Zorro Lang reported a crash in generic/475 if we try to inactivate a
corrupt inode with a NULL attr fork (stack trace shortened somewhat):

RIP: 0010:xfs_bmapi_read+0x311/0xb00 [xfs]
RSP: 0018:ffff888047f9ed68 EFLAGS: 00010202
RAX: dffffc0000000000 RBX: ffff888047f9f038 RCX: 1ffffffff5f99f51
RDX: 0000000000000002 RSI: 0000000000000008 RDI: 0000000000000012
RBP: ffff888002a41f00 R08: ffffed10005483f0 R09: ffffed10005483ef
R10: ffffed10005483ef R11: ffff888002a41f7f R12: 0000000000000004
R13: ffffe8fff53b5768 R14: 0000000000000005 R15: 0000000000000001
FS:  00007f11d44b5b80(0000) GS:ffff888114200000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000ef6000 CR3: 000000002e176003 CR4: 00000000001606e0
Call Trace:
 xfs_dabuf_map.constprop.18+0x696/0xe50 [xfs]
 xfs_da_read_buf+0xf5/0x2c0 [xfs]
 xfs_da3_node_read+0x1d/0x230 [xfs]
 xfs_attr_inactive+0x3cc/0x5e0 [xfs]
 xfs_inactive+0x4c8/0x5b0 [xfs]
 xfs_fs_destroy_inode+0x31b/0x8e0 [xfs]
 destroy_inode+0xbc/0x190
 xfs_bulkstat_one_int+0xa8c/0x1200 [xfs]
 xfs_bulkstat_one+0x16/0x20 [xfs]
 xfs_bulkstat+0x6fa/0xf20 [xfs]
 xfs_ioc_bulkstat+0x182/0x2b0 [xfs]
 xfs_file_ioctl+0xee0/0x12a0 [xfs]
 do_vfs_ioctl+0x193/0x1000
 ksys_ioctl+0x60/0x90
 __x64_sys_ioctl+0x6f/0xb0
 do_syscall_64+0x9f/0x4d0
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f11d39a3e5b

The "obvious" cause is that the attr ifork is null despite the inode
claiming an attr fork having at least one extent, but it's not so
obvious why we ended up with an inode in that state.
Reported-by: NZorro Lang <zlang@redhat.com>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204031Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBill O'Donnell <billodo@redhat.com>

Continue our game of replacing ASSERTs for corrupt ondisk metadata with
EFSCORRUPTED returns.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBill O'Donnell <billodo@redhat.com>

Add an XFS_ICHGTIME_CREATE case to xfs_trans_ichgtime() to keep in
sync with userspace.  (Currently no kernel caller sends this flag.)
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Userspace now has an identical xfs_trans_inode.c which it has already
moved to libxfs/ so do the same move for kernelspace.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

When we're iterating all the attributes using the built-in xattr
iterator, we can use the seen_enough variable to pass error codes back
to the main scrub function instead of flattening them into 0/1.  This
will be used in a more exciting fashion in upcoming patches.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Create a new bulk ireq flag that enables userspace to ask us for a
special inode number instead of interpreting @ino as a literal inode
number.  This enables us to query the root inode easily.

