There is a race between block group removal and block group creation
when the removal is completed by a task running fitrim or scrub. When
this happens we end up failing the block group creation with an error
-EEXIST since we attempt to insert a duplicate block group item key
in the extent tree. That results in a transaction abort.
The race happens like this:
1) Task A is doing a fitrim, and at btrfs_trim_block_group() it freezes
block group X with btrfs_freeze_block_group() (until very recently
that was named btrfs_get_block_group_trimming());
2) Task B starts removing block group X, either because it's now unused
or due to relocation for example. So at btrfs_remove_block_group(),
while holding the chunk mutex and the block group's lock, it sets
the 'removed' flag of the block group and it sets the local variable
'remove_em' to false, because the block group is currently frozen
(its 'frozen' counter is > 0, until very recently this counter was
named 'trimming');
3) Task B unlocks the block group and the chunk mutex;
4) Task A is done trimming the block group and unfreezes the block group
by calling btrfs_unfreeze_block_group() (until very recently this was
named btrfs_put_block_group_trimming()). In this function we lock the
block group and set the local variable 'cleanup' to true because we
were able to decrement the block group's 'frozen' counter down to 0 and
the flag 'removed' is set in the block group.
Since 'cleanup' is set to true, it locks the chunk mutex and removes
the extent mapping representing the block group from the mapping tree;
5) Task C allocates a new block group Y and it picks up the logical address
that block group X had as the logical address for Y, because X was the
block group with the highest logical address and now the second block
group with the highest logical address, the last in the fs mapping tree,
ends at an offset corresponding to block group X's logical address (this
logical address selection is done at volumes.c:find_next_chunk()).
At this point the new block group Y does not have yet its item added
to the extent tree (nor the corresponding device extent items and
chunk item in the device and chunk trees). The new group Y is added to
the list of pending block groups in the transaction handle;
6) Before task B proceeds to removing the block group item for block
group X from the extent tree, which has a key matching:
(X logical offset, BTRFS_BLOCK_GROUP_ITEM_KEY, length)
task C while ending its transaction handle calls
btrfs_create_pending_block_groups(), which finds block group Y and
tries to insert the block group item for Y into the exten tree, which
fails with -EEXIST since logical offset is the same that X had and
task B hasn't yet deleted the key from the extent tree.
This failure results in a transaction abort, producing a stack like
the following:
------------[ cut here ]------------
BTRFS: Transaction aborted (error -17)
WARNING: CPU: 2 PID: 19736 at fs/btrfs/block-group.c:2074 btrfs_create_pending_block_groups+0x1eb/0x260 [btrfs]
Modules linked in: btrfs blake2b_generic xor raid6_pq (...)
CPU: 2 PID: 19736 Comm: fsstress Tainted: G W 5.6.0-rc7-btrfs-next-58 #5
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-59-gc9ba5276e321-prebuilt.qemu.org 04/01/2014
RIP: 0010:btrfs_create_pending_block_groups+0x1eb/0x260 [btrfs]
Code: ff ff ff 48 8b 55 50 f0 48 (...)
RSP: 0018:ffffa4160a1c7d58 EFLAGS: 00010286
RAX: 0000000000000000 RBX: ffff961581909d98 RCX: 0000000000000000
RDX: 0000000000000001 RSI: ffffffffb3d63990 RDI: 0000000000000001
RBP: ffff9614f3356a58 R08: 0000000000000000 R09: 0000000000000001
R10: ffff9615b65b0040 R11: 0000000000000000 R12: ffff961581909c10
R13: ffff9615b0c32000 R14: ffff9614f3356ab0 R15: ffff9614be779000
FS: 00007f2ce2841e80(0000) GS:ffff9615bae00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000555f18780000 CR3: 0000000131d34005 CR4: 00000000003606e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
btrfs_start_dirty_block_groups+0x398/0x4e0 [btrfs]
btrfs_commit_transaction+0xd0/0xc50 [btrfs]
? btrfs_attach_transaction_barrier+0x1e/0x50 [btrfs]
? __ia32_sys_fdatasync+0x20/0x20
iterate_supers+0xdb/0x180
ksys_sync+0x60/0xb0
__ia32_sys_sync+0xa/0x10
do_syscall_64+0x5c/0x280
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f2ce1d4d5b7
Code: 83 c4 08 48 3d 01 (...)
