Skip to contentSkip to conversation

iomap: don't bother unsharing delalloc extents

PATCH Applied Darrick J. Wong 5 participants 9 messages Oct 2, 2024 – Oct 3, 2024

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 iomap: don't bother unsharing delalloc extents

From: Darrick J. Wong <[email protected]>

If unshare encounters a delalloc reservation in the srcmap, that means that the file range isn't shared because delalloc reservations cannot be reflinked. Therefore, don't try to unshare them.

---
 fs/iomap/buffered-io.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/iomap/buffered-io.c +2 -1
index 11ea747228aee..c1c559e0cc07c 100644
@@ -1321,7 +1321,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 		return length;
 
 	/*
-	 * Don't bother with holes or unwritten extents.
+	 * Don't bother with delalloc reservations, holes or unwritten extents.
 	 *
 	 * Note that we use srcmap directly instead of iomap_iter_srcmap as
 	 * unsharing requires providing a separate source map, and the presence
@@ -1330,6 +1330,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 	 * fork for XFS.
 	 */
 	if (iter->srcmap.type == IOMAP_HOLE ||
+	    iter->srcmap.type == IOMAP_DELALLOC ||
 	    iter->srcmap.type == IOMAP_UNWRITTEN)
 		return length;
 
#1 lore raw

PATCH 2/2 iomap: constrain the file range passed to iomap_file_unshare

From: Darrick J. Wong <[email protected]>

File contents can only be shared (i.e. reflinked) below EOF, so it makes no sense to try to unshare ranges beyond EOF. Constrain the file range parameters here so that we don't have to do that in the callers.

Fixes 5f4e5752a8a3 ("fs: add iomap_file_dirty")
---
 fs/dax.c               |    6 +++++-
 fs/iomap/buffered-io.c |    6 +++++-
 2 files changed, 10 insertions(+), 2 deletions(-)
fs/dax.c +5 -1
index becb4a6920c6a..c62acd2812f8d 100644
@@ -1305,11 +1305,15 @@ int dax_file_unshare(struct inode *inode, loff_t pos, loff_t len,
 	struct iomap_iter iter = {
 		.inode		= inode,
 		.pos		= pos,
-		.len		= len,
 		.flags		= IOMAP_WRITE | IOMAP_UNSHARE | IOMAP_DAX,
 	};
+	loff_t size = i_size_read(inode);
 	int ret;
 
+	if (pos < 0 || pos >= size)
+		return 0;
+
+	iter.len = min(len, size - pos);
 	while ((ret = iomap_iter(&iter, ops)) > 0)
 		iter.processed = dax_unshare_iter(&iter);
 	return ret;
fs/iomap/buffered-io.c +5 -1
index c1c559e0cc07c..78ebd265f4259 100644
@@ -1375,11 +1375,15 @@ iomap_file_unshare(struct inode *inode, loff_t pos, loff_t len,
 	struct iomap_iter iter = {
 		.inode		= inode,
 		.pos		= pos,
-		.len		= len,
 		.flags		= IOMAP_WRITE | IOMAP_UNSHARE,
 	};
+	loff_t size = i_size_read(inode);
 	int ret;
 
+	if (pos < 0 || pos >= size)
+		return 0;
+
+	iter.len = min(len, size - pos);
 	while ((ret = iomap_iter(&iter, ops)) > 0)
 		iter.processed = iomap_unshare_iter(&iter);
 	return ret;
Reviewed-by on v1 2/2
#2 lore raw

Re: [PATCH 2/2] iomap: constrain the file range passed to iomap_file_unshare

Quoted 5 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

File contents can only be shared (i.e. reflinked) below EOF, so it makes no sense to try to unshare ranges beyond EOF. Constrain the file range parameters here so that we don't have to do that in the callers.

Looks good:

Reviewed-by on v1 2/2
#3 lore raw

Re: [PATCH 2/2] iomap: constrain the file range passed to iomap_file_unshare

In reply to Darrick J. Wong
Quoted 46 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

File contents can only be shared (i.e. reflinked) below EOF, so it makes no sense to try to unshare ranges beyond EOF. Constrain the file range parameters here so that we don't have to do that in the callers.

