[PATCH 0/2] fuse: inode blocksize fixes for iomap integration

[PATCH 0/2] fuse: inode blocksize fixes for iomap integration

[Joanne Koong]

These 2 patches are fixes for inode blocksize usage in fuse.

The first is neededed to maintain current stat() behavior for clients in the
case where fuse uses cached values for stat, and the second is needed for
fuseblk filesystems as a workaround until fuse implements iomap for reads.
(A previous version of the 2nd patch was submitted earlier [1], which this
current verison supersedes). These two patches are submitted together since
the 2nd patch updates code changed in the 1st patch.

These patches are on top of Christian's vfs.fixes tree.

Thanks,
Joanne

[1] https://lore.kernel.org/linux-fsdevel/20250812014623.2408476-1-joannelkoong@gmail.com/T/#u

Joanne Koong (2):
  fuse: reflect cached blocksize if blocksize was changed
  fuse: fix fuseblk i_blkbits for iomap partial writes

 fs/fuse/dir.c    |  3 ++-
 fs/fuse/fuse_i.h | 14 ++++++++++++++
 fs/fuse/inode.c  | 15 +++++++++++++++
 3 files changed, 31 insertions(+), 1 deletion(-)

-- 
2.47.3

[PATCH 1/2] fuse: reflect cached blocksize if blocksize was changed

[Joanne Koong]

As pointed out by Miklos[1], in the fuse_update_get_attr() path, the
attributes returned to stat may be cached values instead of fresh ones
fetched from the server. In the case where the server returned a
modified blocksize value, we need to cache it and reflect it back to
stat if values are not re-fetched since we now no longer directly change
inode->i_blkbits.

Link: https://lore.kernel.org/linux-fsdevel/CAJfpeguCOxeVX88_zPd1hqziB_C+tmfuDhZP5qO2nKmnb-dTUA@mail.gmail.com/ [1]

Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
Fixes: 542ede096e48 ("fuse: keep inode->i_blkbits constant)
---
 fs/fuse/dir.c    | 1 +
 fs/fuse/fuse_i.h | 6 ++++++
 fs/fuse/inode.c  | 5 +++++
 3 files changed, 12 insertions(+)

diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index 2d817d7cab26..ebee7e0b1cd3 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -1377,6 +1377,7 @@ static int fuse_update_get_attr(struct mnt_idmap *idmap, struct inode *inode,
 		generic_fillattr(idmap, request_mask, inode, stat);
 		stat->mode = fi->orig_i_mode;
 		stat->ino = fi->orig_ino;
+		stat->blksize = 1 << fi->cached_i_blkbits;
 		if (test_bit(FUSE_I_BTIME, &fi->state)) {
 			stat->btime = fi->i_btime;
 			stat->result_mask |= STATX_BTIME;
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index ec248d13c8bf..db44d05c8d02 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -210,6 +210,12 @@ struct fuse_inode {
 	/** Reference to backing file in passthrough mode */
 	struct fuse_backing *fb;
 #endif
+
+	/*
+	 * The underlying inode->i_blkbits value will not be modified,
+	 * so preserve the blocksize specified by the server.
+	 */
+	unsigned char cached_i_blkbits;
 };
 
 /** FUSE inode state bits */
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index 67c2318bfc42..3bfd83469d9f 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -289,6 +289,11 @@ void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr,
 		}
 	}
 
+	if (attr->blksize)
+		fi->cached_i_blkbits = ilog2(attr->blksize);
+	else
+		fi->cached_i_blkbits = inode->i_sb->s_blocksize_bits;
+
 	/*
 	 * Don't set the sticky bit in i_mode, unless we want the VFS
 	 * to check permissions.  This prevents failures due to the
-- 
2.47.3

[PATCH 2/2] fuse: fix fuseblk i_blkbits for iomap partial writes

[Joanne Koong]

On regular fuse filesystems, i_blkbits is set to PAGE_SHIFT which means
any iomap partial writes will mark the entire folio as uptodate. However
fuseblk filesystems work differently and allow the blocksize to be less
than the page size. As such, this may lead to data corruption if fuseblk
sets its blocksize to less than the page size, uses the writeback cache,
and does a partial write, then a read and the read happens before the
write has undergone writeback, since the folio will not be marked
uptodate from the partial write so the read will read in the entire
folio from disk, which will overwrite the partial write.

The long-term solution for this, which will also be needed for fuse to
enable large folios with the writeback cache on, is to have fuse also
use iomap for folio reads, but until that is done, the cleanest
workaround is to use the page size for fuseblk's internal kernel inode
blksize/blkbits values while maintaining current behavior for stat().

This was verified using ntfs-3g:
$ sudo mkfs.ntfs -f -c 512 /dev/vdd1
$ sudo ntfs-3g /dev/vdd1 ~/fuseblk
$ stat ~/fuseblk/hi.txt
IO Block: 512

Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
Fixes: a4c9ab1d4975 ("fuse: use iomap for buffered writes")
---
 fs/fuse/dir.c    |  2 +-
 fs/fuse/fuse_i.h |  8 ++++++++
 fs/fuse/inode.c  | 12 +++++++++++-
 3 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index ebee7e0b1cd3..5c569c3cb53f 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -1199,7 +1199,7 @@ static void fuse_fillattr(struct mnt_idmap *idmap, struct inode *inode,
 	if (attr->blksize != 0)
 		blkbits = ilog2(attr->blksize);
 	else
-		blkbits = inode->i_sb->s_blocksize_bits;
+		blkbits = fc->blkbits;
 
 	stat->blksize = 1 << blkbits;
 }
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index db44d05c8d02..a6aa16422c30 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -975,6 +975,14 @@ struct fuse_conn {
 		/* Request timeout (in jiffies). 0 = no timeout */
 		unsigned int req_timeout;
 	} timeout;
+
+	/*
+	 * This is a workaround until fuse uses iomap for reads.
+	 * For fuseblk servers, this represents the blocksize passed in at
+	 * mount time and for regular fuse servers, this is equivalent to
+	 * inode->i_blkbits.
+	 */
+	unsigned char blkbits;
 };
 
 /*
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index 3bfd83469d9f..7e66419f4556 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -292,7 +292,7 @@ void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr,
 	if (attr->blksize)
 		fi->cached_i_blkbits = ilog2(attr->blksize);
 	else
-		fi->cached_i_blkbits = inode->i_sb->s_blocksize_bits;
+		fi->cached_i_blkbits = fc->blkbits;
 
 	/*
 	 * Don't set the sticky bit in i_mode, unless we want the VFS
@@ -1810,10 +1810,20 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx)
 		err = -EINVAL;
 		if (!sb_set_blocksize(sb, ctx->blksize))
 			goto err;
+		/*
+		 * This is a workaround until fuse hooks into iomap for reads.
+		 * Use PAGE_SIZE for the blocksize else if the writeback cache
+		 * is enabled, buffered writes go through iomap and a read
+		 * may overwrite partially written data if blocksize < PAGE_SIZE
+		 */
+		fc->blkbits = sb->s_blocksize_bits;
+		if (!sb_set_blocksize(sb, PAGE_SIZE))
+			goto err;
 #endif
 	} else {
 		sb->s_blocksize = PAGE_SIZE;
 		sb->s_blocksize_bits = PAGE_SHIFT;
+		fc->blkbits = sb->s_blocksize_bits;
 	}
 
 	sb->s_subtype = ctx->subtype;
-- 
2.47.3