Skip to contentSkip to conversation

btrfs: dynamic and periodic block_group reclaim

PATCH In review Boris Burkov 5 participants 13 messages Jun 17, 2024 – Dec 29, 2025 Btrfs file system

lore
Share & tools
lore
#0 lore raw

v2 cover btrfs: dynamic and periodic block_group reclaim

Btrfs's block_group allocator suffers from a well known problem, that it is capable of eagerly allocating too much space to either data or metadata (most often data, absent bugs) and then later be unable to allocate more space for the other, when needed. When data starves metadata, this can extra painfully result in read only filesystems that need careful manual balancing to fix.

This can be worked around by:
- enabling automatic reclaim
- periodically running balance

The latter is widely deployed via btrfsmaintenance
(https://github.com/kdave/btrfsmaintenance) and the former is used at scale at Meta with good results. However, neither of those solutions is perfect, as they both currently use a fixed threshold. A fixed threshold is vulnerable to workloads that trigger high amounts of reclaim. This has led to btrfsmaintenance setting very conservative thresholds of 5 and 10 percent of data block groups.
(https://github.com/kdave/btrfsmaintenance/commit/edbbfffe592f47c2849a8825f523e2ccc38b15f5)
At Meta, we deal with an elevated level of reclaim which would be desirable to reduce.

This patch set expands on automatic reclaim, adding the ability to set a dynamic reclaim threshold that appropriately scales with the global file system allocation conditions as well as periodic reclaim which runs that reclaim sweep in the cleaner thread. Together, I believe they constitute a robust and general automatic reclaim system that should avoid unfortunate read only filesystems in all but extreme conditions, where space is running quite low anyway and failure is more reasonable.

At a very high level, the dynamic threshold's strategy is to set a fixed target of unallocated block groups (10 block groups) and linearly scale its aggression the further we are from that target. That way we do no automatic reclaim until we actually press against the unallocated target, allowing the allocator to gradually fill fragmented space with new extents, but do claw back space after workloads that use and free a bunch of space, perhaps with fragmentation.

I ran it on three workloads (described in detail on the dynamic reclaim patch) but they are:
1. bounce allocations around X% full.
2. fill up all the way and introduce full fragmentation.
3. write in a fragmented way until the filesystem is just about full. script can be found here:
https://github.com/boryas/scripts/tree/main/fio/reclaim

The important results can be seen here (full results explorable at
https://bur.io/dyn-rec/)

bounce at 30%, higher relocations with a fixed threshold:
https://bur.io/dyn-rec/bounce/reclaims.png
https://bur.io/dyn-rec/bounce/reclaim_bytes.png
https://bur.io/dyn-rec/bounce/unalloc_bytes.png

hard 30% fragmentation, dynamic actually reclaims, relocs not crazy:
https://bur.io/dyn-rec/strict_frag/reclaims.png
https://bur.io/dyn-rec/strict_frag/reclaim_bytes.png
https://bur.io/dyn-rec/strict_frag/unalloc_bytes.png

fill it all the way up in a fragmented way, then keep making allocations:
https://bur.io/dyn-rec/last_gig/reclaims.png
https://bur.io/dyn-rec/last_gig/reclaim_bytes.png
https://bur.io/dyn-rec/last_gig/unalloc_bytes.png
--
Changelog:
v2:
- add reclaim errors counter
- refactor reclaim counter to remove extra else
- account for zone unusable in threshold calculation

Boris Burkov (6):
btrfs: report reclaim stats in sysfs
btrfs: store fs_info on space_info
btrfs: dynamic block_group reclaim threshold
btrfs: periodic block_group reclaim
btrfs: prevent pathological periodic reclaim loops
btrfs: urgent periodic reclaim pass

 fs/btrfs/block-group.c |  42 ++++++--
 fs/btrfs/block-group.h |   1 +
 fs/btrfs/space-info.c  | 240 +++++++++++++++++++++++++++++++++++++++--
 fs/btrfs/space-info.h  |  48 +++++++++
 fs/btrfs/sysfs.c       |  83 +++++++++++++-
 5 files changed, 391 insertions(+), 23 deletions(-)
Signature · 1 line
2.45.2
#1 lore raw

PATCH v2 1/6 btrfs: report reclaim stats in sysfs

When evaluating various reclaim strategies/thresholds against each other, it is useful to collect data about the amount of reclaim happening. Expose a count, error count, and byte count via sysfs per space_info.

Note that this is only for automatic reclaim, not manually invoked balances or other codepaths that use "relocate_block_group"

---
 fs/btrfs/block-group.c | 10 ++++++++++
 fs/btrfs/space-info.h  | 18 ++++++++++++++++++
 fs/btrfs/sysfs.c       |  6 ++++++
 3 files changed, 34 insertions(+)
fs/btrfs/block-group.c +10 -0
index 9f1d328b603e..824fd229d129 100644
@@ -1829,6 +1829,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 	list_sort(NULL, &fs_info->reclaim_bgs, reclaim_bgs_cmp);
 	while (!list_empty(&fs_info->reclaim_bgs)) {
 		u64 zone_unusable;
+		u64 reclaimed;
 		int ret = 0;
 
 		bg = list_first_entry(&fs_info->reclaim_bgs,
@@ -1921,12 +1922,21 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 				div64_u64(bg->used * 100, bg->length),
 				div64_u64(zone_unusable * 100, bg->length));
 		trace_btrfs_reclaim_block_group(bg);
+		reclaimed = bg->used;
 		ret = btrfs_relocate_chunk(fs_info, bg->start);
 		if (ret) {
 			btrfs_dec_block_group_ro(bg);
 			btrfs_err(fs_info, "error relocating chunk %llu",
 				  bg->start);
+			reclaimed = 0;
+			spin_lock(&space_info->lock);
+			space_info->reclaim_errors++;
+			spin_unlock(&space_info->lock);
 		}
+		spin_lock(&space_info->lock);
+		space_info->reclaim_count++;
+		space_info->reclaim_bytes += reclaimed;
+		spin_unlock(&space_info->lock);
 
 next:
 		if (ret) {
fs/btrfs/space-info.h +18 -0
index a733458fd13b..98ea35ae60fe 100644
@@ -165,6 +165,24 @@ struct btrfs_space_info {
 
 	struct kobject kobj;
 	struct kobject *block_group_kobjs[BTRFS_NR_RAID_TYPES];
+
+	/*
+	 * Monotonically increasing counter of block group reclaim attempts
+	 * Exposed in /sys/fs/<uuid>/allocation/<type>/reclaim_count
+	 */
+	u64 reclaim_count;
+
+	/*
+	 * Monotonically increasing counter of reclaimed bytes
+	 * Exposed in /sys/fs/<uuid>/allocation/<type>/reclaim_bytes
+	 */
+	u64 reclaim_bytes;
+
+	/*
+	 * Monotonically increasing counter of reclaim errors
+	 * Exposed in /sys/fs/<uuid>/allocation/<type>/reclaim_errors
+	 */
+	u64 reclaim_errors;
 };
 
 struct reserve_ticket {
fs/btrfs/sysfs.c +6 -0
index af545b6b1190..919c7ba45121 100644
@@ -894,6 +894,9 @@ SPACE_INFO_ATTR(bytes_readonly);
 SPACE_INFO_ATTR(bytes_zone_unusable);
 SPACE_INFO_ATTR(disk_used);
 SPACE_INFO_ATTR(disk_total);
+SPACE_INFO_ATTR(reclaim_count);
+SPACE_INFO_ATTR(reclaim_bytes);
+SPACE_INFO_ATTR(reclaim_errors);
 BTRFS_ATTR_RW(space_info, chunk_size, btrfs_chunk_size_show, btrfs_chunk_size_store);
 BTRFS_ATTR(space_info, size_classes, btrfs_size_classes_show);
 
@@ -949,6 +952,9 @@ static struct attribute *space_info_attrs[] = {
 	BTRFS_ATTR_PTR(space_info, bg_reclaim_threshold),
 	BTRFS_ATTR_PTR(space_info, chunk_size),
 	BTRFS_ATTR_PTR(space_info, size_classes),
+	BTRFS_ATTR_PTR(space_info, reclaim_count),
+	BTRFS_ATTR_PTR(space_info, reclaim_bytes),
+	BTRFS_ATTR_PTR(space_info, reclaim_errors),
 #ifdef CONFIG_BTRFS_DEBUG
 	BTRFS_ATTR_PTR(space_info, force_chunk_alloc),
 #endif
Signature · 1 line
2.45.2
#2 lore raw

PATCH v2 2/6 btrfs: store fs_info on space_info

In reply to Boris Burkov

This is handy when computing space_info dynamic reclaim thresholds where we do not have access to a block group. We could add it to the various functions as a parameter, but it seems reasonable for space_info to have an fs_info pointer.

