Skip to contentSkip to conversation

cifs: Fix validation of WSL-style special files

PATCH In review Pali Rohár 2 participants 10 messages Jun 8, 2025 – Jun 9, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover cifs: Fix validation of WSL-style special files

This patch series fixes validation of WSL special files by Linux SMB client, to be compatible with WSL subsystem.

Pali Rohár (5):
cifs: Remove setting duplicate fattr->cf_dtype from wsl_to_fattr()
function
cifs: Fix validation of EAs for WSL reparse points
cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size
cifs: Query EA $LXMOD in cifs_query_path_info() for WSL reparse points
cifs: Validate presence of EA $LXMOD for WSL reparse points

 fs/smb/client/reparse.c | 40 ++++++++++++++++++++------
 fs/smb/client/smb1ops.c | 62 +++++++++++++++++++++++++++++++++++++++--
 fs/smb/client/smb2pdu.h | 19 ++++++-------
 3 files changed, 101 insertions(+), 20 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/5 cifs: Remove duplicate fattr->cf_dtype assignment from wsl_to_fattr() function

Commit 8bd25b61c5a5 ("smb: client: set correct d_type for reparse DFS/DFSR and mount point") deduplicated assignment of fattr->cf_dtype member from all places to end of the function cifs_reparse_point_to_fattr(). The only one missing place which was not deduplicated is wsl_to_fattr(). Fix it.

Fixes 8bd25b61c5a5 ("smb: client: set correct d_type for reparse DFS/DFSR and mount point")
---
 fs/smb/client/reparse.c | 1 -
 1 file changed, 1 deletion(-)
fs/smb/client/reparse.c +0 -1
index d1d25f5f72ca..5351d3d9d538 100644
@@ -1176,7 +1176,6 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 	if (!have_xattr_dev && (tag == IO_REPARSE_TAG_LX_CHR || tag == IO_REPARSE_TAG_LX_BLK))
 		return false;
 
-	fattr->cf_dtype = S_DT(fattr->cf_mode);
 	return true;
 }
 
Signature · 1 line
2.20.1
#2 lore raw

PATCH 2/5 cifs: Fix validation of EAs for WSL reparse points

In reply to Pali Rohár

When wsl_to_fattr() is called from readdir() then we should skip validation of WSL EAs because readdir FIND_FIRST/NEXT results do not provide EA values at all. For readdir() reply in this case return DT_UNKNOWN type instead of DT_REG (which is the result of failed WSL EAs validation).

When validation is skipped and some required EAs are missing then returns true from wsl_to_fattr() function but do not set fattr->cf_mode. This makes readdir() to return DT_UNKNOWN type in the same way as it is doing NFS reparse point function.

This change fixes readdir() result that for some WSL reparse points returns DT_REG due to missing EAs. After this change it returns DT_UNKNOWN, which is better than returning wrong type.

Fixes ef201e8759d2 ("cifs: Validate EAs for WSL reparse points")
---
 fs/smb/client/reparse.c | 27 ++++++++++++++++++++-------
 1 file changed, 20 insertions(+), 7 deletions(-)
fs/smb/client/reparse.c +20 -7
index 5351d3d9d538..8d989e436517 100644
@@ -1117,27 +1117,39 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 			 u32 tag, struct cifs_fattr *fattr)
 {
 	struct smb2_file_full_ea_info *ea;
+	bool ignore_missing_eas = false;
 	bool have_xattr_dev = false;
+	umode_t reparse_mode_type = 0;
 	u32 next = 0;
 
 	switch (tag) {
 	case IO_REPARSE_TAG_LX_SYMLINK:
-		fattr->cf_mode |= S_IFLNK;
+		reparse_mode_type = S_IFLNK;
 		break;
 	case IO_REPARSE_TAG_LX_FIFO:
-		fattr->cf_mode |= S_IFIFO;
+		reparse_mode_type = S_IFIFO;
 		break;
 	case IO_REPARSE_TAG_AF_UNIX:
-		fattr->cf_mode |= S_IFSOCK;
+		reparse_mode_type = S_IFSOCK;
 		break;
 	case IO_REPARSE_TAG_LX_CHR:
-		fattr->cf_mode |= S_IFCHR;
+		reparse_mode_type = S_IFCHR;
 		break;
 	case IO_REPARSE_TAG_LX_BLK:
-		fattr->cf_mode |= S_IFBLK;
+		reparse_mode_type = S_IFBLK;
 		break;
+	default:
+		return false;
 	}
 
+	/*
+	 * When reparse buffer is not available then this is from readdir() call
+	 * which does not provide EAs. readdir() can return DT_UNKNOWN type,
+	 * which is signaled by no filling the fattr->cf_mode and returning true.
+	 */
+	if (!data->reparse.buf && !data->wsl.eas_len)
+		ignore_missing_eas = true;
+
 	if (!data->wsl.eas_len)
 		goto out;
 
@@ -1162,7 +1174,7 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 			fattr->cf_gid = wsl_make_kgid(cifs_sb, v);
 		else if (!strncmp(name, SMB2_WSL_XATTR_MODE, nlen)) {
 			/* File type in reparse point tag and in xattr mode must match. */
-			if (S_DT(fattr->cf_mode) != S_DT(le32_to_cpu(*(__le32 *)v)))
+			if (S_DT(reparse_mode_type) != S_DT(le32_to_cpu(*(__le32 *)v)))
 				return false;
 			fattr->cf_mode = (umode_t)le32_to_cpu(*(__le32 *)v);
 		} else if (!strncmp(name, SMB2_WSL_XATTR_DEV, nlen)) {
@@ -1174,8 +1186,9 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 
 	/* Major and minor numbers for char and block devices are mandatory. */
 	if (!have_xattr_dev && (tag == IO_REPARSE_TAG_LX_CHR || tag == IO_REPARSE_TAG_LX_BLK))
-		return false;
+		return ignore_missing_eas;
 
+	fattr->cf_mode |= reparse_mode_type;
 	return true;
 }
 
Signature · 1 line
2.20.1
#3 lore raw

PATCH 3/5 cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size

In reply to Pali Rohár

Currently the SMB2_OP_QUERY_WSL_EA checks that response buffer has at least size SMB2_WSL_MIN_QUERY_EA_RESP_SIZE and maximally it is SMB2_WSL_MAX_QUERY_EA_RESP_SIZE.

Constant SMB2_WSL_MIN_QUERY_EA_RESP_SIZE is defined wrongly because it expects that the there are at least 3 EAs. But WSL subsystem has only one mandatory EA: $LXMOD. So fix the SMB2_WSL_MIN_QUERY_EA_RESP_SIZE to be size of the structure of one EA.

Relax also SMB2_WSL_MAX_QUERY_EA_RESP_SIZE, calculate maximum size from the size of the largest EA which is 8 bytes for $LXDEV.

This change allows to recognize WSL CHR and BLK reparse points which have only $LXMOD and $LXDEV EAs (no $LXUID or $LXGID). WSL subsystem recognize such reparse points too.

Fixes ea41367b2a60 ("smb: client: introduce SMB2_OP_QUERY_WSL_EA")
---
 fs/smb/client/smb2pdu.h | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)
fs/smb/client/smb2pdu.h +9 -10
index 3c09a58dfd07..cdf0ab9ddbcd 100644
@@ -425,24 +425,23 @@ struct smb2_create_ea_ctx {
 #define SMB2_WSL_XATTR_MODE		"$LXMOD"
 #define SMB2_WSL_XATTR_DEV		"$LXDEV"
 #define SMB2_WSL_XATTR_NAME_LEN	6
-#define SMB2_WSL_NUM_XATTRS		4
 
 #define SMB2_WSL_XATTR_UID_SIZE	4
 #define SMB2_WSL_XATTR_GID_SIZE	4
 #define SMB2_WSL_XATTR_MODE_SIZE	4
 #define SMB2_WSL_XATTR_DEV_SIZE	8
 
+/* minimal size: at least the smallest EA has to be present */
 #define SMB2_WSL_MIN_QUERY_EA_RESP_SIZE \
-	(ALIGN((SMB2_WSL_NUM_XATTRS - 1) * \
-	       (SMB2_WSL_XATTR_NAME_LEN + 1 + \
-		sizeof(struct smb2_file_full_ea_info)), 4) + \
-	 SMB2_WSL_XATTR_NAME_LEN + 1 + sizeof(struct smb2_file_full_ea_info))
+	(sizeof(struct smb2_file_full_ea_info) + SMB2_WSL_XATTR_NAME_LEN + 1 + 4)
 
+/*
+ * maximal size: all 4 EAs are present,
+ * beginning of each EA structure has to be aligned to 4 bytes,
+ * EAs have different size and can be returned in any other,
+ * use the largest EA size for aligning when calculating maximal size
+ */
 #define SMB2_WSL_MAX_QUERY_EA_RESP_SIZE \
-	(ALIGN(SMB2_WSL_MIN_QUERY_EA_RESP_SIZE + \
-	       SMB2_WSL_XATTR_UID_SIZE + \
-	       SMB2_WSL_XATTR_GID_SIZE + \
-	       SMB2_WSL_XATTR_MODE_SIZE + \
-	       SMB2_WSL_XATTR_DEV_SIZE, 4))
+	4 * ALIGN((sizeof(struct smb2_file_full_ea_info) + SMB2_WSL_XATTR_NAME_LEN + 1 + 8), 4)
 
 #endif				/* _SMB2PDU_H */
Signature · 1 line
2.20.1
#4 lore raw

Re: [PATCH 3/5] cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size

Quoted 16 lines · Pali Rohár

Currently the SMB2_OP_QUERY_WSL_EA checks that response buffer has at least size SMB2_WSL_MIN_QUERY_EA_RESP_SIZE and maximally it is SMB2_WSL_MAX_QUERY_EA_RESP_SIZE.

Constant SMB2_WSL_MIN_QUERY_EA_RESP_SIZE is defined wrongly because it expects that the there are at least 3 EAs. But WSL subsystem has only one mandatory EA: $LXMOD. So fix the SMB2_WSL_MIN_QUERY_EA_RESP_SIZE to be size of the structure of one EA.

Relax also SMB2_WSL_MAX_QUERY_EA_RESP_SIZE, calculate maximum size from the size of the largest EA which is 8 bytes for $LXDEV.

This change allows to recognize WSL CHR and BLK reparse points which have only $LXMOD and $LXDEV EAs (no $LXUID or $LXGID). WSL subsystem recognize such reparse points too.

Quoted 5 lines
Fixes ea41367b2a60 ("smb: client: introduce SMB2_OP_QUERY_WSL_EA")
---
 fs/smb/client/smb2pdu.h | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

If we're querying all those EAs and the file has only $LXMOD, wouldn't the server return empty EAs except for $LXMOD?

#5 lore raw

Re: [PATCH 3/5] cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size

Quoted 18 lines · On Sunday 08 June 2025 18:49:43 Paulo Alcantara wrote:
Quoted 16 lines · Pali Rohár

Currently the SMB2_OP_QUERY_WSL_EA checks that response buffer has at least size SMB2_WSL_MIN_QUERY_EA_RESP_SIZE and maximally it is SMB2_WSL_MAX_QUERY_EA_RESP_SIZE.

Constant SMB2_WSL_MIN_QUERY_EA_RESP_SIZE is defined wrongly because it expects that the there are at least 3 EAs. But WSL subsystem has only one mandatory EA: $LXMOD. So fix the SMB2_WSL_MIN_QUERY_EA_RESP_SIZE to be size of the structure of one EA.

Relax also SMB2_WSL_MAX_QUERY_EA_RESP_SIZE, calculate maximum size from the size of the largest EA which is 8 bytes for $LXDEV.

