drm/i915: Fix lock order reversal in shmem pread path.
authorEric Anholt <eric@anholt.net>
Tue, 10 Mar 2009 18:44:52 +0000 (11:44 -0700)
committerEric Anholt <eric@anholt.net>
Fri, 27 Mar 2009 21:47:21 +0000 (14:47 -0700)
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
drivers/gpu/drm/i915/i915_gem.c

index bdc7326052dfb8f9379a44588f44ef6b5220f2d5..010af908bdb602725b45dcf1a11ce530f913c322 100644 (file)
@@ -136,6 +136,24 @@ i915_gem_create_ioctl(struct drm_device *dev, void *data,
        return 0;
 }
 
+static inline int
+fast_shmem_read(struct page **pages,
+               loff_t page_base, int page_offset,
+               char __user *data,
+               int length)
+{
+       char __iomem *vaddr;
+       int ret;
+
+       vaddr = kmap_atomic(pages[page_base >> PAGE_SHIFT], KM_USER0);
+       if (vaddr == NULL)
+               return -ENOMEM;
+       ret = __copy_to_user_inatomic(data, vaddr + page_offset, length);
+       kunmap_atomic(vaddr, KM_USER0);
+
+       return ret;
+}
+
 static inline int
 slow_shmem_copy(struct page *dst_page,
                int dst_offset,
@@ -163,6 +181,179 @@ slow_shmem_copy(struct page *dst_page,
        return 0;
 }
 
