Merge tag 'iomap-write-race-testing-6.2_2022-11-30' of git://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfs-linux into xfs-6.2-mergeC

xfs: add knobs for testing iomap write race fixes

This series is a followup to Dave Chinner's series entitled
"xfs, iomap: fix data corruption due to stale cached iomaps".
The two patches here add debugging knobs to introduce artificial delays
into the pagecache write and writeback code to facilitate testing of the
iomap invalidation code.  New tracepoints are also introduced so that
fstests can look for the invalidations.

Signed-off-by: Darrick J. Wong <[email protected]>

* tag 'iomap-write-race-testing-6.2_2022-11-30' of git://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfs-linux:
  xfs: add debug knob to slow down write for fun
  xfs: add debug knob to slow down writeback for fun

7 files changed, 149 insertions, 5 deletions

diff --git a/fs/xfs/libxfs/xfs_errortag.h b/fs/xfs/libxfs/xfs_errortag.h
index 580ccbd5aadc..01a9e86b3037 100644
--- a/fs/xfs/libxfs/xfs_errortag.h
+++ b/fs/xfs/libxfs/xfs_errortag.h
@@ -61,7 +61,9 @@
 #define XFS_ERRTAG_LARP					39
 #define XFS_ERRTAG_DA_LEAF_SPLIT			40
 #define XFS_ERRTAG_ATTR_LEAF_TO_NODE			41
-#define XFS_ERRTAG_MAX					42
+#define XFS_ERRTAG_WB_DELAY_MS				42
+#define XFS_ERRTAG_WRITE_DELAY_MS			43
+#define XFS_ERRTAG_MAX					44
 
 /*
  * Random factors for above tags, 1 means always, 2 means 1/2 time, etc.
@@ -107,5 +109,7 @@
 #define XFS_RANDOM_LARP					1
 #define XFS_RANDOM_DA_LEAF_SPLIT			1
 #define XFS_RANDOM_ATTR_LEAF_TO_NODE			1
+#define XFS_RANDOM_WB_DELAY_MS				3000
+#define XFS_RANDOM_WRITE_DELAY_MS			3000
 
 #endif /* __XFS_ERRORTAG_H_ */
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index a22d90af40c8..41734202796f 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -17,6 +17,8 @@
 #include "xfs_bmap.h"
 #include "xfs_bmap_util.h"
 #include "xfs_reflink.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
 
 struct xfs_writepage_ctx {
 	struct iomap_writepage_ctx ctx;
@@ -217,11 +219,17 @@ xfs_imap_valid(
 	 * checked (and found nothing at this offset) could have added
 	 * overlapping blocks.
 	 */
-	if (XFS_WPC(wpc)->data_seq != READ_ONCE(ip->i_df.if_seq))
+	if (XFS_WPC(wpc)->data_seq != READ_ONCE(ip->i_df.if_seq)) {
+		trace_xfs_wb_data_iomap_invalid(ip, &wpc->iomap,
+				XFS_WPC(wpc)->data_seq, XFS_DATA_FORK);
 		return false;
+	}
 	if (xfs_inode_has_cow_data(ip) &&
-	    XFS_WPC(wpc)->cow_seq != READ_ONCE(ip->i_cowfp->if_seq))
+	    XFS_WPC(wpc)->cow_seq != READ_ONCE(ip->i_cowfp->if_seq)) {
+		trace_xfs_wb_cow_iomap_invalid(ip, &wpc->iomap,
+				XFS_WPC(wpc)->cow_seq, XFS_COW_FORK);
 		return false;
+	}
 	return true;
 }
 
@@ -285,6 +293,8 @@ xfs_map_blocks(
 	if (xfs_is_shutdown(mp))
 		return -EIO;
 
+	XFS_ERRORTAG_DELAY(mp, XFS_ERRTAG_WB_DELAY_MS);
+
 	/*
 	 * COW fork blocks can overlap data fork blocks even if the blocks
 	 * aren't shared.  COW I/O always takes precedent, so we must always
diff --git a/fs/xfs/xfs_error.c b/fs/xfs/xfs_error.c
index dea3c0649d2f..713341d246d1 100644
--- a/fs/xfs/xfs_error.c
+++ b/fs/xfs/xfs_error.c
@@ -60,6 +60,8 @@ static unsigned int xfs_errortag_random_default[] = {
 	XFS_RANDOM_LARP,
 	XFS_RANDOM_DA_LEAF_SPLIT,
 	XFS_RANDOM_ATTR_LEAF_TO_NODE,
+	XFS_RANDOM_WB_DELAY_MS,
+	XFS_RANDOM_WRITE_DELAY_MS,
 };
 