This change allows to recognize WSL CHR and BLK reparse points which have only $LXMOD and $LXDEV EAs (no $LXUID or $LXGID). WSL subsystem recognize such reparse points too.

Quoted 8 lines
Quoted 5 lines
Fixes ea41367b2a60 ("smb: client: introduce SMB2_OP_QUERY_WSL_EA")
---
 fs/smb/client/smb2pdu.h | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

If we're querying all those EAs and the file has only $LXMOD, wouldn't the server return empty EAs except for $LXMOD?

We are using FILE_FULL_EA_INFORMATION for querying EAs, which means that always all stored EAs are returned. It is not 4 calls (one by one), but rather one call to return everything at once.

Windows server in this case returns just one EA in its response: $LXMOD EA. And SMB2_WSL_MIN_QUERY_EA_RESP_SIZE specifies that at least 3 EAs must be returned, otherwise check_wsl_eas() throws error and do not try to parse response.

#6 lore raw

Re: [PATCH 3/5] cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size

Quoted 9 lines · Pali Rohár
Quoted 4 lines · On Sunday 08 June 2025 18:49:43 Paulo Alcantara wrote:

Pali Rohár <[email protected]> writes:

If we're querying all those EAs and the file has only $LXMOD, wouldn't the server return empty EAs except for $LXMOD?

We are using FILE_FULL_EA_INFORMATION for querying EAs, which means that always all stored EAs are returned. It is not 4 calls (one by one), but rather one call to return everything at once.

Yes.

Quoted 4 lines

Windows server in this case returns just one EA in its response: $LXMOD EA. And SMB2_WSL_MIN_QUERY_EA_RESP_SIZE specifies that at least 3 EAs must be returned, otherwise check_wsl_eas() throws error and do not try to parse response.

Can you share a trace of the server returning only a single EA in the response when we query $LXUID, $LXGID, $LXMOD and $LXDEV?

What I mean is that we query all those EAs when we find reparse points on non-POSIX mounts, and if the file doesn't have them, the server still returns the EAs but with a zero smb2_file_full_ea_info::ea_value_len. check_wsl_eas() skips the EA when is @vlen zero.

#7 lore raw

Re: [PATCH 3/5] cifs: Fix validation of SMB2_OP_QUERY_WSL_EA response size

Quoted 19 lines · On Sunday 08 June 2025 20:10:24 Paulo Alcantara wrote:
Quoted 9 lines · Pali Rohár
Quoted 4 lines · On Sunday 08 June 2025 18:49:43 Paulo Alcantara wrote:

Pali Rohár <[email protected]> writes:

If we're querying all those EAs and the file has only $LXMOD, wouldn't the server return empty EAs except for $LXMOD?

We are using FILE_FULL_EA_INFORMATION for querying EAs, which means that always all stored EAs are returned. It is not 4 calls (one by one), but rather one call to return everything at once.

Yes.

Quoted 4 lines

Windows server in this case returns just one EA in its response: $LXMOD EA. And SMB2_WSL_MIN_QUERY_EA_RESP_SIZE specifies that at least 3 EAs must be returned, otherwise check_wsl_eas() throws error and do not try to parse response.

Quoted 7 lines

Can you share a trace of the server returning only a single EA in the response when we query $LXUID, $LXGID, $LXMOD and $LXDEV?

What I mean is that we query all those EAs when we find reparse points on non-POSIX mounts, and if the file doesn't have them, the server still returns the EAs but with a zero smb2_file_full_ea_info::ea_value_len. check_wsl_eas() skips the EA when is @vlen zero.

I started recording of tcpdump and I'm not able to reproduce it anymore. That is stupid. So I started investigation what happened there...

And the result is that I have more changes related to EAs on my disk, and I did some tests on all of them. And due to bug in dissector of older wireshark version which I used, I wrongly interpreted the format of network EA buffer. So based on this I calculated all sizes wrongly and introduced bug in my WIP code on my disk. So it my mistake here.

So please drop this one "PATCH 3/5". I re-checked the size limits and they should be correct.

