btrfs: Enhance btrfs_trim_fs function to handle error better

Function btrfs_trim_fs() doesn't handle errors in a consistent way, if error happens when trimming existing block groups, it will skip the remaining blocks and continue to trim unallocated space for each device.

And the return value will only reflect the final error from device trimming.

This patch will fix such behavior by:

1) Recording first error from block group or device trimming So return value will also reflect any error found when trimming. Make developer more aware of the problem.

2) Outputting btrfs warning message for each trimming failure Any error for block group or device trimming will cause btrfs warning kernel message.

3) Continuing trimming if we can
If we failed to trim one block group or device, we could still try next block group or device.

Such behavior can avoid confusion for case like failure to trim the first block group and then only unallocated space is trimmed.

Reported-by Chris Murphy <[email protected]>

Signed-off-by Qu Wenruo <[email protected]>

---
 fs/btrfs/extent-tree.c | 59 ++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 43 insertions(+), 16 deletions(-)

fs/btrfs/extent-tree.c +13 -1

index 309a109069f1..46d65ffb3bd1 100644
@@ -10948,6 +10948,16 @@ static int btrfs_trim_free_extents(struct btrfs_device *device,
 	return ret;
 }
 
+/*
+ * Trim the whole fs, by:
+ * 1) Trimming free space in each block group
+ * 2) Trimming unallocated space in each device
+ *
+ * Will try to continue trimming even if we failed to trim one block group or
+ * device.
+ * The return value will be the error return value of the first error.
+ * Or 0 if nothing wrong happened.
+ */
 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 {
 	struct btrfs_block_group_cache *cache = NULL;

@@ -10958,6 +10968,8 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	u64 end;
 	u64 trimmed = 0;
 	u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
+	int bg_ret = 0;
+	int dev_ret = 0;
 	int ret = 0;
 
 	/*

@@ -10968,7 +10980,7 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	else
 		cache = btrfs_lookup_block_group(fs_info, range->start);
 
-	while (cache) {
+	for (; cache; cache = next_block_group(fs_info, cache)) {
 		if (cache->key.objectid >= (range->start + range->len)) {
 			btrfs_put_block_group(cache);
 			break;

fs/btrfs/extent-tree.c +12 -9

+						   cache->key.objectid, ret);
+					if (!bg_ret)
+						bg_ret = ret;
+					continue;
 				}
 			}
-			ret = btrfs_trim_block_group(cache,
-						     &group_trimmed,
-						     start,
-						     end,
-						     range->minlen);
+			ret = btrfs_trim_block_group(cache, &group_trimmed,
+						start, end, range->minlen);
 
 			trimmed += group_trimmed;
 			if (ret) {
-				btrfs_put_block_group(cache);
-				break;
+				btrfs_warn_rl(fs_info,
+		"failed to trim block group %llu ret %d",
+					   cache->key.objectid, ret);
+				if (!bg_ret)
+					bg_ret = ret;
+				continue;
 			}
 		}
-
-		cache = next_block_group(fs_info, cache);
 	}
 
 	mutex_lock(&fs_info->fs_devices->device_list_mutex);

fs/btrfs/extent-tree.c +6 -1

+			if (!dev_ret)
+				dev_ret = ret;
 			break;
+		}
 
 		trimmed += group_trimmed;
 	}
 	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 
 	range->len = trimmed;
-	return ret;
+	if (bg_ret)
+		return bg_ret;
+	return dev_ret;
 }
 
 /*

Signature · 6 lines

2.15.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Quoted 13 lines · Qu Wenruo

Function btrfs_trim_fs() doesn't handle errors in a consistent way, if error happens when trimming existing block groups, it will skip the remaining blocks and continue to trim unallocated space for each device.

And the return value will only reflect the final error from device trimming.

This patch will fix such behavior by:

1) Recording first error from block group or device trimming So return value will also reflect any error found when trimming. Make developer more aware of the problem.

Quoted 3 lines

2) Outputting btrfs warning message for each trimming failure Any error for block group or device trimming will cause btrfs warning kernel message.

I think this could become too noisy, trimming failures are soft errors IMO, so it should be enough to report all errors cumulatively per-device.

Quoted 4 lines

3) Continuing trimming if we can
If we failed to trim one block group or device, we could still try next block group or device.

Right, best-effort.

Quoted 43 lines

Such behavior can avoid confusion for case like failure to trim the first block group and then only unallocated space is trimmed.

Reported-by Chris Murphy <[email protected]>

Signed-off-by Qu Wenruo <[email protected]>

---
 fs/btrfs/extent-tree.c | 59 ++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 43 insertions(+), 16 deletions(-)

fs/btrfs/extent-tree.c +12 -0

index 309a109069f1..46d65ffb3bd1 100644
@@ -10948,6 +10948,16 @@ static int btrfs_trim_free_extents(struct btrfs_device *device,
 	return ret;
 }
 
+/*
+ * Trim the whole fs, by:
+ * 1) Trimming free space in each block group
+ * 2) Trimming unallocated space in each device
+ *
+ * Will try to continue trimming even if we failed to trim one block group or
+ * device.
+ * The return value will be the error return value of the first error.
+ * Or 0 if nothing wrong happened.
+ */
 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 {
 	struct btrfs_block_group_cache *cache = NULL;

