Skip to contentSkip to conversation

xfs: clean up pointer sign differences

PATCH In review Dave Chinner 2 participants 18 messages Jan 19, 2010 XFS filesystem

lore
Share & tools
lore
#0 lore raw

cover xfs: clean up pointer sign differences

This patch series cleans up all the pointer sign warnings when they are turned back on in the XFS build. These are the kernel equivalents of the libxfs changes I posted a last week. On Christoph's suggestion, I've made the changes for the kernel code so that we can fix the libxfs warnings simply by resyncing the userspce code with the kernel code.

Hence this makes all of the directory and attribute names and values unsigned chars and uses that consistently within XFS. The on disk formats are all unsigned, but the VFS is a mix of signed and unsigned interfaces.

The series converts all the names and values at the highest level to unsigned so the rest of the code doesn't need to juggle different signedness. It cleans up all the places where we have cast pointers, makes all functions take unsigned strings and so on.

Finally, it turns on the pointer sign warnings again (which are actually turned off globally in the build) so that we don't reintroduce new problems due to a lack of warnings being produced by the compiler.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#1 lore raw

PATCH 1/8 xfs: directory names are unsigned

Convert the struct xfs_name to use unsigned chars for the name strings to match both what is stored on disk (__uint8_t) and what the VFS expects (unsigned char).

---
 fs/xfs/xfs_types.h |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
fs/xfs/xfs_types.h +2 -2
index d725428..b099045 100644
@@ -151,8 +151,8 @@ typedef enum {
 } xfs_btnum_t;
 
 struct xfs_name {
-	const char	*name;
-	int		len;
+	const unsigned char	*name;
+	int			len;
 };
 
 #endif	/* __XFS_TYPES_H__ */
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#3 lore raw

PATCH 2/8 xfs: convert DM ops to use unsigned char names

In reply to Dave Chinner

dmops uses a signed char for it's namespace event. To be consistent with the rest of the code, convert them to unsigned char for the namespace string.

---
 fs/xfs/xfs_mount.h |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
fs/xfs/xfs_mount.h +2 -1
index e62fd1c..f4d1441 100644
@@ -78,7 +78,8 @@ typedef int	(*xfs_send_destroy_t)(struct xfs_inode *, dm_right_t);
 typedef int	(*xfs_send_namesp_t)(dm_eventtype_t, struct xfs_mount *,
 			struct xfs_inode *, dm_right_t,
 			struct xfs_inode *, dm_right_t,
-			const char *, const char *, mode_t, int, int);
+			const unsigned char *, const unsigned char *,
+			mode_t, int, int);
 typedef int	(*xfs_send_mount_t)(struct xfs_mount *, dm_right_t,
 			char *, char *);
 typedef void	(*xfs_send_unmount_t)(struct xfs_mount *, struct xfs_inode *,
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 2/8
#4 lore raw

Re: [PATCH 2/8] xfs: convert DM ops to use unsigned char names

Quoted 3 lines · Dave Chinner

dmops uses a signed char for it's namespace event. To be consistent with the rest of the code, convert them to unsigned char for the namespace string.

Looks good,

possibly we should just get rid of the dm hooks now that they've been dead for years..

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#5 lore raw

PATCH 3/8 xfs: convert dirnameops to unsigned char names

In reply to Dave Chinner

To be consistent across the codebase, convert the dirnameops to pass the directory names by unsigned char strings.

---
 fs/xfs/xfs_da_btree.c |    4 ++--
 fs/xfs/xfs_da_btree.h |    5 +++--
 fs/xfs/xfs_dir2.c     |    4 ++--
 3 files changed, 7 insertions(+), 6 deletions(-)
fs/xfs/xfs_da_btree.c +2 -2
index c0c8869..0ca556b 100644
@@ -1534,8 +1534,8 @@ xfs_da_hashname(const __uint8_t *name, int namelen)
 enum xfs_dacmp
 xfs_da_compname(
 	struct xfs_da_args *args,
-	const char 	*name,
-	int 		len)
+	const unsigned char *name,
+	int		len)
 {
 	return (args->namelen == len && memcmp(args->name, name, len) == 0) ?
 					XFS_CMP_EXACT : XFS_CMP_DIFFERENT;
fs/xfs/xfs_da_btree.h +3 -2
index 30cd08f..fe9f5a8 100644
@@ -209,7 +209,8 @@ typedef struct xfs_da_state {
  */
 struct xfs_nameops {
 	xfs_dahash_t	(*hashname)(struct xfs_name *);
-	enum xfs_dacmp	(*compname)(struct xfs_da_args *, const char *, int);
+	enum xfs_dacmp	(*compname)(struct xfs_da_args *,
+					const unsigned char *, int);
 };
 
 
@@ -260,7 +261,7 @@ int	xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
 
 uint xfs_da_hashname(const __uint8_t *name_string, int name_length);
 enum xfs_dacmp xfs_da_compname(struct xfs_da_args *args,
-				const char *name, int len);
+				const unsigned char *name, int len);
 
 
 xfs_da_state_t *xfs_da_state_alloc(void);
fs/xfs/xfs_dir2.c +2 -2
index 93634a7..c21c527 100644
@@ -66,8 +66,8 @@ xfs_ascii_ci_hashname(
 STATIC enum xfs_dacmp
 xfs_ascii_ci_compname(
 	struct xfs_da_args *args,
-	const char	*name,
-	int 		len)
+	const unsigned char *name,
+	int		len)
 {
 	enum xfs_dacmp	result;
 	int		i;
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 3/8
#6 lore raw

Re: [PATCH 3/8] xfs: convert dirnameops to unsigned char names

Quoted 4 lines · Dave Chinner

To be consistent across the codebase, convert the dirnameops to pass the directory names by unsigned char strings.

