sched_job)
 {
         struct amdgpu_job *job = to_amdgpu_job(sched_job);
-        struct amdgpu_sync *sync = &job->ibs->sync;
         struct amdgpu_vm *vm = job->ibs->vm;
 
-        struct fence *fence = amdgpu_sync_get_fence(sync);
+        struct fence *fence = amdgpu_sync_get_fence(&job->sync);
 
         if (fence == NULL && vm && !job->ibs->grabbed_vmid) {
                 struct amdgpu_ring *ring = job->ring;
                 int r;
 
-                r = amdgpu_vm_grab_id(vm, ring, sync,
+                r = amdgpu_vm_grab_id(vm, ring, &job->sync,
                                       &job->base.s_fence->base);
                 if (r)
                         DRM_ERROR("Error getting VM ID (%d)\n", r);
                 else
                         job->ibs->grabbed_vmid = true;
 
-                fence = amdgpu_sync_get_fence(sync);
+                fence = amdgpu_sync_get_fence(&job->sync);
         }
 
         return fence;
@@ -132,9 +134,16 @@ static struct fence *amdgpu_job_run(struct amd_sched_job *sched_job)
                 return NULL;
         }
         job = to_amdgpu_job(sched_job);
+
+        r = amdgpu_sync_wait(&job->sync);
+        if (r) {
+                DRM_ERROR("failed to sync wait (%d)\n", r);
+                return NULL;
+        }
+
         trace_amdgpu_sched_run_job(job);
-        r = amdgpu_ib_schedule(job->ring, job->num_ibs, job->ibs,
-                               job->owner, &fence);
+        r = amdgpu_ib_schedule(job->ring, job->num_ibs, job->ibs, job->owner,
+                               job->sync.last_vm_update, &fence);
         if (r) {
                 DRM_ERROR("Error scheduling IBs (%d)\n", r);
                 goto err;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 21c1a18c6d48..e47d5188c886 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -1032,7 +1032,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring,
                 return r;
 
         if (resv) {
-                r = amdgpu_sync_resv(adev, &job->ibs[0].sync, resv,
+                r = amdgpu_sync_resv(adev, &job->sync, resv,
                                      AMDGPU_FENCE_OWNER_UNDEFINED);
                 if (r) {
                         DRM_ERROR("sync failed (%d).\n", r);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 00b608b6c8c4..c536630580f8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -874,7 +874,7 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 
         if (direct) {
                 r = amdgpu_ib_schedule(ring, 1, ib,
-                                       AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+                                       AMDGPU_FENCE_OWNER_UNDEFINED, NULL, &f);
                 if (r)
                         goto err_free;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
index 97c22212d048..fb2ce3ed9aab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -411,7 +411,8 @@ int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
         for (i = ib->length_dw; i < ib_size_dw; ++i)
                 ib->ptr[i] = 0x0;
 
-        r = amdgpu_ib_schedule(ring, 1, ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err;
 
@@ -473,7 +474,8 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 
         if (direct) {
                 r = amdgpu_ib_schedule(ring, 1, ib,
-                                       AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+                                       AMDGPU_FENCE_OWNER_UNDEFINED,
+                                       NULL, &f);
                 if (r)
                         goto err;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index fb003089f73c..b291b1a4611a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -473,7 +473,8 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 
         if (ib->length_dw != 0) {
                 amdgpu_ring_pad_ib(ring, ib);
-                amdgpu_sync_resv(adev, &ib->sync, pd->tbo.resv, AMDGPU_FENCE_OWNER_VM);
+                amdgpu_sync_resv(adev, &job->sync, pd->tbo.resv,
+                                 AMDGPU_FENCE_OWNER_VM);
                 WARN_ON(ib->length_dw > ndw);
                 r = amdgpu_job_submit(job, ring, AMDGPU_FENCE_OWNER_VM, &fence);
                 if (r)
@@ -714,7 +715,7 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 
         ib = &job->ibs[0];
 
-        r = amdgpu_sync_resv(adev, &ib->sync, vm->page_directory->tbo.resv,
+        r = amdgpu_sync_resv(adev, &job->sync, vm->page_directory->tbo.resv,
                              owner);
         if (r)
                 goto error_free;
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index 6004dce23dc1..47fba3142a5a 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -633,7 +633,8 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
         ib.ptr[3] = 1;
         ib.ptr[4] = 0xDEADBEEF;
         ib.length_dw = 5;
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err1;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 4dac79ac9d3d..fd6796657051 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -2641,7 +2641,8 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
         ib.ptr[2] = 0xDEADBEEF;
         ib.length_dw = 3;
 
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err2;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 4a7708541723..415da6e100cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -709,7 +709,8 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
         ib.ptr[2] = 0xDEADBEEF;
         ib.length_dw = 3;
 
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err2;
 
@@ -1264,7 +1265,8 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
         ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
 
         /* shedule the ib on the ring */
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r) {
                 DRM_ERROR("amdgpu: ib submit failed (%d).\n", r);
                 goto fail;
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index 6a048b0b5fa7..423be6e10d01 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -691,7 +691,8 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
         ib.ptr[7] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
         ib.length_dw = 8;
 
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err1;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index 446aa016e96e..960462a9e941 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -842,7 +842,8 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
         ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
         ib.length_dw = 8;
 
-        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+        r = amdgpu_ib_schedule(ring, 1, &ib, AMDGPU_FENCE_OWNER_UNDEFINED,
+                               NULL, &f);
         if (r)
                 goto err1;