RSP: 002b:00007ffd8b558c58 EFLAGS: 00000202 ORIG_RAX: 00000000000000a2
RAX: ffffffffffffffda RBX: 000000000000002c RCX: 00007f2ce1d4d5b7
RDX: 00000000ffffffff RSI: 00000000186ba07b RDI: 000000000000002c
RBP: 0000555f17b9e520 R08: 0000000000000012 R09: 000000000000ce00
R10: 0000000000000078 R11: 0000000000000202 R12: 0000000000000032
R13: 0000000051eb851f R14: 00007ffd8b558cd0 R15: 0000555f1798ec20
irq event stamp: 0
hardirqs last enabled at (0): [<0000000000000000>] 0x0
hardirqs last disabled at (0): [<ffffffffb2abdedf>] copy_process+0x74f/0x2020
softirqs last enabled at (0): [<ffffffffb2abdedf>] copy_process+0x74f/0x2020
softirqs last disabled at (0): [<0000000000000000>] 0x0
---[ end trace bd7c03622e0b0a9c ]---
Fix this simply by making btrfs_remove_block_group() remove the block
group's item from the extent tree before it flags the block group as
removed. Also make the free space deletion from the free space tree
before flagging the block group as removed, to avoid a similar race
with adding and removing free space entries for the free space tree.
Fixes: 04216820fe ("Btrfs: fix race between fs trimming and block group remove/allocation")
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When removing a block group, if we fail to delete the block group's item
from the extent tree, we jump to the 'out' label and end up decrementing
the block group's reference count once only (by 1), resulting in a counter
leak because the block group at that point was already removed from the
block group cache rbtree - so we have to decrement the reference count
twice, once for the rbtree and once for our lookup at the start of the
function.
There is a second bug where if removing the free space tree entries (the
call to remove_block_group_free_space()) fails we end up jumping to the
'out_put_group' label but end up decrementing the reference count only
once, when we should have done it twice, since we have already removed
the block group from the block group cache rbtree. This happens because
the reference count decrement for the rbtree reference happens after
attempting to remove the free space tree entries, which is far away from
the place where we remove the block group from the rbtree.
To make things less error prone, decrement the reference count for the
rbtree immediately after removing the block group from it. This also
eleminates the need for two different exit labels on error, renaming
'out_put_label' to just 'out' and removing the old 'out'.
Fixes: f6033c5e33 ("btrfs: fix block group leak when removing fails")
CC: stable@vger.kernel.org # 4.4+
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
disk-io.h is included more than once in block-group.c, remove it.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
Signed-off-by: David Sterba <dsterba@suse.com>
The name of this function contains the word "cache", which is left from
the times where btrfs_block_group was called btrfs_block_group_cache.
Now this "cache" doesn't match anything, and we have better namings for
functions like read/insert/remove_block_group_item().
Rename it to update_block_group_item().
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently the block group item insert is pretty straight forward, fill
the block group item structure and insert it into extent tree.
However the incoming skinny block group feature is going to change this,
so this patch will refactor insertion into a new function,
insert_block_group_item(), to make the incoming feature easier to add.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When deleting a block group item, it's pretty straight forward, just
delete the item pointed by the key. However it will not be that
straight-forward for incoming skinny block group item.
So refactor the block group item deletion into a new function,
remove_block_group_item(), also to make the already lengthy
btrfs_remove_block_group() a little shorter.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Structure btrfs_block_group has the following members which are
currently read from on-disk block group item and key:
- length - from item key
- used
- flags - from block group item
However for incoming skinny block group tree, we are going to read those
members from different sources.
This patch will refactor such read by:
- Don't initialize btrfs_block_group::length at allocation
Caller should initialize them manually.
Also to avoid possible (well, only two callers) missing
initialization, add extra ASSERT() in btrfs_add_block_group_cache().
- Refactor length/used/flags initialization into one function
The new function, fill_one_block_group() will handle the
initialization of such members.
- Use btrfs_block_group::length to replace key::offset
Since skinny block group item would have a different meaning for its
key offset.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Regular block group items in extent tree are scattered inside the huge
tree, thus forward readahead makes no sense.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The helpers btrfs_freeze_block_group() and btrfs_unfreeze_block_group()
used to be named btrfs_get_block_group_trimming() and
btrfs_put_block_group_trimming() respectively.
