Skip to contentSkip to conversation

xfs: add missing cmap->br_state = XFS_EXT_NORM update

PATCH In review Gao Xiang 2 participants 5 messages Feb 17, 2022 – Mar 1, 2022 XFS filesystem

lore
Share & tools
lore

Gao Xiang

author
#0 lore raw

PATCH 1/1 xfs: add missing cmap->br_state = XFS_EXT_NORM update

COW extents are already converted into written real extents after xfs_reflink_convert_cow_locked(), therefore cmap->br_state should reflect it.

Otherwise, there is another necessary unwritten convertion triggered in xfs_dio_write_end_io() for direct I/O cases.

---
From the logic itself and runtime tracing, IMO, it seems true. Kindly correct me here if my understanding is wrong.

 fs/xfs/xfs_reflink.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
fs/xfs/xfs_reflink.c +4 -1
index 75bd2e03cd5b..5f0a364739a5 100644
@@ -424,7 +424,10 @@ xfs_reflink_allocate_cow(
 	if (!convert_now || cmap->br_state == XFS_EXT_NORM)
 		return 0;
 	trace_xfs_reflink_convert_cow(ip, cmap);
-	return xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	error = xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	if (!error)
+		cmap->br_state = XFS_EXT_NORM;
+	return error;
 
 out_trans_cancel:
 	xfs_trans_cancel(tp);
Signature · 1 line
2.24.4
#1 lore raw
Quoted 8 lines · Gao Xiang

COW extents are already converted into written real extents after xfs_reflink_convert_cow_locked(), therefore cmap->br_state should reflect it.

Otherwise, there is another necessary unwritten convertion triggered in xfs_dio_write_end_io() for direct I/O cases.

I /think/ this looks ok. Does it test ok too? AFAICT nothing in the iomap/writeback machinery cares the incorrect state because we always set IOMAP_F_SHARED (which triggers COW) so I think this is simply a fix for directio, like you said?

--D

Quoted 26 lines

---
From the logic itself and runtime tracing, IMO, it seems true. Kindly correct me here if my understanding is wrong.

 fs/xfs/xfs_reflink.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
fs/xfs/xfs_reflink.c +4 -1
index 75bd2e03cd5b..5f0a364739a5 100644
@@ -424,7 +424,10 @@ xfs_reflink_allocate_cow(
 	if (!convert_now || cmap->br_state == XFS_EXT_NORM)
 		return 0;
 	trace_xfs_reflink_convert_cow(ip, cmap);
-	return xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	error = xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	if (!error)
+		cmap->br_state = XFS_EXT_NORM;
+	return error;
 
 out_trans_cancel:
 	xfs_trans_cancel(tp);
Signature · 1 line
2.24.4
author
#2 lore raw

Hi Darrick,

Quoted 8 lines · Darrick J. Wong
Quoted 7 lines · Gao Xiang

COW extents are already converted into written real extents after xfs_reflink_convert_cow_locked(), therefore cmap->br_state should reflect it.

Otherwise, there is another necessary unwritten convertion triggered in xfs_dio_write_end_io() for direct I/O cases.

Quoted 6 lines
Quoted 1 line

I /think/ this looks ok. Does it test ok too? AFAICT nothing in the iomap/writeback machinery cares the incorrect state because we always set IOMAP_F_SHARED (which triggers COW) so I think this is simply a fix for directio, like you said?

Yeah, I think so, from the code logic buffered i/o seems no impacted... And the unnecessary unwritten convertion under direct i/o takes noticeable extra overhead in our workloads...

I checked my last night xfstests, it seems it stops unexpectedly (maybe due to some environmental problem). I will rerun tests today and feedback later.

Thanks,
Gao Xiang

Quoted 29 lines

--D

Quoted 26 lines

---
From the logic itself and runtime tracing, IMO, it seems true. Kindly correct me here if my understanding is wrong.

 fs/xfs/xfs_reflink.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
fs/xfs/xfs_reflink.c +4 -1
index 75bd2e03cd5b..5f0a364739a5 100644
@@ -424,7 +424,10 @@ xfs_reflink_allocate_cow(
 	if (!convert_now || cmap->br_state == XFS_EXT_NORM)
 		return 0;
 	trace_xfs_reflink_convert_cow(ip, cmap);
-	return xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	error = xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
+	if (!error)
+		cmap->br_state = XFS_EXT_NORM;
+	return error;
 
 out_trans_cancel:
 	xfs_trans_cancel(tp);
Signature · 1 line
2.24.4
author
#3 lore raw
Quoted 18 lines · Gao Xiang

Hi Darrick,

Quoted 8 lines · Darrick J. Wong
Quoted 7 lines · Gao Xiang

COW extents are already converted into written real extents after xfs_reflink_convert_cow_locked(), therefore cmap->br_state should reflect it.

Otherwise, there is another necessary unwritten convertion triggered in xfs_dio_write_end_io() for direct I/O cases.

Quoted 6 lines
Quoted 1 line

I /think/ this looks ok. Does it test ok too? AFAICT nothing in the iomap/writeback machinery cares the incorrect state because we always set IOMAP_F_SHARED (which triggers COW) so I think this is simply a fix for directio, like you said?

Quoted 7 lines

Yeah, I think so, from the code logic buffered i/o seems no impacted... And the unnecessary unwritten convertion under direct i/o takes noticeable extra overhead in our workloads...

I checked my last night xfstests, it seems it stops unexpectedly (maybe due to some environmental problem). I will rerun tests today and feedback later.

On my test environment machine, with linux 5.17-rc4, both w/ and w/o the patch, it fails:

generic/594 generic/600 xfs/158 xfs/160

I think no issue with this patch.

Thanks,
Gao Xiang

author
#4 lore raw
Quoted 27 lines · Gao Xiang
Quoted 18 lines · Gao Xiang

Hi Darrick,

Quoted 8 lines · Darrick J. Wong
Quoted 7 lines · Gao Xiang

COW extents are already converted into written real extents after xfs_reflink_convert_cow_locked(), therefore cmap->br_state should reflect it.

Otherwise, there is another necessary unwritten convertion triggered in xfs_dio_write_end_io() for direct I/O cases.

Quoted 6 lines
Quoted 1 line

I /think/ this looks ok. Does it test ok too? AFAICT nothing in the iomap/writeback machinery cares the incorrect state because we always set IOMAP_F_SHARED (which triggers COW) so I think this is simply a fix for directio, like you said?

Quoted 7 lines

Yeah, I think so, from the code logic buffered i/o seems no impacted... And the unnecessary unwritten convertion under direct i/o takes noticeable extra overhead in our workloads...

I checked my last night xfstests, it seems it stops unexpectedly (maybe due to some environmental problem). I will rerun tests today and feedback later.

Quoted 6 lines

On my test environment machine, with linux 5.17-rc4, both w/ and w/o the patch, it fails:

generic/594 generic/600 xfs/158 xfs/160

I think no issue with this patch.

ping.. it seems a simple fix for direct I/O..

Thanks,
Gao Xiang

Quoted 3 lines

Thanks,
Gao Xiang