xfs: remote attribute blocks aren't really userdata

When adding a new remote attribute, we write the attribute to the
new extent before the allocation transaction is committed. This
means we cannot reuse busy extents as that violates crash
consistency semantics. Hence we currently treat remote attribute
extent allocation like userdata because it has the same overwrite
ordering constraints as userdata.

Unfortunately, this also allows the allocator to incorrectly apply
extent size hints to the remote attribute extent allocation. This
results in interesting failures, such as transaction block
reservation overruns and in-memory inode attribute fork corruption.

To fix this, we need to separate the busy extent reuse configuration
from the userdata configuration. This changes the definition of
XFS_BMAPI_METADATA slightly - it now means that allocation is
metadata and reuse of busy extents is acceptible due to the metadata
ordering semantics of the journal. If this flag is not set, it
means the allocation is that has unordered data writeback, and hence
busy extent reuse is not allowed. It no longer implies the
allocation is for user data, just that the data write will not be
strictly ordered. This matches the semantics for both user data
and remote attribute block allocation.

As such, This patch changes the "userdata" field to a "datatype"
field, and adds a "no busy reuse" flag to the field.
When we detect an unordered data extent allocation, we immediately set
the no reuse flag. We then set the "user data" flags based on the
inode fork we are allocating the extent to. Hence we only set
userdata flags on data fork allocations now and consider attribute
fork remote extents to be an unordered metadata extent.

The result is that remote attribute extents now have the expected
allocation semantics, and the data fork allocation behaviour is
completely unchanged.

It should be noted that there may be other ways to fix this (e.g.
use ordered metadata buffers for the remote attribute extent data
write) but they are more invasive and difficult to validate both
from a design and implementation POV. Hence this patch takes the
simple, obvious route to fixing the problem...

Reported-and-tested-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>

1 files changed, 26 insertions, 15 deletions

diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
index b060bca93402..06d1201b4718 100644
--- a/fs/xfs/libxfs/xfs_bmap.c
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -3347,7 +3347,8 @@ xfs_bmap_adjacent(
 
 	mp = ap->ip->i_mount;
 	nullfb = *ap->firstblock == NULLFSBLOCK;
-	rt = XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata;
+	rt = XFS_IS_REALTIME_INODE(ap->ip) &&
+		xfs_alloc_is_userdata(ap->datatype);
 	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, *ap->firstblock);
 	/*
 	 * If allocating at eof, and there's a previous real block,
@@ -3622,7 +3623,7 @@ xfs_bmap_btalloc(
 {
 	xfs_mount_t	*mp;		/* mount point structure */
 	xfs_alloctype_t	atype = 0;	/* type for allocation routines */
-	xfs_extlen_t	align;		/* minimum allocation alignment */
+	xfs_extlen_t	align = 0;	/* minimum allocation alignment */
 	xfs_agnumber_t	fb_agno;	/* ag number of ap->firstblock */
 	xfs_agnumber_t	ag;
 	xfs_alloc_arg_t	args;
@@ -3645,7 +3646,8 @@ xfs_bmap_btalloc(
 	else if (mp->m_dalign)
 		stripe_align = mp->m_dalign;
 
-	align = ap->userdata ? xfs_get_extsz_hint(ap->ip) : 0;
+	if (xfs_alloc_is_userdata(ap->datatype))
+		align = xfs_get_extsz_hint(ap->ip);
 	if (unlikely(align)) {
 		error = xfs_bmap_extsize_align(mp, &ap->got, &ap->prev,
 						align, 0, ap->eof, 0, ap->conv,
@@ -3658,7 +3660,8 @@ xfs_bmap_btalloc(
 	nullfb = *ap->firstblock == NULLFSBLOCK;
 	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, *ap->firstblock);
 	if (nullfb) {
-		if (ap->userdata && xfs_inode_is_filestream(ap->ip)) {
+		if (xfs_alloc_is_userdata(ap->datatype) &&
+		    xfs_inode_is_filestream(ap->ip)) {
 			ag = xfs_filestream_lookup_ag(ap->ip);
 			ag = (ag != NULLAGNUMBER) ? ag : 0;
 			ap->blkno = XFS_AGB_TO_FSB(mp, ag, 0);
@@ -3698,7 +3701,8 @@ xfs_bmap_btalloc(
 		 * enough for the request.  If one isn't found, then adjust
 		 * the minimum allocation size to the largest space found.
 		 */
-		if (ap->userdata && xfs_inode_is_filestream(ap->ip))
+		if (xfs_alloc_is_userdata(ap->datatype) &&
+		    xfs_inode_is_filestream(ap->ip))
 			error = xfs_bmap_btalloc_filestreams(ap, &args, &blen);
 		else
 			error = xfs_bmap_btalloc_nullfb(ap, &args, &blen);
@@ -3782,8 +3786,8 @@ xfs_bmap_btalloc(
 	args.minleft = ap->minleft;
 	args.wasdel = ap->wasdel;
 	args.isfl = 0;
-	args.userdata = ap->userdata;
-	if (ap->userdata & XFS_ALLOC_USERDATA_ZERO)
+	args.datatype = ap->datatype;
+	if (ap->datatype & XFS_ALLOC_USERDATA_ZERO)
 		args.ip = ap->ip;
 
 	error = xfs_alloc_vextent(&args);
@@ -3877,7 +3881,8 @@ STATIC int
 xfs_bmap_alloc(
 	struct xfs_bmalloca	*ap)	/* bmap alloc argument struct */
 {
-	if (XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata)
+	if (XFS_IS_REALTIME_INODE(ap->ip) &&
+	    xfs_alloc_is_userdata(ap->datatype))
 		return xfs_bmap_rtalloc(ap);
 	return xfs_bmap_btalloc(ap);
 }
@@ -4287,15 +4292,21 @@ xfs_bmapi_allocate(
 	}
 
 	/*
-	 * Indicate if this is the first user data in the file, or just any
-	 * user data. And if it is userdata, indicate whether it needs to
-	 * be initialised to zero during allocation.
+	 * Set the data type being allocated. For the data fork, the first data
+	 * in the file is treated differently to all other allocations. For the
+	 * attribute fork, we only need to ensure the allocated range is not on
+	 * the busy list.
 	 */
 	if (!(bma->flags & XFS_BMAPI_METADATA)) {
-		bma->userdata = (bma->offset == 0) ?
-			XFS_ALLOC_INITIAL_USER_DATA : XFS_ALLOC_USERDATA;
+		bma->datatype = XFS_ALLOC_NOBUSY;
+		if (whichfork == XFS_DATA_FORK) {
+			if (bma->offset == 0)
+				bma->datatype |= XFS_ALLOC_INITIAL_USER_DATA;
+			else
+				bma->datatype |= XFS_ALLOC_USERDATA;
+		}
 		if (bma->flags & XFS_BMAPI_ZERO)
-			bma->userdata |= XFS_ALLOC_USERDATA_ZERO;
+			bma->datatype |= XFS_ALLOC_USERDATA_ZERO;
 	}
 
 	bma->minlen = (bma->flags & XFS_BMAPI_CONTIG) ? bma->length : 1;
@@ -4565,7 +4576,7 @@ xfs_bmapi_write(
 	bma.tp = tp;
 	bma.ip = ip;
 	bma.total = total;
-	bma.userdata = 0;
+	bma.datatype = 0;
 	bma.dfops = dfops;
 	bma.firstblock = firstblock;