The reason for adding the ability to query specifically the root
directory inode is that certain programs (xfsdump and xfsrestore) want
to confirm when they've been pointed to the root directory.  The
userspace code assumes the root directory is always the first result
from calling bulkstat with lastino == 0, but this isn't true if the
(initial btree roots + initial AGFL + inode alignment padding) is itself
long enough to be allocated to new inodes if all of those blocks should
happen to be free at the same time.  Rather than make userspace guess
at internal filesystem state, we provide a direct query.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Add a new xfs_bulk_ireq flag to constrain the iteration to a single AG.
If the passed-in startino value is zero then we start with the first
inode in the AG that the user passes in; otherwise, we iterate only
within the same AG as the passed-in inode.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Wire up the v5 INUMBERS ioctl.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Wire up the new v5 BULKSTAT ioctl.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Introduce a new "v5" inode group structure that fixes the alignment
and padding problems of the existing structure.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Introduce a new version of the in-core bulkstat structure that supports
our new v5 format features.  This structure also fills the gaps in the
previous structure.  We leave wiring up the ioctls for the next patch.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Remove xfs_bstat_t, xfs_fsop_bulkreq_t, xfs_inogrp_t, and similarly
named compat typedefs.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Create a new iterator function to simplify walking inodes in an XFS
filesystem.  This new iterator will replace the existing open-coded
walking that goes on in various places.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Currently, xfs doesn't have generic error codes defined for "stop
iterating"; we just reuse the XFS_BTREE_QUERY_* return values.  This
looks a little weird if we're not actually iterating a btree index.
Before we start adding more iterators, we should create general
XFS_ITER_{CONTINUE,ABORT} return values and define the XFS_BTREE_QUERY_*
ones from that.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Instead of a magic flag for xfs_trans_alloc, just ensure all callers
that can't relclaim through the file system use memalloc_nofs_save to
set the per-task nofs flag.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

There are many, many xfs header files which are included but
unneeded (or included twice) in the xfs code, so remove them.

nb: xfs_linux.h includes about 9 headers for everyone, so those
explicit includes get removed by this.  I'm not sure what the
preference is, but if we wanted explicit includes everywhere,
a followup patch could remove those xfs_*.h includes from
xfs_linux.h and move them into the files that need them.
Or it could be left as-is.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

When we're writing out a fresh new AG, make sure that we don't list an
internal log as free and that we create the rmap for the region.  growfs
never does this, but we will need it when we hook up mkfs.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Refactor the code that populates the free space btrees of a new AG so
that we can avoid code duplication once things start getting
complicated.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

xfs_alloc_ag_vextent_small() doesn't update the output parameters in
the event of an AGFL allocation. Instead, it updates the
xfs_alloc_arg structure directly to complete the allocation.

Update both args and the output params to provide consistent
behavior for future callers.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The small allocation helper is implemented in a way that is fairly
tightly integrated to the existing allocation algorithms. It expects
a cntbt cursor beyond the end of the tree, attempts to locate the
last record in the tree and only attempts an AGFL allocation if the
cntbt is empty.

The upcoming generic algorithm doesn't rely on the cntbt processing
of this function. It will only call this function when the cntbt
doesn't have a big enough extent or is empty and thus AGFL
allocation is the only remaining option. Tweak
xfs_alloc_ag_vextent_small() to handle a NULL cntbt cursor and skip
the cntbt logic. This facilitates use by the existing allocation
code and new code that only requires an AGFL allocation attempt.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Move the small allocation helper further up in the file to avoid the
need for a function declaration. The remaining declarations will be
removed by followup patches. No functional changes.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

xfs_alloc_ag_vextent_small() is kind of a mess. Clean it up in
preparation for future changes. No functional changes.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

We need to derive the mount pointer from a buffer in a lot of place.
Add a direct pointer to short cut the pointer chasing.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The inode geometry structure isn't related to ondisk format; it's
support for the mount structure.  Move it to xfs_shared.h.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

There are several functions which take a flag argument that is
only ever passed as "0," so remove these arguments.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NBill O'Donnell <billodo@redhat.com>
Reviewed-by: NAllison Collins <allison.henderson@oracle.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The flags value is always passed as 0 so remove the argument.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Finish converting all the old inode_cluster_size >> inopblog users to
inodes_per_cluster.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

inode_cluster_size is supposed to represent the size (in bytes) of an
inode cluster buffer.  We avoid having to handle multiple clusters per
filesystem block on filesystems with large blocks by openly rounding
this value up to 1 FSB when necessary.  However, we never reset
inode_cluster_size to reflect this new rounded value, which adds to the
potential for mistakes in calculating geometries.