---
 fs/btrfs/space-info.c | 1 +
 fs/btrfs/space-info.h | 1 +
 2 files changed, 2 insertions(+)
fs/btrfs/space-info.c +1 -0
index 0283ee9bf813..7384286c5058 100644
@@ -232,6 +232,7 @@ static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 	if (!space_info)
 		return -ENOMEM;
 
+	space_info->fs_info = info;
 	for (i = 0; i < BTRFS_NR_RAID_TYPES; i++)
 		INIT_LIST_HEAD(&space_info->block_groups[i]);
 	init_rwsem(&space_info->groups_sem);
fs/btrfs/space-info.h +1 -0
index 98ea35ae60fe..25edfd453b27 100644
@@ -94,6 +94,7 @@ enum btrfs_flush_state {
 };
 
 struct btrfs_space_info {
+	struct btrfs_fs_info *fs_info;
 	spinlock_t lock;
 
 	u64 total_bytes;	/* total bytes in the space,
Signature · 1 line
2.45.2
#3 lore raw

PATCH v2 3/6 btrfs: dynamic block_group reclaim threshold

In reply to Boris Burkov

We can currently recover allocated block_groups by:
- explicitly starting balance operations
- "auto reclaim" via bg_reclaim_threshold

The latter works by checking against a fixed threshold on frees. If we pass from above the threshold to below, relocation triggers and the block group will get reclaimed by the cleaner thread (assuming it is still eligible)

Picking a threshold is challenging. Too high, and you end up trying to reclaim very full block_groups which is quite costly, and you don't do reclaim on block_groups that don't get quite THAT full, but could still be quite fragmented and stranding a lot of space. Too low, and you similarly miss out on reclaim even if you badly need it to avoid running out of unallocated space, if you have heavily fragmented block groups living above the threshold.

No matter the threshold, it suffers from a workload that happens to bounce around that threshold, which can introduce arbitrary amounts of reclaim waste.

To improve this situation, introduce a dynamic threshold. The basic idea behind this threshold is that it should be very lax when there is plenty of unallocated space, and increasingly aggressive as we approach zero unallocated space. To that end, it sets a target for unallocated space (10 chunks) and then linearly increases the threshold as the amount of space short of the target we are increases. The formula is:
(target - unalloc) / target

I tested this by running it on three interesting workloads:
1. bounce allocations around X% full.
2. fill up all the way and introduce full fragmentation.
3. write in a fragmented way until the filesystem is just about full.

1. and 2. attack the weaknesses of a fixed threshold; fixed either works perfectly or fully falls apart, depending on the threshold. Dynamic always handles these cases well.

3. attacks dynamic by checking whether it is too zealous to reclaim in conditions with low unallocated and low unused. It tends to claw back 1GiB of unallocated fairly aggressively, but not much more. Early versions of dynamic threshold struggled on this test.

Additional work could be done to intelligently ratchet up the urgency of reclaim in very low unallocated conditions. Existing mechanisms are already useless in that case anyway.

---
 fs/btrfs/block-group.c |  18 ++++---
 fs/btrfs/space-info.c  | 115 +++++++++++++++++++++++++++++++++++++----
 fs/btrfs/space-info.h  |   8 +++
 fs/btrfs/sysfs.c       |  43 ++++++++++++++-
 4 files changed, 164 insertions(+), 20 deletions(-)
fs/btrfs/block-group.c +10 -8
index 824fd229d129..c3313697475f 100644
@@ -1764,24 +1764,21 @@ static inline bool btrfs_should_reclaim(struct btrfs_fs_info *fs_info)
 
 static bool should_reclaim_block_group(struct btrfs_block_group *bg, u64 bytes_freed)
 {
-	const struct btrfs_space_info *space_info = bg->space_info;
-	const int reclaim_thresh = READ_ONCE(space_info->bg_reclaim_threshold);
+	const int thresh_pct = btrfs_calc_reclaim_threshold(bg->space_info);
+	u64 thresh_bytes = mult_perc(bg->length, thresh_pct);
 	const u64 new_val = bg->used;
 	const u64 old_val = new_val + bytes_freed;
-	u64 thresh;
 
-	if (reclaim_thresh == 0)
+	if (thresh_bytes == 0)
 		return false;
 
-	thresh = mult_perc(bg->length, reclaim_thresh);
-
 	/*
 	 * If we were below the threshold before don't reclaim, we are likely a
 	 * brand new block group and we don't want to relocate new block groups.
 	 */
-	if (old_val < thresh)
+	if (old_val < thresh_bytes)
 		return false;
-	if (new_val >= thresh)
+	if (new_val >= thresh_bytes)
 		return false;
 	return true;
 }
@@ -1843,6 +1840,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		/* Don't race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 
+		spin_lock(&space_info->lock);
 		spin_lock(&bg->lock);
 		if (bg->reserved || bg->pinned || bg->ro) {
 			/*
@@ -1852,6 +1850,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			 * this block group.
 			 */
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 		}
@@ -1870,6 +1869,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			if (!btrfs_test_opt(fs_info, DISCARD_ASYNC))
 				btrfs_mark_bg_unused(bg);
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 
@@ -1886,10 +1886,12 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		 */
 		if (!should_reclaim_block_group(bg, bg->length)) {
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 		}
 		spin_unlock(&bg->lock);
+		spin_unlock(&space_info->lock);
 
 		/*
 		 * Get out fast, in case we're read-only or unmounting the
fs/btrfs/space-info.c +104 -11

145 lines not on this page · raw message

fs/btrfs/space-info.h +8 -0
index 25edfd453b27..2cac771321c7 100644
@@ -184,6 +184,12 @@ struct btrfs_space_info {
 	 * Exposed in /sys/fs/<uuid>/allocation/<type>/reclaim_errors
 	 */
 	u64 reclaim_errors;
+
+	/*
+	 * If true, use the dynamic relocation threshold, instead of the
+	 * fixed bg_reclaim_threshold.
+	 */
+	bool dynamic_reclaim;
 };
 
 struct reserve_ticket {
@@ -266,4 +272,6 @@ void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info);
 void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 
+int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info);
+
 #endif /* BTRFS_SPACE_INFO_H */
fs/btrfs/sysfs.c +42 -1

68 lines not on this page · raw message

Signature · 1 line
2.45.2
#4 lore raw

Re: [PATCH v2 3/6] btrfs: dynamic block_group reclaim threshold

Quoted 156 lines · Boris Burkov

We can currently recover allocated block_groups by:
- explicitly starting balance operations
- "auto reclaim" via bg_reclaim_threshold

The latter works by checking against a fixed threshold on frees. If we pass from above the threshold to below, relocation triggers and the block group will get reclaimed by the cleaner thread (assuming it is still eligible)

Picking a threshold is challenging. Too high, and you end up trying to reclaim very full block_groups which is quite costly, and you don't do reclaim on block_groups that don't get quite THAT full, but could still be quite fragmented and stranding a lot of space. Too low, and you similarly miss out on reclaim even if you badly need it to avoid running out of unallocated space, if you have heavily fragmented block groups living above the threshold.

