xfs: flush speculative space allocations when we run out of quota

If a fs modification (creation, file write, reflink, etc.) is unable to
reserve enough quota to handle the modification, try clearing whatever
space the filesystem might have been hanging onto in the hopes of
speeding up the filesystem.  The flushing behavior will become
particularly important when we add deferred inode inactivation because
that will increase the amount of space that isn't actively tied to user
data.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

10 files changed, 146 insertions, 8 deletions

diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
index e62fb5216341..ce463e3b61af 100644
--- a/fs/xfs/xfs_bmap_util.c
+++ b/fs/xfs/xfs_bmap_util.c
@@ -761,6 +761,7 @@ xfs_alloc_file_space(
 	 */
 	while (allocatesize_fsb && !error) {
 		xfs_fileoff_t	s, e;
+		bool		cleared_space = false;
 
 		/*
 		 * Determine space reservations for data/realtime.
@@ -803,6 +804,7 @@ xfs_alloc_file_space(
 		/*
 		 * Allocate and setup the transaction.
 		 */
+retry:
 		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks,
 				resrtextents, 0, &tp);
 
@@ -819,6 +821,20 @@ xfs_alloc_file_space(
 		xfs_ilock(ip, XFS_ILOCK_EXCL);
 		error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks,
 						      0, quota_flag);
+		/*
+		 * We weren't able to reserve enough quota to handle fallocate.
+		 * Flush any disk space that was being held in the hopes of
+		 * speeding up the filesystem.  We hold the IOLOCK so we cannot
+		 * do a synchronous scan.
+		 */
+		if ((error == -ENOSPC || error == -EDQUOT) && !cleared_space) {
+			xfs_trans_cancel(tp);
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			cleared_space = xfs_inode_free_quota_blocks(ip, false);
+			if (cleared_space)
+				goto retry;
+			goto error2;
+		}
 		if (error)
 			goto error1;
 
@@ -857,6 +873,7 @@ error0:	/* unlock inode, unreserve quota blocks, cancel trans */
 error1:	/* Just cancel transaction */
 	xfs_trans_cancel(tp);
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+error2:
 	return error;
 }
 
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index d45bda4d4f39..5b04d3dc200a 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -660,7 +660,7 @@ write_retry:
 	 */
 	if (ret == -EDQUOT && !cleared_space) {
 		xfs_iunlock(ip, iolock);
-		cleared_space = xfs_inode_free_quota_blocks(ip);
+		cleared_space = xfs_inode_free_quota_blocks(ip, true);
 		if (cleared_space)
 			goto write_retry;
 		iolock = 0;
diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index d954e37af5d0..5dffa87973b4 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -1536,7 +1536,8 @@ xfs_icache_free_eofblocks(
  */
 bool
 xfs_inode_free_quota_blocks(
-	struct xfs_inode	*ip)
+	struct xfs_inode	*ip,
+	bool			sync)
 {
 	struct xfs_eofblocks	eofb = {0};
 	struct xfs_dquot	*dq;
@@ -1546,7 +1547,9 @@ xfs_inode_free_quota_blocks(
 	 * Run a sync scan to increase effectiveness and use the union filter to
 	 * cover all applicable quotas in a single scan.
 	 */
-	eofb.eof_flags = XFS_EOF_FLAGS_UNION | XFS_EOF_FLAGS_SYNC;
+	eofb.eof_flags = XFS_EOF_FLAGS_UNION;
+	if (sync)
+		eofb.eof_flags |= XFS_EOF_FLAGS_SYNC;
 
 	if (XFS_IS_UQUOTA_ENFORCED(ip->i_mount)) {
 		dq = xfs_inode_dquot(ip, XFS_DQ_USER);
@@ -1578,6 +1581,8 @@ xfs_inode_free_quota_blocks(
 	if (!do_work)
 		return false;
 
