Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit d5550c63 authored by Chris Mason's avatar Chris Mason
Browse files

Btrfs: don't lock bits in the extent tree during writepage 



At writepage time, we have the page locked and we have the
extent_map entry for this extent pinned in the extent_map tree.
So, the page can't go away and its mapping can't change.

There is no need for the extra extent_state lock bits during writepage.

Signed-off-by: default avatarChris Mason <chris.mason@oracle.com>
parent 2c64c53d

fs/btrfs/extent_io.c

+0 −21
Original line number Original line Diff line number Diff line
@@ -2219,16 +2219,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, 
			goto done_unlocked;

			goto done_unlocked;

		}

		}

	}

	}

	lock_extent_bits(tree, start, page_end, 0, &cached_state, GFP_NOFS);



	unlock_start = start;



	if (tree->ops && tree->ops->writepage_start_hook) {

	if (tree->ops && tree->ops->writepage_start_hook) {

		ret = tree->ops->writepage_start_hook(page, start,

		ret = tree->ops->writepage_start_hook(page, start,

						      page_end);

						      page_end);

		if (ret == -EAGAIN) {

		if (ret == -EAGAIN) {

			unlock_extent_cached(tree, start, page_end,

					     &cached_state, GFP_NOFS);

			redirty_page_for_writepage(wbc, page);

			redirty_page_for_writepage(wbc, page);

			update_nr_written(page, wbc, nr_written);

			update_nr_written(page, wbc, nr_written);

			unlock_page(page);

			unlock_page(page);
@@ -2244,13 +2238,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, 
	update_nr_written(page, wbc, nr_written + 1);

	update_nr_written(page, wbc, nr_written + 1);





	end = page_end;

	end = page_end;

	if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0))

		printk(KERN_ERR "btrfs delalloc bits after lock_extent\n");



	if (last_byte <= start) {

	if (last_byte <= start) {

		clear_extent_bit(tree, start, page_end,

				 EXTENT_LOCKED | EXTENT_DIRTY,

				 1, 0, NULL, GFP_NOFS);

		if (tree->ops && tree->ops->writepage_end_io_hook)

		if (tree->ops && tree->ops->writepage_end_io_hook)

			tree->ops->writepage_end_io_hook(page, start,

			tree->ops->writepage_end_io_hook(page, start,

							 page_end, NULL, 1);

							 page_end, NULL, 1);
@@ -2262,8 +2250,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, 




	while (cur <= end) {

	while (cur <= end) {

		if (cur >= last_byte) {

		if (cur >= last_byte) {

			unlock_extent_cached(tree, unlock_start, page_end,

					     &cached_state, GFP_NOFS);

			if (tree->ops && tree->ops->writepage_end_io_hook)

			if (tree->ops && tree->ops->writepage_end_io_hook)

				tree->ops->writepage_end_io_hook(page, cur,

				tree->ops->writepage_end_io_hook(page, cur,

							 page_end, NULL, 1);

							 page_end, NULL, 1);
@@ -2295,10 +2281,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, 
		 */

		 */

		if (compressed || block_start == EXTENT_MAP_HOLE ||

		if (compressed || block_start == EXTENT_MAP_HOLE ||

		    block_start == EXTENT_MAP_INLINE) {

		    block_start == EXTENT_MAP_INLINE) {

			unlock_extent_cached(tree, unlock_start,

					     cur + iosize - 1, &cached_state,

					     GFP_NOFS);



			/*

			/*

			 * end_io notification does not happen here for

			 * end_io notification does not happen here for

			 * compressed extents

			 * compressed extents
@@ -2366,9 +2348,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, 
		set_page_writeback(page);

		set_page_writeback(page);

		end_page_writeback(page);

		end_page_writeback(page);

	}

	}

	if (unlock_start <= page_end)

		unlock_extent_cached(tree, unlock_start, page_end,

				     &cached_state, GFP_NOFS);

	unlock_page(page);

	unlock_page(page);





done_unlocked:

done_unlocked: