Merge tag 'drm-intel-fixes-2017-06-27' of git://anongit.freedesktop.org/git/drm-intel into drm-fixes

Just a few minor fixes. Important one is the execbuf async fix (aka
ANDROID_native_sync). There was another patch for a display coherency
corner case on APL, but we've random-walked in that space too much,
and the cherry-pick looked really invasive.

* tag 'drm-intel-fixes-2017-06-27' of git://anongit.freedesktop.org/git/drm-intel:
  drm/i915: Disable EXEC_OBJECT_ASYNC when doing relocations
  drm/i915: Hold struct_mutex for per-file stats in debugfs/i915_gem_object
  drm/i915: Retire the VMA's fence tracker before unbinding

3 files changed, 24 insertions, 4 deletions

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index d689e511744e..4bd1467c17b1 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -292,6 +292,8 @@ static int per_file_stats(int id, void *ptr, void *data)
         struct file_stats *stats = data;
         struct i915_vma *vma;
 
+        lockdep_assert_held(&obj->base.dev->struct_mutex);
+
         stats->count++;
         stats->total += obj->base.size;
         if (!obj->bind_count)
@@ -476,6 +478,8 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
                 struct drm_i915_gem_request *request;
                 struct task_struct *task;
 
+                mutex_lock(&dev->struct_mutex);
+
                 memset(&stats, 0, sizeof(stats));
                 stats.file_priv = file->driver_priv;
                 spin_lock(&file->table_lock);
@@ -487,7 +491,6 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
                  * still alive (e.g. get_pid(current) => fork() => exit()).
                  * Therefore, we need to protect this ->comm access using RCU.
                  */
-                mutex_lock(&dev->struct_mutex);
                 request = list_first_entry_or_null(&file_priv->mm.request_list,
                                                    struct drm_i915_gem_request,
                                                    client_link);
@@ -497,6 +500,7 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
                                 PIDTYPE_PID);
                 print_file_stats(m, task ? task->comm : "<unknown>", stats);
                 rcu_read_unlock();
+
                 mutex_unlock(&dev->struct_mutex);
         }
         mutex_unlock(&dev->filelist_mutex);
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index a3e59c8ef27b..9ad13eeed904 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -546,11 +546,12 @@ repeat:
 }
 
 static int
-i915_gem_execbuffer_relocate_entry(struct drm_i915_gem_object *obj,
+i915_gem_execbuffer_relocate_entry(struct i915_vma *vma,
                                    struct eb_vmas *eb,
                                    struct drm_i915_gem_relocation_entry *reloc,
                                    struct reloc_cache *cache)
 {
+        struct drm_i915_gem_object *obj = vma->obj;
         struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
         struct drm_gem_object *target_obj;
         struct drm_i915_gem_object *target_i915_obj;
@@ -628,6 +629,16 @@ i915_gem_execbuffer_relocate_entry(struct drm_i915_gem_object *obj,
                 return -EINVAL;
         }
 
+        /*
+         * If we write into the object, we need to force the synchronisation
+         * barrier, either with an asynchronous clflush or if we executed the
+         * patching using the GPU (though that should be serialised by the
+         * timeline). To be completely sure, and since we are required to
+         * do relocations we are already stalling, disable the user's opt
+         * of our synchronisation.
+         */
+        vma->exec_entry->flags &= ~EXEC_OBJECT_ASYNC;
+
         ret = relocate_entry(obj, reloc, cache, target_offset);
         if (ret)
                 return ret;
@@ -678,7 +689,7 @@ i915_gem_execbuffer_relocate_vma(struct i915_vma *vma,
                 do {
                         u64 offset = r->presumed_offset;
 
-                        ret = i915_gem_execbuffer_relocate_entry(vma->obj, eb, r, &cache);
+                        ret = i915_gem_execbuffer_relocate_entry(vma, eb, r, &cache);
                         if (ret)
                                 goto out;
 
@@ -726,7 +737,7 @@ i915_gem_execbuffer_relocate_vma_slow(struct i915_vma *vma,
 
         reloc_cache_init(&cache, eb->i915);
         for (i = 0; i < entry->relocation_count; i++) {
-                ret = i915_gem_execbuffer_relocate_entry(vma->obj, eb, &relocs[i], &cache);
+                ret = i915_gem_execbuffer_relocate_entry(vma, eb, &relocs[i], &cache);
                 if (ret)
                         break;
         }
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 1aba47024656..f066e2d785f5 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -650,6 +650,11 @@ int i915_vma_unbind(struct i915_vma *vma)
                                 break;
                 }
 
+                if (!ret) {
+                        ret = i915_gem_active_retire(&vma->last_fence,
+                                                     &vma->vm->i915->drm.struct_mutex);
+                }
+
                 __i915_vma_unpin(vma);
                 if (ret)
                         return ret;