No matter the threshold, it suffers from a workload that happens to bounce around that threshold, which can introduce arbitrary amounts of reclaim waste.

To improve this situation, introduce a dynamic threshold. The basic idea behind this threshold is that it should be very lax when there is plenty of unallocated space, and increasingly aggressive as we approach zero unallocated space. To that end, it sets a target for unallocated space (10 chunks) and then linearly increases the threshold as the amount of space short of the target we are increases. The formula is:
(target - unalloc) / target

I tested this by running it on three interesting workloads:
1. bounce allocations around X% full.
2. fill up all the way and introduce full fragmentation.
3. write in a fragmented way until the filesystem is just about full.

1. and 2. attack the weaknesses of a fixed threshold; fixed either works perfectly or fully falls apart, depending on the threshold. Dynamic always handles these cases well.

3. attacks dynamic by checking whether it is too zealous to reclaim in conditions with low unallocated and low unused. It tends to claw back 1GiB of unallocated fairly aggressively, but not much more. Early versions of dynamic threshold struggled on this test.

Additional work could be done to intelligently ratchet up the urgency of reclaim in very low unallocated conditions. Existing mechanisms are already useless in that case anyway.

---
 fs/btrfs/block-group.c |  18 ++++---
 fs/btrfs/space-info.c  | 115 +++++++++++++++++++++++++++++++++++++----
 fs/btrfs/space-info.h  |   8 +++
 fs/btrfs/sysfs.c       |  43 ++++++++++++++-
 4 files changed, 164 insertions(+), 20 deletions(-)
fs/btrfs/block-group.c +10 -8
index 824fd229d129..c3313697475f 100644
@@ -1764,24 +1764,21 @@ static inline bool btrfs_should_reclaim(struct btrfs_fs_info *fs_info)
 
 static bool should_reclaim_block_group(struct btrfs_block_group *bg, u64 bytes_freed)
 {
-	const struct btrfs_space_info *space_info = bg->space_info;
-	const int reclaim_thresh = READ_ONCE(space_info->bg_reclaim_threshold);
+	const int thresh_pct = btrfs_calc_reclaim_threshold(bg->space_info);
+	u64 thresh_bytes = mult_perc(bg->length, thresh_pct);
 	const u64 new_val = bg->used;
 	const u64 old_val = new_val + bytes_freed;
-	u64 thresh;
 
-	if (reclaim_thresh == 0)
+	if (thresh_bytes == 0)
 		return false;
 
-	thresh = mult_perc(bg->length, reclaim_thresh);
-
 	/*
 	 * If we were below the threshold before don't reclaim, we are likely a
 	 * brand new block group and we don't want to relocate new block groups.
 	 */
-	if (old_val < thresh)
+	if (old_val < thresh_bytes)
 		return false;
-	if (new_val >= thresh)
+	if (new_val >= thresh_bytes)
 		return false;
 	return true;
 }
@@ -1843,6 +1840,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		/* Don't race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 
+		spin_lock(&space_info->lock);
 		spin_lock(&bg->lock);
 		if (bg->reserved || bg->pinned || bg->ro) {
 			/*
@@ -1852,6 +1850,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			 * this block group.
 			 */
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 		}
@@ -1870,6 +1869,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			if (!btrfs_test_opt(fs_info, DISCARD_ASYNC))
 				btrfs_mark_bg_unused(bg);
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 
@@ -1886,10 +1886,12 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		 */
 		if (!should_reclaim_block_group(bg, bg->length)) {
 			spin_unlock(&bg->lock);
+			spin_unlock(&space_info->lock);
 			up_write(&space_info->groups_sem);
 			goto next;
 		}
 		spin_unlock(&bg->lock);
+		spin_unlock(&space_info->lock);
 
 		/*
 		 * Get out fast, in case we're read-only or unmounting the
fs/btrfs/space-info.c +9 -0
index 7384286c5058..0d13282dac05 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include <linux/minmax.h>
 #include "misc.h"
 #include "ctree.h"
 #include "space-info.h"
@@ -190,6 +191,8 @@ void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
  */
 #define BTRFS_DEFAULT_ZONED_RECLAIM_THRESH			(75)
 
+#define BTRFS_UNALLOC_BLOCK_GROUP_TARGET			(10ULL)
+
 /*
  * Calculate chunk size depending on volume type (regular or zoned).
  */
@@ -341,11 +344,27 @@ struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
 	return NULL;
 }
 
+static u64 calc_effective_data_chunk_size(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_space_info *data_sinfo;
+	u64 data_chunk_size;
+	/*
+	 * Calculate the data_chunk_size, space_info->chunk_size is the

fs/btrfs/space-info.c struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info, · line 362

Quoted 10 lines
fs/btrfs/space-info.c +10 -0
+	 * "optimal" chunk size based on the fs size.  However when we actually
+	 * allocate the chunk we will strip this down further, making it no more
+	 * than 10% of the disk or 1G, whichever is smaller.
+	 */
+	data_sinfo = btrfs_find_space_info(fs_info, BTRFS_BLOCK_GROUP_DATA);
+	data_chunk_size = min(data_sinfo->chunk_size,
+			      mult_perc(fs_info->fs_devices->total_rw_bytes, 10));
+	return min_t(u64, data_chunk_size, SZ_1G);
+
+}

I know this is copied from the previous code. But, the logic is wrong on zoned mode. We always use data_sinfo->chunk_size which is zone_size.

I was working on a fix in the old code, and a patch is almost ready. Since the fix should be backported, it would be easier if I could put my fix patch before this series in the for-next branch.

#5 lore raw

PATCH v2 4/6 btrfs: periodic block_group reclaim

In reply to Boris Burkov

We currently employ a edge-triggered block group reclaim strategy which marks block groups for reclaim as they free down past a threshold.

With a dynamic threshold, this is worse than doing it in a level-triggered fashion periodically. That is because the reclaim itself happens periodically, so the threshold at that point in time is what really matters, not the threshold at freeing time. If we mark the reclaim in a big pass, then sort by usage and do reclaim, we also benefit from a negative feedback loop preventing unnecessary reclaims as we crunch through the "best" candidates.

Since this is quite a different model, it requires some additional support. The edge triggered reclaim has a good heuristic for not reclaiming fresh block groups, so we need to replace that with a typical GC sweep mark which skips block groups that have seen an allocation since the last sweep.

---
 fs/btrfs/block-group.c |  2 ++
 fs/btrfs/block-group.h |  1 +
 fs/btrfs/space-info.c  | 51 ++++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/space-info.h  |  7 ++++++
 fs/btrfs/sysfs.c       | 34 ++++++++++++++++++++++++++++
 5 files changed, 95 insertions(+)
fs/btrfs/block-group.c +2 -0
index c3313697475f..6bcf24f2ac79 100644
@@ -1974,6 +1974,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 
 void btrfs_reclaim_bgs(struct btrfs_fs_info *fs_info)
 {
+	btrfs_reclaim_sweep(fs_info);
 	spin_lock(&fs_info->unused_bgs_lock);
 	if (!list_empty(&fs_info->reclaim_bgs))
 		queue_work(system_unbound_wq, &fs_info->reclaim_bgs_work);
@@ -3672,6 +3673,7 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		old_val += num_bytes;
 		cache->used = old_val;
 		cache->reserved -= num_bytes;
+		cache->reclaim_mark = 0;
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
fs/btrfs/block-group.h +1 -0
index 85e2d4cd12dc..8656b38f1fa5 100644
@@ -263,6 +263,7 @@ struct btrfs_block_group {
 	struct work_struct zone_finish_work;
 	struct extent_buffer *last_eb;
 	enum btrfs_block_group_size_class size_class;
+	u64 reclaim_mark;
 };
 
 static inline u64 btrfs_block_group_end(struct btrfs_block_group *block_group)
fs/btrfs/space-info.c +51 -0

54 lines not on this page · raw message

fs/btrfs/space-info.h +7 -0
index 2cac771321c7..ae4a1f7d5856 100644
@@ -190,6 +190,12 @@ struct btrfs_space_info {
 	 * fixed bg_reclaim_threshold.
 	 */
 	bool dynamic_reclaim;
+
+	/*
+	 * Periodically check all block groups against the reclaim
+	 * threshold in the cleaner thread.
+	 */
+	bool periodic_reclaim;
 };
 
 struct reserve_ticket {
@@ -273,5 +279,6 @@ void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info);
+int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info);
 
 #endif /* BTRFS_SPACE_INFO_H */
fs/btrfs/sysfs.c +34 -0
index 360d6093476f..c58cea0da597 100644
@@ -973,6 +973,39 @@ BTRFS_ATTR_RW(space_info, dynamic_reclaim,
 	      btrfs_sinfo_dynamic_reclaim_show,
 	      btrfs_sinfo_dynamic_reclaim_store);
 
+static ssize_t btrfs_sinfo_periodic_reclaim_show(struct kobject *kobj,
+						struct kobj_attribute *a,
+						char *buf)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+
+	return sysfs_emit(buf, "%d\n", READ_ONCE(space_info->periodic_reclaim));
+}
+
+static ssize_t btrfs_sinfo_periodic_reclaim_store(struct kobject *kobj,
+						 struct kobj_attribute *a,
+						 const char *buf, size_t len)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	int periodic_reclaim;
+	int ret;
+
+	ret = kstrtoint(buf, 10, &periodic_reclaim);
+	if (ret)
+		return ret;
+
+	if (periodic_reclaim < 0)
+		return -EINVAL;
+
+	WRITE_ONCE(space_info->periodic_reclaim, periodic_reclaim != 0);
+
+	return len;
+}
+
+BTRFS_ATTR_RW(space_info, periodic_reclaim,
+	      btrfs_sinfo_periodic_reclaim_show,
+	      btrfs_sinfo_periodic_reclaim_store);
+
 /*
  * Allocation information about block group types.
  *
@@ -996,6 +1029,7 @@ static struct attribute *space_info_attrs[] = {
 	BTRFS_ATTR_PTR(space_info, reclaim_count),
 	BTRFS_ATTR_PTR(space_info, reclaim_bytes),
 	BTRFS_ATTR_PTR(space_info, reclaim_errors),
+	BTRFS_ATTR_PTR(space_info, periodic_reclaim),
 #ifdef CONFIG_BTRFS_DEBUG
 	BTRFS_ATTR_PTR(space_info, force_chunk_alloc),
 #endif
Signature · 1 line
2.45.2
#6 lore raw

PATCH v2 5/6 btrfs: prevent pathological periodic reclaim loops

In reply to Boris Burkov

Periodic reclaim runs the risk of getting stuck in a state where it keeps reclaiming the same block group over and over. This can happen if
1. reclaiming that block_group fails
2. reclaiming that block_group fails to move any extents into existing block_groups and just allocates a fresh chunk and moves everything.

Currently, 1. is a very tight loop inside the reclaim worker. That is critical for edge triggered reclaim or else we risk forgetting about a reclaimable group. On the other hand, with level triggered reclaim we can break out of that loop and get it later.

With that fixed, 2. applies to both failures and "successes" with no progress. If we have done a periodic reclaim on a space_info and nothing has changed in that space_info, there is not much point to trying again, so don't, until enough space gets free, which we capture with a heuristic of needing to net free 1 chunk.

---
 fs/btrfs/block-group.c | 12 ++++++---
 fs/btrfs/space-info.c  | 56 ++++++++++++++++++++++++++++++++++++------
 fs/btrfs/space-info.h  | 14 +++++++++++
 3 files changed, 71 insertions(+), 11 deletions(-)
fs/btrfs/block-group.c +9 -3
index 6bcf24f2ac79..ba9afb94e7ce 100644
@@ -1933,6 +1933,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			reclaimed = 0;
 			spin_lock(&space_info->lock);
 			space_info->reclaim_errors++;
+			if (READ_ONCE(space_info->periodic_reclaim))
+				space_info->periodic_reclaim_ready = false;
 			spin_unlock(&space_info->lock);
 		}
 		spin_lock(&space_info->lock);
@@ -1941,7 +1943,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret) {
+		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
 			/* Refcount held by the reclaim_bgs list after splice. */
 			btrfs_get_block_group(bg);
 			list_add_tail(&bg->bg_list, &retry_list);
@@ -3677,6 +3679,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, -num_bytes);
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
 	} else {
@@ -3686,8 +3690,10 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
 		space_info->bytes_used -= num_bytes;
 		space_info->disk_used -= num_bytes * factor;
-
-		reclaim = should_reclaim_block_group(cache, num_bytes);
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, num_bytes);
+		else
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
fs/btrfs/space-info.c +48 -8

89 lines not on this page · raw message

fs/btrfs/space-info.h +14 -0
index ae4a1f7d5856..4db8a0267c16 100644
@@ -196,6 +196,17 @@ struct btrfs_space_info {
 	 * threshold in the cleaner thread.
 	 */
 	bool periodic_reclaim;
+
+	/*
+	 * Periodic reclaim should be a no-op if a space_info hasn't
+	 * freed any space since the last time we tried.
+	 */
+	bool periodic_reclaim_ready;
+
+	/*
+	 * Net bytes freed or allocated since the last reclaim pass.
+	 */
+	s64 reclaimable_bytes;
 };
 
 struct reserve_ticket {
@@ -278,6 +289,9 @@ void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info);
 void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes);
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready);
+bool btrfs_should_periodic_reclaim(struct btrfs_space_info *space_info);
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info);
 int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info);
 
Signature · 1 line
2.45.2
#7 lore raw

Re: [PATCH v2 5/6] btrfs: prevent pathological periodic reclaim loops

Quoted 116 lines · Boris Burkov

Periodic reclaim runs the risk of getting stuck in a state where it keeps reclaiming the same block group over and over. This can happen if
1. reclaiming that block_group fails
2. reclaiming that block_group fails to move any extents into existing block_groups and just allocates a fresh chunk and moves everything.

Currently, 1. is a very tight loop inside the reclaim worker. That is critical for edge triggered reclaim or else we risk forgetting about a reclaimable group. On the other hand, with level triggered reclaim we can break out of that loop and get it later.

With that fixed, 2. applies to both failures and "successes" with no progress. If we have done a periodic reclaim on a space_info and nothing has changed in that space_info, there is not much point to trying again, so don't, until enough space gets free, which we capture with a heuristic of needing to net free 1 chunk.

---
 fs/btrfs/block-group.c | 12 ++++++---
 fs/btrfs/space-info.c  | 56 ++++++++++++++++++++++++++++++++++++------
 fs/btrfs/space-info.h  | 14 +++++++++++
 3 files changed, 71 insertions(+), 11 deletions(-)
fs/btrfs/block-group.c +9 -3
index 6bcf24f2ac79..ba9afb94e7ce 100644
@@ -1933,6 +1933,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			reclaimed = 0;
 			spin_lock(&space_info->lock);
 			space_info->reclaim_errors++;
+			if (READ_ONCE(space_info->periodic_reclaim))
+				space_info->periodic_reclaim_ready = false;
 			spin_unlock(&space_info->lock);
 		}
 		spin_lock(&space_info->lock);
@@ -1941,7 +1943,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret) {
+		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
 			/* Refcount held by the reclaim_bgs list after splice. */
 			btrfs_get_block_group(bg);
 			list_add_tail(&bg->bg_list, &retry_list);
@@ -3677,6 +3679,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, -num_bytes);
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
 	} else {
@@ -3686,8 +3690,10 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
 		space_info->bytes_used -= num_bytes;
 		space_info->disk_used -= num_bytes * factor;
-
-		reclaim = should_reclaim_block_group(cache, num_bytes);
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, num_bytes);
+		else
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
fs/btrfs/space-info.c +11 -5
index ff92ad26ffa2..e7a2aa751f8f 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include "linux/spinlock.h"
 #include <linux/minmax.h>
 #include "misc.h"
 #include "ctree.h"
@@ -1899,7 +1900,9 @@ static u64 calc_pct_ratio(u64 x, u64 y)
  */
 static u64 calc_unalloc_target(struct btrfs_fs_info *fs_info)
 {
-	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * calc_effective_data_chunk_size(fs_info);
+	u64 chunk_sz = calc_effective_data_chunk_size(fs_info);
+
+	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * chunk_sz;
 }
 
 /*
@@ -1935,14 +1938,13 @@ static int calc_dynamic_reclaim_threshold(struct btrfs_space_info *space_info)
 	u64 unused = alloc - used;
 	u64 want = target > unalloc ? target - unalloc : 0;
 	u64 data_chunk_size = calc_effective_data_chunk_size(fs_info);
-	/* Cast to int is OK because want <= target */
-	int ratio = calc_pct_ratio(want, target);
 
-	/* If we have no unused space, don't bother, it won't work anyway */
+	/* If we have no unused space, don't bother, it won't work anyway. */
 	if (unused < data_chunk_size)
 		return 0;
 
-	return ratio;
+	/* Cast to int is OK because want <= target. */
+	return calc_pct_ratio(want, target);
 }
 
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info)
@@ -1984,6 +1986,46 @@ static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes)
+{
+	u64 chunk_sz = calc_effective_data_chunk_size(space_info->fs_info);
+

fs/btrfs/space-info.c static int do_reclaim_sweep(struct btrfs_fs_info *fs_info, · line 2002

Quoted 10 lines
fs/btrfs/space-info.c +10 -0
+	assert_spin_locked(&space_info->lock);
+	space_info->reclaimable_bytes += bytes;
+
+	if (space_info->reclaimable_bytes >= chunk_sz)
+		btrfs_set_periodic_reclaim_ready(space_info, true);
+}
+
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready)
+{
+	assert_spin_locked(&space_info->lock);

This is essentially

BUG_ON(!locked(spin_lock));

instead use

lockdep_assert_held()

which will just yell at us so we can fix it. Thanks,

Josef

#8 lore raw

Re: [PATCH v2 5/6] btrfs: prevent pathological periodic reclaim loops

Quoted 130 lines · Josef Bacik
Quoted 116 lines · Boris Burkov

Periodic reclaim runs the risk of getting stuck in a state where it keeps reclaiming the same block group over and over. This can happen if
1. reclaiming that block_group fails
2. reclaiming that block_group fails to move any extents into existing block_groups and just allocates a fresh chunk and moves everything.

Currently, 1. is a very tight loop inside the reclaim worker. That is critical for edge triggered reclaim or else we risk forgetting about a reclaimable group. On the other hand, with level triggered reclaim we can break out of that loop and get it later.

With that fixed, 2. applies to both failures and "successes" with no progress. If we have done a periodic reclaim on a space_info and nothing has changed in that space_info, there is not much point to trying again, so don't, until enough space gets free, which we capture with a heuristic of needing to net free 1 chunk.

---
 fs/btrfs/block-group.c | 12 ++++++---
 fs/btrfs/space-info.c  | 56 ++++++++++++++++++++++++++++++++++++------
 fs/btrfs/space-info.h  | 14 +++++++++++
 3 files changed, 71 insertions(+), 11 deletions(-)
fs/btrfs/block-group.c +9 -3
index 6bcf24f2ac79..ba9afb94e7ce 100644
@@ -1933,6 +1933,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			reclaimed = 0;
 			spin_lock(&space_info->lock);
 			space_info->reclaim_errors++;
+			if (READ_ONCE(space_info->periodic_reclaim))
+				space_info->periodic_reclaim_ready = false;
 			spin_unlock(&space_info->lock);
 		}
 		spin_lock(&space_info->lock);
@@ -1941,7 +1943,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret) {
+		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
 			/* Refcount held by the reclaim_bgs list after splice. */
 			btrfs_get_block_group(bg);
 			list_add_tail(&bg->bg_list, &retry_list);
@@ -3677,6 +3679,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, -num_bytes);
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
 	} else {
@@ -3686,8 +3690,10 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
 		space_info->bytes_used -= num_bytes;
 		space_info->disk_used -= num_bytes * factor;
-
-		reclaim = should_reclaim_block_group(cache, num_bytes);
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, num_bytes);
+		else
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
fs/btrfs/space-info.c +11 -5
index ff92ad26ffa2..e7a2aa751f8f 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include "linux/spinlock.h"
 #include <linux/minmax.h>
 #include "misc.h"
 #include "ctree.h"
@@ -1899,7 +1900,9 @@ static u64 calc_pct_ratio(u64 x, u64 y)
  */
 static u64 calc_unalloc_target(struct btrfs_fs_info *fs_info)
 {
-	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * calc_effective_data_chunk_size(fs_info);
+	u64 chunk_sz = calc_effective_data_chunk_size(fs_info);
+
+	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * chunk_sz;
 }
 
 /*
@@ -1935,14 +1938,13 @@ static int calc_dynamic_reclaim_threshold(struct btrfs_space_info *space_info)
 	u64 unused = alloc - used;
 	u64 want = target > unalloc ? target - unalloc : 0;
 	u64 data_chunk_size = calc_effective_data_chunk_size(fs_info);
-	/* Cast to int is OK because want <= target */
-	int ratio = calc_pct_ratio(want, target);
 
-	/* If we have no unused space, don't bother, it won't work anyway */
+	/* If we have no unused space, don't bother, it won't work anyway. */
 	if (unused < data_chunk_size)
 		return 0;
 
-	return ratio;
+	/* Cast to int is OK because want <= target. */
+	return calc_pct_ratio(want, target);
 }
 
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info)
@@ -1984,6 +1986,46 @@ static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes)
+{
+	u64 chunk_sz = calc_effective_data_chunk_size(space_info->fs_info);
+

fs/btrfs/space-info.c static int do_reclaim_sweep(struct btrfs_fs_info *fs_info, · line 2002

Quoted 10 lines
fs/btrfs/space-info.c +10 -0
+	assert_spin_locked(&space_info->lock);
+	space_info->reclaimable_bytes += bytes;
+
+	if (space_info->reclaimable_bytes >= chunk_sz)
+		btrfs_set_periodic_reclaim_ready(space_info, true);
+}
+
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready)
+{
+	assert_spin_locked(&space_info->lock);

This is essentially

Quoted 7 lines

BUG_ON(!locked(spin_lock));

instead use

lockdep_assert_held()

which will just yell at us so we can fix it. Thanks,

Also documented

https://btrfs.readthedocs.io/en/latest/dev/Development-notes.html#locking

#9 lore raw

Re: [PATCH v2 5/6] btrfs: prevent pathological periodic reclaim loops

In reply to Boris Burkov
Quoted 29 lines

Periodic reclaim runs the risk of getting stuck in a state where it keeps reclaiming the same block group over and over. This can happen if
1. reclaiming that block_group fails
2. reclaiming that block_group fails to move any extents into existing block_groups and just allocates a fresh chunk and moves everything.

Currently, 1. is a very tight loop inside the reclaim worker. That is critical for edge triggered reclaim or else we risk forgetting about a reclaimable group. On the other hand, with level triggered reclaim we can break out of that loop and get it later.

With that fixed, 2. applies to both failures and "successes" with no progress. If we have done a periodic reclaim on a space_info and nothing has changed in that space_info, there is not much point to trying again, so don't, until enough space gets free, which we capture with a heuristic of needing to net free 1 chunk.

---
 fs/btrfs/block-group.c | 12 ++++++---
 fs/btrfs/space-info.c  | 56 ++++++++++++++++++++++++++++++++++++------
 fs/btrfs/space-info.h  | 14 +++++++++++
 3 files changed, 71 insertions(+), 11 deletions(-)
fs/btrfs/block-group.c +0 -0
index 6bcf24f2ac79..ba9afb94e7ce 100644

fs/btrfs/block-group.c void btrfs_reclaim_bgs_work(struct work_struct *work) · line 1937

Quoted 5 lines
fs/btrfs/block-group.c +2 -0
@@ -1933,6 +1933,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			reclaimed = 0;
 			spin_lock(&space_info->lock);
 			space_info->reclaim_errors++;
+			if (READ_ONCE(space_info->periodic_reclaim))
+				space_info->periodic_reclaim_ready = false;

I wonder why we're not clearing the reclaimble_bytes count here.

Quoted 76 lines
fs/btrfs/block-group.c +7 -3
 			spin_unlock(&space_info->lock);
 		}
 		spin_lock(&space_info->lock);
@@ -1941,7 +1943,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret) {
+		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
 			/* Refcount held by the reclaim_bgs list after splice. */
 			btrfs_get_block_group(bg);
 			list_add_tail(&bg->bg_list, &retry_list);
@@ -3677,6 +3679,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, -num_bytes);
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
 	} else {
@@ -3686,8 +3690,10 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
 		space_info->bytes_used -= num_bytes;
 		space_info->disk_used -= num_bytes * factor;
-
-		reclaim = should_reclaim_block_group(cache, num_bytes);
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, num_bytes);
+		else
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
fs/btrfs/space-info.c +7 -5
index ff92ad26ffa2..e7a2aa751f8f 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include "linux/spinlock.h"
 #include <linux/minmax.h>
 #include "misc.h"
 #include "ctree.h"
@@ -1899,7 +1900,9 @@ static u64 calc_pct_ratio(u64 x, u64 y)
  */
 static u64 calc_unalloc_target(struct btrfs_fs_info *fs_info)
 {
-	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * calc_effective_data_chunk_size(fs_info);
+	u64 chunk_sz = calc_effective_data_chunk_size(fs_info);
+
+	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * chunk_sz;
 }
 
 /*
@@ -1935,14 +1938,13 @@ static int calc_dynamic_reclaim_threshold(struct btrfs_space_info *space_info)
 	u64 unused = alloc - used;
 	u64 want = target > unalloc ? target - unalloc : 0;
 	u64 data_chunk_size = calc_effective_data_chunk_size(fs_info);
-	/* Cast to int is OK because want <= target */
-	int ratio = calc_pct_ratio(want, target);
 
-	/* If we have no unused space, don't bother, it won't work anyway */
+	/* If we have no unused space, don't bother, it won't work anyway. */
 	if (unused < data_chunk_size)
 		return 0;
 
-	return ratio;
+	/* Cast to int is OK because want <= target. */
+	return calc_pct_ratio(want, target);
 }
 
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info)
@@ -1984,6 +1986,46 @@ static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 	return 0;

fs/btrfs/space-info.c static int do_reclaim_sweep(struct btrfs_fs_info *fs_info, · line 1996

Quoted 10 lines
fs/btrfs/space-info.c +8 -0
 }
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes)
+{
+	u64 chunk_sz = calc_effective_data_chunk_size(space_info->fs_info);
+
+	assert_spin_locked(&space_info->lock);
+	space_info->reclaimable_bytes += bytes;
+
+	if (space_info->reclaimable_bytes >= chunk_sz)

We're comparing s64 with u64 here, and it won't work as expected.

Even after fixing this by changing chunk_sz to s64, it will not work as expected in the following case:

- We're filling the disk so reclaimable_bytes is always negative.
- There's less than 10G unallocated so dynamic_reclaim kicked in.
- periodic_reclaim will never work since reclaimable_bytes is always negetive.

Quoted 51 lines
fs/btrfs/space-info.c +33 -3
+		btrfs_set_periodic_reclaim_ready(space_info, true);
+}
+
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready)
+{
+	assert_spin_locked(&space_info->lock);
+	if (!READ_ONCE(space_info->periodic_reclaim))
+		return;
+	if (ready != space_info->periodic_reclaim_ready) {
+		space_info->periodic_reclaim_ready = ready;
+		if (!ready)
+			space_info->reclaimable_bytes = 0;
+	}
+}
+
+bool btrfs_should_periodic_reclaim(struct btrfs_space_info *space_info)
+{
+	bool ret;
+
+	if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return false;
+	if (!READ_ONCE(space_info->periodic_reclaim))
+		return false;
+
+	spin_lock(&space_info->lock);
+	ret = space_info->periodic_reclaim_ready;
+	btrfs_set_periodic_reclaim_ready(space_info, false);
+	spin_unlock(&space_info->lock);
+
+	return ret;
+}
+
 int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info)
 {
 	int ret;
@@ -1991,9 +2033,7 @@ int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info)
 	struct btrfs_space_info *space_info;
 
 	list_for_each_entry(space_info, &fs_info->space_info, list) {
-		if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
-			continue;
-		if (!READ_ONCE(space_info->periodic_reclaim))
+		if (!btrfs_should_periodic_reclaim(space_info))
 			continue;
 		for (raid = 0; raid < BTRFS_NR_RAID_TYPES; raid++) {
 			ret = do_reclaim_sweep(fs_info, space_info, raid);
fs/btrfs/space-info.h +0 -0
index ae4a1f7d5856..4db8a0267c16 100644

fs/btrfs/space-info.h struct btrfs_space_info { · line 204

Quoted 9 lines
fs/btrfs/space-info.h +6 -0
@@ -196,6 +196,17 @@ struct btrfs_space_info {
 	 * threshold in the cleaner thread.
 	 */
 	bool periodic_reclaim;
+
+	/*
+	 * Periodic reclaim should be a no-op if a space_info hasn't
+	 * freed any space since the last time we tried.
+	 */
+	bool periodic_reclaim_ready;

Also, I wonder why we need this bool flag. I think we care more about if reclaimable_bytes' value is more than 1G when calling btrfs_should_periodic_reclaim() rather than if it has been more than 1G during two calls of btrfs_should_periodic_reclaim().

Thanks,
Sun YangKai

Quoted 20 lines
fs/btrfs/space-info.h +8 -0
+
+	/*
+	 * Net bytes freed or allocated since the last reclaim pass.
+	 */
+	s64 reclaimable_bytes;
 };
 
 struct reserve_ticket {
@@ -278,6 +289,9 @@ void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info);
 void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes);
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready);
+bool btrfs_should_periodic_reclaim(struct btrfs_space_info *space_info);
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info);
 int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info);
 
Signature · 1 line
2.45.2
#10 lore raw

Re: [PATCH v2 5/6] btrfs: prevent pathological periodic reclaim loops

Quoted 24 lines · Sun Yangkai
Quoted 24 lines

Periodic reclaim runs the risk of getting stuck in a state where it keeps reclaiming the same block group over and over. This can happen if
1. reclaiming that block_group fails
2. reclaiming that block_group fails to move any extents into existing block_groups and just allocates a fresh chunk and moves everything.

Currently, 1. is a very tight loop inside the reclaim worker. That is critical for edge triggered reclaim or else we risk forgetting about a reclaimable group. On the other hand, with level triggered reclaim we can break out of that loop and get it later.

With that fixed, 2. applies to both failures and "successes" with no progress. If we have done a periodic reclaim on a space_info and nothing has changed in that space_info, there is not much point to trying again, so don't, until enough space gets free, which we capture with a heuristic of needing to net free 1 chunk.

---
 fs/btrfs/block-group.c | 12 ++++++---
 fs/btrfs/space-info.c  | 56 ++++++++++++++++++++++++++++++++++++------
 fs/btrfs/space-info.h  | 14 +++++++++++
 3 files changed, 71 insertions(+), 11 deletions(-)
Quoted 13 lines

fs/btrfs/block-group.c void btrfs_reclaim_bgs_work(struct work_struct *work) · line 1937

Quoted 10 lines
fs/btrfs/block-group.c +2 -0
index 6bcf24f2ac79..ba9afb94e7ce 100644
@@ -1933,6 +1933,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			reclaimed = 0;
 			spin_lock(&space_info->lock);
 			space_info->reclaim_errors++;
+			if (READ_ONCE(space_info->periodic_reclaim))
+				space_info->periodic_reclaim_ready = false;

I wonder why we're not clearing the reclaimble_bytes count here.

As far as I can tell, it's an oversight. I think it ought to use btrfs_set_reclaim_ready(fs_info, false) here. However, FWIW, the reclaimable bytes already got reset when we checked whether reclaim was ready, so this would extra re-clear it after. It honestly might make the most sense to just get rid of this logic here entirely, as it's pretty redundant with setting ready=false at the start of each invocation of periodic_reclaim.

Quoted 78 lines
Quoted 78 lines
fs/btrfs/block-group.c +7 -3
 			spin_unlock(&space_info->lock);
 		}
 		spin_lock(&space_info->lock);
@@ -1941,7 +1943,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret) {
+		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
 			/* Refcount held by the reclaim_bgs list after splice. */
 			btrfs_get_block_group(bg);
 			list_add_tail(&bg->bg_list, &retry_list);
@@ -3677,6 +3679,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		space_info->bytes_reserved -= num_bytes;
 		space_info->bytes_used += num_bytes;
 		space_info->disk_used += num_bytes * factor;
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, -num_bytes);
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
 	} else {
@@ -3686,8 +3690,10 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
 		space_info->bytes_used -= num_bytes;
 		space_info->disk_used -= num_bytes * factor;
-
-		reclaim = should_reclaim_block_group(cache, num_bytes);
+		if (READ_ONCE(space_info->periodic_reclaim))
+			btrfs_space_info_update_reclaimable(space_info, num_bytes);
+		else
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 
 		spin_unlock(&cache->lock);
 		spin_unlock(&space_info->lock);
fs/btrfs/space-info.c +7 -5
index ff92ad26ffa2..e7a2aa751f8f 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include "linux/spinlock.h"
 #include <linux/minmax.h>
 #include "misc.h"
 #include "ctree.h"
@@ -1899,7 +1900,9 @@ static u64 calc_pct_ratio(u64 x, u64 y)
  */
 static u64 calc_unalloc_target(struct btrfs_fs_info *fs_info)
 {
-	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * calc_effective_data_chunk_size(fs_info);
+	u64 chunk_sz = calc_effective_data_chunk_size(fs_info);
+
+	return BTRFS_UNALLOC_BLOCK_GROUP_TARGET * chunk_sz;
 }
 
 /*
@@ -1935,14 +1938,13 @@ static int calc_dynamic_reclaim_threshold(struct btrfs_space_info *space_info)
 	u64 unused = alloc - used;
 	u64 want = target > unalloc ? target - unalloc : 0;
 	u64 data_chunk_size = calc_effective_data_chunk_size(fs_info);
-	/* Cast to int is OK because want <= target */
-	int ratio = calc_pct_ratio(want, target);
 
-	/* If we have no unused space, don't bother, it won't work anyway */
+	/* If we have no unused space, don't bother, it won't work anyway. */
 	if (unused < data_chunk_size)
 		return 0;
 
-	return ratio;
+	/* Cast to int is OK because want <= target. */
+	return calc_pct_ratio(want, target);
 }
 
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info)
@@ -1984,6 +1986,46 @@ static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
Quoted 10 lines

fs/btrfs/space-info.c static int do_reclaim_sweep(struct btrfs_fs_info *fs_info, · line 1996

Quoted 8 lines
fs/btrfs/space-info.c +8 -0
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes)
+{
+	u64 chunk_sz = calc_effective_data_chunk_size(space_info->fs_info);
+
+	assert_spin_locked(&space_info->lock);
+	space_info->reclaimable_bytes += bytes;
+
+	if (space_info->reclaimable_bytes >= chunk_sz)

We're comparing s64 with u64 here, and it won't work as expected.

Good catch. Since we could do a bunch of allocation and make reclaimable_bytes negative, we need to check for negative reclaimable_bytes first, as the coercion to a giant u64 is actually quite possible in practice.

Quoted 8 lines

Even after fixing this by changing chunk_sz to s64, it will not work as expected in the following case:

- We're filling the disk so reclaimable_bytes is always negative.
- There's less than 10G unallocated so dynamic_reclaim kicked in.
- periodic_reclaim will never work since reclaimable_bytes is always negetive.

Yes, I agree that this case will thrash the reclaim if we keep allocating constantly (without going enospc).

Good catch! Hopefully that's what is happening in your report on the other email.

Quoted 57 lines
Quoted 57 lines
fs/btrfs/space-info.c +33 -3
+		btrfs_set_periodic_reclaim_ready(space_info, true);
+}
+
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready)
+{
+	assert_spin_locked(&space_info->lock);
+	if (!READ_ONCE(space_info->periodic_reclaim))
+		return;
+	if (ready != space_info->periodic_reclaim_ready) {
+		space_info->periodic_reclaim_ready = ready;
+		if (!ready)
+			space_info->reclaimable_bytes = 0;
+	}
+}
+
+bool btrfs_should_periodic_reclaim(struct btrfs_space_info *space_info)
+{
+	bool ret;
+
+	if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return false;
+	if (!READ_ONCE(space_info->periodic_reclaim))
+		return false;
+
+	spin_lock(&space_info->lock);
+	ret = space_info->periodic_reclaim_ready;
+	btrfs_set_periodic_reclaim_ready(space_info, false);
+	spin_unlock(&space_info->lock);
+
+	return ret;
+}
+
 int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info)
 {
 	int ret;
@@ -1991,9 +2033,7 @@ int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info)
 	struct btrfs_space_info *space_info;
 
 	list_for_each_entry(space_info, &fs_info->space_info, list) {
-		if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
-			continue;
-		if (!READ_ONCE(space_info->periodic_reclaim))
+		if (!btrfs_should_periodic_reclaim(space_info))
 			continue;
 		for (raid = 0; raid < BTRFS_NR_RAID_TYPES; raid++) {
 			ret = do_reclaim_sweep(fs_info, space_info, raid);
fs/btrfs/space-info.h +3 -0
index ae4a1f7d5856..4db8a0267c16 100644
@@ -196,6 +196,17 @@ struct btrfs_space_info {
 	 * threshold in the cleaner thread.
 	 */
 	bool periodic_reclaim;
+
+	/*
+	 * Periodic reclaim should be a no-op if a space_info hasn't
Quoted 8 lines

fs/btrfs/space-info.h struct btrfs_space_info { · line 204

Quoted 3 lines
fs/btrfs/space-info.h +3 -0
+	 * freed any space since the last time we tried.
+	 */
+	bool periodic_reclaim_ready;

Also, I wonder why we need this bool flag. I think we care more about if reclaimable_bytes' value is more than 1G when calling btrfs_should_periodic_reclaim() rather than if it has been more than 1G during two calls of btrfs_should_periodic_reclaim().

