Commit c4cf5261 ("bio: skip atomic inc/dec of ->bi_remaining for
non-chains") regressed all existing callers that followed this pattern:
1) saving a bio's original bi_end_io
2) wiring up an intermediate bi_end_io
3) restoring the original bi_end_io from intermediate bi_end_io
4) calling bio_endio() to execute the restored original bi_end_io
The regression was due to BIO_CHAIN only ever getting set if
bio_inc_remaining() is called. For the above pattern it isn't set until
step 3 above (step 2 would've needed to establish BIO_CHAIN). As such
the first bio_endio(), in step 2 above, never decremented __bi_remaining
before calling the intermediate bi_end_io -- leaving __bi_remaining with
the value 1 instead of 0. When bio_inc_remaining() occurred during step
3 it brought it to a value of 2. When the second bio_endio() was
called, in step 4 above, it should've called the original bi_end_io but
it didn't because there was an extra reference that wasn't dropped (due
to atomic operations being optimized away since BIO_CHAIN wasn't set
upfront).
Fix this issue by removing the __bi_remaining management complexity for
all callers that use the above pattern -- bio_chain() is the only
interface that _needs_ to be concerned with __bi_remaining. For the
above pattern callers just expect the bi_end_io they set to get called!
Remove bio_endio_nodec() and also remove all bio_inc_remaining() calls
that aren't associated with the bio_chain() interface.
Also, the bio_inc_remaining() interface has been moved local to bio.c.
Fixes: c4cf5261 ("bio: skip atomic inc/dec of ->bi_remaining for non-chains")
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Struct bio has a reference count that controls when it can be freed.
Most uses cases is allocating the bio, which then returns with a
single reference to it, doing IO, and then dropping that single
reference. We can remove this atomic_dec_and_test() in the completion
path, if nobody else is holding a reference to the bio.
If someone does call bio_get() on the bio, then we flag the bio as
now having valid count and that we must properly honor the reference
count when it's being put.
Tested-by: Robert Elliott <elliott@hp.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Use generic io stats accounting help functions (generic_{start,end}_io_acct)
to simplify io stat accounting.
Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Acked-by: Kent Overstreet <kmo@datera.io>
Signed-off-by: Jens Axboe <axboe@fb.com>
Clear QUEUE_FLAG_ADD_RANDOM in all block drivers that set
QUEUE_FLAG_NONROT.
Historically, all block devices have automatically made entropy
contributions. But as previously stated in commit e2e1a148 ("block: add
sysfs knob for turning off disk entropy contributions"):
- On SSD disks, the completion times aren't as random as they
are for rotational drives. So it's questionable whether they
should contribute to the random pool in the first place.
- Calling add_disk_randomness() has a lot of overhead.
There are more reliable sources for randomness than non-rotational block
devices. From a security perspective it is better to err on the side of
caution than to allow entropy contributions from unreliable "random"
sources.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
this is needed for the queue/block device we created (it's done by
blk_cleanup_queue() which we do call) - but calling it for the block devices we
only opened is pointless.
Change-Id: I53dfded14ed15b9581d10ca8399d5e1b3abbf9f2
Since bch_is_open will iterate linked list bch_cache_sets and
uncached_devices, it needs bch_register_lock.
Signed-off-by: Jianjian Huo <samuel.huo@gmail.com>
time_stats::btree_gc_max_duration_mc is not bit shifted by 8
Fixes BUG #138
Change-Id: I44fc6e1d0579674016acc533f1a546b080e5371a
Signed-off-by: Surbhi Palande <sap@daterainc.com>
If register_cache_set() failed, we would touch ca->set after
it had already been freed. Also, fix an assertion to catch
this.
Change-Id: I748e5f5b223e2d9b2602075dec2f997cced2394d
If we goto out_nocoalesce after we free new_nodes[0], we end up freeing
new_nodes[0] again. This was generating a lockdep warning. The fix is
to set new_nodes[0] to NULL, since the out_nocoalesce path safely
ignores NULL entries in the new_nodes array.
This regression was introduced in 2d7f9531.
Change-Id: I76564d7257800583214376b4bacf236cda90c89c
When running with multiple cache devices, if one of the devices has a completely
empty journal but we'd already found some journal entries on a previosu device
we'd go into an infinite loop.
Change-Id: I1dcdc0d738192746de28f40e8b08825b0dea5e2b
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
There's no point in blocking on these allocations, since our fallback paths will
probably go faster than blocking.