#8 lore raw

PATCH 4/5 cifs: Query EA $LXMOD in cifs_query_path_info() for WSL reparse points

In reply to Pali Rohár

EA $LXMOD is required for WSL non-symlink reparse points.

Fixes ef86ab131d91 ("cifs: Fix querying of WSL CHR and BLK reparse points over SMB1")
---
 fs/smb/client/smb1ops.c | 62 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 60 insertions(+), 2 deletions(-)
fs/smb/client/smb1ops.c +60 -2
index df955cecdd4c..1f9166143ab1 100644
@@ -659,14 +659,72 @@ static int cifs_query_path_info(const unsigned int xid,
 	}
 
 #ifdef CONFIG_CIFS_XATTR
+	/*
+	 * For non-symlink WSL reparse points it is required to fetch
+	 * EA $LXMOD which contains in its S_DT part the mandatory file type.
+	 */
+	if (!rc && data->reparse_point) {
+		struct smb2_file_full_ea_info *ea;
+		u32 next = 0;
+
+		ea = (struct smb2_file_full_ea_info *)data->wsl.eas;
+		do {
+			ea = (void *)((u8 *)ea + next);
+			next = le32_to_cpu(ea->next_entry_offset);
+		} while (next);
+		if (le16_to_cpu(ea->ea_value_length)) {
+			ea->next_entry_offset = cpu_to_le32(ALIGN(sizeof(*ea) +
+						ea->ea_name_length + 1 +
+						le16_to_cpu(ea->ea_value_length), 4));
+			ea = (void *)((u8 *)ea + le32_to_cpu(ea->next_entry_offset));
+		}
+
+		rc = CIFSSMBQAllEAs(xid, tcon, full_path, SMB2_WSL_XATTR_MODE,
+				    &ea->ea_data[SMB2_WSL_XATTR_NAME_LEN + 1],
+				    SMB2_WSL_XATTR_MODE_SIZE, cifs_sb);
+		if (rc == SMB2_WSL_XATTR_MODE_SIZE) {
+			ea->next_entry_offset = cpu_to_le32(0);
+			ea->flags = 0;
+			ea->ea_name_length = SMB2_WSL_XATTR_NAME_LEN;
+			ea->ea_value_length = cpu_to_le16(SMB2_WSL_XATTR_MODE_SIZE);
+			memcpy(&ea->ea_data[0], SMB2_WSL_XATTR_MODE, SMB2_WSL_XATTR_NAME_LEN + 1);
+			data->wsl.eas_len += ALIGN(sizeof(*ea) + SMB2_WSL_XATTR_NAME_LEN + 1 +
+						   SMB2_WSL_XATTR_MODE_SIZE, 4);
+			rc = 0;
+		} else if (rc >= 0) {
+			/* It is an error if EA $LXMOD has wrong size. */
+			rc = -EINVAL;
+		} else {
+			/*
+			 * In all other cases ignore error if fetching
+			 * of EA $LXMOD failed. It is needed only for
+			 * non-symlink WSL reparse points and wsl_to_fattr()
+			 * handle the case when EA is missing.
+			 */
+			rc = 0;
+		}
+	}
+
 	/*
 	 * For WSL CHR and BLK reparse points it is required to fetch
 	 * EA $LXDEV which contains major and minor device numbers.
 	 */
 	if (!rc && data->reparse_point) {
 		struct smb2_file_full_ea_info *ea;
+		u32 next = 0;
 
