Merge tag 'drm-intel-next-fixes-2015-09-02' into drm-intel-next-queued

Backmerge -fixes since there's more DDI-E related cleanups on top of
the pile of -fixes for skl that just landed for 4.3.

Conflicts:
	drivers/gpu/drm/i915/intel_display.c
	drivers/gpu/drm/i914/intel_dp.c
	drivers/gpu/drm/i915/intel_lrc.c

Conflicts are all fairly harmless adjacent line stuff.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

1 files changed, 178 insertions, 96 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 9a4e3b63f1cb..a78a206e176e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -127,16 +127,16 @@ struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev,
 /**
  * amdgpu_vm_grab_id - allocate the next free VMID
  *
- * @ring: ring we want to submit job to
  * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
  *
- * Allocate an id for the vm (cayman+).
- * Returns the fence we need to sync to (if any).
+ * Allocate an id for the vm, adding fences to the sync obj as necessary.
  *
- * Global and local mutex must be locked!
+ * Global mutex must be locked!
  */
-struct amdgpu_fence *amdgpu_vm_grab_id(struct amdgpu_ring *ring,
-                                       struct amdgpu_vm *vm)
+int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+                      struct amdgpu_sync *sync)
 {
         struct amdgpu_fence *best[AMDGPU_MAX_RINGS] = {};
         struct amdgpu_vm_id *vm_id = &vm->ids[ring->idx];
@@ -148,7 +148,7 @@ struct amdgpu_fence *amdgpu_vm_grab_id(struct amdgpu_ring *ring,
         /* check if the id is still valid */
         if (vm_id->id && vm_id->last_id_use &&
             vm_id->last_id_use == adev->vm_manager.active[vm_id->id])
-                return NULL;
+                return 0;
 
         /* we definately need to flush */
         vm_id->pd_gpu_addr = ~0ll;
@@ -161,7 +161,7 @@ struct amdgpu_fence *amdgpu_vm_grab_id(struct amdgpu_ring *ring,
                         /* found a free one */
                         vm_id->id = i;
                         trace_amdgpu_vm_grab_id(i, ring->idx);
-                        return NULL;
+                        return 0;
                 }
 
                 if (amdgpu_fence_is_earlier(fence, best[fence->ring->idx])) {
@@ -172,15 +172,19 @@ struct amdgpu_fence *amdgpu_vm_grab_id(struct amdgpu_ring *ring,
 
         for (i = 0; i < 2; ++i) {
                 if (choices[i]) {
+                        struct amdgpu_fence *fence;
+
+                        fence  = adev->vm_manager.active[choices[i]];
                         vm_id->id = choices[i];
+
                         trace_amdgpu_vm_grab_id(choices[i], ring->idx);
-                        return adev->vm_manager.active[choices[i]];
+                        return amdgpu_sync_fence(ring->adev, sync, &fence->base);
                 }
         }
 
         /* should never happen */
         BUG();
-        return NULL;
+        return -EINVAL;
 }
 
 /**
@@ -200,13 +204,15 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring,
 {
         uint64_t pd_addr = amdgpu_bo_gpu_offset(vm->page_directory);
         struct amdgpu_vm_id *vm_id = &vm->ids[ring->idx];
+        struct amdgpu_fence *flushed_updates = vm_id->flushed_updates;
 
-        if (pd_addr != vm_id->pd_gpu_addr || !vm_id->flushed_updates ||
-            amdgpu_fence_is_earlier(vm_id->flushed_updates, updates)) {
+        if (pd_addr != vm_id->pd_gpu_addr || !flushed_updates ||
+            (updates && amdgpu_fence_is_earlier(flushed_updates, updates))) {
 
                 trace_amdgpu_vm_flush(pd_addr, ring->idx, vm_id->id);
-                amdgpu_fence_unref(&vm_id->flushed_updates);
-                vm_id->flushed_updates = amdgpu_fence_ref(updates);
+                vm_id->flushed_updates = amdgpu_fence_ref(
+                        amdgpu_fence_later(flushed_updates, updates));
+                amdgpu_fence_unref(&flushed_updates);
                 vm_id->pd_gpu_addr = pd_addr;
                 amdgpu_ring_emit_vm_flush(ring, vm_id->id, vm_id->pd_gpu_addr);
         }
@@ -300,6 +306,16 @@ static void amdgpu_vm_update_pages(struct amdgpu_device *adev,
         }
 }
 
