drm/nvc0/fb: fix crash when different mutex is used to protect same list

Fixes regression introduced in commit 861d2107
"drm/nouveau/fb: merge fb/vram and port to subdev interfaces"

nv50_fb_vram_{new,del} functions were changed to use
nouveau_subdev->mutex instead of the old nouveau_mm->mutex.
nvc0_fb_vram_new still uses the nouveau_mm->mutex, but nvc0 doesn't
have its own fb_vram_del function, using nv50_fb_vram_del instead.
Because of this, on nvc0 a different mutex ends up being used to protect
additions and deletions to the same list.

This patch is a -stable candidate for 3.7.
Signed-off-by: Aleksi Torhamo <aleksi@torhamo.net>
Reported-by: Roy Spliet <r.spliet@student.tudelft.nl>
Tested-by: Roy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Cc: stable@vger.kernel.org

drivers/gpu/drm/nouveau/core/subdev/fb/nvc0.c

@@ -145,14 +145,14 @@ nvc0_fb_vram_new(struct nouveau_fb *pfb, u64 size, u32 align, u32 ncmin,
 	mem->memtype = type;
 	mem->size = size;
 
-	mutex_lock(&mm->mutex);
+	mutex_lock(&pfb->base.mutex);
 	do {
 		if (back)
 			ret = nouveau_mm_tail(mm, 1, size, ncmin, align, &r);
 		else
 			ret = nouveau_mm_head(mm, 1, size, ncmin, align, &r);
 		if (ret) {
-			mutex_unlock(&mm->mutex);
+			mutex_unlock(&pfb->base.mutex);
 			pfb->ram.put(pfb, &mem);
 			return ret;
 		}
@@ -160,7 +160,7 @@ nvc0_fb_vram_new(struct nouveau_fb *pfb, u64 size, u32 align, u32 ncmin,
 		list_add_tail(&r->rl_entry, &mem->regions);
 		size -= r->length;
 	} while (size);
-	mutex_unlock(&mm->mutex);
+	mutex_unlock(&pfb->base.mutex);
 
 	r = list_first_entry(&mem->regions, struct nouveau_mm_node, rl_entry);
 	mem->offset = (u64)r->offset << 12;