At the time they were added to free-space-cache.c, by commit e33e17ee10
("btrfs: add missing discards when unpinning extents with -o discard")
because all the trimming related functions were in free-space-cache.c.
Now that the helpers were renamed and are used in scrub context as well,
move them to block-group.c, a much more logical location for them.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Back in 2014, commit 04216820fe ("Btrfs: fix race between fs trimming
and block group remove/allocation"), I added the 'trimming' member to the
block group structure. Its purpose was to prevent races between trimming
and block group deletion/allocation by pinning the block group in a way
that prevents its logical address and device extents from being reused
while trimming is in progress for a block group, so that if another task
deletes the block group and then another task allocates a new block group
that gets the same logical address and device extents while the trimming
task is still in progress.
After the previous fix for scrub (patch "btrfs: fix a race between scrub
and block group removal/allocation"), scrub now also has the same needs that
trimming has, so the member name 'trimming' no longer makes sense.
Since there is already a 'pinned' member in the block group that refers
to space reservations (pinned bytes), rename the member to 'frozen',
add a comment on top of it to describe its general purpose and rename
the helpers to increment and decrement the counter as well, to match
the new member name.
The next patch in the series will move the helpers into a more suitable
file (from free-space-cache.c to block-group.c).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
At clean_pinned_extents(), whether we end up returning success or failure,
we pretty much have to do the same things:
1) unlock unused_bg_unpin_mutex
2) decrement reference count on the previous transaction
We also call btrfs_dec_block_group_ro() in case of failure, but that is
better done in its caller, btrfs_delete_unused_bgs(), since its the
caller that calls inc_block_group_ro(), so it should be responsible for
the decrement operation, as it is in case any of the other functions it
calls fail.
So move the call to btrfs_dec_block_group_ro() from clean_pinned_extents()
into btrfs_delete_unused_bgs() and unify the error and success return
paths for clean_pinned_extents(), reducing duplicated code and making it
simpler.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For unlink transactions and block group removal
btrfs_start_transaction_fallback_global_rsv will first try to start an
ordinary transaction and if it fails it will fall back to reserving the
required amount by stealing from the global reserve. This is problematic
because of all the same reasons we had with previous iterations of the
ENOSPC handling, thundering herd. We get a bunch of failures all at
once, everybody tries to allocate from the global reserve, some win and
some lose, we get an ENSOPC.
Fix this behavior by introducing BTRFS_RESERVE_FLUSH_ALL_STEAL. It's
used to mark unlink reservation. To fix this we need to integrate this
logic into the normal ENOSPC infrastructure. We still go through all of
the normal flushing work, and at the moment we begin to fail all the
tickets we try to satisfy any tickets that are allowed to steal by
stealing from the global reserve. If this works we start the flushing
system over again just like we would with a normal ticket satisfaction.
This serializes our global reserve stealing, so we don't have the
thundering herd problem.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_remove_block_group() invokes btrfs_lookup_block_group(), which
returns a local reference of the block group that contains the given
bytenr to "block_group" with increased refcount.
When btrfs_remove_block_group() returns, "block_group" becomes invalid,
so the refcount should be decreased to keep refcount balanced.
The reference counting issue happens in several exception handling paths
of btrfs_remove_block_group(). When those error scenarios occur such as
btrfs_alloc_path() returns NULL, the function forgets to decrease its
refcnt increased by btrfs_lookup_block_group() and will cause a refcnt
leak.
Fix this issue by jumping to "out_put_group" label and calling
btrfs_put_block_group() when those error scenarios occur.
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When cleaning pinned extents right before deleting an unused block group,
we check if there's still a previous transaction running and if so we
increment its reference count before using it for cleaning pinned ranges
in its pinned extents iotree. However we ended up never decrementing the
reference count after using the transaction, resulting in a memory leak.
Fix it by decrementing the reference count.
Fixes: fe119a6eeb ("btrfs: switch to per-transaction pinned extents")
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Whenever we add a ticket to a space_info object we increment the object's
reclaim_size counter witht the ticket's bytes, and we decrement it with
the corresponding amount only when we are able to grant the requested
space to the ticket. When we are not able to grant the space to a ticket,
or when the ticket is removed due to a signal (e.g. an application has
received sigterm from the terminal) we never decrement the counter with
the corresponding bytes from the ticket. This leak can result in the
space reclaim code to later do much more work than necessary. So fix it
by decrementing the counter when those two cases happen as well.