Change-Id: I733ca202c25cb36bde02607a0a60552229a4241c
this was very wrong - mempool_alloc() only guarantees success with GFP_WAIT.
bcache uses GFP_NOWAIT in various other places where we have a fallback,
circuits must've gotten crossed when writing this code or something.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
There were two issues here:
- writeback thread did not start until the device first became dirty
- writeback thread used uninterruptible sleep once running
Without this patch I see kernel warnings printed and a load average of
1.52 after booting my test VM. With this patch the warnings are gone and
the load average is near 0.00 as expected.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Tested:
- sometimes bcache_tier test would hang on startup with a failure
to allocate the btree root -- no longer seeing this
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
while loop was executing infinitely.
This fix ends the while loop gracefully.
Signed-off-by: Surbhi Palande <sap@daterainc.com>
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
After detaching a backing device from a cache set, a bit wasn't getting
reset meaning the second detach wouldn't work correctly.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Uninlined nested functions can cause crashes when using ftrace, as they don't
follow the normal calling convention and confuse the ftrace function graph
tracer as it examines the stack.
Also, nested functions are supported as a gcc extension, but may fail on other
compilers (e.g. llvm).
Signed-off-by: John Sheu <john.sheu@gmail.com>
gc_gen was a temporary used to recalculate last_gc, but since we only need
bucket->last_gc when gc isn't running (gc_mark_valid = 1), we can just update
last_gc directly.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This was originally added as at optimization that for various reasons isn't
needed anymore, but it does add a lot of nasty corner cases (and it was
responsible for some recently fixed bugs). Just get rid of it now.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This changes the bucket allocation reserves to use _real_ reserves - separate
freelists - instead of watermarks, which if nothing else makes the current code
saner to reason about and is going to be important in the future when we add
support for multiple btrees.
It also adds btree_check_reserve(), which checks (and locks) the reserves for
both bucket allocation and memory allocation for btree nodes; the old code just
kinda sorta assumed that since (e.g. for btree node splits) it had the root
locked and that meant no other threads could try to make use of the same
reserve; this technically should have been ok for memory allocation (we should
always have a reserve for memory allocation (the btree node cache is used as a
reserve and we preallocate it)), but multiple btrees will mean that locking the
root won't be sufficient anymore, and for the bucket allocation reserve it was
technically possible for the old code to deadlock.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
With the locking rework in the last patch, this shouldn't be needed anymore -
btree_node_write_work() only takes b->write_lock which is never held for very
long.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Add a new lock, b->write_lock, which is required to actually modify - or write -
a btree node; this lock is only held for short durations.
This means we can write out a btree node without taking b->lock, which _is_ held
for long durations - solving a deadlock when btree_flush_write() (from the
journalling code) is called with a btree node locked.
Right now just occurs in bch_btree_set_root(), but with an upcoming journalling
rework is going to happen a lot more.
This also turns b->lock is now more of a read/intent lock instead of a
read/write lock - but not completely, since it still blocks readers. May turn it
into a real intent lock at some point in the future.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This isn't a bulletproof fix; btree_node_free() -> bch_bucket_free() puts the
bucket on the unused freelist, where it can be reused right away without any
ordering requirements. It would be better to wait on at least a journal write to
go down before reusing the bucket. bch_btree_set_root() does this, and inserting
into non leaf nodes is completely synchronous so we should be ok, but future
patches are just going to get rid of the unused freelist - it was needed in the
past for various reasons but shouldn't be anymore.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This means the garbage collection code can better check for data and metadata
pointers to the same buckets.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This will potentially save us an allocation when we've got inode/dirent bkeys
that don't fit in the keylist's inline keys.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Change the invalidate tracepoint to indicate how much data we're invalidating,
and change the alloc tracepoints to indicate what offset they're for.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Deadlock happened because a foreground write slept, waiting for a bucket
to be allocated. Normally the gc would mark buckets available for invalidation.
But the moving_gc was stuck waiting for outstanding writes to complete.
These writes used the bcache_wq, the same queue foreground writes used.
This fix gives moving_gc its own work queue, so it was still finish moving
even if foreground writes are stuck waiting for allocation. It also makes
work queue a parameter to the data_insert path, so moving_gc can use its
workqueue for writes.
Signed-off-by: Nicholas Swenson <nks@daterainc.com>
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
The on disk bucket gens are allowed to be out of date, when we reuse buckets
that didn't have any live data in them. To deal with this, the initial gc has to
update the bucket gen when we find a pointer gen newer than the bucket's gen.
Unfortunately we weren't doing this for pointers in the journal that we're about
to replay.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
On recovery we weren't correctly keeping track of what journal buckets had open
journal entries, thus it was possible for them to be overwritten until we'd
written all new journal entries.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
The code was using sectors to count the number of sectors it was zeroing... but
then it passed it to bio_advance()... after it had been set to 0. Amusing...
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
