Commit 7cddbb43 authored by Mark Harmstone's avatar Mark Harmstone Committed by David Sterba
Browse files

btrfs: handle discarding fully-remapped block groups 



Discard normally works by iterating over the free-space entries of a
block group. This doesn't work for fully-remapped block groups, as we
removed their free-space entries when we started relocation.

For sync discard, call btrfs_discard_extent() when we commit the
transaction in which the last identity remap was removed.

For async discard, add a new function btrfs_trim_fully_remapped_block_group()
to be called by the discard worker, which iterates over the block
group's range using the normal async discard rules. Once we reach the
end, remove the chunk's stripes and device extents to get back its free
space.

Reviewed-by: default avatarBoris Burkov <boris@bur.io>
Signed-off-by: default avatarMark Harmstone <mark@harmstone.com>
Signed-off-by: default avatarDavid Sterba <dsterba@suse.com>
parent 81e5a455

fs/btrfs/block-group.c

+17 −13
Original line number Diff line number Diff line
@@ -4792,12 +4792,15 @@ void btrfs_mark_bg_fully_remapped(struct btrfs_block_group *bg, 
{

	struct btrfs_fs_info *fs_info = trans->fs_info;





	if (btrfs_test_opt(fs_info, DISCARD_ASYNC)) {

		btrfs_discard_queue_work(&fs_info->discard_ctl, bg);

	} else {

		spin_lock(&fs_info->unused_bgs_lock);

		/*

	 * The block group might already be on the unused_bgs list, remove it

	 * if it is. It'll get readded after the async discard worker finishes,

	 * or in btrfs_handle_fully_remapped_bgs() if we're not using async

	 * discard.

		 * The block group might already be on the unused_bgs list,

		 * remove it if it is. It'll get readded after

		 * btrfs_handle_fully_remapped_bgs() finishes.

		 */

		if (!list_empty(&bg->bg_list))

			list_del(&bg->bg_list);
@@ -4807,3 +4810,4 @@ void btrfs_mark_bg_fully_remapped(struct btrfs_block_group *bg, 
		list_add_tail(&bg->bg_list, &fs_info->fully_remapped_bgs);

		spin_unlock(&fs_info->unused_bgs_lock);

	}

}

fs/btrfs/block-group.h

+1 −0
Original line number Diff line number Diff line
@@ -49,6 +49,7 @@ enum btrfs_discard_state { 
	BTRFS_DISCARD_EXTENTS,

	BTRFS_DISCARD_BITMAPS,

	BTRFS_DISCARD_RESET_CURSOR,

	BTRFS_DISCARD_FULLY_REMAPPED,

};



/*

fs/btrfs/discard.c

+45 −7
Original line number Diff line number Diff line
@@ -215,6 +215,25 @@ static struct btrfs_block_group *find_next_block_group( 
	return ret_block_group;

}



/*

 * Check whether a block group is empty.

 *

 * "Empty" here means that there are no extents physically located within the

 * device extents corresponding to this block group.

 *

 * For a remapped block group, this means that all of its identity remaps have

 * been removed. For a non-remapped block group, this means that no extents

 * have an address within its range, and that nothing has been remapped to be

 * within it.

 */

static bool block_group_is_empty(const struct btrfs_block_group *bg)

{

	if (bg->flags & BTRFS_BLOCK_GROUP_REMAPPED)

		return bg->identity_remap_count == 0;



	return bg->used == 0 && bg->remap_bytes == 0;

}



/*

 * Look up next block group and set it for use.

 *
@@ -241,8 +260,10 @@ static struct btrfs_block_group *peek_discard_list( 
	block_group = find_next_block_group(discard_ctl, now);



	if (block_group && now >= block_group->discard_eligible_time) {

		const bool empty = block_group_is_empty(block_group);



		if (block_group->discard_index == BTRFS_DISCARD_INDEX_UNUSED &&

		    block_group->used != 0) {

		    !empty) {

			if (btrfs_is_block_group_data_only(block_group)) {

				__add_to_discard_list(discard_ctl, block_group);

				/*
@@ -267,9 +288,14 @@ static struct btrfs_block_group *peek_discard_list( 
		}

		if (block_group->discard_state == BTRFS_DISCARD_RESET_CURSOR) {

			block_group->discard_cursor = block_group->start;



			if (block_group->flags & BTRFS_BLOCK_GROUP_REMAPPED && empty) {

				block_group->discard_state = BTRFS_DISCARD_FULLY_REMAPPED;

			} else {

				block_group->discard_state = BTRFS_DISCARD_EXTENTS;

