xfs: online repair of realtime summariesrepair-rtsummary_2021-10-22

Repair the realtime summary data by constructing a new rtsummary file in
the scrub temporary file, then atomically swapping the contents.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>

8 files changed, 324 insertions, 3 deletions

diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 7455c1360548..7573fb05b244 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -187,6 +187,10 @@ xfs-y				+= $(addprefix scrub/, \
 				   tempfile.o \
 				   )
 
+xfs-$(CONFIG_XFS_RT)		+= $(addprefix scrub/, \
+				   rtsummary_repair.o \
+				   )
+
 xfs-$(CONFIG_XFS_QUOTA)		+= $(addprefix scrub/, \
 				   quota_repair.o \
 				   quotacheck_repair.o \
diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
index 8d04afffc21b..f98d2763918b 100644
--- a/fs/xfs/scrub/repair.c
+++ b/fs/xfs/scrub/repair.c
@@ -42,6 +42,7 @@
 #include "scrub/trace.h"
 #include "scrub/repair.h"
 #include "scrub/bitmap.h"
+#include "scrub/xfile.h"
 
 /*
  * Attempt to repair some metadata, if the metadata is corrupt and userspace
@@ -167,9 +168,21 @@ int
 xrep_roll_trans(
 	struct xfs_scrub	*sc)
 {
+	int			error;
+
 	if (!sc->ip)
 		return xrep_roll_ag_trans(sc);
-	return xfs_trans_roll_inode(&sc->tp, sc->ip);
+
+	/*
+	 * Roll the transaction with the inode we're fixing and the temp inode,
+	 * so that neither can pin the log.
+	 */
+	if (sc->tempip)
+		xfs_trans_log_inode(sc->tp, sc->tempip, XFS_ILOG_CORE);
+	error = xfs_trans_roll_inode(&sc->tp, sc->ip);
+	if (sc->tempip)
+		xfs_trans_ijoin(sc->tp, sc->tempip, 0);
+	return error;
 }
 
 /*
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 6a8b85dd71be..e72a97f50ab2 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -72,6 +72,7 @@ int xrep_reset_perag_resv(struct xfs_scrub *sc);
 int xrep_bmap(struct xfs_scrub *sc, int whichfork, bool allow_unwritten);
 int xrep_metadata_inode_forks(struct xfs_scrub *sc);
 int xrep_setup_ag_rmapbt(struct xfs_scrub *sc);
+int xrep_setup_rtsummary(struct xfs_scrub *sc, unsigned int *resblks);
 
 void xrep_ag_btcur_init(struct xfs_scrub *sc, struct xchk_ag *sa);
 int xrep_ag_init(struct xfs_scrub *sc, struct xfs_perag *pag,
@@ -107,6 +108,12 @@ int xrep_quotacheck(struct xfs_scrub *sc);
 # define xrep_quotacheck		xrep_notsupported
 #endif /* CONFIG_XFS_QUOTA */
 
+#ifdef CONFIG_XFS_RT
+int xrep_rtsummary(struct xfs_scrub *sc);
+#else
+# define xrep_rtsummary			xrep_notsupported
+#endif /* CONFIG_XFS_RT */
+
 struct xrep_newbt_resv {
 	/* Link to list of extents that we've reserved. */
 	struct list_head	list;
@@ -209,6 +216,13 @@ xrep_setup_ag_rmapbt(
 	return xchk_setup_ag_btree(sc, false);
 }
 
+static inline int
+xrep_setup_rtsummary(struct xfs_scrub *sc, unsigned int *resblks)
+{
+	*resblks = 0;
+	return 0;
+}
+
 #define xrep_revalidate_allocbt		(NULL)
 #define xrep_revalidate_iallocbt	(NULL)
 
@@ -228,6 +242,7 @@ xrep_setup_ag_rmapbt(
 #define xrep_quota			xrep_notsupported
 #define xrep_quotacheck			xrep_notsupported
 #define xrep_fscounters			xrep_notsupported
+#define xrep_rtsummary			xrep_notsupported
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/rtsummary.c b/fs/xfs/scrub/rtsummary.c
index f5c400559a45..6c0ca58e82dc 100644
--- a/fs/xfs/scrub/rtsummary.c
+++ b/fs/xfs/scrub/rtsummary.c
@@ -21,6 +21,8 @@
 #include "scrub/common.h"
 #include "scrub/trace.h"
 #include "scrub/xfile.h"
+#include "scrub/repair.h"
+#include "scrub/tempfile.h"
 