+static int amdgpu_vm_free_job(
+        struct amdgpu_cs_parser *sched_job)
+{
+        int i;
+        for (i = 0; i < sched_job->num_ibs; i++)
+                amdgpu_ib_free(sched_job->adev, &sched_job->ibs[i]);
+        kfree(sched_job->ibs);
+        return 0;
+}
+
 /**
  * amdgpu_vm_clear_bo - initially clear the page dir/table
  *
@@ -310,7 +326,8 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
                               struct amdgpu_bo *bo)
 {
         struct amdgpu_ring *ring = adev->vm_manager.vm_pte_funcs_ring;
-        struct amdgpu_ib ib;
+        struct fence *fence = NULL;
+        struct amdgpu_ib *ib;
         unsigned entries;
         uint64_t addr;
         int r;
@@ -330,24 +347,33 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
         addr = amdgpu_bo_gpu_offset(bo);
         entries = amdgpu_bo_size(bo) / 8;
 
-        r = amdgpu_ib_get(ring, NULL, entries * 2 + 64, &ib);
-        if (r)
+        ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+        if (!ib)
                 goto error_unreserve;
 
-        ib.length_dw = 0;
-
-        amdgpu_vm_update_pages(adev, &ib, addr, 0, entries, 0, 0, 0);
-        amdgpu_vm_pad_ib(adev, &ib);
-        WARN_ON(ib.length_dw > 64);
-
-        r = amdgpu_ib_schedule(adev, 1, &ib, AMDGPU_FENCE_OWNER_VM);
+        r = amdgpu_ib_get(ring, NULL, entries * 2 + 64, ib);
         if (r)
                 goto error_free;
 
-        amdgpu_bo_fence(bo, ib.fence, true);
-
+        ib->length_dw = 0;
+
+        amdgpu_vm_update_pages(adev, ib, addr, 0, entries, 0, 0, 0);
+        amdgpu_vm_pad_ib(adev, ib);
+        WARN_ON(ib->length_dw > 64);
+        r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+                                                 &amdgpu_vm_free_job,
+                                                 AMDGPU_FENCE_OWNER_VM,
+                                                 &fence);
+        if (!r)
+                amdgpu_bo_fence(bo, fence, true);
+        fence_put(fence);
+        if (amdgpu_enable_scheduler) {
+                amdgpu_bo_unreserve(bo);
+                return 0;
+        }
 error_free:
-        amdgpu_ib_free(adev, &ib);
+        amdgpu_ib_free(adev, ib);
+        kfree(ib);
 
 error_unreserve:
         amdgpu_bo_unreserve(bo);
@@ -400,7 +426,9 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
         uint32_t incr = AMDGPU_VM_PTE_COUNT * 8;
         uint64_t last_pde = ~0, last_pt = ~0;
         unsigned count = 0, pt_idx, ndw;
-        struct amdgpu_ib ib;
+        struct amdgpu_ib *ib;
+        struct fence *fence = NULL;
+
         int r;
 
         /* padding, etc. */
@@ -413,10 +441,14 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
         if (ndw > 0xfffff)
                 return -ENOMEM;
 
-        r = amdgpu_ib_get(ring, NULL, ndw * 4, &ib);
+        ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+        if (!ib)
+                return -ENOMEM;
+
+        r = amdgpu_ib_get(ring, NULL, ndw * 4, ib);
         if (r)
                 return r;
-        ib.length_dw = 0;
+        ib->length_dw = 0;
 
         /* walk over the address space and update the page directory */
         for (pt_idx = 0; pt_idx <= vm->max_pde_used; ++pt_idx) {
@@ -436,7 +468,7 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
                     ((last_pt + incr * count) != pt)) {
 
                         if (count) {
-                                amdgpu_vm_update_pages(adev, &ib, last_pde,
+                                amdgpu_vm_update_pages(adev, ib, last_pde,
                                                        last_pt, count, incr,
                                                        AMDGPU_PTE_VALID, 0);
                         }
@@ -450,23 +482,37 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
         }
 
         if (count)
-                amdgpu_vm_update_pages(adev, &ib, last_pde, last_pt, count,
+                amdgpu_vm_update_pages(adev, ib, last_pde, last_pt, count,
                                        incr, AMDGPU_PTE_VALID, 0);
 
