linux-xfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: darrick.wong@oracle.com
Cc: linux-xfs@vger.kernel.org
Subject: [PATCH 5/6] xfs: flush speculative space allocations when we run out of quota
Date: Tue, 31 Dec 2019 17:07:46 -0800	[thread overview]
Message-ID: <157784086638.1361522.14471831853437220735.stgit@magnolia> (raw)
In-Reply-To: <157784083298.1361522.7064886067520069080.stgit@magnolia>

From: Darrick J. Wong <darrick.wong@oracle.com>

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>
---
 fs/xfs/xfs_bmap_util.c |   17 +++++++++++++++++
 fs/xfs/xfs_file.c      |    2 +-
 fs/xfs/xfs_icache.c    |    9 +++++++--
 fs/xfs/xfs_icache.h    |    2 +-
 fs/xfs/xfs_inode.c     |   17 +++++++++++++++++
 fs/xfs/xfs_ioctl.c     |    2 ++
 fs/xfs/xfs_iomap.c     |   19 +++++++++++++++++++
 fs/xfs/xfs_reflink.c   |   45 +++++++++++++++++++++++++++++++++++++++++----
 fs/xfs/xfs_trace.c     |    1 +
 fs/xfs/xfs_trace.h     |   40 ++++++++++++++++++++++++++++++++++++++++
 10 files changed, 146 insertions(+), 8 deletions(-)


diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
index f96c1d9a68a7..72965a22ad1d 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 @@ xfs_alloc_file_space(
 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 26e3edc07445..6b37d0435e3a 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -655,7 +655,7 @@ xfs_file_buffered_aio_write(
 	 */
 	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 3a88014fc259..c17ac64c0896 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 1f1c40b20fa6..d11857125f45 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -2285,6 +2285,8 @@ 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 c1befb899911..1eed29139b58 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 de451235c4ee..6bff74ab49a0 100644
--- a/fs/xfs/xfs_reflink.c
+++ b/fs/xfs/xfs_reflink.c
@@ -353,13 +353,14 @@ 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) {
@@ -378,6 +379,7 @@ 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);
@@ -402,6 +404,23 @@ 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;
 
@@ -443,6 +462,7 @@ xfs_reflink_allocate_cow(
 			XFS_QMOPT_RES_REGBLKS);
 out_trans_cancel:
 	xfs_trans_cancel(tp);
+out:
 	return error;
 }
 
@@ -1005,6 +1025,7 @@ 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;
@@ -1020,21 +1041,37 @@ 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 222b313f61b0..8307d188aea7 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


  parent reply	other threads:[~2020-01-01  1:09 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-01  1:07 [PATCH v2 0/6] xfs: try harder to reclaim space when we run out Darrick J. Wong
2020-01-01  1:07 ` [PATCH 1/6] xfs: refactor messy xfs_inode_free_quota_* functions Darrick J. Wong
2020-01-01  1:07 ` [PATCH 2/6] xfs: don't stall cowblocks scan if we can't take locks Darrick J. Wong
2020-01-01  1:07 ` [PATCH 3/6] xfs: move inode flush to a workqueue Darrick J. Wong
2020-01-01  1:07 ` [PATCH 4/6] xfs: xfs_inode_free_quota_blocks should scan project quota Darrick J. Wong
2020-01-01  1:07 ` Darrick J. Wong [this message]
2020-01-01  1:07 ` [PATCH 6/6] xfs: flush speculative space allocations when we run out of space Darrick J. Wong
2021-01-11 23:22 [PATCHSET v2 0/6] xfs: try harder to reclaim space when we run out Darrick J. Wong
2021-01-11 23:23 ` [PATCH 5/6] xfs: flush speculative space allocations when we run out of quota Darrick J. Wong
2021-01-12  1:22   ` Dave Chinner
2021-01-12  1:31     ` Darrick J. Wong
2021-01-12  1:40       ` Dave Chinner
2021-01-12  2:18         ` Darrick J. Wong
2021-01-14 22:10           ` Darrick J. Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=157784086638.1361522.14471831853437220735.stgit@magnolia \
    --to=darrick.wong@oracle.com \
    --cc=linux-xfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).