 /*
  * Realtime Summary
@@ -38,8 +40,15 @@ xchk_setup_rtsummary(
 	struct xfs_scrub	*sc)
 {
 	struct xfs_mount	*mp = sc->mp;
+	unsigned int		resblks = 0;
 	int			error;
 
+	if (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR) {
+		error = xrep_setup_rtsummary(sc, &resblks);
+		if (error)
+			return error;
+	}
+
 	/*
 	 * Create an xfile to construct a new rtsummary file.  The xfile allows
 	 * us to avoid pinning kernel memory for this purpose.
@@ -48,7 +57,7 @@ xchk_setup_rtsummary(
 	if (IS_ERR(sc->xfile))
 		return PTR_ERR(sc->xfile);
 
-	error = xchk_trans_alloc(sc, 0);
+	error = xchk_trans_alloc(sc, resblks);
 	if (error)
 		return error;
 
diff --git a/fs/xfs/scrub/rtsummary_repair.c b/fs/xfs/scrub/rtsummary_repair.c
new file mode 100644
index 000000000000..1e671d395221
--- /dev/null
+++ b/fs/xfs/scrub/rtsummary_repair.c
@@ -0,0 +1,103 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2021 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <djwong@kernel.org>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_btree.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_rtalloc.h"
+#include "xfs_inode.h"
+#include "xfs_bit.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_swapext.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+#include "scrub/tempfile.h"
+
+/* Set us up to repair the rtsummary file. */
+int
+xrep_setup_rtsummary(
+	struct xfs_scrub	*sc,
+	unsigned int		*resblks)
+{
+	struct xfs_mount	*mp = sc->mp;
+	unsigned long long	blocks;
+	int			error;
+
+	error = xrep_tempfile_create(sc, S_IFREG);
+	if (error)
+		return error;
+
+	/*
+	 * If we're doing a repair, we reserve enough blocks to write out a
+	 * completely new summary file, plus twice as many blocks as we would
+	 * need if we can only allocate one block per data fork mapping.  This
+	 * should cover the preallocation of the temporary file and swapping
+	 * the extent mappings.
+	 *
+	 * We cannot use xfs_swapext_estimate because we have not yet
+	 * constructed the replacement rtsummary and therefore do not know how
+	 * many extents it will use.  By the time we do, we will have a dirty
+	 * transaction (which we cannot drop because we cannot drop the
+	 * rtsummary ILOCK) and cannot ask for more reservation.
+	 */
+	blocks = XFS_B_TO_FSB(mp, mp->m_rsumsize);
+	blocks += xfs_bmbt_calc_size(mp, blocks) * 2;
+	if (blocks > UINT_MAX)
+		return -EOPNOTSUPP;
+
+	*resblks = blocks;
+	return 0;
+}
+
+/* Repair the realtime summary. */
+int
+xrep_rtsummary(
+	struct xfs_scrub	*sc)
+{
+	struct xfs_swapext_req	req = { .req_flags = 0 };
+	int			error;
+
+	/* Make sure any problems with the fork are fixed. */
+	error = xrep_metadata_inode_forks(sc);
+	if (error)
+		return error;
+
+	/*
+	 * Trylock the temporary file.  We had better be the only ones holding
+	 * onto this inode...
+	 */
+	if (!xrep_tempfile_ilock_nowait(sc, XFS_ILOCK_EXCL))
+		return -EAGAIN;
+
+	/* Make sure we have space allocated for the entire summary file. */
+	xfs_trans_ijoin(sc->tp, sc->ip, 0);
+	xfs_trans_ijoin(sc->tp, sc->tempip, 0);
+	error = xrep_tempfile_prealloc(sc, 0,
+			XFS_B_TO_FSB(sc->mp, sc->mp->m_rsumsize));
+	if (error)
+		return error;
+
+	/* Copy the rtsummary file that we generated. */
+	error = xrep_tempfile_copyin(sc, &xfs_rtbuf_ops,
+			XFS_BLFT_RTSUMMARY_BUF, sc->mp->m_rsumsize);
+	if (error)
+		return error;
+
+	/* Now swap the extents. */
+	req.ip1 = sc->tempip;
+	req.ip2 = sc->ip;
+	req.whichfork = XFS_DATA_FORK;
+	req.blockcount = XFS_B_TO_FSB(sc->mp, sc->mp->m_rsumsize);
+	return xfs_swapext(&sc->tp, &req);
+}
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 1ba85edbf694..c52bbe25068b 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -335,7 +335,7 @@ static const struct xchk_meta_ops meta_scrub_ops[] = {
 		.setup	= xchk_setup_rtsummary,
 		.scrub	= xchk_rtsummary,
 		.has	= xfs_has_realtime,
-		.repair	= xrep_notsupported,
+		.repair	= xrep_rtsummary,
 	},
 	[XFS_SCRUB_TYPE_UQUOTA] = {	/* user quota */
 		.type	= ST_FS,
diff --git a/fs/xfs/scrub/tempfile.c b/fs/xfs/scrub/tempfile.c
index 5ff376d228db..09dbefa62049 100644
--- a/fs/xfs/scrub/tempfile.c
+++ b/fs/xfs/scrub/tempfile.c
@@ -14,14 +14,17 @@
 #include "xfs_inode.h"
 #include "xfs_ialloc.h"
 #include "xfs_quota.h"
+#include "xfs_bmap.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_trans_space.h"
 #include "xfs_dir2.h"
 #include "xfs_xchgrange.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
+#include "scrub/repair.h"
 #include "scrub/trace.h"
 #include "scrub/tempfile.h"
+#include "scrub/xfile.h"
 
