This adds a new method of doing btree updates - a straight write buffer,
implemented as a flat fixed size array.
This is only useful when we don't need to read from the btree in order
to do the update, and when reading is infrequent - perfect for the LRU
btree.
This will make LRU btree updates fast enough that we'll be able to use
it for persistently indexing buckets by fragmentation, which will be a
massive boost to copygc performance.
Changes:
- A new btree_insert_type enum, for btree_insert_entries. Specifies
btree, btree key cache, or btree write buffer.
- bch2_trans_update_buffered(): updates via the btree write buffer
don't need a btree path, so we need a new update path.
- Transaction commit path changes:
The update to the btree write buffer both mutates global, and can
fail if there isn't currently room. Therefore we do all write buffer
updates in the transaction all at once, and also if it fails we have
to revert filesystem usage counter changes.
If there isn't room we flush the write buffer in the transaction
commit error path and retry.
- A new persistent option, for specifying the number of entries in the
write buffer.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Recursive transaction commits are occasionally necessary - in
particular, for the upcoming btree write buffer's flush path.
This avoids bugs due to trans->flags being accidentally mutated
mid-commit, which can cause c->writes refcount leaks.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This patch
- Adds a mechanism for queuing up journal entries prior to the journal
being started, which will be used for early journal log messages
- Adds bch2_fs_log_msg() and improves bch2_trans_log_msg(), which now
take format strings. bch2_fs_log_msg() can be used before or after
the journal has been started, and will use the appropriate mechanism.
- Deletes the now obsolete bch2_journal_log_msg()
- And adds more log messages to the recovery path - messages for
journal/filesystem started, journal entries being blacklisted, and
journal replay starting/finishing.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Now that we have an error path plumbed through, there's no need to be
using bch2_btree_node_lock_write_nofail().
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Now that we have error codes, with subtypes, we can switch to our own
error code for transaction restarts - and even better, a distinct error
code for each transaction restart reason: clearer code and better
debugging.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This will help us improve nested transactions - we need to add
assertions that whenever an inner transaction handles a restart, it
still returns -EINTR to the outer transaction.
This also adds nested_lockrestart_do() and nested_commit_do() which use
the new counters to correctly return -EINTR when the transaction was
restarted.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This adds a new helper, bch2_trans_run(), that runs a function with a
btree_transaction context but without handling transaction restarts.
We're adding checks for nested transaction restart handling: when an
inner transaction handles a transaction restart it will still have to
return it to the outer transaction, or else assertions will be popped in
the outer transaction.
But some places don't need restart handling at the outer scope, so this
helper does what they need.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Better/more descriptive naming, and prep for adding
nested_lockrestart_do() and nested_commit_do().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Since journal reclaim -> btree key cache flushing may require the
allocation of new btree nodes, it has an implicit dependency on copygc
in order to make forward progress - so we should avoid blocking copygc
unless the journal is really close to full.
This introduces watermarks to replace our single MAY_GET_UNRESERVED bit
in the journal, and adds a watermark for copygc and plumbs it through.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Add a new helper for logging messages to the journal - a new debugging
tool, an alternative to trace_printk().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
- Updates to non key cache iterators will now be transparently
redirected to the key cache for cached btrees.
- Except when creating new keys: then the update goes to underlying
btree
For for iterating over a cached btree to work, we need to ensure that if
a key exists in the key cache, it also exists in the btree - otherwise
the iterator code will skip past it and not check the key cache.
Otherwise, for consistency, all updates should go to the same place -
the key cache.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
More prep work for getting rid of the in-memory bucket array: now that
we have BTREE_ITER_WITH_JOURNAL, the allocator code can do ntree lookups
before journal replay is finished, and there's no longer any need for it
to get allocation information from the in-memory bucket array.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
With BTREE_ITER_FILTER_SNAPSHOTS, we have to distinguish between the
path where the key was found, and the path for inserting into the
current snapshot. This adds a new field to struct btree_iter for saving
a path for the current snapshot, and plumbs it through
bch2_trans_update().
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
With snapshots, bch2_trans_update() has to check if we need a whitout,
which can cause a transaction restart, so this is important now.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
We have been getting away from handling transaction restarts locally -
convert bch2_btree_node_rewrite() to the newer style.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Now that peek_node()/next_node() are converted to return errors
directly, we don't need bch2_trans_exit() to return errors - it's
cleaner this way and wasn't used much anymore.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This updates the fsck algorithms to handle snapshots - meaning there
will be multiple versions of the same key (extents, inodes, dirents,
xattrs) in different snapshots, and we have to carefully consider which
keys are visible in which snapshot.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This splits btree_iter into two components: btree_iter is now the
externally visible componont, and it points to a btree_path which is now
reference counted.
This means we no longer have to clone iterators up front if they might
be mutated - btree_path can be shared by multiple iterators, and cloned
if an iterator would mutate a shared btree_path. This will help us use
iterators more efficiently, as well as slimming down the main long lived
state in btree_trans, and significantly cleans up the logic for iterator
lifetimes.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
With the recent transaction restart changes, it's no longer needed - all
transaction commits have BTREE_INSERT_NOUNLOCK semantics.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Upcoming patch will require that a transaction restart is always
immediately followed by bch2_trans_begin().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This closes a significant hole (and last known hole) in our ability to
verify metadata. Previously, since btree nodes are log structured, we
couldn't detect lost btree writes that weren't the first write to a
given node. Additionally, this seems to have lead to some significant
metadata corruption on multi device filesystems with metadata
replication: since a write may have made it to one device and not
another, if we read that btree node back from the replica that did have
that write and started appending after that point, the other replica
would have a gap in the bset entries and reading from that replica
wouldn't find the rest of the bsets.