 struct xfs_errortag_attr {
@@ -175,6 +177,8 @@ XFS_ERRORTAG_ATTR_RW(ag_resv_fail, XFS_ERRTAG_AG_RESV_FAIL);
 XFS_ERRORTAG_ATTR_RW(larp,		XFS_ERRTAG_LARP);
 XFS_ERRORTAG_ATTR_RW(da_leaf_split,	XFS_ERRTAG_DA_LEAF_SPLIT);
 XFS_ERRORTAG_ATTR_RW(attr_leaf_to_node,	XFS_ERRTAG_ATTR_LEAF_TO_NODE);
+XFS_ERRORTAG_ATTR_RW(wb_delay_ms,	XFS_ERRTAG_WB_DELAY_MS);
+XFS_ERRORTAG_ATTR_RW(write_delay_ms,	XFS_ERRTAG_WRITE_DELAY_MS);
 
 static struct attribute *xfs_errortag_attrs[] = {
 	XFS_ERRORTAG_ATTR_LIST(noerror),
@@ -218,6 +222,8 @@ static struct attribute *xfs_errortag_attrs[] = {
 	XFS_ERRORTAG_ATTR_LIST(larp),
 	XFS_ERRORTAG_ATTR_LIST(da_leaf_split),
 	XFS_ERRORTAG_ATTR_LIST(attr_leaf_to_node),
+	XFS_ERRORTAG_ATTR_LIST(wb_delay_ms),
+	XFS_ERRORTAG_ATTR_LIST(write_delay_ms),
 	NULL,
 };
 ATTRIBUTE_GROUPS(xfs_errortag);
@@ -268,6 +274,19 @@ xfs_errortag_valid(
 }
 
 bool
+xfs_errortag_enabled(
+	struct xfs_mount	*mp,
+	unsigned int		tag)
+{
+	if (!mp->m_errortag)
+		return false;
+	if (!xfs_errortag_valid(tag))
+		return false;
+
+	return mp->m_errortag[tag] != 0;
+}
+
+bool
 xfs_errortag_test(
 	struct xfs_mount	*mp,
 	const char		*expression,
diff --git a/fs/xfs/xfs_error.h b/fs/xfs/xfs_error.h
index 5191e9145e55..dbe6c37dc697 100644
--- a/fs/xfs/xfs_error.h
+++ b/fs/xfs/xfs_error.h
@@ -45,6 +45,18 @@ extern bool xfs_errortag_test(struct xfs_mount *mp, const char *expression,
 		const char *file, int line, unsigned int error_tag);
 #define XFS_TEST_ERROR(expr, mp, tag)		\
 	((expr) || xfs_errortag_test((mp), #expr, __FILE__, __LINE__, (tag)))
+bool xfs_errortag_enabled(struct xfs_mount *mp, unsigned int tag);
+#define XFS_ERRORTAG_DELAY(mp, tag)		\
+	do { \
+		might_sleep(); \
+		if (!xfs_errortag_enabled((mp), (tag))) \
+			break; \
+		xfs_warn_ratelimited((mp), \
+"Injecting %ums delay at file %s, line %d, on filesystem \"%s\"", \
+				(mp)->m_errortag[(tag)], __FILE__, __LINE__, \
+				(mp)->m_super->s_id); \
+		mdelay((mp)->m_errortag[(tag)]); \
+	} while (0)
 
 extern int xfs_errortag_get(struct xfs_mount *mp, unsigned int error_tag);
 extern int xfs_errortag_set(struct xfs_mount *mp, unsigned int error_tag,
@@ -55,6 +67,7 @@ extern int xfs_errortag_clearall(struct xfs_mount *mp);
 #define xfs_errortag_init(mp)			(0)
 #define xfs_errortag_del(mp)
 #define XFS_TEST_ERROR(expr, mp, tag)		(expr)
+#define XFS_ERRORTAG_DELAY(mp, tag)		((void)0)
 #define xfs_errortag_set(mp, tag, val)		(ENOSYS)
 #define xfs_errortag_add(mp, tag)		(ENOSYS)
 #define xfs_errortag_clearall(mp)		(ENOSYS)
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 1bdd7afc1010..1005f1e36545 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -27,6 +27,8 @@
 #include "xfs_dquot_item.h"
 #include "xfs_dquot.h"
 #include "xfs_reflink.h"
+#include "xfs_error.h"
+#include "xfs_errortag.h"
 