+	trace_xfs_inode_free_quota_blocks(ip->i_mount, &eofb, _RET_IP_);
+
 	xfs_icache_free_eofblocks(ip->i_mount, &eofb);
 	xfs_icache_free_cowblocks(ip->i_mount, &eofb);
 	return true;
diff --git a/fs/xfs/xfs_icache.h b/fs/xfs/xfs_icache.h
index 6727b792ee40..835c735301b8 100644
--- a/fs/xfs/xfs_icache.h
+++ b/fs/xfs/xfs_icache.h
@@ -57,7 +57,7 @@ long xfs_reclaim_inodes_nr(struct xfs_mount *mp, int nr_to_scan);
 
 void xfs_inode_set_reclaim_tag(struct xfs_inode *ip);
 
-bool xfs_inode_free_quota_blocks(struct xfs_inode *ip);
+bool xfs_inode_free_quota_blocks(struct xfs_inode *ip, bool sync);
 
 void xfs_inode_set_eofblocks_tag(struct xfs_inode *ip);
 void xfs_inode_clear_eofblocks_tag(struct xfs_inode *ip);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 9ab13d258e4b..7269243edfb3 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1143,6 +1143,7 @@ xfs_create(
 	struct xfs_dquot	*gdqp = NULL;
 	struct xfs_dquot	*pdqp = NULL;
 	struct xfs_trans_res	*tres;
+	bool			cleared_space = false;
 	uint			resblks;
 
 	trace_xfs_create(dp, name);
@@ -1176,6 +1177,7 @@ xfs_create(
 	 * the case we'll drop the one we have and get a more
 	 * appropriate transaction later.
 	 */
+retry:
 	error = xfs_trans_alloc(mp, tres, resblks, 0, 0, &tp);
 	if (error == -ENOSPC) {
 		/* flush outstanding delalloc blocks and retry */
@@ -1193,6 +1195,21 @@ xfs_create(
 	 */
 	error = xfs_trans_reserve_quota(tp, mp, udqp, gdqp,
 						pdqp, resblks, 1, 0);
+	/*
+	 * We weren't able to reserve enough quota to handle adding the inode.
+	 * Flush any disk space that was being held in the hopes of speeding up
+	 * the filesystem.
+	 */
+	if ((error == -EDQUOT || error == -ENOSPC) && !cleared_space) {
+		xfs_trans_cancel(tp);
+		if (unlock_dp_on_error)
+			xfs_iunlock(dp, XFS_ILOCK_EXCL);
+		unlock_dp_on_error = false;
+		cleared_space = xfs_inode_free_quota_blocks(dp, true);
+		if (cleared_space)
+			goto retry;
+		goto out_release_inode;
+	}
 	if (error)
 		goto out_trans_cancel;
 
diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
index dfaf66dc916b..6abaea5dd608 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -2299,6 +2299,8 @@ xfs_file_ioctl(
 		if (error)
 			return error;
 
+		trace_xfs_ioc_free_eofblocks(mp, &keofb, _RET_IP_);
+
 		return xfs_icache_free_eofblocks(mp, &keofb);
 	}
 
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 0eb4da50d075..20ec4622e43e 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -28,6 +28,7 @@
 #include "xfs_dquot.h"
 #include "xfs_reflink.h"
 #include "xfs_health.h"
+#include "xfs_icache.h"
 
 #define XFS_ALLOC_ALIGN(mp, off) \
 	(((off) >> mp->m_allocsize_log) << mp->m_allocsize_log)
@@ -202,6 +203,7 @@ xfs_iomap_write_direct(
 	int			error;
 	int			bmapi_flags = XFS_BMAPI_PREALLOC;
 	uint			tflags = 0;
+	bool			cleared_space = false;
 
 	ASSERT(count_fsb > 0);
 