Fixes 5f4e5752a8a3 ("fs: add iomap_file_dirty")
---
 fs/dax.c               |    6 +++++-
 fs/iomap/buffered-io.c |    6 +++++-
 2 files changed, 10 insertions(+), 2 deletions(-)
fs/dax.c +5 -1
index becb4a6920c6a..c62acd2812f8d 100644
@@ -1305,11 +1305,15 @@ int dax_file_unshare(struct inode *inode, loff_t pos, loff_t len,
 	struct iomap_iter iter = {
 		.inode		= inode,
 		.pos		= pos,
-		.len		= len,
 		.flags		= IOMAP_WRITE | IOMAP_UNSHARE | IOMAP_DAX,
 	};
+	loff_t size = i_size_read(inode);
 	int ret;
 
+	if (pos < 0 || pos >= size)
+		return 0;
+
+	iter.len = min(len, size - pos);
 	while ((ret = iomap_iter(&iter, ops)) > 0)
 		iter.processed = dax_unshare_iter(&iter);
 	return ret;
fs/iomap/buffered-io.c +0 -1
index c1c559e0cc07c..78ebd265f4259 100644
@@ -1375,11 +1375,15 @@ iomap_file_unshare(struct inode *inode, loff_t pos, loff_t len,
 	struct iomap_iter iter = {
 		.inode		= inode,
 		.pos		= pos,
-		.len		= len,
 		.flags		= IOMAP_WRITE | IOMAP_UNSHARE,
 	};

fs/iomap/buffered-io.c iomap_file_unshare(struct inode *inode, loff_t pos, loff_t len, · line 1389

Quoted 10 lines
fs/iomap/buffered-io.c +5 -0
+	loff_t size = i_size_read(inode);
 	int ret;
 
+	if (pos < 0 || pos >= size)
+		return 0;
+
+	iter.len = min(len, size - pos);
 	while ((ret = iomap_iter(&iter, ops)) > 0)
 		iter.processed = iomap_unshare_iter(&iter);
 	return ret;

Heh. This was pretty much my local fix when I was testing fsx unshare range, so LGTM. Apologies, I probably should have just sent it out. It just seemed like Julian was 90% there, but then review went off the rails and I guess I lost interest. Anyways, thanks for the fix:

#4 lore raw

Re: [PATCH 2/2] iomap: constrain the file range passed to iomap_file_unshare

In reply to Darrick J. Wong
Quoted 50 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

File contents can only be shared (i.e. reflinked) below EOF, so it makes no sense to try to unshare ranges beyond EOF.  Constrain the file range parameters here so that we don't have to do that in the callers.

Fixes 5f4e5752a8a3 ("fs: add iomap_file_dirty")

---
 fs/dax.c               |    6 +++++-
 fs/iomap/buffered-io.c |    6 +++++-

 2 files changed, 10 insertions(+), 2 deletions(-)
fs/dax.c +0 -1
index becb4a6920c6a..c62acd2812f8d 100644
@@ -1305,11 +1305,15 @@ int dax_file_unshare(struct inode *inode, loff_t
 pos, loff_t len,
         struct iomap_iter iter = {
                 .inode          = inode,
                 .pos            = pos,
-               .len            = len,

                .flags          = IOMAP_WRITE | IOMAP_UNSHARE | IOMAP_DAX,
        };
+       loff_t size = i_size_read(inode);
        int ret;

+       if (pos < 0 || pos >= size)
+               return 0;
+
+       iter.len = min(len, size - pos);
        while ((ret = iomap_iter(&iter, ops)) > 0) iter.processed = dax_unshare_iter(&iter); return ret;

fs/iomap/buffered-io.c +1 -1
index c1c559e0cc07c..78ebd265f4259 100644
@@ -1375,11 +1375,15 @@ iomap_file_unshare(struct inode *inode, loff_t
 pos, loff_t len,
         struct iomap_iter iter = {
                 .inode          = inode,
                 .pos            = pos,
-               .len            = len,
                 .flags          = IOMAP_WRITE | IOMAP_UNSHARE,
         };
+       loff_t size = i_size_read(inode);
         int ret;
Quoted 8 lines


+       if (pos < 0 || pos >= size)
+               return 0;
+
+       iter.len = min(len, size - pos);
        while ((ret = iomap_iter(&iter, ops)) > 0) iter.processed = iomap_unshare_iter(&iter); return ret;

Sorry I didn’t update the patch sooner—I was planning to get to it after wrapping up my week-long vacation... Anyway, thanks for the patch, it looks great!

