Now explicitly allocate and free the buckets_nouse bitmap - this is
going to be used for online fsck.

To go RW when we haven't check allocations, we'll do a much slimmed down
version that just initializes the buckets_nouse bitmaps.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

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

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

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

by using bucket_m_to_alloc() more, we can get some nice code cleanup.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

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

this was from metadata only gc - we don't need it anymore
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

It should be FS_IOC32_GETFLAGS instead of FS_IOC_GETFLAGS in
compat ioctl.
Signed-off-by: Youling Tang <tangyouling@kylinos.cn>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

In bch2_link_trans(), if bch2_inode_nlink_inc() fails, it needs to
call bch2_trans_iter_exit() in the error path.
Signed-off-by: Youling Tang <tangyouling@kylinos.cn>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

The vfs[1] documentation describes free_inode as follows:
```
free_inode
    this method is called from RCU callback. If you use call_rcu()
    in ->destroy_inode to free ‘struct inode’ memory, then it’s
    better to release memory in this method.
```

free_inode will be called by the RCU callback, so it might be better
to move the inode free operation to destroy_inode.

Similar to commit ae6b47b5 ("fs/ntfs3: Change destroy_inode to
free_inode").

Link:
[1]: https://www.kernel.org/doc/html/latest/filesystems/vfs.htmlSigned-off-by: Youling Tang <tangyouling@kylinos.cn>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

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

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

On recovery from clean shutdown we don't typically read the journal, but
we still want to avoid overwriting existing entries in the journal for
list_journal debugging.

Thus, add some fields to the member info section so we can remember
where we left off.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

When compiling the bcachefs-tools, the following compilation warning
is reported:
    libbcachefs/snapshot.c: In function ‘bch2_reconstruct_snapshots’:
    libbcachefs/snapshot.c:915:19: warning: ‘tree_id’ may be used uninitialized in this function [-Wmaybe-uninitialized]
      915 |  snapshot->v.tree = cpu_to_le32(tree_id);
    libbcachefs/snapshot.c:903:6: note: ‘tree_id’ was declared here
      903 |  u32 tree_id;
       |      ^~~~~~~

This is a false alert, because @tree_id is changed in
bch2_snapshot_tree_create after it returns 0. And if this function
returns other value, @tree_id wouldn't be used. Thus there should
be nothing wrong in logical.

Although the report itself is a false alert, we can still make it more
explicit by setting the initial value of @tree_id to 0 (an invalid
tree ID).

Fixes: a292be3b ("bcachefs: Reconstruct missing snapshot nodes")
Signed-off-by: Hongbo Li <lihongbo22@huawei.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

When building for a 32-bit target, for which 'size_t' is 'unsigned int',
there are two warnings around mismatched format specifiers and argument
types:

  In file included from fs/bcachefs/vstructs.h:5,
                   from fs/bcachefs/bcachefs_format.h:79,
                   from fs/bcachefs/bcachefs.h:207,
                   from fs/bcachefs/btree_key_cache.c:3:
  fs/bcachefs/btree_key_cache.c: In function 'bch2_btree_key_cache_to_text':
  fs/bcachefs/btree_key_cache.c:1046:25: error: format '%lu' expects argument of type 'long unsigned int', but argument 3 has type 'size_t' {aka 'unsigned int'} [-Werror=format=]
   1046 |         prt_printf(out, "nonpcpu freelist:\t%lu\r\n",   bc->nr_freed_nonpcpu);
        |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~    ~~~~~~~~~~~~~~~~~~~~
        |                                                           |
        |                                                           size_t {aka unsigned int}
  fs/bcachefs/util.h:192:63: note: in definition of macro 'prt_printf'
    192 | #define prt_printf(_out, ...)           bch2_prt_printf(_out, __VA_ARGS__)
        |                                                               ^~~~~~~~~~~
  fs/bcachefs/btree_key_cache.c:1046:47: note: format string is defined here
   1046 |         prt_printf(out, "nonpcpu freelist:\t%lu\r\n",   bc->nr_freed_nonpcpu);
        |                                             ~~^
        |                                               |
        |                                               long unsigned int
        |                                             %u
  fs/bcachefs/btree_key_cache.c:1047:25: error: format '%lu' expects argument of type 'long unsigned int', but argument 3 has type 'size_t' {aka 'unsigned int'} [-Werror=format=]
   1047 |         prt_printf(out, "pcpu freelist:\t%lu\r\n",      bc->nr_freed_pcpu);
        |                         ^~~~~~~~~~~~~~~~~~~~~~~~~       ~~~~~~~~~~~~~~~~~
        |                                                           |
        |                                                           size_t {aka unsigned int}
  fs/bcachefs/util.h:192:63: note: in definition of macro 'prt_printf'
    192 | #define prt_printf(_out, ...)           bch2_prt_printf(_out, __VA_ARGS__)
        |                                                               ^~~~~~~~~~~
  fs/bcachefs/btree_key_cache.c:1047:44: note: format string is defined here
   1047 |         prt_printf(out, "pcpu freelist:\t%lu\r\n",      bc->nr_freed_pcpu);
        |                                          ~~^
        |                                            |
        |                                            long unsigned int
        |                                          %u
  cc1: all warnings being treated as error

Use the proper 'size_t' specifier, '%zu', to clear up the warnings for
these platforms.

