xfs: widen inode delalloc block counter to 64-bits
authorDarrick J. Wong <darrick.wong@oracle.com>
Wed, 17 Apr 2019 23:30:24 +0000 (16:30 -0700)
committerDarrick J. Wong <darrick.wong@oracle.com>
Tue, 23 Apr 2019 15:36:23 +0000 (08:36 -0700)
commit394aafdc15da01c049f580e097806ee12ad6900e
tree9fe0b82f4a0abeafdef75f1c07b7e3d3e7fac83b
parent903b1fc2737f1c7086420db0a1e550ba0b479a0c
xfs: widen inode delalloc block counter to 64-bits

Widen the incore inode's i_delayed_blks counter to be a 64-bit integer.
This is necessary to fix an integer overflow problem that can be
reproduced easily now that we use the counter to track blocks that are
assigned to the inode in memory but not on disk.  This includes actual
delalloc reservations as well as real extents in the COW fork that
are waiting to be remapped into the data fork.

These 'delayed mapping' blocks can easily exceed 2^32 blocks if one
creates a very large sparse file of size approximately 2^33 bytes with
one byte written every 2^23 bytes, sets a very large COW extent size
hint of 2^23 blocks, reflinks the first file into a second file, and
then writes a single byte every 2^23 blocks in the original file.

When this happens, we'll try to create approximately 1024 2^23 extent
reservations in the COW fork, which will overflow the counter and cause
problems.

Note that on x64 we end up filling a 4-byte gap in the structure so this
doesn't increase the incore size.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Allison Collins <allison.henderson@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
fs/xfs/xfs_inode.h
fs/xfs/xfs_qm.c