Fix another deadlock related to the merge path; previously, we switched
to always running merges at a lower watermark (because they are
noncritical); but when we run at a lower watermark we also need to run
nonblocking or we've introduced a new deadlock.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Reported-and-tested-by: s@m-h.ug

discard_new_inode() is the wrong interface to use when we need to free
an inode that was never inserted into the inode hash table; we can
bypass the whole iput() -> evict() path and replace it with
__destroy_inode(); kmem_cache_free() - this fixes a WARN_ON() about
I_NEW.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

bch2_journal_write() was incorrectly waiting on earlier journal writes
synchronously; this usually worked because most of the time we'd be
running in the context of a thread that did a journal_buf_put(), but
sometimes we'd be running out of the same workqueue that completes those
prior journal writes.

Additionally, this makes sure to punt to a workqueue before submitting
preflushes - we really don't want to be calling submit_bio() in the main
transaction commit path.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Freeing key cache items is a multi stage process; we need to wait for an
SRCU grace period to elapse, and we handle this ourselves - partially to
avoid callback overhead, but primarily so that when allocating we can
first allocate from the freed items waiting for an SRCU grace period.

Previously, the shrinker was counting the items on the 'waiting for SRCU
grace period' lists as items being scanned, but this meant that too many
items waiting for an SRCU grace period could prevent it from doing any
work at all.

After this, we're seeing that items skipped due to the accessed bit are
the main cause of the shrinker not making any progress, and we actually
want the key cache shrinker to run quite aggressively because reclaimed
items will still generally be found (more compactly) in the btree node
cache - so we also tweak the shrinker to not count those against
nr_to_scan.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

this stores the SRCU sequence number, which we use to check if an SRCU
barrier has elapsed; this is a partial fix for the key cache shrinker
not actually freeing.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Fix a missing bounds check in superblock validation.

Note that we don't yet have repair code for this case - repair code for
individual items is generally low priority, since the whole superblock
is checksummed, validated prior to write, and we have backups.
Reported-by: lei lu <llfamsec@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

if the buffer is virtually mapped it won't be a single bvec
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

KEY_TYPE_error is left behind when we have to delete all pointers in an
extent in fsck; it allows errors to be correctly returned by reads
later.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Fixes: 27c15ed2 bcachefs: bch_member.btree_allocated_bitmap
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This fixes a deadlock when journal replay has many keys to insert that
were from fsck, not the journal.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Interior nodes are not really needed, when we have to scan - but if this
pops up for leaf nodes we'll need a real heuristic.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

When building for 32-bit platforms, for which size_t is 'unsigned int',
there is a warning from a format string in validate_bset_keys():

  fs/bcachefs/btree_io.c: In function 'validate_bset_keys':
  fs/bcachefs/btree_io.c:891:34: error: format '%lu' expects argument of type 'long unsigned int', but argument 12 has type 'unsigned int' [-Werror=format=]
    891 |                                  "bad k->u64s %u (min %u max %lu)", k->u64s,
        |                                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  fs/bcachefs/btree_io.c:603:32: note: in definition of macro 'btree_err'
    603 |                                msg, ##__VA_ARGS__);                     \
        |                                ^~~
  fs/bcachefs/btree_io.c:887:21: note: in expansion of macro 'btree_err_on'
    887 |                 if (btree_err_on(!bkeyp_u64s_valid(&b->format, k),
        |                     ^~~~~~~~~~~~
  fs/bcachefs/btree_io.c:891:64: note: format string is defined here
    891 |                                  "bad k->u64s %u (min %u max %lu)", k->u64s,
        |                                                              ~~^
        |                                                                |
        |                                                                long unsigned int
        |                                                              %u
  cc1: all warnings being treated as errors

BKEY_U64s is size_t so the entire expression is promoted to size_t. Use
the '%zu' specifier so that there is no warning regardless of the width
of size_t.

Fixes: 031ad9e7 ("bcachefs: Check for packed bkeys that are too big")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404130747.wH6Dd23p-lkp@intel.com/
Closes: https://lore.kernel.org/oe-kbuild-all/202404131536.HdAMBOVc-lkp@intel.com/Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

We need to initialize the stdio redirects before they're used.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This is part of a larger series cleaning up the semantics of
should_be_locked and adding assertions around it; if we don't need an
iterator/path anymore, it clearly doesn't need to be locked.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

In __bch2_read_super(), if kstrdup() fails, it needs to release memory
in sb->holder, fix to call bch2_free_super() in the error path.
Signed-off-by: Chao Yu <chao@kernel.org>
Reviewed-by: Hongbo Li <lihongbo22@huawei.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This adds a small (64 bit) per-device bitmap that tracks ranges that
have btree nodes, for accelerating btree node scan if it is ever needed.

- New helpers, bch2_dev_btree_bitmap_marked() and
  bch2_dev_bitmap_mark(), for checking and updating the bitmap

- Interior btree update path updates the bitmaps when required

- The check_allocations pass has a new fsck_err check,
  btree_bitmap_not_marked

- New on disk format version, mi_btree_mitmap, which indicates the new
  bitmap is present

- Upgrade table lists the required recovery pass and expected fsck error

- Btree node scan uses the bitmap to skip ranges if we're on the new
  version
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Add a sysfs knob for immediately flushing the entire journal.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

We shouldn't be doing the unlock/relock dance when we're not using a
path - this fixes an assertion pop when called from btree node scan.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

For forwards compatibilyt, we allow bkeys of unknown type in leaf nodes;
we can simply ignore metadata we don't understand. Pointers to btree
nodes must always be of known types, howwever.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

The btree write buffer flush fastpath that avoids the main transaction
commit path had the unfortunate side effect of not doing btree node
merging.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

There's been a bug in the btree write buffer where it wasn't triggering
btree node merges - and leaving behind a bunch of nearly empty btree
nodes.

Then during journal replay, when updates to the backpointers btree
aren't using the btree write buffer (because we require synchronization
with journal replay), we end up doing those merges all at once.

Then if it's the interior update path running them, we deadlock because
those run with the highest watermark.

There's no real need for the interior update path to be doing btree node
merges; other code paths can handle that at lower watermarks.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This fixes a deadlock where the interior update path during journal
replay ends up doing a ton of merges on the backpointers btree, and
deadlocking.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

bch2_journal_flush_seq requires us to have a write ref
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

btree_key_can_insert_cached() should be checking the watermark -
BCH_TRANS_COMMIT_journal_replay really means nonblocking mode when
watermark < reclaim, it was being used incorrectly.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This fixes a bug where we fail to start when upgrading/downgrading
because we forgot we needed to go rw.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

many nodes found during scan will be old nodes, overwritten by newer
nodes
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

.to_text() functions need to work on key values that didn't pass .valid
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Need to check if we have a valid bucket before checking if ptr is stale
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

add missing validation; fixes assertion pop in bkey unpack
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

The btree paths array is now dynamically resizable - and as well the
btree_insert_entries array, as it needs to be the same size.

The merge path (and interior update path) allocates new btree paths,
thus can trigger a resize; thus we need to not retain direct pointers
after invoking merge; similarly when running btree node triggers.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

It turns out - btree splits happen with the rest of the transaction
still locked, to avoid unnecessary restarts, which means using nofail
doesn't work here - we can deadlock.

Fortunately, we now have the ability to return errors here.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

We weren't respecting trans->journal_replay_not_finished - we shouldn't
be searching the journal keys unless we have a ref on them.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>