+/**
+ * This is the fast shmem pread path, which attempts to copy_from_user directly
+ * from the backing pages of the object to the user's address space.  On a
+ * fault, it fails so we can fall back to i915_gem_shmem_pwrite_slow().
+ */
+static int
+i915_gem_shmem_pread_fast(struct drm_device *dev, struct drm_gem_object *obj,
+                         struct drm_i915_gem_pread *args,
+                         struct drm_file *file_priv)
+{
+       struct drm_i915_gem_object *obj_priv = obj->driver_private;
+       ssize_t remain;
+       loff_t offset, page_base;
+       char __user *user_data;
+       int page_offset, page_length;
+       int ret;
+
+       user_data = (char __user *) (uintptr_t) args->data_ptr;
+       remain = args->size;
+
+       mutex_lock(&dev->struct_mutex);
+
+       ret = i915_gem_object_get_pages(obj);
+       if (ret != 0)
+               goto fail_unlock;
+
+       ret = i915_gem_object_set_cpu_read_domain_range(obj, args->offset,
+                                                       args->size);
+       if (ret != 0)
+               goto fail_put_pages;
+
+       obj_priv = obj->driver_private;
+       offset = args->offset;
+
+       while (remain > 0) {
+               /* Operation in this page
+                *
+                * page_base = page offset within aperture
+                * page_offset = offset within page
+                * page_length = bytes to copy for this page
+                */
+               page_base = (offset & ~(PAGE_SIZE-1));
+               page_offset = offset & (PAGE_SIZE-1);
+               page_length = remain;
+               if ((page_offset + remain) > PAGE_SIZE)
+                       page_length = PAGE_SIZE - page_offset;
+
+               ret = fast_shmem_read(obj_priv->pages,
+                                     page_base, page_offset,
+                                     user_data, page_length);
+               if (ret)
+                       goto fail_put_pages;
+
+               remain -= page_length;
+               user_data += page_length;
+               offset += page_length;
+       }
+
+fail_put_pages:
+       i915_gem_object_put_pages(obj);
+fail_unlock:
+       mutex_unlock(&dev->struct_mutex);
+
+       return ret;
+}
+
+/**
+ * This is the fallback shmem pread path, which allocates temporary storage
+ * in kernel space to copy_to_user into outside of the struct_mutex, so we
+ * can copy out of the object's backing pages while holding the struct mutex
+ * and not take page faults.
+ */
+static int
+i915_gem_shmem_pread_slow(struct drm_device *dev, struct drm_gem_object *obj,
+                         struct drm_i915_gem_pread *args,
+                         struct drm_file *file_priv)
+{
+       struct drm_i915_gem_object *obj_priv = obj->driver_private;
+       struct mm_struct *mm = current->mm;
+       struct page **user_pages;
+       ssize_t remain;
+       loff_t offset, pinned_pages, i;
+       loff_t first_data_page, last_data_page, num_pages;
+       int shmem_page_index, shmem_page_offset;
+       int data_page_index,  data_page_offset;
+       int page_length;
+       int ret;
+       uint64_t data_ptr = args->data_ptr;
+
+       remain = args->size;
+
+       /* Pin the user pages containing the data.  We can't fault while
+        * holding the struct mutex, yet we want to hold it while
+        * dereferencing the user data.
+        */
+       first_data_page = data_ptr / PAGE_SIZE;
+       last_data_page = (data_ptr + args->size - 1) / PAGE_SIZE;
+       num_pages = last_data_page - first_data_page + 1;
+
+       user_pages = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
+       if (user_pages == NULL)
+               return -ENOMEM;
+
+       down_read(&mm->mmap_sem);
+       pinned_pages = get_user_pages(current, mm, (uintptr_t)args->data_ptr,
+                                     num_pages, 0, 0, user_pages, NULL);
+       up_read(&mm->mmap_sem);
+       if (pinned_pages < num_pages) {
+               ret = -EFAULT;
+               goto fail_put_user_pages;
+       }
+
+       mutex_lock(&dev->struct_mutex);
+
+       ret = i915_gem_object_get_pages(obj);
+       if (ret != 0)
+               goto fail_unlock;
+
+       ret = i915_gem_object_set_cpu_read_domain_range(obj, args->offset,
+                                                       args->size);
+       if (ret != 0)
+               goto fail_put_pages;
+
+       obj_priv = obj->driver_private;
+       offset = args->offset;
+
+       while (remain > 0) {
+               /* Operation in this page
+                *
+                * shmem_page_index = page number within shmem file
+                * shmem_page_offset = offset within page in shmem file
+                * data_page_index = page number in get_user_pages return
+                * data_page_offset = offset with data_page_index page.
+                * page_length = bytes to copy for this page
+                */
+               shmem_page_index = offset / PAGE_SIZE;
+               shmem_page_offset = offset & ~PAGE_MASK;
+               data_page_index = data_ptr / PAGE_SIZE - first_data_page;
+               data_page_offset = data_ptr & ~PAGE_MASK;
+
+               page_length = remain;
+               if ((shmem_page_offset + page_length) > PAGE_SIZE)
+                       page_length = PAGE_SIZE - shmem_page_offset;
+               if ((data_page_offset + page_length) > PAGE_SIZE)
+                       page_length = PAGE_SIZE - data_page_offset;
+
+               ret = slow_shmem_copy(user_pages[data_page_index],
+                                     data_page_offset,
+                                     obj_priv->pages[shmem_page_index],
+                                     shmem_page_offset,
+                                     page_length);
+               if (ret)
+                       goto fail_put_pages;
+
+               remain -= page_length;
+               data_ptr += page_length;
+               offset += page_length;
+       }
+
+fail_put_pages:
+       i915_gem_object_put_pages(obj);
+fail_unlock:
+       mutex_unlock(&dev->struct_mutex);
+fail_put_user_pages:
+       for (i = 0; i < pinned_pages; i++) {
+               SetPageDirty(user_pages[i]);
+               page_cache_release(user_pages[i]);
+       }
+       kfree(user_pages);
+
+       return ret;
+}
+
 /**
  * Reads data from the object referenced by handle.
  *
@@ -175,8 +366,6 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
        struct drm_i915_gem_pread *args = data;
        struct drm_gem_object *obj;
        struct drm_i915_gem_object *obj_priv;
-       ssize_t read;
-       loff_t offset;
        int ret;
 
        obj = drm_gem_object_lookup(dev, file_priv, args->handle);
@@ -194,33 +383,13 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
                return -EINVAL;
        }
 
-       mutex_lock(&dev->struct_mutex);
-
-       ret = i915_gem_object_set_cpu_read_domain_range(obj, args->offset,
-                                                       args->size);
-       if (ret != 0) {
-               drm_gem_object_unreference(obj);
-               mutex_unlock(&dev->struct_mutex);
-               return ret;
-       }
-
-       offset = args->offset;
-
-       read = vfs_read(obj->filp, (char __user *)(uintptr_t)args->data_ptr,
-                       args->size, &offset);
-       if (read != args->size) {
-               drm_gem_object_unreference(obj);
-               mutex_unlock(&dev->struct_mutex);
-               if (read < 0)
-                       return read;
-               else
-                       return -EINVAL;
-       }
+       ret = i915_gem_shmem_pread_fast(dev, obj, args, file_priv);
+       if (ret != 0)
+               ret = i915_gem_shmem_pread_slow(dev, obj, args, file_priv);
 
        drm_gem_object_unreference(obj);
-       mutex_unlock(&dev->struct_mutex);
 
-       return 0;
+       return ret;
 }
 
 /* This is the fast write path which cannot handle