 /*
  * Create a temporary file for reconstructing metadata, with the intention of
@@ -211,3 +214,166 @@ xrep_tempfile_rele(
 	xfs_irele(sc->tempip);
 	sc->tempip = NULL;
 }
+
+/*
+ * Make sure that the given range of the data fork of the temporary file is
+ * mapped to written blocks.  The caller must ensure that both inodes are
+ * joined to the transaction.
+ */
+int
+xrep_tempfile_prealloc(
+	struct xfs_scrub	*sc,
+	xfs_fileoff_t		off,
+	xfs_filblks_t		len)
+{
+	struct xfs_bmbt_irec	map;
+	xfs_fileoff_t		end = off + len;
+	int			nmaps;
+	int			error = 0;
+
+	ASSERT(sc->tempip != NULL);
+	ASSERT(!XFS_NOT_DQATTACHED(sc->mp, sc->tempip));
+
+	while (off < len) {
+		/*
+		 * If we have a real extent mapping this block then we're
+		 * in ok shape.
+		 */
+		nmaps = 1;
+		error = xfs_bmapi_read(sc->tempip, off, end - off, &map, &nmaps,
+				XFS_DATA_FORK);
+		if (error)
+			break;
+
+		if (nmaps == 1 && xfs_bmap_is_written_extent(&map)) {
+			off += map.br_startblock;
+			continue;
+		}
+
+		/*
+		 * If we find a delalloc reservation then something is very
+		 * very wrong.  Bail out.
+		 */
+		if (map.br_startblock == DELAYSTARTBLOCK)
+			return -EFSCORRUPTED;
+
+		/*
+		 * Make sure this rtsum block has a real zeroed extent
+		 * allocated to it.
+		 */
+		nmaps = 1;
+		error = xfs_bmapi_write(sc->tp, sc->tempip, off, end - off,
+				XFS_BMAPI_CONVERT | XFS_BMAPI_ZERO, 0, &map,
+				&nmaps);
+		if (error)
+			break;
+
+		error = xrep_roll_trans(sc);
+		if (error)
+			break;
+		off += map.br_startblock;
+	}
+
+	return error;
+}
+
+/*
+ * Write a number of bytes from the xfile into the temp file.  The copybuf must
+ * be large enough to hold one filesystem block's worth of data.  The caller
+ * must join both inodes to the transaction.
+ */
+int
+xrep_tempfile_copyin(
+	struct xfs_scrub	*sc,
+	const struct xfs_buf_ops *ops,
+	enum xfs_blft		type,
+	xfs_fileoff_t		isize)
+{
+	LIST_HEAD(buffers_list);
+	struct xfs_bmbt_irec	map;
+	struct xfs_mount	*mp = sc->mp;
+	struct xfs_buf		*bp;
+	xfs_rtblock_t		off = 0;
+	loff_t			pos = 0;
+	unsigned int		nr_buffers = 0;
+	int			nmaps;
+	int			error = 0;
+
+	ASSERT(S_ISREG(VFS_I(sc->tempip)->i_mode));
+
+	for (; pos < isize; pos += mp->m_sb.sb_blocksize, off++) {
+		size_t		count;
+
+		/* Read block mapping for this file block. */
+		nmaps = 1;
+		error = xfs_bmapi_read(sc->tempip, off, 1, &map, &nmaps, 0);
+		if (error)
+			goto out;
+		if (nmaps == 0 || !xfs_bmap_is_written_extent(&map)) {
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+
+		/* Get the metadata buffer for this offset in the file. */
