btrfs: convert cow_file_range() to take a folio
authorJosef Bacik <josef@toxicpanda.com>
Wed, 24 Jul 2024 20:37:29 +0000 (16:37 -0400)
committerDavid Sterba <dsterba@suse.com>
Tue, 10 Sep 2024 14:51:15 +0000 (16:51 +0200)
Convert this to take a folio and pass it into all of the various cleanup
functions.  Update the callers to pass in a folio instead.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/inode.c

index fd32324..6047fd3 100644 (file)
@@ -1307,21 +1307,21 @@ u64 btrfs_get_extent_allocation_hint(struct btrfs_inode *inode, u64 start,
  * allocate extents on disk for the range, and create ordered data structs
  * in ram to track those extents.
  *
- * locked_page is the page that writepage had locked already.  We use
+ * locked_folio is the folio that writepage had locked already.  We use
  * it to make sure we don't do extra locks or unlocks.
  *
- * When this function fails, it unlocks all pages except @locked_page.
+ * When this function fails, it unlocks all pages except @locked_folio.
  *
  * When this function successfully creates an inline extent, it returns 1 and
- * unlocks all pages including locked_page and starts I/O on them.
- * (In reality inline extents are limited to a single page, so locked_page is
+ * unlocks all pages including locked_folio and starts I/O on them.
+ * (In reality inline extents are limited to a single page, so locked_folio is
  * the only page handled anyway).
  *
  * When this function succeed and creates a normal extent, the page locking
  * status depends on the passed in flags:
  *
  * - If @keep_locked is set, all pages are kept locked.
- * - Else all pages except for @locked_page are unlocked.
+ * - Else all pages except for @locked_folio are unlocked.
  *
  * When a failure happens in the second or later iteration of the
  * while-loop, the ordered extents created in previous iterations are kept
@@ -1330,8 +1330,8 @@ u64 btrfs_get_extent_allocation_hint(struct btrfs_inode *inode, u64 start,
  * example.
  */
 static noinline int cow_file_range(struct btrfs_inode *inode,
-                                  struct page *locked_page, u64 start, u64 end,
-                                  u64 *done_offset,
+                                  struct folio *locked_folio, u64 start,
+                                  u64 end, u64 *done_offset,
                                   bool keep_locked, bool no_inline)
 {
        struct btrfs_root *root = inode->root;
@@ -1364,9 +1364,8 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
 
        if (!no_inline) {
                /* lets try to make an inline extent */
-               ret = cow_file_range_inline(inode, page_folio(locked_page),
-                                           start, end, 0, BTRFS_COMPRESS_NONE,
-                                           NULL, false);
+               ret = cow_file_range_inline(inode, locked_folio, start, end, 0,
+                                           BTRFS_COMPRESS_NONE, NULL, false);
                if (ret <= 0) {
                        /*
                         * We succeeded, return 1 so the caller knows we're done
@@ -1502,7 +1501,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
                page_ops |= PAGE_SET_ORDERED;
 
                extent_clear_unlock_delalloc(inode, start, start + ram_size - 1,
-                                            page_folio(locked_page), &cached,
+                                            locked_folio, &cached,
                                             EXTENT_LOCKED | EXTENT_DELALLOC,
                                             page_ops);
                if (num_bytes < cur_alloc_size)
@@ -1555,14 +1554,13 @@ out_unlock:
         * function.
         *
         * However, in case of @keep_locked, we still need to unlock the pages
-        * (except @locked_page) to ensure all the pages are unlocked.
+        * (except @locked_folio) to ensure all the pages are unlocked.
         */
        if (keep_locked && orig_start < start) {
-               if (!locked_page)
+               if (!locked_folio)
                        mapping_set_error(inode->vfs_inode.i_mapping, ret);
                extent_clear_unlock_delalloc(inode, orig_start, start - 1,
-                                            page_folio(locked_page), NULL, 0,
-                                            page_ops);
+                                            locked_folio, NULL, 0, page_ops);
        }
 
        /*
@@ -1585,8 +1583,7 @@ out_unlock:
        if (extent_reserved) {
                extent_clear_unlock_delalloc(inode, start,
                                             start + cur_alloc_size - 1,
-                                            page_folio(locked_page), &cached,
-                                            clear_bits,
+                                            locked_folio, &cached, clear_bits,
                                             page_ops);
                btrfs_qgroup_free_data(inode, NULL, start, cur_alloc_size, NULL);
                start += cur_alloc_size;
@@ -1600,9 +1597,8 @@ out_unlock:
         */
        if (start < end) {
                clear_bits |= EXTENT_CLEAR_DATA_RESV;
-               extent_clear_unlock_delalloc(inode, start, end,
-                                            page_folio(locked_page), &cached,
-                                            clear_bits, page_ops);
+               extent_clear_unlock_delalloc(inode, start, end, locked_folio,
+                                            &cached, clear_bits, page_ops);
                btrfs_qgroup_free_data(inode, NULL, start, cur_alloc_size, NULL);
        }
        return ret;
@@ -1755,7 +1751,7 @@ static noinline int run_delalloc_cow(struct btrfs_inode *inode,
        int ret;
 
        while (start <= end) {
-               ret = cow_file_range(inode, &locked_folio->page, start, end,
+               ret = cow_file_range(inode, locked_folio, start, end,
                                     &done_offset, true, false);
                if (ret)
                        return ret;
@@ -1837,7 +1833,8 @@ static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page,
         * is written out and unlocked directly and a normal NOCOW extent
         * doesn't work.
         */
-       ret = cow_file_range(inode, locked_page, start, end, NULL, false, true);
+       ret = cow_file_range(inode, page_folio(locked_page), start, end, NULL,
+                            false, true);
        ASSERT(ret != 1);
        return ret;
 }
@@ -2314,8 +2311,8 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page
                ret = run_delalloc_cow(inode, page_folio(locked_page), start,
                                       end, wbc, true);
        else
-               ret = cow_file_range(inode, locked_page, start, end, NULL,
-                                    false, false);
+               ret = cow_file_range(inode, page_folio(locked_page), start, end,
+                                    NULL, false, false);
 
 out:
        if (ret < 0)