			}

		}

	}

	if (block_group) {

		btrfs_get_block_group(block_group);

		discard_ctl->block_group = block_group;
@@ -373,7 +399,7 @@ void btrfs_discard_queue_work(struct btrfs_discard_ctl *discard_ctl, 
	if (!block_group || !btrfs_test_opt(block_group->fs_info, DISCARD_ASYNC))

		return;



	if (block_group->used == 0 && block_group->remap_bytes == 0)

	if (block_group_is_empty(block_group))

		add_to_discard_unused_list(discard_ctl, block_group);

	else

		add_to_discard_list(discard_ctl, block_group);
@@ -470,7 +496,7 @@ static void btrfs_finish_discard_pass(struct btrfs_discard_ctl *discard_ctl, 
{

	remove_from_discard_list(discard_ctl, block_group);



	if (block_group->used == 0) {

	if (block_group_is_empty(block_group)) {

		if (btrfs_is_free_space_trimmed(block_group))

			btrfs_mark_bg_unused(block_group);

		else
@@ -524,7 +550,8 @@ static void btrfs_discard_workfn(struct work_struct *work) 
	/* Perform discarding */

	minlen = discard_minlen[discard_index];



	if (discard_state == BTRFS_DISCARD_BITMAPS) {

	switch (discard_state) {

	case BTRFS_DISCARD_BITMAPS: {

		u64 maxlen = 0;



		/*
@@ -541,17 +568,28 @@ static void btrfs_discard_workfn(struct work_struct *work) 
				       btrfs_block_group_end(block_group),

				       minlen, maxlen, true);

		discard_ctl->discard_bitmap_bytes += trimmed;

	} else {



		break;

	}



	case BTRFS_DISCARD_FULLY_REMAPPED:

		btrfs_trim_fully_remapped_block_group(block_group);

		break;



	default:

		btrfs_trim_block_group_extents(block_group, &trimmed,

				       block_group->discard_cursor,

				       btrfs_block_group_end(block_group),

				       minlen, true);

		discard_ctl->discard_extent_bytes += trimmed;



		break;

	}



	/* Determine next steps for a block_group */

	if (block_group->discard_cursor >= btrfs_block_group_end(block_group)) {

		if (discard_state == BTRFS_DISCARD_BITMAPS) {

		if (discard_state == BTRFS_DISCARD_BITMAPS ||

		    discard_state == BTRFS_DISCARD_FULLY_REMAPPED) {

			btrfs_finish_discard_pass(discard_ctl, block_group);

		} else {

			block_group->discard_cursor = block_group->start;

fs/btrfs/extent-tree.c

+2 −0
Original line number Diff line number Diff line
@@ -2905,6 +2905,8 @@ void btrfs_handle_fully_remapped_bgs(struct btrfs_fs_info *fs_info) 
		list_del_init(&bg->bg_list);

		spin_unlock(&fs_info->unused_bgs_lock);



		btrfs_discard_extent(fs_info, bg->start, bg->length, NULL, false);



		ret = btrfs_complete_bg_remapping(bg);

		if (ret) {

			btrfs_put_block_group(bg);

fs/btrfs/free-space-cache.c

+33 −0
Original line number Diff line number Diff line
@@ -29,6 +29,7 @@ 
#include "file-item.h"

#include "file.h"

#include "super.h"

#include "relocation.h"



#define BITS_PER_BITMAP		(PAGE_SIZE * 8UL)

#define MAX_CACHE_BYTES_PER_GIG	SZ_64K
@@ -3066,6 +3067,11 @@ bool btrfs_is_free_space_trimmed(struct btrfs_block_group *block_group) 
	struct rb_node *node;

	bool ret = true;



	if (block_group->flags & BTRFS_BLOCK_GROUP_REMAPPED &&

	    block_group->identity_remap_count == 0) {

		return true;

	}



	spin_lock(&ctl->tree_lock);

	node = rb_first(&ctl->free_space_offset);
@@ -3834,6 +3840,33 @@ static int trim_no_bitmap(struct btrfs_block_group *block_group, 
	return ret;

}



void btrfs_trim_fully_remapped_block_group(struct btrfs_block_group *bg)

{

	struct btrfs_fs_info *fs_info = bg->fs_info;

	struct btrfs_discard_ctl *discard_ctl = &fs_info->discard_ctl;

	int ret = 0;

	u64 bytes, trimmed;

	const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);

	u64 end = btrfs_block_group_end(bg);



	bytes = end - bg->discard_cursor;



	if (max_discard_size &&

	    bytes >= (max_discard_size + BTRFS_ASYNC_DISCARD_MIN_FILTER))

		bytes = max_discard_size;



	ret = btrfs_discard_extent(fs_info, bg->discard_cursor, bytes, &trimmed, false);

	if (ret)

		return;



	bg->discard_cursor += trimmed;



	if (bg->discard_cursor < end)

		return;



	btrfs_complete_bg_remapping(bg);

}



/*

 * If we break out of trimming a bitmap prematurely, we should reset the

 * trimming bit.  In a rather contrived case, it's possible to race here so