+		error = xfs_trans_get_buf(sc->tp, mp->m_ddev_targp,
+				XFS_FSB_TO_DADDR(mp, map.br_startblock),
+				mp->m_bsize, 0, &bp);
+		if (error)
+			goto out;
+		bp->b_ops = ops;
+		xfs_trans_buf_set_type(sc->tp, bp, type);
+
+		/* Read in a block's worth of data from the xfile. */
+		count = min_t(loff_t, isize - pos, mp->m_sb.sb_blocksize);
+		error = xfile_obj_load(sc->xfile, bp->b_addr, count, pos);
+		if (error) {
+			xfs_trans_brelse(sc->tp, bp);
+			goto out;
+		}
+
+		/*
+		 * Put this buffer on the delwri list so we can write them all
+		 * out in batches.
+		 */
+		xfs_buf_delwri_queue(bp, &buffers_list);
+		xfs_trans_brelse(sc->tp, bp);
+		nr_buffers++;
+
+		/*
+		 * If we have more than 256K of data to write out, flush it to
+		 * disk so we don't use up too much memory.
+		 */
+		if (XFS_FSB_TO_B(mp, nr_buffers) > 262144) {
+			error = xfs_buf_delwri_submit(&buffers_list);
+			if (error)
+				goto out;
+			nr_buffers = 0;
+		}
+	}
+
+	/*
+	 * Write the new blocks to disk.  If the ordered list isn't empty after
+	 * that, then something went wrong and we have to fail.  This should
+	 * never happen, but we'll check anyway.
+	 */
+	error = xfs_buf_delwri_submit(&buffers_list);
+	if (error)
+		goto out;
+	if (!list_empty(&buffers_list)) {
+		ASSERT(list_empty(&buffers_list));
+		return -EIO;
+	}
+
+	/* Set the new inode size, if needed. */
+	if (sc->tempip->i_disk_size != isize) {
+		sc->tempip->i_disk_size = isize;
+		i_size_write(VFS_I(sc->tempip), isize);
+		xfs_trans_log_inode(sc->tp, sc->tempip, XFS_ILOG_CORE);
+		return xrep_roll_trans(sc);
+	}
+
+	return 0;
+out:
+	xfs_buf_delwri_cancel(&buffers_list);
+	return error;
+}
diff --git a/fs/xfs/scrub/tempfile.h b/fs/xfs/scrub/tempfile.h
index a6a4c8d6a373..4b9d56399cc2 100644
--- a/fs/xfs/scrub/tempfile.h
+++ b/fs/xfs/scrub/tempfile.h
@@ -13,6 +13,17 @@ void xrep_tempfile_rele(struct xfs_scrub *sc);
 void xrep_tempfile_ilock(struct xfs_scrub *sc, unsigned int ilock_flags);
 bool xrep_tempfile_ilock_nowait(struct xfs_scrub *sc, unsigned int ilock_flags);
 void xrep_tempfile_iunlock(struct xfs_scrub *sc, unsigned int ilock_flags);
+
+int xrep_tempfile_prealloc(struct xfs_scrub *sc, xfs_fileoff_t off,
+		xfs_filblks_t len);
+
+enum xfs_blft;
+
+typedef int (*xrep_tempfile_copyin_getbuf_fn)(struct xfs_scrub *sc,
+		xfs_fileoff_t off, struct xfs_buf **bpp);
+int xrep_tempfile_copyin(struct xfs_scrub *sc,
+		const struct xfs_buf_ops *ops, enum xfs_blft type,
+		xfs_fileoff_t isize);
 #else
 # define xrep_tempfile_rele(sc)
 #endif /* CONFIG_XFS_ONLINE_REPAIR */