Looks good,

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#7 lore raw

PATCH 4/8 xfs: make xfs_dir_cilookup_result use unsigned char

In reply to Dave Chinner

For consistency with the result of the code.

---
 fs/xfs/xfs_dir2.c |    2 +-
 fs/xfs/xfs_dir2.h |    4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)
fs/xfs/xfs_dir2.c +1 -1
index c21c527..3a8c6ba 100644
@@ -247,7 +247,7 @@ xfs_dir_createname(
 int
 xfs_dir_cilookup_result(
 	struct xfs_da_args *args,
-	const char	*name,
+	const unsigned char *name,
 	int		len)
 {
 	if (args->cmpresult == XFS_CMP_DIFFERENT)
fs/xfs/xfs_dir2.h +2 -2
index 1d9ef96..74a3b10 100644
@@ -100,7 +100,7 @@ extern int xfs_dir2_isleaf(struct xfs_trans *tp, struct xfs_inode *dp,
 extern int xfs_dir2_shrink_inode(struct xfs_da_args *args, xfs_dir2_db_t db,
 				struct xfs_dabuf *bp);
 
-extern int xfs_dir_cilookup_result(struct xfs_da_args *args, const char *name,
-				int len);
+extern int xfs_dir_cilookup_result(struct xfs_da_args *args,
+				const unsigned char *name, int len);
 
 #endif	/* __XFS_DIR2_H__ */
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 4/8
#8 lore raw

Re: [PATCH 4/8] xfs: make xfs_dir_cilookup_result use unsigned char

Quoted 13 lines · Dave Chinner

For consistency with the result of the code.

---
 fs/xfs/xfs_dir2.c |    2 +-
 fs/xfs/xfs_dir2.h |    4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)
fs/xfs/xfs_dir2.c +0 -0
index c21c527..3a8c6ba 100644

fs/xfs/xfs_dir2.c xfs_dir_createname( · line 252

Quoted 7 lines
fs/xfs/xfs_dir2.c +1 -1
@@ -247,7 +247,7 @@ xfs_dir_createname(
 int
 xfs_dir_cilookup_result(
 	struct xfs_da_args *args,
-	const char	*name,
+	const unsigned char *name,
 	int		len)
 {

Given that the value we copy things into is unsigned anyway this looks correct,

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#9 lore raw

PATCH 5/8 xfs: xfs_buf_iomove() doesn't care about signedness

In reply to Dave Chinner

xfs_buf_iomove() uses xfs_caddr_t as it's parameter types, but it doesn't care about the signedness of the variables as it is just copying the data. Change the prototype to use void * so that we don't get sign warnings at call sites.

---
 fs/xfs/linux-2.6/xfs_buf.c |    2 +-
 fs/xfs/linux-2.6/xfs_buf.h |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
fs/xfs/linux-2.6/xfs_buf.c +1 -1
index 158fad4..efd745b 100644
@@ -1443,7 +1443,7 @@ xfs_buf_iomove(
 	xfs_buf_t		*bp,	/* buffer to process		*/
 	size_t			boff,	/* starting buffer offset	*/
 	size_t			bsize,	/* length to copy		*/
-	caddr_t			data,	/* data address			*/
+	void			*data,	/* data address			*/
 	xfs_buf_rw_t		mode)	/* read/write/zero flag		*/
 {
 	size_t			bend, cpoff, csize;
fs/xfs/linux-2.6/xfs_buf.h +1 -1
index 9a29d18..4f2ad66 100644
@@ -243,7 +243,7 @@ extern void xfs_buf_ioend(xfs_buf_t *,	int);
 extern void xfs_buf_ioerror(xfs_buf_t *, int);
 extern int xfs_buf_iorequest(xfs_buf_t *);
 extern int xfs_buf_iowait(xfs_buf_t *);
-extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, xfs_caddr_t,
+extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, void *,
 				xfs_buf_rw_t);
 
 static inline int xfs_buf_iostrategy(xfs_buf_t *bp)
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 5/8
#10 lore raw

Re: [PATCH 5/8] xfs: xfs_buf_iomove() doesn't care about signedness

Quoted 4 lines · Dave Chinner

xfs_buf_iomove() uses xfs_caddr_t as it's parameter types, but it doesn't care about the signedness of the variables as it is just copying the data. Change the prototype to use void * so that we don't get sign warnings at call sites.