-        if (ib.length_dw != 0) {
-                amdgpu_vm_pad_ib(adev, &ib);
-                amdgpu_sync_resv(adev, &ib.sync, pd->tbo.resv, AMDGPU_FENCE_OWNER_VM);
-                WARN_ON(ib.length_dw > ndw);
-                r = amdgpu_ib_schedule(adev, 1, &ib, AMDGPU_FENCE_OWNER_VM);
-                if (r) {
-                        amdgpu_ib_free(adev, &ib);
-                        return r;
-                }
-                amdgpu_bo_fence(pd, ib.fence, true);
+        if (ib->length_dw != 0) {
+                amdgpu_vm_pad_ib(adev, ib);
+                amdgpu_sync_resv(adev, &ib->sync, pd->tbo.resv, AMDGPU_FENCE_OWNER_VM);
+                WARN_ON(ib->length_dw > ndw);
+                r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+                                                         &amdgpu_vm_free_job,
+                                                         AMDGPU_FENCE_OWNER_VM,
+                                                         &fence);
+                if (r)
+                        goto error_free;
+
+                amdgpu_bo_fence(pd, fence, true);
+                fence_put(vm->page_directory_fence);
+                vm->page_directory_fence = fence_get(fence);
+                fence_put(fence);
+        }
+
+        if (!amdgpu_enable_scheduler || ib->length_dw == 0) {
+                amdgpu_ib_free(adev, ib);
+                kfree(ib);
         }
-        amdgpu_ib_free(adev, &ib);
 
         return 0;
+
+error_free:
+        amdgpu_ib_free(adev, ib);
+        kfree(ib);
+        return r;
 }
 
 /**
@@ -640,7 +686,7 @@ static int amdgpu_vm_update_ptes(struct amdgpu_device *adev,
  */
 static void amdgpu_vm_fence_pts(struct amdgpu_vm *vm,
                                 uint64_t start, uint64_t end,
-                                struct amdgpu_fence *fence)
+                                struct fence *fence)
 {
         unsigned i;
 
@@ -670,12 +716,13 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
                                        struct amdgpu_vm *vm,
                                        struct amdgpu_bo_va_mapping *mapping,
                                        uint64_t addr, uint32_t gtt_flags,
-                                       struct amdgpu_fence **fence)
+                                       struct fence **fence)
 {
         struct amdgpu_ring *ring = adev->vm_manager.vm_pte_funcs_ring;
         unsigned nptes, ncmds, ndw;
         uint32_t flags = gtt_flags;
-        struct amdgpu_ib ib;
+        struct amdgpu_ib *ib;
+        struct fence *f = NULL;
         int r;
 
         /* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
@@ -722,46 +769,65 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
         if (ndw > 0xfffff)
                 return -ENOMEM;
 
-        r = amdgpu_ib_get(ring, NULL, ndw * 4, &ib);
-        if (r)
+        ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+        if (!ib)
+                return -ENOMEM;
+
+        r = amdgpu_ib_get(ring, NULL, ndw * 4, ib);
+        if (r) {
+                kfree(ib);
                 return r;
-        ib.length_dw = 0;
+        }
+
+        ib->length_dw = 0;
 
         if (!(flags & AMDGPU_PTE_VALID)) {
                 unsigned i;
 
                 for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
                         struct amdgpu_fence *f = vm->ids[i].last_id_use;
-                        amdgpu_sync_fence(&ib.sync, f);
+                        r = amdgpu_sync_fence(adev, &ib->sync, &f->base);
+                        if (r)
+                                return r;
                 }
         }
 
-        r = amdgpu_vm_update_ptes(adev, vm, &ib, mapping->it.start,
+        r = amdgpu_vm_update_ptes(adev, vm, ib, mapping->it.start,
                                   mapping->it.last + 1, addr + mapping->offset,
                                   flags, gtt_flags);
 
         if (r) {
-                amdgpu_ib_free(adev, &ib);
+                amdgpu_ib_free(adev, ib);
+                kfree(ib);
                 return r;
         }
 
