On failure to get a write lock (because we had a conflicting read lock),
we need to make sure to upgrade the read lock to an intent lock - or we
could end up spinning.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

The bch2_btree_path_upgrade() call was failing and tripping an assert -
path->level + 1 is in this case not necessarily exactly what we want,
fix it by upgrading exactly the locks we want.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This allows triggers to distinguish between a key entering the btree -
i.e. being called from the trans commit path - vs. being called on a key
that already exists, i.e. by GC.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This helps to unify the interface between bch2_mark_key() and
bch2_trans_mark_key() - and it also gives access to the journal
reservation and journal seq in the mark_key path.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

- The backpointer that ec_stripe_update_ptrs() uses now needs to include
  the snapshot ID, which means we have to change where we add the
  backpointer to after getting the snapshot ID for the new extents

- ec_stripe_update_ptrs() needs to be calling bch2_trans_begin()

- improve error message in bch2_mark_stripe()
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

When the old or new key doesn't exist, we should still pass in a deleted
key with the correct pos. This fixes a bug in the ec code, when
bch2_mark_stripe() was looking up the wrong in-memory stripe.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This tweaks the journal code to always act as if there's space available
in nochanges mode, when we're not going to be doing any writes. This
helps in recovering filesystems that won't mount because they need
journal replay and the journal has gotten stuck.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

The fsck code has been handling transaction restarts locally, to avoid
calling fsck_err() multiple times (and asking the user/logging the error
multiple times) on transaction restart.

However, with our improving assertions about iterator validity, this
isn't working anymore - the code wasn't entirely correct, in ways that
are fine for now but are going to matter once we start wanting online
fsck.

This code converts much of the fsck code to handle transaction restarts
in a more rigorously correct way - moving restart handling up to the top
level of check_dirent, check_xattr and others - at the cost of logging
errors multiple times on transaction restart.

Fixing the issues with logging errors multiple times is probably going
to require memoizing calls to fsck_err() - we'll leave that for future
improvements.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Was popping an assertion on !BTREE_ITER_ALL_SNAPSHOTS iters when getting
to the end.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This fixes building in userspace - code that's coupled to the kernel VFS
interface should live in fs.c
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

nochanges mode is often used for getting data off of otherwise
nonrecoverable filesystems, which is often because of errors hit during
fsck.

Don't force version upgrade & fsck in nochanges mode, so that it's more
likely to mount.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Back when we relied on the journal sequence number blacklist machinery
for consistency between btree and the journal, we needed to ensure a new
journal entry was written before any btree writes were done. But, this
had the side effect of consuming some space in the journal prior to
doing journal replay - which could lead to a very wedged filesystem,
since we don't yet have a way to grow the journal prior to going RW.

Fortunately, the journal sequence number blacklist machinery isn't
needed anymore, as btree node pointers now record the numer of sectors
currently written to that node - that code should all be ripped out.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Snapshot deletion needs to become a multi step process, where we unlink,
then tear down the page cache, then delete the subvolume - the deleting
flag is equivalent to an inode with i_nlink = 0.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

We should always print out the key we were marking.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

It seems some users have reflink pointers which span many indirect
extents, from a short window in time when merging of reflink pointers
was allowed.

Now, we're seeing transaction path overflows in fix_reflink_p(), the
code path to clear out the reflink_p fields now used for front/back pad
- but, we don't actually need to be running triggers in that path, which
is an easy partial fix.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

for_each_btree_key() now calls bch2_trans_begin() as needed; that means,
we can also call it when we're in danger of overflowing transaction
paths.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

The way __bch2_mark_reflink_p returns errors was clashing with returning
the number of sectors processed - we weren't returning FSCK_ERR_EXIT
correctly.

Fix this by only using the return code for errors, which actually ends
up simplifying the overall logic.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This fixes a possible race where we fail to remove a device because of
btree nodes still on it, that are being deleted by in flight btree
updates.
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>

We were modifying state, then return -EINTR, causing us to skip nodes -
ouch.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

But we don't need to call it from outside the btree iterator code
anymore, since it's called by bch2_trans_begin() and
bch2_btree_path_traverse().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This is a hacky but effective fix to device usage stats for superblock
and journal being wrong on a newly added device (following the comment
that already told us how it needed to be done!)
Reported-by: Chris Webb <chris@arachsys.com>
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

readdir() in a directory with many subvolumes could overflow transaction
paths - this is a simple hack around the issue.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

The back merging case wasn't returning errors correctly.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This changes the on disk format for dirents that point to subvols so
that they also record the subvolid of the parent subvol, so that we can
filter them out in other subvolumes.

This also updates the dirent code to do that filtering, and in
particular tweaks the rename code - we need to ensure that there's only
ever one dirent (counting multiplicities in different snapshots) that
point to a subvolume.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Code that uses for_each_btree_key often wants transaction restarts to be
handled locally and not returned. Originally, we wouldn't return
transaction restarts if there was a single iterator in the transaction -
the reasoning being if there weren't other iterators being invalidated,
and the current iterator was being advanced/retraversed, there weren't
any locks or iterators we were required to preserve.

But with the btree_path conversion that approach doesn't work anymore -
even when we're using for_each_btree_key() with a single iterator there
will still be two paths in the transaction, since we now always preserve
the path at the pos the iterator was initialized at - the reason being
that on restart we often restart from the same place.

And it turns out there's now a lot of for_each_btree_key() uses that _do
not_ want transaction restarts handled locally, and should be returning
them.

This patch splits out for_each_btree_key_norestart() and
for_each_btree_key_continue_norestart(), and converts existing users as
appropriate. for_each_btree_key(), for_each_btree_key_continue(), and
for_each_btree_node() now handle transaction restarts themselves by
calling bch2_trans_begin() when necessary - and the old hack to not
return transaction restarts when there's a single path in the
transaction has been deleted.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

It's not an error if we don't have cached data - skip BCH_DATA_cached in
bch2_have_enough_devs().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

This fixes a bug where subsequently doing creates with the same name
fails.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

check_path() wasn't checking the snapshot ID when checking for directory
structure loops - so, snapshots would cause us to detect a loop that
wasn't actually a loop.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

When a reflink pointer points to only part of an indirect extent, and
then that indirect extent is fragmented (e.g. by copygc), if the reflink
pointer only points to one of the fragments we leak a reference.

Fix this by storing front/back pad values in reflink pointers - when
inserting reflink pointesr, we initialize them to cover the full range
of the indirect extents we reference.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

We had a bug where reflink_p pointers weren't being initialized to 0,
and when we started using the second word, things broke badly.

This patch revs the on disk format version and adds cleanup code to zero
out the second word of reflink_p pointers before we start using it.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

It shouldn't be necessary when we're only using a single iterator and
not doing updates, but that's harder to debug at the moment.
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 changes for_each_btree_node() to work like for_each_btree_key(),
and to that end bch2_btree_iter_peek_node() and next_node() also return
error ptrs.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

When a reflink pointer points to an indirect extent that doesn't exist,
we need to replace it with a KEY_TYPE_error key.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Checking of directory structure across subvolumes was broken - we need
to look up the snapshot ID of the parent subvolume when crossing subvol
boundaries.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Subvolume deletion doesn't flush & evict the btree key cache - ideally
it would, but that's tricky, so instead bch2_subvolume_create() needs to
make sure the slot doesn't exist in the key cache to avoid triggering
assertions.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

Type size_t is architecture-specific. Fix warnings for some non-amd64
arches.
Signed-off-by: Brett Holman <bholman.devel@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This bug was only discovered when we started using the 2nd word in the
val, which should have been zeroed out as those fields had never been
used before - ouch.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>