Fixes: f2d47ec26af5 ("bcachefs: Btree key cache instrumentation")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

When building with clang's -Wincompatible-function-pointer-types-strict
(a warning designed to catch potential kCFI failures at build time),
there are several warnings along the lines of:

  fs/bcachefs/bkey_methods.c:118:2: error: incompatible function pointer types initializing 'int (*)(struct btree_trans *, enum btree_id, unsigned int, struct bkey_s_c, struct bkey_s, enum btree_iter_update_trigger_flags)' with an expression of type 'int (struct btree_trans *, enum btree_id, unsigned int, struct bkey_s_c, struct bkey_s, unsigned int)' [-Werror,-Wincompatible-function-pointer-types-strict]
    118 |         BCH_BKEY_TYPES()
        |         ^~~~~~~~~~~~~~~~
  fs/bcachefs/bcachefs_format.h:394:2: note: expanded from macro 'BCH_BKEY_TYPES'
    394 |         x(inode,                8)                      \
        |         ^~~~~~~~~~~~~~~~~~~~~~~~~~
  fs/bcachefs/bkey_methods.c:117:41: note: expanded from macro 'x'
    117 | #define x(name, nr) [KEY_TYPE_##name]   = bch2_bkey_ops_##name,
        |                                           ^~~~~~~~~~~~~~~~~~~~
  <scratch space>:277:1: note: expanded from here
    277 | bch2_bkey_ops_inode
        | ^~~~~~~~~~~~~~~~~~~
  fs/bcachefs/inode.h:26:13: note: expanded from macro 'bch2_bkey_ops_inode'
     26 |         .trigger        = bch2_trigger_inode,           \
      |                           ^~~~~~~~~~~~~~~~~~

There are several functions that did not have their flags parameter
converted to 'enum btree_iter_update_trigger_flags' in the recent
unification, which will cause kCFI failures at runtime because the
types, while ABI compatible (hence no warning from the non-strict
version of this warning), do not match exactly.

Fix up these functions (as well as a few other obvious functions that
should have it, even if there are no warnings currently) to resolve the
warnings and potential kCFI runtime failures.

Fixes: 31e4ef3280c8 ("bcachefs: iter/update/trigger/str_hash flag cleanup")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This unifies the online and offline btree gc passes; we're not yet
running it online.

We now iterate over one level of the btree at a time - the same as
check_extents_to_backpointers(); this ordering preserves order of keys
regardless of btree splits and merges, which will be important when we
re-enable online gc.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

Currently, the reflink_p gc trigger does repair as well - turning a
reflink_p key into an error key if the reflink_v it points to doesn't
exist.

This won't work with online check/repair, because the repair path once
online will be subject to transaction restarts, but BTREE_TRIGGER_gc is
not idempotant - we can't run it multiple times if we get a transaction
restart.

So we need to split these paths; to do so this patch calls
check_fix_ptrs() by a new general path - a new trigger type,
BTREE_TRIGGER_check_repair.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

looping when we change a bucket gen is not ideal - it means we risk
failing if we'd go into an infinite loop, and it's better to make
forward progress even if fsck doesn't fix everything.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

If we hit an inconsistency when updating allocation information, we
don't want to fail the update if it's for a deletion - only if it's for
a new key.

Rename check_bucket_ref() -> bucket_ref_update() so we can centralize
the logic to do this.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This eliminates some duplicated logic, and the gc path now handles
stripe updates and deletions - we need this since soon we're bringing
back runtime gc.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Start to work on unifying mark_stripe_bucket() and
trans_mark_stripe_bucket(); first, clean up all the unnecessary and
gratuitious differences.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

We're working on potentially unifying bch2_check_bucket_ref() and
bch2_check_fix_ptrs() - or at least eliminating gratuitious differences.

Most immediately, there's a bunch of cleanups to be done regarding
BCH_DATA_stripe.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

There's no need to be using new_inode(); we can skip all that
indirection and make the code easier to follow.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Since btree_ptr_v2, we no longer require the journal seq blacklist table
for skipping blacklisted bsets (btree node entries); the pointer to a
given node indicates how much data is present.

Therefore there's no longer any need for journal seq blacklist gc to
walk the btree - we can prune entries older than journal last_seq.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

prep work for making the allocator try to keep btree nodes within the
existing member info btree allocated bitmap
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

It turns out the btree key cache shrinker wasn't actually reclaiming
anything, prior to the previous patch. This adds instrumentation so that
if we have further issues we can see what's going on.

Specifically, sysfs internal/btree_key_cache is greatly expanded with
new counters, and the SRCU sequence numbers of the first 10 entries on
each pending freelist, and we also add trigger_btree_key_cache_shrink
for testing without having to prune all the system caches.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Common code doesn't test the error flag, so we don't need to set it in
bcachefs.  We can use folio_end_read() to combine the setting (or not)
of the uptodate flag and clearing the lock flag.

Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Brian Foster <bfoster@redhat.com>
Cc: linux-bcachefs@vger.kernel.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

This is a nice cleanup - and we've also been having problems with
kthread creation in the mount path.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

Make things easier when we're debugging long fsck runs - persist the
work that successful recovery passes did.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

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

Originally, when deleting snapshots we didn't collapse redundant
snapshot nodes; thus, the notion of a class of equivalent snapshot nodes
leaked into fsck.

Now we do, so snapshot ID equivalence classes are purely local to
snapshot deletion.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>