@@ -241,6 +243,7 @@ xfs_iomap_write_direct(
 			resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0) << 1;
 		}
 	}
+retry:
 	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, resrtextents,
 			tflags, &tp);
 	if (error)
@@ -249,6 +252,22 @@ xfs_iomap_write_direct(
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
 
 	error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
+	/*
+	 * We weren't able to reserve enough quota for the direct write.
+	 * Flush any disk space that was being held in the hopes of speeding up
+	 * the filesystem.  Historically, we expected callers to have
+	 * preallocated all the space before a direct write, but this is not an
+	 * absolute requirement.  We still hold the IOLOCK so we cannot do a
+	 * sync scan.
+	 */
+	if ((error == -ENOSPC || error == -EDQUOT) && !cleared_space) {
+		xfs_trans_cancel(tp);
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		cleared_space = xfs_inode_free_quota_blocks(ip, false);
+		if (cleared_space)
+			goto retry;
+		return error;
+	}
 	if (error)
 		goto out_trans_cancel;
 
diff --git a/fs/xfs/xfs_reflink.c b/fs/xfs/xfs_reflink.c
index b0ce04ffd3cd..85b5f8671de7 100644
--- a/fs/xfs/xfs_reflink.c
+++ b/fs/xfs/xfs_reflink.c
@@ -351,13 +351,14 @@ xfs_reflink_allocate_cow(
 	bool			convert_now)
 {
 	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_trans	*tp;
 	xfs_fileoff_t		offset_fsb = imap->br_startoff;
 	xfs_filblks_t		count_fsb = imap->br_blockcount;
-	struct xfs_trans	*tp;
-	int			nimaps, error = 0;
-	bool			found;
 	xfs_filblks_t		resaligned;
 	xfs_extlen_t		resblks = 0;
+	bool			found;
+	bool			cleared_space = false;
+	int			nimaps, error = 0;
 
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
 	if (!ip->i_cowfp) {
@@ -376,6 +377,7 @@ xfs_reflink_allocate_cow(
 	resblks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
 
 	xfs_iunlock(ip, *lockmode);
+retry:
 	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, 0, 0, &tp);
 	*lockmode = XFS_ILOCK_EXCL;
 	xfs_ilock(ip, *lockmode);
@@ -400,6 +402,23 @@ xfs_reflink_allocate_cow(
 
 	error = xfs_trans_reserve_quota_nblks(tp, ip, resblks, 0,
 			XFS_QMOPT_RES_REGBLKS);
+	/*
+	 * We weren't able to reserve enough quota to handle copy on write.
+	 * Flush any disk space that was being held in the hopes of speeding up
+	 * the filesystem.  We potentially hold the IOLOCK so we cannot do a
+	 * synchronous scan.
+	 */
+	if ((error == -ENOSPC || error == -EDQUOT) && !cleared_space) {
+		xfs_trans_cancel(tp);
+		xfs_iunlock(ip, *lockmode);
+		*lockmode = 0;
+		cleared_space = xfs_inode_free_quota_blocks(ip, false);
+		if (cleared_space)
+			goto retry;
+		*lockmode = XFS_ILOCK_EXCL;
+		xfs_ilock(ip, *lockmode);
+		goto out;
+	}
 	if (error)
 		goto out_trans_cancel;
 
@@ -441,6 +460,7 @@ out_unreserve:
 			XFS_QMOPT_RES_REGBLKS);
 out_trans_cancel:
 	xfs_trans_cancel(tp);
+out:
 	return error;
 }
 
@@ -1003,6 +1023,7 @@ xfs_reflink_remap_extent(
 	xfs_filblks_t		rlen;
 	xfs_filblks_t		unmap_len;
 	xfs_off_t		newlen;
+	bool			cleared_space = false;
 	int			error;
 
 	unmap_len = irec->br_startoff + irec->br_blockcount - destoff;
@@ -1018,21 +1039,37 @@ xfs_reflink_remap_extent(
 
 	/* Start a rolling transaction to switch the mappings */
 	resblks = XFS_EXTENTADD_SPACE_RES(ip->i_mount, XFS_DATA_FORK);
+retry:
 	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, 0, 0, &tp);
 	if (error)
 		goto out;
 