Fixes: db161806dc ("btrfs: account ticket size at add/delete time")
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The space_info list is normally RCU protected and should be traversed
with rcu_read_lock held. There's a warning
[29.104756] WARNING: suspicious RCU usage
[29.105046] 5.6.0-rc4-next-20200305 #1 Not tainted
[29.105231] -----------------------------
[29.105401] fs/btrfs/block-group.c:2011 RCU-list traversed in non-reader section!!
pointing out that the locking is missing in btrfs_read_block_groups.
However this is not necessary as the list traversal happens at mount
time when there's no other thread potentially accessing the list.
To fix the warning and for consistency let's add the RCU lock/unlock,
the code won't be affected much as it's doing some lightweight
operations.
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Madhuparna Bhowmik <madhuparnabhowmik10@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This commit flips the switch to start tracking/processing pinned extents
on a per-transaction basis. It mostly replaces all references from
btrfs_fs_info::(pinned_extents|freed_extents[]) to
btrfs_transaction::pinned_extents.
Two notable modifications that warrant explicit mention are changing
clean_pinned_extents to get a reference to the previously running
transaction. The other one is removal of call to
btrfs_destroy_pinned_extent since transactions are going to be cleaned
in btrfs_cleanup_one_transaction.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Next patch is going to refactor how pinned extents are tracked which
will necessitate changing this code. To ease that work and contain the
changes factor the code now in preparation, this will also help review.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The status of aborted transaction can change between calls and it needs
to be accessed by READ_ONCE. Add a helper that also wraps the unlikely
hint.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We are incorrectly dropping the raid56 and raid1c34 incompat flags when
there are still raid56 and raid1c34 block groups, not when we do not any
of those anymore. The logic just got unintentionally broken after adding
the support for the raid1c34 modes.
Fix this by clear the flags only if we do not have block groups with the
respective profiles.
Fixes: 9c907446dc ("btrfs: drop incompat bit for raid1c34 after last block group is gone")
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
inc_block_group_ro does a calculation to see if we have enough room left
over if we mark this block group as read only in order to see if it's ok
to mark the block group as read only.
The problem is this calculation _only_ works for data, where our used is
always less than our total. For metadata we will overcommit, so this
will almost always fail for metadata.
Fix this by exporting btrfs_can_overcommit, and then see if we have
enough space to remove the remaining free space in the block group we
are trying to mark read only. If we do then we can mark this block
group as read only.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For some reason we've translated the do_chunk_alloc that goes into
btrfs_inc_block_group_ro to force in inc_block_group_ro, but these are
two different things.
force for inc_block_group_ro is used when we are forcing the block group
read only no matter what, for example when the underlying chunk is
marked read only. We need to not do the space check here as this block
group needs to be read only.
btrfs_inc_block_group_ro() has a do_chunk_alloc flag that indicates that
we need to pre-allocate a chunk before marking the block group read
only. This has nothing to do with forcing, and in fact we _always_ want
to do the space check in this case, so unconditionally pass false for
force in this case.
Then fixup inc_block_group_ro to honor force as it's expected and
documented to do.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Move variables to appropriate scope. Remove last BUG_ON in the function
and rework error handling accordingly. Make the duplicate detection code
more straightforward. Use in_range macro. And give variables more
descriptive name by explicitly distinguishing between IO stripe size
(size recorded in the chunk item) and data stripe size (the size of
an actual stripe, constituting a logical chunk/block group).
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
It's used only during initial block group reading to map physical
address of super block to a list of logical ones. Make it private to
block-group.c, add proper kernel doc and ensure it's exported only for
tests.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_get_alloc_profile() is a simple wrapper over get_alloc_profile().
The only difference is btrfs_get_alloc_profile() is visible to other
functions in btrfs while get_alloc_profile() is static and thus only
visible to functions in block-group.c.