-        amdgpu_vm_pad_ib(adev, &ib);
-        WARN_ON(ib.length_dw > ndw);
+        amdgpu_vm_pad_ib(adev, ib);
+        WARN_ON(ib->length_dw > ndw);
+        r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+                                                 &amdgpu_vm_free_job,
+                                                 AMDGPU_FENCE_OWNER_VM,
+                                                 &f);
+        if (r)
+                goto error_free;
 
-        r = amdgpu_ib_schedule(adev, 1, &ib, AMDGPU_FENCE_OWNER_VM);
-        if (r) {
-                amdgpu_ib_free(adev, &ib);
-                return r;
-        }
         amdgpu_vm_fence_pts(vm, mapping->it.start,
-                            mapping->it.last + 1, ib.fence);
+                            mapping->it.last + 1, f);
         if (fence) {
-                amdgpu_fence_unref(fence);
-                *fence = amdgpu_fence_ref(ib.fence);
+                fence_put(*fence);
+                *fence = fence_get(f);
+        }
+        fence_put(f);
+        if (!amdgpu_enable_scheduler) {
+                amdgpu_ib_free(adev, ib);
+                kfree(ib);
         }
-        amdgpu_ib_free(adev, &ib);
-
         return 0;
+
+error_free:
+        amdgpu_ib_free(adev, ib);
+        kfree(ib);
+        return r;
 }
 
 /**
@@ -794,21 +860,25 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
                 addr = 0;
         }
 
-        if (addr == bo_va->addr)
-                return 0;
-
         flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem);
 
-        list_for_each_entry(mapping, &bo_va->mappings, list) {
+        spin_lock(&vm->status_lock);
+        if (!list_empty(&bo_va->vm_status))
+                list_splice_init(&bo_va->valids, &bo_va->invalids);
+        spin_unlock(&vm->status_lock);
+
+        list_for_each_entry(mapping, &bo_va->invalids, list) {
                 r = amdgpu_vm_bo_update_mapping(adev, vm, mapping, addr,
                                                 flags, &bo_va->last_pt_update);
                 if (r)
                         return r;
         }
 
-        bo_va->addr = addr;
         spin_lock(&vm->status_lock);
+        list_splice_init(&bo_va->invalids, &bo_va->valids);
         list_del_init(&bo_va->vm_status);
+        if (!mem)
+                list_add(&bo_va->vm_status, &vm->cleared);
         spin_unlock(&vm->status_lock);
 
         return 0;
@@ -861,7 +931,7 @@ int amdgpu_vm_clear_invalids(struct amdgpu_device *adev,
                              struct amdgpu_vm *vm, struct amdgpu_sync *sync)
 {
         struct amdgpu_bo_va *bo_va = NULL;
-        int r;
+        int r = 0;
 
         spin_lock(&vm->status_lock);
         while (!list_empty(&vm->invalidated)) {
@@ -878,8 +948,9 @@ int amdgpu_vm_clear_invalids(struct amdgpu_device *adev,
         spin_unlock(&vm->status_lock);
 
         if (bo_va)
-                amdgpu_sync_fence(sync, bo_va->last_pt_update);
-        return 0;
+                r = amdgpu_sync_fence(adev, sync, bo_va->last_pt_update);
+
+        return r;
 }
 
 /**
@@ -907,10 +978,10 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
         }
         bo_va->vm = vm;
         bo_va->bo = bo;
-        bo_va->addr = 0;
         bo_va->ref_count = 1;
         INIT_LIST_HEAD(&bo_va->bo_list);
-        INIT_LIST_HEAD(&bo_va->mappings);
+        INIT_LIST_HEAD(&bo_va->valids);
+        INIT_LIST_HEAD(&bo_va->invalids);
         INIT_LIST_HEAD(&bo_va->vm_status);
 
         mutex_lock(&vm->mutex);
@@ -999,12 +1070,10 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
         mapping->offset = offset;
         mapping->flags = flags;
 
-        list_add(&mapping->list, &bo_va->mappings);
+        list_add(&mapping->list, &bo_va->invalids);
         interval_tree_insert(&mapping->it, &vm->va);
         trace_amdgpu_vm_bo_map(bo_va, mapping);
 