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	xfs_trans_ijoin(tp, ip, 0);
 
 	/* If we're not just clearing space, then do we have enough quota? */
 	if (real_extent) {
 		error = xfs_trans_reserve_quota_nblks(tp, ip,
 				irec->br_blockcount, 0, XFS_QMOPT_RES_REGBLKS);
+		/*
+		 * We weren't able to reserve enough quota for the remapping.
+		 * Flush any disk space that was being held in the hopes of
+		 * speeding up the filesystem.  We still hold the IOLOCK so we
+		 * cannot do a sync scan.
+		 */
+		if ((error == -ENOSPC || error == -EDQUOT) && !cleared_space) {
+			xfs_trans_cancel(tp);
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			cleared_space = xfs_inode_free_quota_blocks(ip, false);
+			if (cleared_space)
+				goto retry;
+			goto out;
+		}
 		if (error)
 			goto out_cancel;
 	}
 
+	xfs_trans_ijoin(tp, ip, 0);
+
 	trace_xfs_reflink_remap(ip, irec->br_startoff,
 				irec->br_blockcount, irec->br_startblock);
 
diff --git a/fs/xfs/xfs_trace.c b/fs/xfs/xfs_trace.c
index 9b5e58a92381..09c643643604 100644
--- a/fs/xfs/xfs_trace.c
+++ b/fs/xfs/xfs_trace.c
@@ -28,6 +28,7 @@
 #include "xfs_log_recover.h"
 #include "xfs_filestream.h"
 #include "xfs_fsmap.h"
+#include "xfs_icache.h"
 
 /*
  * We include this last to have the helpers above available for the trace
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 69e8605f9f97..8fb42a155f54 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -36,6 +36,7 @@ struct xfs_owner_info;
 struct xfs_trans_res;
 struct xfs_inobt_rec_incore;
 union xfs_btree_ptr;
+struct xfs_eofblocks;
 
 DECLARE_EVENT_CLASS(xfs_attr_list_class,
 	TP_PROTO(struct xfs_attr_list_context *ctx),
@@ -3740,6 +3741,45 @@ TRACE_EVENT(xfs_btree_bload_block,
 		  __entry->nr_records)
 )
 
+DECLARE_EVENT_CLASS(xfs_eofblocks_class,
+	TP_PROTO(struct xfs_mount *mp, struct xfs_eofblocks *eofb,
+		 unsigned long caller_ip),
+	TP_ARGS(mp, eofb, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(__u32, flags)
+		__field(uint32_t, uid)
+		__field(uint32_t, gid)
+		__field(prid_t, prid)
+		__field(__u64, min_file_size)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+                __entry->flags = eofb->eof_flags;
+                __entry->uid = xfs_kuid_to_uid(eofb->eof_uid);
+                __entry->gid = xfs_kgid_to_gid(eofb->eof_gid);
+                __entry->prid = eofb->eof_prid;
+                __entry->min_file_size = eofb->eof_min_file_size;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d flags 0x%x uid %u gid %u prid %u minsize %llu caller %pS",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+                  __entry->flags,
+                  __entry->uid,
+                  __entry->gid,
+                  __entry->prid,
+                  __entry->min_file_size,
+		  (char *)__entry->caller_ip)
+);
+#define DEFINE_EOFBLOCKS_EVENT(name)	\
+DEFINE_EVENT(xfs_eofblocks_class, name,	\
+	TP_PROTO(struct xfs_mount *mp, struct xfs_eofblocks *eofb, \
+		 unsigned long caller_ip), \
+	TP_ARGS(mp, eofb, caller_ip))
+DEFINE_EOFBLOCKS_EVENT(xfs_ioc_free_eofblocks);
+DEFINE_EOFBLOCKS_EVENT(xfs_inode_free_quota_blocks);
+
 #endif /* _TRACE_XFS_H */
 
 #undef TRACE_INCLUDE_PATH