smb/client: fix emulated insert range handling
Subsystems (2)
cover smb/client: fix emulated insert range handling
From: Huiwen He <[email protected]>
This series fixes two issues in the SMB3 FALLOC_FL_INSERT_RANGE emulation.
Patch 1 fixes missing new EOF validation, which allowed RLIMIT_FSIZE to be bypassed.
Patch 2 preserves hole semantics for inserted ranges, fixing xfstests generic/064 against Samba and ksmbd.
In my Windows Server testing, aligned 64 KiB and 128 KiB insert ranges were reported as holes, but generic/064 still fails because it uses 4 KiB inserts. Azure Files may not pass the test because it does not support sparse files.
Thanks,
Huiwen
Huiwen He (2):
smb/client: validate new EOF for insert range smb/client: mark file sparse before emulating insert range
fs/smb/client/smb2ops.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-)
Signature · 1 line
2.43.0
PATCH 1/2 smb/client: validate new EOF for insert range
From: Huiwen He <[email protected]>
smb3_insert_range() does not check if the new file size (i_size + len) is valid. This allows FALLOC_FL_INSERT_RANGE to bypass RLIMIT_FSIZE, exceed s_maxbytes, or produce a size outside the loff_t range.
Use check_add_overflow() to calculate the new EOF. Validate it with inode_newsize_ok() before modifying the file.
Reproducer, using a file on a CIFS mount:
bash -c '
FILE=/mnt/cifs/repro
trap "" SIGXFSZ
ulimit -f 3072 # RLIMIT_FSIZE = 3 MiB
# A regular write is stopped at 3 MiB.
dd if=/dev/zero of="$FILE" bs=1M count=4 status=none stat -c "size after write: %s" "$FILE"
# Insert 2 MiB into a 2 MiB file. truncate -s 2M "$FILE" fallocate -i -o 0 -l 2M "$FILE" stat -c "size after insert: %s" "$FILE"
'
Before this change, the regular write stops at the 3 MiB limit, but insert range grows the file to 4 MiB:
dd: error writing '/mnt/cifs/repro': File too large size after write: 3145728 size after insert: 4194304
After this change, insert range also fails at the limit and leaves the 2 MiB file unchanged:
dd: error writing '/mnt/cifs/repro': File too large size after write: 3145728 fallocate: fallocate failed: File too large size after insert: 2097152
--- fs/smb/client/smb2ops.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-)
fs/smb/client/smb2ops.c +10 -2
@@ -3985,7 +3985,8 @@ static long smb3_insert_range(struct file *file, struct cifs_tcon *tcon,struct cifsFileInfo *cfile = file->private_data;struct inode *inode =file_inode (file);struct cifsInodeInfo *cifsi =CIFS_I (inode); -__u64 count, old_eof, new_eof; +u64 count; +loff_t old_eof, new_eof; xid =get_xid ();
@@ -3995,8 +3996,15 @@ static long smb3_insert_range(struct file *file, struct cifs_tcon *tcon, goto out; } +if (check_add_overflow (old_eof, len, &new_eof)) { + rc = -EFBIG ; + goto out; + } + rc =inode_newsize_ok (inode, new_eof); +if (rc) + goto out; + count = old_eof - off; - new_eof = old_eof + len;filemap_invalidate_lock (inode->i_mapping); rc =filemap_write_and_wait_range (inode->i_mapping, off, new_eof -1 );
Signature · 1 line
2.43.0
PATCH 2/2 smb/client: mark file sparse before emulating insert range
In reply to Huiwen HeFrom: Huiwen He <[email protected]>
The SMB client emulates FALLOC_FL_INSERT_RANGE with SET_EOF, COPYCHUNK and SET_ZERO_DATA.
SET_ZERO_DATA creates a hole only when the file is sparse. On a non-sparse file, it clears the inserted range but leaves its blocks allocated, causing the extent count check in xfstests generic/064 to fail.
Try to mark the file sparse before starting the emulation. As with existing file extension paths, ignore SET_SPARSE errors and continue with the existing zero-filled behavior.
Tested with xfstests generic/064 against Samba and ksmbd.
--- fs/smb/client/smb2ops.c | 6 ++++++ 1 file changed, 6 insertions(+)
fs/smb/client/smb2ops.c +6 -0
@@ -4006,6 +4006,12 @@ static long smb3_insert_range(struct file *file, struct cifs_tcon *tcon, count = old_eof - off; +/* +* SET_ZERO_DATA creates a hole only in a sparse file. Try to mark the file +* sparse, but continue the emulation regardless of the result. +*/ +smb2_set_sparse (xid, tcon, cfile, inode, true); +filemap_invalidate_lock (inode->i_mapping); rc =filemap_write_and_wait_range (inode->i_mapping, off, new_eof -1 );if (rc <0 )
Signature · 1 line
2.43.0
Namjae Jeon
maintainerRe: [PATCH 2/2] smb/client: mark file sparse before emulating insert range
fs/smb/client/smb2ops.c static long smb3_insert_range(struct file *file, struct cifs_tcon *tcon, · line 4013
Quoted 13 lines
fs/smb/client/smb2ops.c +5
-0@@ -4006,6 +4006,12 @@ static long smb3_insert_range(struct file *file, struct cifs_tcon *tcon, count = old_eof - off; +/* +* SET_ZERO_DATA creates a hole only in a sparse file. Try to mark the file +* sparse, but continue the emulation regardless of the result. +*/ +smb2_set_sparse (xid, tcon, cfile, inode, true);
Shouldn't this error be propagated? Unlike ordinary file extension, FALLOC_FL_INSERT_RANGE is required to create an unallocated hole. If SET_SPARSE fails, SET_ZERO_DATA only writes allocated zeroes, but the operation still returns success...
Re: [PATCH 2/2] smb/client: mark file sparse before emulating insert range
Hi Namjae,
Thanks for the review. I have also found a few other issues in the FALLOC_FL_INSERT_RANGE path and am working on further improvements to this series. I will send a v2 as soon as it is ready.
Thanks,
Huiwen
Quoted 4 lines · Namjae Jeon
Shouldn't this error be propagated? Unlike ordinary file extension,
FALLOC_FL_INSERT_RANGEis required to create an unallocated hole. IfSET_SPARSEfails,SET_ZERO_DATAonly writes allocated zeroes, but the operation still returns success...