@@ -10958,6 +10968,8 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	u64 end;
 	u64 trimmed = 0;
 	u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
+	int bg_ret = 0;
+	int dev_ret = 0;
 	int ret = 0;
 
 	/*

@@ -10968,7 +10980,7 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	else
 		cache = btrfs_lookup_block_group(fs_info, range->start);
 

fs/btrfs/extent-tree.c int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range) · line 10998

Quoted 13 lines

fs/btrfs/extent-tree.c +3 -3

-	while (cache) {
+	for (; cache; cache = next_block_group(fs_info, cache)) {
 		if (cache->key.objectid >= (range->start + range->len)) {
 			btrfs_put_block_group(cache);
 			break;

@@ -10982,29 +10994,36 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 			if (!block_group_cache_done(cache)) {
 				ret = cache_block_group(cache, 0);
 				if (ret) {
-					btrfs_put_block_group(cache);
-					break;
+					btrfs_warn_rl(fs_info,
+		"failed to cache block group %llu ret %d",

If this is meant for a developer as you say in the changelog, then btrfs_debug is better.

fs/btrfs/extent-tree.c int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range) · line 11007

Quoted 12 lines

fs/btrfs/extent-tree.c +6 -2

+						   cache->key.objectid, ret);
+					if (!bg_ret)
+						bg_ret = ret;
+					continue;
 				}
 				ret = wait_block_group_cache_done(cache);
 				if (ret) {
-					btrfs_put_block_group(cache);
-					break;
+					btrfs_warn_rl(fs_info,
+		"failed to wait cache for block group %llu ret %d",

The message wording is confusing, if this is another message for developer, the function name can be printed.

Quoted 31 lines

fs/btrfs/extent-tree.c +12 -9

+						   cache->key.objectid, ret);
+					if (!bg_ret)
+						bg_ret = ret;
+					continue;
 				}
 			}
-			ret = btrfs_trim_block_group(cache,
-						     &group_trimmed,
-						     start,
-						     end,
-						     range->minlen);
+			ret = btrfs_trim_block_group(cache, &group_trimmed,
+						start, end, range->minlen);
 
 			trimmed += group_trimmed;
 			if (ret) {
-				btrfs_put_block_group(cache);
-				break;
+				btrfs_warn_rl(fs_info,
+		"failed to trim block group %llu ret %d",
+					   cache->key.objectid, ret);
+				if (!bg_ret)
+					bg_ret = ret;
+				continue;
 			}
 		}
-
-		cache = next_block_group(fs_info, cache);
 	}
 

fs/btrfs/extent-tree.c int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range) · line 11037

Quoted 10 lines

fs/btrfs/extent-tree.c +4 -1

 	mutex_lock(&fs_info->fs_devices->device_list_mutex);

@@ -11012,15 +11031,23 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	list_for_each_entry(device, devices, dev_alloc_list) {
 		ret = btrfs_trim_free_extents(device, range->minlen,
 					      &group_trimmed);
-		if (ret)
+		if (ret) {
+			btrfs_warn_rl(fs_info,
+		"failed to trim unallocated space for devid %llu ret %d",
+				      device->devid, ret);

So the idea is to print one message here, with devid, number of errors and how many bytes were skipped.

Quoted 25 lines

fs/btrfs/extent-tree.c +6 -1

+			if (!dev_ret)
+				dev_ret = ret;
 			break;
+		}
 
 		trimmed += group_trimmed;
 	}
 	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 
 	range->len = trimmed;
-	return ret;
+	if (bg_ret)
+		return bg_ret;
+	return dev_ret;
 }
 
 /*

Signature · 6 lines

2.15.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 18 lines · David Sterba

Quoted 16 lines · Qu Wenruo

Function btrfs_trim_fs() doesn't handle errors in a consistent way, if error happens when trimming existing block groups, it will skip the remaining blocks and continue to trim unallocated space for each device.

And the return value will only reflect the final error from device trimming.

This patch will fix such behavior by:

1) Recording first error from block group or device trimming So return value will also reflect any error found when trimming. Make developer more aware of the problem.

2) Outputting btrfs warning message for each trimming failure Any error for block group or device trimming will cause btrfs warning kernel message.

Quoted 3 lines

I think this could become too noisy, trimming failures are soft errors IMO, so it should be enough to report all errors cumulatively per-device.

Although block group trimming errors are not that obvious to be reported at per-device base.

Quoted 4 lines

Quoted 4 lines

fs/btrfs/extent-tree.c +0 -0

@@ -11012,15 +11031,23 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	list_for_each_entry(device, devices, dev_alloc_list) {
 		ret = btrfs_trim_free_extents(device, range->minlen,
 					      &group_trimmed);

Quoted 8 lines

fs/btrfs/extent-tree.c int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range) · line 11037

Quoted 5 lines

fs/btrfs/extent-tree.c +4 -1

-		if (ret)
+		if (ret) {
+			btrfs_warn_rl(fs_info,
+		"failed to trim unallocated space for devid %llu ret %d",
+				      device->devid, ret);

So the idea is to print one message here, with devid, number of errors and how many bytes were skipped.

About number of errors, did you mean also accounting the errors found in block group trimming?
In that case, it may not be related to all devices of a block group.

So this doesn't look appropriate to account block group errors into device error.