Let's just fold get_alloc_profile() into btrfs_get_alloc_profile() to
get rid of the unnecessary second function.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Johannes Thumshirn <jth@kernel.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
block_group removal is a little tricky. It can race with the extent
allocator, the cleaner thread, and balancing. The current path is for a
block_group to be added to the unused_bgs list. Then, when the cleaner
thread comes around, it starts a transaction and then proceeds with
removing the block_group. Extents that are pinned are subsequently
removed from the pinned trees and then eventually a discard is issued
for the entire block_group.
Async discard introduces another player into the game, the discard
workqueue. While it has none of the racing issues, the new problem is
ensuring we don't leave free space untrimmed prior to forgetting the
block_group. This is handled by placing fully free block_groups on a
separate discard queue. This is necessary to maintain discarding order
as in the future we will slowly trim even fully free block_groups. The
ordering helps us make progress on the same block_group rather than say
the last fully freed block_group or needing to search through the fully
freed block groups at the beginning of a list and insert after.
The new order of events is a fully freed block group gets placed on the
unused discard queue first. Once it's processed, it will be placed on
the unusued_bgs list and then the original sequence of events will
happen, just without the final whole block_group discard.
The mount flags can change when processing unused_bgs, so when flipping
from DISCARD to DISCARD_ASYNC, the unused_bgs must be punted to the
discard_list to be trimmed. If we flip off DISCARD_ASYNC, we punt
free block groups on the discard_list to the unused_bg queue which will
do the final discard for us.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Dennis Zhou <dennis@kernel.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When discard is enabled, everytime a pinned extent is released back to
the block_group's free space cache, a discard is issued for the extent.
This is an overeager approach when it comes to discarding and helping
the SSD maintain enough free space to prevent severe garbage collection
situations.
This adds the beginning of async discard. Instead of issuing a discard
prior to returning it to the free space, it is just marked as untrimmed.
The block_group is then added to a LRU which then feeds into a workqueue
to issue discards at a much slower rate. Full discarding of unused block
groups is still done and will be addressed in a future patch of the
series.
For now, we don't persist the discard state of extents and bitmaps.
Therefore, our failure recovery mode will be to consider extents
untrimmed. This lets us handle failure and unmounting as one in the
same.
On a number of Facebook webservers, I collected data every minute
accounting the time we spent in btrfs_finish_extent_commit() (col. 1)
and in btrfs_commit_transaction() (col. 2). btrfs_finish_extent_commit()
is where we discard extents synchronously before returning them to the
free space cache.
discard=sync:
p99 total per minute p99 total per minute
Drive | extent_commit() (ms) | commit_trans() (ms)
---------------------------------------------------------------
Drive A | 434 | 1170
Drive B | 880 | 2330
Drive C | 2943 | 3920
Drive D | 4763 | 5701
discard=async:
p99 total per minute p99 total per minute
Drive | extent_commit() (ms) | commit_trans() (ms)
--------------------------------------------------------------
Drive A | 134 | 956
Drive B | 64 | 1972
Drive C | 59 | 1032
Drive D | 62 | 1200
While it's not great that the stats are cumulative over 1m, all of these
servers are running the same workload and and the delta between the two
are substantial. We are spending significantly less time in
btrfs_finish_extent_commit() which is responsible for discarding.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Dennis Zhou <dennis@kernel.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This series introduces async discard which will use the flag
DISCARD_ASYNC, so rename the original flag to DISCARD_SYNC as it is
synchronously done in transaction commit.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dennis Zhou <dennis@kernel.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is a relic from a time before we had a proper reservation mechanism
and you could end up with really full chunks at chunk allocation time.
This doesn't make sense anymore, so just kill it.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
When running btrfs/072 with only one online CPU, it has a pretty high
chance to fail:
btrfs/072 12s ... _check_dmesg: something found in dmesg (see xfstests-dev/results//btrfs/072.dmesg)
- output mismatch (see xfstests-dev/results//btrfs/072.out.bad)
--- tests/btrfs/072.out 2019-10-22 15:18:14.008965340 +0800
+++ /xfstests-dev/results//btrfs/072.out.bad 2019-11-14 15:56:45.877152240 +0800
@@ -1,2 +1,3 @@
QA output created by 072
Silence is golden
+Scrub find errors in "-m dup -d single" test
...