This is probably an accident of how the logic evolved as I iterated on the design. The first version was to set ready false on a failure then set it true on a deallocation, then I "enhanced" it with the reclaimable_bytes logic to make it more conservative.

I think getting rid of periodic_reclaim_ready and just doing it off of bytes_reclaimable (which still gets reset after every reclaim attempt) would make sense.

Thanks,
Boris

Quoted 25 lines

Thanks,
Sun YangKai

Quoted 20 lines
fs/btrfs/space-info.h +8 -0
+
+	/*
+	 * Net bytes freed or allocated since the last reclaim pass.
+	 */
+	s64 reclaimable_bytes;
 };
 
 struct reserve_ticket {
@@ -278,6 +289,9 @@ void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info);
 void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 
+void btrfs_space_info_update_reclaimable(struct btrfs_space_info *space_info, s64 bytes);
+void btrfs_set_periodic_reclaim_ready(struct btrfs_space_info *space_info, bool ready);
+bool btrfs_should_periodic_reclaim(struct btrfs_space_info *space_info);
 int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info);
 int btrfs_reclaim_sweep(struct btrfs_fs_info *fs_info);
 
Signature · 1 line
2.45.2
#11 lore raw

PATCH v2 6/6 btrfs: urgent periodic reclaim pass

In reply to Boris Burkov

Periodic reclaim attempts to avoid block_groups seeing active use with a sweep mark that gets cleared on allocation and set on a sweep. In urgent conditions where we have very little unallocated space (less than one chunk used by the threshold calculation for the unallocated target), we want to be able to override this mechanism.

Introduce a second pass that only happens if we fail to find a reclaim candidate and reclaim is urgent. In that case, do a second pass where all block groups are eligible.

---
 fs/btrfs/space-info.c | 35 ++++++++++++++++++++++++++++++++++-
 1 file changed, 34 insertions(+), 1 deletion(-)
fs/btrfs/space-info.c +34 -1
index e7a2aa751f8f..95e65d5163ab 100644
@@ -1956,17 +1956,35 @@ int btrfs_calc_reclaim_threshold(struct btrfs_space_info *space_info)
 	return READ_ONCE(space_info->bg_reclaim_threshold);
 }
 
+/*
+ * Under "urgent" reclaim, we will reclaim even fresh block groups that have
+ * recently seen successful allocations, as we are desperate to reclaim
+ * whatever we can to avoid ENOSPC in a transaction leading to a readonly fs.
+ */
+static bool is_reclaim_urgent(struct btrfs_space_info *space_info)
+{
+	struct btrfs_fs_info *fs_info = space_info->fs_info;
+	u64 unalloc = atomic64_read(&fs_info->free_chunk_space);
+	u64 data_chunk_size = calc_effective_data_chunk_size(fs_info);
+
+	return unalloc < data_chunk_size;
+}
+
 static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 			    struct btrfs_space_info *space_info, int raid)
 {
 	struct btrfs_block_group *bg;
 	int thresh_pct;
+	bool try_again = true;
+	bool urgent;
 
 	spin_lock(&space_info->lock);
+	urgent = is_reclaim_urgent(space_info);
 	thresh_pct = btrfs_calc_reclaim_threshold(space_info);
 	spin_unlock(&space_info->lock);
 
 	down_read(&space_info->groups_sem);
+again:
 	list_for_each_entry(bg, &space_info->block_groups[raid], list) {
 		u64 thresh;
 		bool reclaim = false;
@@ -1974,14 +1992,29 @@ static int do_reclaim_sweep(struct btrfs_fs_info *fs_info,
 		btrfs_get_block_group(bg);
 		spin_lock(&bg->lock);
 		thresh = mult_perc(bg->length, thresh_pct);
-		if (bg->used < thresh && bg->reclaim_mark)
+		if (bg->used < thresh && bg->reclaim_mark) {
+			try_again = false;
 			reclaim = true;
+		}
 		bg->reclaim_mark++;
 		spin_unlock(&bg->lock);
 		if (reclaim)
 			btrfs_mark_bg_to_reclaim(bg);
 		btrfs_put_block_group(bg);
 	}
+
+	/*
+	 * In situations where we are very motivated to reclaim (low unalloc)
+	 * use two passes to make the reclaim mark check best effort.
+	 *
+	 * If we have any staler groups, we don't touch the fresher ones, but if we
+	 * really need a block group, do take a fresh one.
+	 */
+	if (try_again && urgent) {
+		try_again = false;
+		goto again;
+	}
+
 	up_read(&space_info->groups_sem);
 	return 0;
 }
Signature · 1 line
2.45.2
Reviewed-by on the v2 series
#12 lore raw
In reply to Boris Burkov
Quoted 59 lines · Boris Burkov

Btrfs's block_group allocator suffers from a well known problem, that it is capable of eagerly allocating too much space to either data or metadata (most often data, absent bugs) and then later be unable to allocate more space for the other, when needed. When data starves metadata, this can extra painfully result in read only filesystems that need careful manual balancing to fix.

This can be worked around by:
- enabling automatic reclaim
- periodically running balance

The latter is widely deployed via btrfsmaintenance
(https://github.com/kdave/btrfsmaintenance) and the former is used at scale at Meta with good results. However, neither of those solutions is perfect, as they both currently use a fixed threshold. A fixed threshold is vulnerable to workloads that trigger high amounts of reclaim. This has led to btrfsmaintenance setting very conservative thresholds of 5 and 10 percent of data block groups.
(https://github.com/kdave/btrfsmaintenance/commit/edbbfffe592f47c2849a8825f523e2ccc38b15f5)
At Meta, we deal with an elevated level of reclaim which would be desirable to reduce.

This patch set expands on automatic reclaim, adding the ability to set a dynamic reclaim threshold that appropriately scales with the global file system allocation conditions as well as periodic reclaim which runs that reclaim sweep in the cleaner thread. Together, I believe they constitute a robust and general automatic reclaim system that should avoid unfortunate read only filesystems in all but extreme conditions, where space is running quite low anyway and failure is more reasonable.

At a very high level, the dynamic threshold's strategy is to set a fixed target of unallocated block groups (10 block groups) and linearly scale its aggression the further we are from that target. That way we do no automatic reclaim until we actually press against the unallocated target, allowing the allocator to gradually fill fragmented space with new extents, but do claw back space after workloads that use and free a bunch of space, perhaps with fragmentation.

I ran it on three workloads (described in detail on the dynamic reclaim patch) but they are:
1. bounce allocations around X% full.
2. fill up all the way and introduce full fragmentation.
3. write in a fragmented way until the filesystem is just about full. script can be found here:
https://github.com/boryas/scripts/tree/main/fio/reclaim

The important results can be seen here (full results explorable at
https://bur.io/dyn-rec/)

bounce at 30%, higher relocations with a fixed threshold:
https://bur.io/dyn-rec/bounce/reclaims.png
https://bur.io/dyn-rec/bounce/reclaim_bytes.png
https://bur.io/dyn-rec/bounce/unalloc_bytes.png

hard 30% fragmentation, dynamic actually reclaims, relocs not crazy:
https://bur.io/dyn-rec/strict_frag/reclaims.png
https://bur.io/dyn-rec/strict_frag/reclaim_bytes.png
https://bur.io/dyn-rec/strict_frag/unalloc_bytes.png

Quoted 5 lines

fill it all the way up in a fragmented way, then keep making allocations:
https://bur.io/dyn-rec/last_gig/reclaims.png
https://bur.io/dyn-rec/last_gig/reclaim_bytes.png
https://bur.io/dyn-rec/last_gig/unalloc_bytes.png

These results are great, once you fix up the one comment I had you can add

to the whole series. Thanks,

Josef