But, since updates to interior btree nodes are now journalled, we can
close this hole by updating pointers to btree nodes after every write
with the currently written number of sectors, without negatively
affecting performance. This means we will always detect lost or corrupt
metadata - it also means that our btree is now a curious hybrid of COW
and non COW btrees, with all the benefits of both (excluding
complexity).
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
btree_trans should always be passed when we have one - iter->trans is
disfavoured. This mainly updates old code in btree_update_interior.c,
some of which predates btree_trans.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Add a new flag to control assertions about updating to internal snapshot
nodes, that normally should not be written to - to be used in an
upcoming patch.
Also do some renaming - trigger_flags is now update_flags.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Now that extent handling has been lifted to bch2_trans_update(), we
don't need to keep two different lists of updates.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
This patch adds self healing functionality for btree nodes - if we
notice a problem when reading a btree node, we just rewrite it.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Now that we have inode backpointers, we can simplify checking directory
structure: instead of doing a DFS from the filesystem root and then
checking if we found everything, we can iterate over every inode and see
if we can go up until we get to the root.
This patch also has a number of fixes and simplifications for the inode
backpointer checks. Also, it turns out we don't actually need the
BCH_INODE_BACKPTR_UNTRUSTED flag.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Previously, we were using BTREE_INSERT_RESERVE in a lot of places where
it no longer makes sense.
- we now have more open_buckets than we used to, and the reserves work
better, so we shouldn't need to use BTREE_INSERT_RESERVE just because
we're holding open_buckets pinned anymore.
- We have the btree key cache for updates to the alloc btree, meaning
we no longer need the btree reserve to ensure the allocator can make
forward progress.
This means that we should only need a reserve for btree updates to
ensure that copygc can make forward progress.
Since it's now just for copygc, we can also fold RESERVE_BTREE into
RESERVE_MOVINGGC (the allocator's freelist reserve).
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This helps reduce stack usage by avoiding multiple btree_trans on the
stack.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This introduces a new kind of btree iterator, cached iterators, which
point to keys cached in a hash table. The cache also acts as a write
cache - in the update path, we journal the update but defer updating the
btree until the cached entry is flushed by journal reclaim.
Cache coherency is for now up to the users to handle, which isn't ideal
but should be good enough for now.
These new iterators will be used for updating inodes and alloc info (the
alloc and stripes btrees).
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Previously, the btree has always been self contained and internally
consistent on disk without anything from the journal - the journal just
contained pointers to the btree roots.
However, this meant that btree node split or compact operations - i.e.
anything that changes btree node topology and involves updates to
interior nodes - would require that interior btree node to be written
immediately, which means emitting a btree node write that's mostly empty
(using 4k of space on disk if the filesystemm blocksize is 4k to only
write perhaps ~100 bytes of new keys).
More importantly, this meant most btree node writes had to be FUA, and
consumer drives have a history of slow and/or buggy FUA support - other
filesystes have been bit by this.
This patch changes the interior btree update path to journal updates to
interior nodes, after the writes for the new btree nodes have completed.
Best of all, it turns out to simplify the interior node update path
somewhat.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Ever since the btree code was first written, handling of overwriting
existing extents - including partially overwriting and splittin existing
extents - was handled as part of the core btree insert path. The modern
transaction and iterator infrastructure didn't exist then, so that was
the only way for it to be done.
This patch moves that outside of the core btree code to a pass that runs
at transaction commit time.
This is a significant simplification to the btree code and overall
reduction in code size, but more importantly it gets us much closer to
the core btree code being completely independent of extents and is
important prep work for snapshots.
This introduces a new feature bit; the old and new extent update models
are incompatible when the filesystem needs journal replay.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
All iterators should be released now with bch2_trans_iter_put(), so
TRANS_RESET_ITERS shouldn't be needed anymore, and TRANS_RESET_MEM is
always used.
Also convert more code to __bch2_trans_do().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This is partly prep work for introducing bch_btree_ptr_v2, but it'll
also be a bit of a performance boost by moving the full key out of the
hot part of struct btree.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Previously, when doing multiple update in the same transaction commit
that overwrote each other, we relied on doing the updates in the same
order as the bch2_trans_update() calls in order to get the correct
result. But that wasn't correct for triggers; bch2_trans_mark_update()
when marking overwrites would do the wrong thing because it hadn't seen
the update that was being overwritten.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The trigger flags really belong with individual btree_insert_entries,
not the transaction commit flags - this splits out those flags and
unifies them with the BCH_BUCKET_MARK flags. Todo - split out
btree_trigger.c from buckets.c
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This should be private to btree_update_leaf.c, and we might end up
removing it.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This was causing a bug with transaction iterators overflowing; now, if
triggers have to be reexecuted we always return -EINTR and retry from
the start of the transaction.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
BTREE_INSERT_ATOMIC should really be the default mode, and there's not
that much code that doesn't need it - so this is prep work for getting
rid of the flag.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The btree_trans struct needs to memoize/cache btree iterators, so that
on transaction restart we don't have to completely redo btree lookups,
and so that we can do them all at once in the correct order when the
transaction had to restart to avoid a deadlock.
This switches the btree iterator lookups to work based on iterator
position, instead of trying to match them up based on the stack trace.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