And with the following call trace:
BTRFS info (device dm-5): scrub: started on devid 1
------------[ cut here ]------------
BTRFS: Transaction aborted (error -27)
WARNING: CPU: 0 PID: 55087 at fs/btrfs/block-group.c:1890 btrfs_create_pending_block_groups+0x3e6/0x470 [btrfs]
CPU: 0 PID: 55087 Comm: btrfs Tainted: G W O 5.4.0-rc1-custom+ #13
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
RIP: 0010:btrfs_create_pending_block_groups+0x3e6/0x470 [btrfs]
Call Trace:
__btrfs_end_transaction+0xdb/0x310 [btrfs]
btrfs_end_transaction+0x10/0x20 [btrfs]
btrfs_inc_block_group_ro+0x1c9/0x210 [btrfs]
scrub_enumerate_chunks+0x264/0x940 [btrfs]
btrfs_scrub_dev+0x45c/0x8f0 [btrfs]
btrfs_ioctl+0x31a1/0x3fb0 [btrfs]
do_vfs_ioctl+0x636/0xaa0
ksys_ioctl+0x67/0x90
__x64_sys_ioctl+0x43/0x50
do_syscall_64+0x79/0xe0
entry_SYSCALL_64_after_hwframe+0x49/0xbe
---[ end trace 166c865cec7688e7 ]---
[CAUSE]
The error number -27 is -EFBIG, returned from the following call chain:
btrfs_end_transaction()
|- __btrfs_end_transaction()
|- btrfs_create_pending_block_groups()
|- btrfs_finish_chunk_alloc()
|- btrfs_add_system_chunk()
This happens because we have used up all space of
btrfs_super_block::sys_chunk_array.
The root cause is, we have the following bad loop of creating tons of
system chunks:
1. The only SYSTEM chunk is being scrubbed
It's very common to have only one SYSTEM chunk.
2. New SYSTEM bg will be allocated
As btrfs_inc_block_group_ro() will check if we have enough space
after marking current bg RO. If not, then allocate a new chunk.
3. New SYSTEM bg is still empty, will be reclaimed
During the reclaim, we will mark it RO again.
4. That newly allocated empty SYSTEM bg get scrubbed
We go back to step 2, as the bg is already mark RO but still not
cleaned up yet.
If the cleaner kthread doesn't get executed fast enough (e.g. only one
CPU), then we will get more and more empty SYSTEM chunks, using up all
the space of btrfs_super_block::sys_chunk_array.
[FIX]
Since scrub/dev-replace doesn't always need to allocate new extent,
especially chunk tree extent, so we don't really need to do chunk
pre-allocation.
To break above spiral, here we introduce a new parameter to
btrfs_inc_block_group(), @do_chunk_alloc, which indicates whether we
need extra chunk pre-allocation.
For relocation, we pass @do_chunk_alloc=true, while for scrub, we pass
@do_chunk_alloc=false.
This should keep unnecessary empty chunks from popping up for scrub.
Also, since there are two parameters for btrfs_inc_block_group_ro(),
add more comment for it.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The type name is misleading, a single entry is named 'cache' while this
normally means a collection of objects. Rename that everywhere. Also the
identifier was quite long, making function prototypes harder to format.
Suggested-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For read_one_block_group(), its only caller has already got the item key
to search next block group item.
So we can use that key directly without doing our own convertion on
stack.
Also, since that key used in btrfs_read_block_groups() is vital for
block group item search, add 'const' keyword for that parameter to
prevent read_one_block_group() to modify it.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Refactor the work inside the loop of btrfs_read_block_groups() into one
separate function, read_one_block_group().
This allows read_one_block_group to be reused for later BG_TREE feature.
The refactor does the following extra fix:
- Use btrfs_fs_incompat() to replace open-coded feature check
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When there are no raid1c3 or raid1c4 block groups left after balance
(either convert or with other filters applied), remove the incompat bit.
This is already done for RAID56, do the same for RAID1C34.
Signed-off-by: David Sterba <dsterba@suse.com>
The on-disk format of block group item makes use of the key that stores
the offset and length. This is further used in the code, although this
makes thing harder to understand. The key is also packed so the
offset/length is not properly aligned as u64.
Add start (key.objectid) and length (key.offset) members to block group
and remove the embedded key. When the item is searched or written, a
local variable for key is used.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
All accessors defined by BTRFS_SETGET_STACK_FUNCS contain _stack_ in the
name, the block group ones were not following that scheme, so let's
switch them.
