The early-early allocation path, bch2_bucket_alloc_new_fs(), is no
longer needed - and inconsistencies around new_fs_bucket_idx have been a
frequent source of bugs.
Reported-by: syzbot+592425844580a6598410@syzkaller.appspotmail.com
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
btree_root entries for unknown btree IDs are created during recovery,
before reading those btree roots.
But btree_node_scan may find btree nodes with unknown btree IDs when we
haven't seen roots for those btrees.
Reported-by: syzbot+1f202d4da221ec6ebf8e@syzkaller.appspotmail.com
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
If we rewind recovery to run topology repair, that causes
accounting_read to run twice.
This fixes accounting being double counted.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Accounting keys that reference invalid devices are corrected by fsck,
they shouldn't cause an emergency shutdown.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Instead of throwing standard error codes, we should be throwing
dedicated private error codes, this greatly improves debugability.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The discard bucket fastpath previously was using its own code for
discarding buckets and clearing them in the need_discard btree, which
didn't have any of the consistency checks of the main discard path.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Per reports of performance issues on mixed multi device filesystems
where we're issuing too much IO to the spinning rust - tweak this
algorithm.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
- bch2_backpointer_del()
- bch2_backpointer_maybe_flush()
Kill a bit of open coding and make sure we're properly handling the
btree write buffer.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bch_backpointer.bucket_offset is going away - it's no longer needed
since we no longer store backpointers in alloc keys, the same
information is in the key position itself.
And we'll be reclaiming the space in bch_backpointer for the bucket
generation number.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bch2_get_btree_in_memory_pos() will return positions that refer directly
to the btree it's checking will fit in memory - i.e. backpointer
positions, not buckets.
This also means check_bp_exists() no longer has to refer to the device,
and we can delete some code.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Since we no longer store backpointers in alloc keys, there's no reason
not to pass around bkey_i_backpointers; this means we don't have to pass
the bucket pos separately.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
_noerror means don't produce inconsistent errors, so it should be using
bch2_dev_rcu_noerror().
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
86a494c8ee ("bcachefs: Kill bch2_get_next_backpointer()") dropped some
things the tracepoint emitted because bch2_evacuate_bucket() no longer
looks at the alloc key - but we did want at least some of that.
We still no longer look at the alloc key so we can't report on the
fragmentation number, but that's a direct function of dirty_sectors and
a copygc concern anyways - copygc should get its own tracepoint that
includes information from the fragmentation LRU.
But we can report on the number of sectors we moved and the bucket size.
Co-developed-by: Piotr Zalewski <pZ010001011111@proton.me>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The journal_keys array can't be substantially modified after we go RW,
because lookups need to be able to check it locklessly - thus we're
limited on what we can do when a key in the journal has been
overwritten.
This is a problem when there's many overwrites to skip over for peek()
operations. To fix this, add tracking of ranges of overwrites: we create
a range entry when there's more than one contiguous whiteout.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
To help ameloriate issues with peek operations having to skip over
deletions in the journal - just bail out if all we're doing is
prefetching btree nodes.
Since btree node prefetching runs every time we iterate to a new node,
and has to sequentially scan ahead, this avoids another O(n^2).
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
There's an unavoidable issue with btree lookups when we're overlaying
journal keys and the journal has many deletions for keys present in the
btree - peek operations will have to iterate over all those deletions to
find the next live key to return.
This is mainly a problem for lookups in interior nodes, if we have to
traverse to a leaf. Looking up an insert position in a leaf (for journal
replay) doesn't have to find the next live key, but walking down the
btree does.
So to ameloriate this, change journal key sort ordering so that we
replay keys from roots and interior nodes first.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
We don't allocate the mempools for compression/decompression unless we
need them - but that means there's an inconsistency to check for.
Reported-by: syzbot+cb3fbcfb417448cfd278@syzkaller.appspotmail.com
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
gzip and zstd require different decompress workspace sizes, and if we
start with one and then start using the other at runtime we may not get
the correct size
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
type includes lz4 and lz4_old, which do not get different compression
workspaces, and incompressible, a fake type - BCH_COMPRESSION_OPTS() is
the correct enum to use.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Since for quite some time backpointers have only been stored in the
backpointers btree, not alloc keys (an aborted experiment, support for
which has been removed) - we can replace get_next_backpointer() with
simple btree iteration.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
try_alloc_bucket() has a "safety" check, which avoids allocating a
bucket if there's any backpointers present.
But backpointers are not the source of truth for live data in a bucket,
the bucket sector counts are; this check was fairly useless, and we're
also deferring backpointers checks from fsck to runtime in the near
future.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
With extents and snapshots, for slightly different reasons, we may have
to search forwards to find a key that compares equal to iter->pos (i.e.
a key that peek_prev() should return, as it returns keys <= iter->pos).
peek_slot() does this, and is an easy way to fix this case.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
A user contributed a filessytem dump, where the dump was actually
corrupted (due to being taken while the filesystem was online), but
which exposed an interesting bug in fsck - reconstruct_inode().
When itearting in BTREE_ITER_filter_snapshots mode, it's required to
give an end position for the iteration and it can't span inode numbers;
continuing into the next inode might mean we start seeing keys from a
different snapshot tree, that the is_ancestor() checks always filter,
thus we're never able to return a key and stop iterating.
Backwards iteration never implemented the end position because nothing
else needed it - except for reconstuct_inode().
Additionally, backwards iteration is now able to overlay keys from the
journal, which will be useful if we ever decide to start doing journal
replay in the background.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Factor out a common helper, need_discard_or_freespace_err(), which is
now used by both fsck and the runtime checks, and can repair.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
check_discard_freespace_key() was doing all the same checks as
try_alloc_bucket(), but with repair.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Change it to a normal fsck_err() - meaning it'll get repaired at runtime
when that's flipped on.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
To avoid tragic loss in the event of transient errors (i.e., a btree
node topology error that was later corrected by btree node scan), we
can't delete reflink pointers to correct errors.
This adds a new error bit to bch_reflink_p, indicating that it is known
to point to a missing indirect extent, and the error has already been
reported.
Indirect extent lookups now use bch2_lookup_indirect_extent(), which on
error reports it as a fsck_err() and sets the error bit, and clears it
if necessary on succesful lookup.
This also gets rid of the bch2_inconsistent_error() call in
__bch2_read_indirect_extent, and in the reflink_p trigger: part of the
online self healing project.
An on disk format change isn't necessary here: setting the error bit
will be interpreted by older versions as pointing to a different index,
which will also be missing - which is fine.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Better repair for reflink pointers, as well as propagating new inode
options to indirect extents, are going to require a few extra bits
bch_reflink_p: so claim a few from the high end of the destination
index.
Also add some missing bounds checking.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
If we find an error that indicates that we need to run fsck, we can
specify that directly with run_explicit_recovery_pass().
These are now log_fsck_err() calls: we're just logging in the superblock
that an error occurred - and possibly doing an emergency shutdown,
depending on policy.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
alloc key validation ensures that if a bucket is in need_discard state
the sector counts are all zero - we don't have to check for that.
The NEED_INC_GEN check appears to be dead code, as well: we only see
buckets in the need_discard btree, and it's an error if they aren't in
the need_discard state.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