-        bo_va->addr = 0;
-
         /* Make sure the page tables are allocated */
         saddr >>= amdgpu_vm_block_size;
         eaddr >>= amdgpu_vm_block_size;
@@ -1085,17 +1154,27 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 {
         struct amdgpu_bo_va_mapping *mapping;
         struct amdgpu_vm *vm = bo_va->vm;
+        bool valid = true;
 
         saddr /= AMDGPU_GPU_PAGE_SIZE;
 
-        list_for_each_entry(mapping, &bo_va->mappings, list) {
+        list_for_each_entry(mapping, &bo_va->valids, list) {
                 if (mapping->it.start == saddr)
                         break;
         }
 
-        if (&mapping->list == &bo_va->mappings) {
-                amdgpu_bo_unreserve(bo_va->bo);
-                return -ENOENT;
+        if (&mapping->list == &bo_va->valids) {
+                valid = false;
+
+                list_for_each_entry(mapping, &bo_va->invalids, list) {
+                        if (mapping->it.start == saddr)
+                                break;
+                }
+
+                if (&mapping->list == &bo_va->invalids) {
+                        amdgpu_bo_unreserve(bo_va->bo);
+                        return -ENOENT;
+                }
         }
 
         mutex_lock(&vm->mutex);
@@ -1103,12 +1182,10 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
         interval_tree_remove(&mapping->it, &vm->va);
         trace_amdgpu_vm_bo_unmap(bo_va, mapping);
 
-        if (bo_va->addr) {
-                /* clear the old address */
+        if (valid)
                 list_add(&mapping->list, &vm->freed);
-        } else {
+        else
                 kfree(mapping);
-        }
         mutex_unlock(&vm->mutex);
         amdgpu_bo_unreserve(bo_va->bo);
 
@@ -1139,16 +1216,19 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
         list_del(&bo_va->vm_status);
         spin_unlock(&vm->status_lock);
 
-        list_for_each_entry_safe(mapping, next, &bo_va->mappings, list) {
+        list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
                 list_del(&mapping->list);
                 interval_tree_remove(&mapping->it, &vm->va);
                 trace_amdgpu_vm_bo_unmap(bo_va, mapping);
-                if (bo_va->addr)
-                        list_add(&mapping->list, &vm->freed);
-                else
-                        kfree(mapping);
+                list_add(&mapping->list, &vm->freed);
         }
-        amdgpu_fence_unref(&bo_va->last_pt_update);
+        list_for_each_entry_safe(mapping, next, &bo_va->invalids, list) {
+                list_del(&mapping->list);
+                interval_tree_remove(&mapping->it, &vm->va);
+                kfree(mapping);
+        }
+
+        fence_put(bo_va->last_pt_update);
         kfree(bo_va);
 
         mutex_unlock(&vm->mutex);
@@ -1169,12 +1249,10 @@ void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
         struct amdgpu_bo_va *bo_va;
 
         list_for_each_entry(bo_va, &bo->va, bo_list) {
-                if (bo_va->addr) {
-                        spin_lock(&bo_va->vm->status_lock);
-                        list_del(&bo_va->vm_status);
+                spin_lock(&bo_va->vm->status_lock);
+                if (list_empty(&bo_va->vm_status))
                         list_add(&bo_va->vm_status, &bo_va->vm->invalidated);
-                        spin_unlock(&bo_va->vm->status_lock);
-                }
+                spin_unlock(&bo_va->vm->status_lock);
         }
 }
 
@@ -1202,6 +1280,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
         vm->va = RB_ROOT;
         spin_lock_init(&vm->status_lock);
         INIT_LIST_HEAD(&vm->invalidated);
+        INIT_LIST_HEAD(&vm->cleared);
         INIT_LIST_HEAD(&vm->freed);
 
         pd_size = amdgpu_vm_directory_size(adev);
@@ -1215,6 +1294,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
                 return -ENOMEM;
         }
 
+        vm->page_directory_fence = NULL;
+
         r = amdgpu_bo_create(adev, pd_size, align, true,
                              AMDGPU_GEM_DOMAIN_VRAM, 0,
                              NULL, &vm->page_directory);
@@ -1263,6 +1344,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
         kfree(vm->page_tables);
 
         amdgpu_bo_unref(&vm->page_directory);
+        fence_put(vm->page_directory_fence);
 
         for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
                 amdgpu_fence_unref(&vm->ids[i].flushed_updates);