Fix this by setting inode_cluster_size to reflect the rounded-up size if
needed, and special-case the few places in the sparse inodes code where
we actually need the smaller value to validate on-disk metadata.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

Migrate all of the inode geometry setup code from xfs_mount.c into a
single libxfs function that we can share with xfsprogs.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

Separate the inode geometry information into a distinct structure.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

It turns out that the log can consume nearly all the space in an AG, and
when this happens this it's possible that there will be less free space
in the AG than the reservation would try to hide.  On a debug kernel
this can trigger an ASSERT in xfs/250:

XFS: Assertion failed: xfs_perag_resv(pag, XFS_AG_RESV_METADATA)->ar_reserved + xfs_perag_resv(pag, XFS_AG_RESV_RMAPBT)->ar_reserved <= pag->pagf_freeblks + pag->pagf_flcount, file: fs/xfs/libxfs/xfs_ag_resv.c, line: 319

The log is permanently allocated, so we know we're never going to have
to expand the btrees to hold any records associated with the log space.
We therefore can treat the space as if it doesn't exist.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NEric Sandeen <sandeen@redhat.com>

There are several functions which have no opportunity to return
an error, and don't contain any ASSERTs which could be argued
to be better constructed as error cases.  So, make them voids
to simplify the callers.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

Teach online scrub how to check the filesystem summary counters.  We use
the incore delalloc block counter along with the incore AG headers to
compute expected values for fdblocks, icount, and ifree, and then check
that the percpu counter is within a certain threshold of the expected
value.  This is done to avoid having to freeze or otherwise lock the
filesystem, which means that we're only checking that the counters are
fairly close, not that they're exactly correct.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

During testing of xfs/141 on a V4 filesystem, I observed some
inconsistent behavior with regards to resources that are held (i.e.
remain locked) across a defer roll.  The transaction roll always gives
the defer roll function a new transaction, even if committing the old
transaction fails.  However, the defer roll function only rejoins the
held resources if the transaction commit succeedied.  This means that
callers of defer roll have to figure out whether the held resources are
attached to the transaction being passed back.

Worse yet, if the defer roll was part of a defer finish call, we have a
third possibility: the defer finish could pass back a dirty transaction
with dirty held resources and an error code.

The only sane way to handle all of these scenarios is to require that
the code that held the resource either cancel the transaction before
unlocking and releasing the resources, or use functions that detach
resources from a transaction properly (e.g.  xfs_trans_brelse) if they
need to drop the reference before committing or cancelling the
transaction.

In order to make this so, change the defer roll code to join held
resources to the new transaction unconditionally and fix all the bhold
callers to release the held buffers correctly.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Add a percpu counter to track the number of blocks directly reserved for
delayed allocations on the data device.  This counter (in contrast to
i_delayed_blks) does not track allocated CoW staging extents or anything
going on with the realtime device.  It will be used in the upcoming
summary counter scrub function to check the free block counts without
having to freeze the filesystem or walk all the inodes to find the
delayed allocations.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>

Block allocation requires a permanent transaction for deferred AGFL
frees.  Add an assert in the block allocation path to make explicit and
obvious to future callers the requirement of a transaction with a
permanent reservation.
Reported-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
[darrick: split this out from the previous patch per hch request]
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The growdata transaction is used by growfs operations to increase
the data size of the filesystem. Part of this sequence involves
extending the size of the last preexisting AG in the fs, if
necessary. This is implemented by freeing the newly available
physical range to the AG.

tr_growdata is not a permanent transaction, however, and block
allocation transactions must be permanent to handle deferred frees
of AGFL blocks. If the grow operation extends an existing AG that
requires AGFL fixing, assert failures occur due to a populated dfops
list on a non-permanent transaction and the AGFL free does not
occur. This is reproduced (rarely) by xfs/104.

Change tr_growdata to a permanent transaction with a default log
count. This increases initial transaction reservation size, but
growfs is an infrequent and non-performance critical operation and
so should have minimal impact.
Reported-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
[darrick: add a comment to the assert]
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Use space in the bulkstat ioctl structure to report any problems
observed with the inode.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Use the AG geometry info ioctl to report health status too.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Use our newly expanded geometry structure to report the overall fs and
realtime health status.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>