 		ea = (struct smb2_file_full_ea_info *)data->wsl.eas;
+		do {
+			ea = (void *)((u8 *)ea + next);
+			next = le32_to_cpu(ea->next_entry_offset);
+		} while (next);
+		if (le16_to_cpu(ea->ea_value_length)) {
+			ea->next_entry_offset = cpu_to_le32(ALIGN(sizeof(*ea) +
+						ea->ea_name_length + 1 +
+						le16_to_cpu(ea->ea_value_length), 4));
+			ea = (void *)((u8 *)ea + le32_to_cpu(ea->next_entry_offset));
+		}
+
 		rc = CIFSSMBQAllEAs(xid, tcon, full_path, SMB2_WSL_XATTR_DEV,
 				    &ea->ea_data[SMB2_WSL_XATTR_NAME_LEN + 1],
 				    SMB2_WSL_XATTR_DEV_SIZE, cifs_sb);
@@ -676,8 +734,8 @@ static int cifs_query_path_info(const unsigned int xid,
 			ea->ea_name_length = SMB2_WSL_XATTR_NAME_LEN;
 			ea->ea_value_length = cpu_to_le16(SMB2_WSL_XATTR_DEV_SIZE);
 			memcpy(&ea->ea_data[0], SMB2_WSL_XATTR_DEV, SMB2_WSL_XATTR_NAME_LEN + 1);
-			data->wsl.eas_len = sizeof(*ea) + SMB2_WSL_XATTR_NAME_LEN + 1 +
-					    SMB2_WSL_XATTR_DEV_SIZE;
+			data->wsl.eas_len += ALIGN(sizeof(*ea) + SMB2_WSL_XATTR_NAME_LEN + 1 +
+						   SMB2_WSL_XATTR_MODE_SIZE, 4);
 			rc = 0;
 		} else if (rc >= 0) {
 			/* It is an error if EA $LXDEV has wrong size. */
Signature · 1 line
2.20.1
#9 lore raw

PATCH 5/5 cifs: Validate presence of EA $LXMOD for WSL reparse points

In reply to Pali Rohár

S_DT part of EA $LXMOD is mandatory for all WSL reparse points except the WSL symlink and Win32 socket. Microsoft WSL subsystem does not recognize them without EA $LXMOD too, and treat such inodes as regular files.

Fixes ef201e8759d2 ("cifs: Validate EAs for WSL reparse points")
Fixes 78e26bec4d6d ("smb: client: parse uid, gid, mode and dev from WSL reparse points")
---
 fs/smb/client/reparse.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)
fs/smb/client/reparse.c +12 -0
index 8d989e436517..3dffd5f0dc07 100644
@@ -1118,6 +1118,7 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 {
 	struct smb2_file_full_ea_info *ea;
 	bool ignore_missing_eas = false;
+	bool have_xattr_mode = false;
 	bool have_xattr_dev = false;
 	umode_t reparse_mode_type = 0;
 	u32 next = 0;
@@ -1177,6 +1178,7 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 			if (S_DT(reparse_mode_type) != S_DT(le32_to_cpu(*(__le32 *)v)))
 				return false;
 			fattr->cf_mode = (umode_t)le32_to_cpu(*(__le32 *)v);
+			have_xattr_mode = true;
 		} else if (!strncmp(name, SMB2_WSL_XATTR_DEV, nlen)) {
 			fattr->cf_rdev = reparse_mkdev(v);
 			have_xattr_dev = true;
@@ -1188,6 +1190,16 @@ static bool wsl_to_fattr(struct cifs_open_info_data *data,
 	if (!have_xattr_dev && (tag == IO_REPARSE_TAG_LX_CHR || tag == IO_REPARSE_TAG_LX_BLK))
 		return ignore_missing_eas;
 
+	/*
+	 * S_DT part of xattr MODE is mandatory for all WSL reparse points except the WSL symlink.
+	 * Microsoft WSL does not recognize them without xattr MODE too (except the WSL symlink).
+	 * IO_REPARSE_TAG_AF_UNIX is here an exception because this reparse point is used by both
+	 * WSL subsystem and native NT/WinAPI subsystems. And NT/WinAPI creates AF UNIX socket
+	 * without the xattr MODE and recognize it also without the xattr MODE.
+	 */
+	if (!have_xattr_mode && (tag != IO_REPARSE_TAG_AF_UNIX && tag != IO_REPARSE_TAG_LX_SYMLINK))
+		return ignore_missing_eas;
+
 	fattr->cf_mode |= reparse_mode_type;
 	return true;
 }
Signature · 1 line
2.20.1