Signed-off-by: David Sterba <dsterba@suse.com>
The members ::used and ::flags are now in the block group cache
structure, the last one is chunk_objectid, but that's set to a fixed
value and otherwise unused. The item is constructed from a local
variable before write, so we can remove the embedded one from block
group.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The flags are read from the item that's embedded to block group struct,
but the item will be removed. Use the ::flags after read and before
write.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For unknown reasons, the member 'used' in the block group struct is
stored in the b-tree item and accessed everywhere using the special
accessor helper. Let's unify it and make it a regular member and only
update the item before writing it to the tree.
The item is still being used for flags and chunk_objectid, there's some
duplication until the item is removed in following patches.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When free'ing extents in a block group we check to see if the block
group is not cached, and then cache it if we need to. However we'll
just carry on as long as we're loading the cache. This is problematic
because we are dirtying the block group here. If we are fast enough we
could do a transaction commit and clear the free space cache while we're
still loading the space cache in another thread. This truncates the
free space inode, which will keep it from loading the space cache.
Fix this by using the btrfs_block_group_cache_done helper so that we try
to load the space cache unconditionally here, which will result in the
caller waiting for the fast caching to complete and keep us from
truncating the free space inode.
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Commit 4617ea3a52 (" Btrfs: fix necessary chunk tree space calculation
when allocating a chunk") removed the is_allocation argument from
check_system_chunk, since the formula for reserving the necessary space
for allocation or removing a chunk would be the same.
So, rework the comment by removing the mention of is_allocation
argument.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Commit 9e0af23764 ("Btrfs: fix task hang under heavy compressed
write") worked around the issue that a recycled work item could get a
false dependency on the original work item due to how the workqueue code
guarantees non-reentrancy. It did so by giving different work functions
to different types of work.
However, the fixes in the previous few patches are more complete, as
they prevent a work item from being recycled at all (except for a tiny
window that the kernel workqueue code handles for us). This obsoletes
the previous fix, so we don't need the unique helpers for correctness.
The only other reason to keep them would be so they show up in stack
traces, but they always seem to be optimized to a tail call, so they
don't show up anyways. So, let's just get rid of the extra indirection.
While we're here, rename normal_work_helper() to the more informative
btrfs_work_helper().
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In btrfs_read_block_groups(), if we have an invalid block group which
has mixed type (DATA|METADATA) while the fs doesn't have MIXED_GROUPS
feature, we error out without freeing the block group cache.
This patch will add the missing btrfs_put_block_group() to prevent
memory leak.
Note for stable backports: the file to patch in versions <= 5.3 is
fs/btrfs/extent-tree.c
Fixes: 49303381f1 ("Btrfs: bail out if block group has different mixed flag")
CC: stable@vger.kernel.org # 4.9+
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
I noticed when folding the trace_btrfs_space_reservation() tracepoint
into the btrfs_space_info_update_* helpers that we didn't emit a
tracepoint when doing btrfs_add_reserved_bytes(). I know this is
because we were swapping bytes_may_use for bytes_reserved, so in my mind
there was no reason to have the tracepoint there. But now there is
because we always emit the unreserve for the bytes_may_use side, and
this would have broken if compression was on anyway. Add a tracepoint
to cover the bytes_reserved counter so the math still comes out right.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We duplicate this tracepoint everywhere we call these helpers, so update
the helper to have the tracepoint as well.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_calc_trunc_metadata_size differs from trans_metadata_size in that
it doesn't take into account any splitting at the levels, because
truncate will never split nodes. However truncate _and_ changing will
never split nodes, so rename btrfs_calc_trunc_metadata_size to
btrfs_calc_metadata_size. Also btrfs_calc_trans_metadata_size is purely
for inserting items, so rename this to btrfs_calc_insert_metadata_size.
Making these clearer will help when I start using them differently in
upcoming patches.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These were renamed and exported to facilitate logical migration of
different code chunks into block-group.c. Now that all the users are in
one file go ahead and rename them back, move the code around, and make
them static.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This can now be easily migrated as well.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ refresh on top of sysfs cleanups ]
Signed-off-by: David Sterba <dsterba@suse.com>
These feel more at home in block-group.c.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ refresh, adjust btrfs_get_alloc_profile exports ]
Signed-off-by: David Sterba <dsterba@suse.com>
Filipe Manana