Thanks,

Signature · 1 line
Julian Sun <[email protected]>
Reviewed-by on v1 1/2
#6 lore raw
In reply to Darrick J. Wong
Quoted 26 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

If unshare encounters a delalloc reservation in the srcmap, that means that the file range isn't shared because delalloc reservations cannot be reflinked. Therefore, don't try to unshare them.

---
 fs/iomap/buffered-io.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/iomap/buffered-io.c +1 -1
index 11ea747228aee..c1c559e0cc07c 100644
@@ -1321,7 +1321,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 		return length;
 
 	/*
-	 * Don't bother with holes or unwritten extents.
+	 * Don't bother with delalloc reservations, holes or unwritten extents.
 	 *
 	 * Note that we use srcmap directly instead of iomap_iter_srcmap as
 	 * unsharing requires providing a separate source map, and the presence

fs/iomap/buffered-io.c static loff_t iomap_unshare_iter(struct iomap_iter *iter) · line 1336

Quoted 7 lines
fs/iomap/buffered-io.c +1 -0
@@ -1330,6 +1330,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 	 * fork for XFS.
 	 */
 	if (iter->srcmap.type == IOMAP_HOLE ||
+	    iter->srcmap.type == IOMAP_DELALLOC ||
 	    iter->srcmap.type == IOMAP_UNWRITTEN)
 		return length;
 

IIUC in the case of shared blocks srcmap always refers to the data fork (so delalloc in the COW fork is not an issue). If so:

#7 lore raw
Quoted 29 lines · Brian Foster
Quoted 28 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

If unshare encounters a delalloc reservation in the srcmap, that means that the file range isn't shared because delalloc reservations cannot be reflinked. Therefore, don't try to unshare them.

---
 fs/iomap/buffered-io.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/iomap/buffered-io.c +1 -1
index 11ea747228aee..c1c559e0cc07c 100644
@@ -1321,7 +1321,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 		return length;
 
 	/*
-	 * Don't bother with holes or unwritten extents.
+	 * Don't bother with delalloc reservations, holes or unwritten extents.
 	 *
 	 * Note that we use srcmap directly instead of iomap_iter_srcmap as
 	 * unsharing requires providing a separate source map, and the presence
@@ -1330,6 +1330,7 @@ static loff_t iomap_unshare_iter(struct iomap_iter *iter)
 	 * fork for XFS.
 	 */
 	if (iter->srcmap.type == IOMAP_HOLE ||
Quoted 8 lines

fs/iomap/buffered-io.c static loff_t iomap_unshare_iter(struct iomap_iter *iter) · line 1336

Quoted 5 lines
fs/iomap/buffered-io.c +1 -0
+	    iter->srcmap.type == IOMAP_DELALLOC ||
 	    iter->srcmap.type == IOMAP_UNWRITTEN)
 		return length;
 

IIUC in the case of shared blocks srcmap always refers to the data fork (so delalloc in the COW fork is not an issue). If so:

Yep.

Quoted 1 line

Thanks!

--D

maintainer Applied
#8 lore raw
In reply to Darrick J. Wong
Quoted 5 lines · Darrick J. Wong

If unshare encounters a delalloc reservation in the srcmap, that means that the file range isn't shared because delalloc reservations cannot be reflinked. Therefore, don't try to unshare them.

Applied to the vfs.fixes branch of the vfs/vfs.git tree. Patches in the vfs.fixes branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a new review to the original patch series allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the patch has now been applied. If possible patch trailers will be updated.

Note that commit hashes shown below are subject to change due to rebase, trailer updates or similar. If in doubt, please check the listed branch.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: vfs.fixes

[1/2] iomap: don't bother unsharing delalloc extents

      https://git.kernel.org/vfs/vfs/c/f7a4874d977b
[2/2] iomap: constrain the file range passed to iomap_file_unshare
      https://git.kernel.org/vfs/vfs/c/a311a08a4237