 #define XFS_ALLOC_ALIGN(mp, off) \
 	(((off) >> mp->m_allocsize_log) << mp->m_allocsize_log)
@@ -71,8 +73,16 @@ xfs_iomap_valid(
 	struct inode		*inode,
 	const struct iomap	*iomap)
 {
-	return iomap->validity_cookie ==
-			xfs_iomap_inode_sequence(XFS_I(inode), iomap->flags);
+	struct xfs_inode	*ip = XFS_I(inode);
+
+	if (iomap->validity_cookie !=
+			xfs_iomap_inode_sequence(ip, iomap->flags)) {
+		trace_xfs_iomap_invalid(ip, iomap);
+		return false;
+	}
+
+	XFS_ERRORTAG_DELAY(ip->i_mount, XFS_ERRTAG_WRITE_DELAY_MS);
+	return true;
 }
 
 const struct iomap_page_ops xfs_iomap_page_ops = {
diff --git a/fs/xfs/xfs_trace.c b/fs/xfs/xfs_trace.c
index d269ef57ff01..8a5dc1538aa8 100644
--- a/fs/xfs/xfs_trace.c
+++ b/fs/xfs/xfs_trace.c
@@ -34,6 +34,8 @@
 #include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 #include "xfs_error.h"
+#include <linux/iomap.h>
+#include "xfs_iomap.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 372d871bccc5..421d1e504ac4 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -3352,6 +3352,92 @@ DEFINE_EVENT(xfs_inode_irec_class, name, \
 	TP_PROTO(struct xfs_inode *ip, struct xfs_bmbt_irec *irec), \
 	TP_ARGS(ip, irec))
 
+/* inode iomap invalidation events */
+DECLARE_EVENT_CLASS(xfs_wb_invalid_class,
+	TP_PROTO(struct xfs_inode *ip, const struct iomap *iomap, unsigned int wpcseq, int whichfork),
+	TP_ARGS(ip, iomap, wpcseq, whichfork),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(u64, addr)
+		__field(loff_t, pos)
+		__field(u64, len)
+		__field(u16, type)
+		__field(u16, flags)
+		__field(u32, wpcseq)
+		__field(u32, forkseq)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->addr = iomap->addr;
+		__entry->pos = iomap->offset;
+		__entry->len = iomap->length;
+		__entry->type = iomap->type;
+		__entry->flags = iomap->flags;
+		__entry->wpcseq = wpcseq;
+		__entry->forkseq = READ_ONCE(xfs_ifork_ptr(ip, whichfork)->if_seq);
+	),
+	TP_printk("dev %d:%d ino 0x%llx pos 0x%llx addr 0x%llx bytecount 0x%llx type 0x%x flags 0x%x wpcseq 0x%x forkseq 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->pos,
+		  __entry->addr,
+		  __entry->len,
+		  __entry->type,
+		  __entry->flags,
+		  __entry->wpcseq,
+		  __entry->forkseq)
+);
+#define DEFINE_WB_INVALID_EVENT(name) \
+DEFINE_EVENT(xfs_wb_invalid_class, name, \
+	TP_PROTO(struct xfs_inode *ip, const struct iomap *iomap, unsigned int wpcseq, int whichfork), \
+	TP_ARGS(ip, iomap, wpcseq, whichfork))
+DEFINE_WB_INVALID_EVENT(xfs_wb_cow_iomap_invalid);
+DEFINE_WB_INVALID_EVENT(xfs_wb_data_iomap_invalid);
+
+DECLARE_EVENT_CLASS(xfs_iomap_invalid_class,
+	TP_PROTO(struct xfs_inode *ip, const struct iomap *iomap),
+	TP_ARGS(ip, iomap),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(u64, addr)
+		__field(loff_t, pos)
+		__field(u64, len)
+		__field(u64, validity_cookie)
+		__field(u64, inodeseq)
+		__field(u16, type)
+		__field(u16, flags)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->addr = iomap->addr;
+		__entry->pos = iomap->offset;
+		__entry->len = iomap->length;
+		__entry->validity_cookie = iomap->validity_cookie;
+		__entry->type = iomap->type;
+		__entry->flags = iomap->flags;
+		__entry->inodeseq = xfs_iomap_inode_sequence(ip, iomap->flags);
+	),
+	TP_printk("dev %d:%d ino 0x%llx pos 0x%llx addr 0x%llx bytecount 0x%llx type 0x%x flags 0x%x validity_cookie 0x%llx inodeseq 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->pos,
+		  __entry->addr,
+		  __entry->len,
+		  __entry->type,
+		  __entry->flags,
+		  __entry->validity_cookie,
+		  __entry->inodeseq)
+);
+#define DEFINE_IOMAP_INVALID_EVENT(name) \
+DEFINE_EVENT(xfs_iomap_invalid_class, name, \
+	TP_PROTO(struct xfs_inode *ip, const struct iomap *iomap), \
+	TP_ARGS(ip, iomap))
+DEFINE_IOMAP_INVALID_EVENT(xfs_iomap_invalid);
+
 /* refcount/reflink tracepoint definitions */
 
 /* reflink tracepoints */