Looks good,

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#11 lore raw

PATCH 6/8 xfs: convert attr to use unsigned names

In reply to Dave Chinner

To be consistent with the directory code, the attr code should use unsigned names. Convert the names from the vfs at the highest level to unsigned, and ænsure they are consistenly used as unsigned down to disk.

---
 fs/xfs/linux-2.6/xfs_acl.c     |   11 ++++++-----
 fs/xfs/linux-2.6/xfs_ioctl.c   |   18 +++++++++---------
 fs/xfs/linux-2.6/xfs_ioctl.h   |   12 ++++++------
 fs/xfs/linux-2.6/xfs_ioctl32.c |    4 ++--
 fs/xfs/linux-2.6/xfs_iops.c    |    4 ++--
 fs/xfs/linux-2.6/xfs_xattr.c   |   27 +++++++++++++++++++--------
 fs/xfs/xfs_acl.h               |    4 ++--
 fs/xfs/xfs_attr.c              |   38 +++++++++++++++++++++++---------------
 fs/xfs/xfs_attr.h              |    2 +-
 fs/xfs/xfs_attr_leaf.c         |   28 ++++++++++++++--------------
 fs/xfs/xfs_attr_sf.h           |    2 +-
 fs/xfs/xfs_vnodeops.h          |   10 +++++-----
 12 files changed, 90 insertions(+), 70 deletions(-)
fs/xfs/linux-2.6/xfs_acl.c +6 -5
index 883ca5a..bf85bbe 100644
@@ -106,7 +106,7 @@ xfs_get_acl(struct inode *inode, int type)
 	struct posix_acl *acl;
 	struct xfs_acl *xfs_acl;
 	int len = sizeof(struct xfs_acl);
-	char *ea_name;
+	unsigned char *ea_name;
 	int error;
 
 	acl = get_cached_acl(inode, type);
@@ -133,7 +133,8 @@ xfs_get_acl(struct inode *inode, int type)
 	if (!xfs_acl)
 		return ERR_PTR(-ENOMEM);
 
-	error = -xfs_attr_get(ip, ea_name, (char *)xfs_acl, &len, ATTR_ROOT);
+	error = -xfs_attr_get(ip, ea_name, (unsigned char *)xfs_acl,
+							&len, ATTR_ROOT);
 	if (error) {
 		/*
 		 * If the attribute doesn't exist make sure we have a negative
@@ -162,7 +163,7 @@ STATIC int
 xfs_set_acl(struct inode *inode, int type, struct posix_acl *acl)
 {
 	struct xfs_inode *ip = XFS_I(inode);
-	char *ea_name;
+	unsigned char *ea_name;
 	int error;
 
 	if (S_ISLNK(inode->i_mode))
@@ -194,7 +195,7 @@ xfs_set_acl(struct inode *inode, int type, struct posix_acl *acl)
 			(sizeof(struct xfs_acl_entry) *
 			 (XFS_ACL_MAX_ENTRIES - acl->a_count));
 
-		error = -xfs_attr_set(ip, ea_name, (char *)xfs_acl,
+		error = -xfs_attr_set(ip, ea_name, (unsigned char *)xfs_acl,
 				len, ATTR_ROOT);
 
 		kfree(xfs_acl);
@@ -262,7 +263,7 @@ xfs_set_mode(struct inode *inode, mode_t mode)
 }
 
 static int
-xfs_acl_exists(struct inode *inode, char *name)
+xfs_acl_exists(struct inode *inode, unsigned char *name)
 {
 	int len = sizeof(struct xfs_acl);
 
fs/xfs/linux-2.6/xfs_ioctl.c +9 -9
index a034cf6..3906e85 100644
@@ -447,12 +447,12 @@ xfs_attrlist_by_handle(
 int
 xfs_attrmulti_attr_get(
 	struct inode		*inode,
-	char			*name,
-	char			__user *ubuf,
+	unsigned char		*name,
+	unsigned char		__user *ubuf,
 	__uint32_t		*len,
 	__uint32_t		flags)
 {
-	char			*kbuf;
+	unsigned char		*kbuf;
 	int			error = EFAULT;
 
 	if (*len > XATTR_SIZE_MAX)
@@ -476,12 +476,12 @@ xfs_attrmulti_attr_get(
 int
 xfs_attrmulti_attr_set(
 	struct inode		*inode,
-	char			*name,
-	const char		__user *ubuf,
+	unsigned char		*name,
+	const unsigned char	__user *ubuf,
 	__uint32_t		len,
 	__uint32_t		flags)
 {
-	char			*kbuf;
+	unsigned char		*kbuf;
 	int			error = EFAULT;
 
 	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
@@ -501,7 +501,7 @@ xfs_attrmulti_attr_set(
 int
 xfs_attrmulti_attr_remove(
 	struct inode		*inode,
-	char			*name,
+	unsigned char		*name,
 	__uint32_t		flags)
 {
 	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
@@ -519,7 +519,7 @@ xfs_attrmulti_by_handle(
 	xfs_fsop_attrmulti_handlereq_t am_hreq;
 	struct dentry		*dentry;
 	unsigned int		i, size;
-	char			*attr_name;
+	unsigned char		*attr_name;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -XFS_ERROR(EPERM);
@@ -547,7 +547,7 @@ xfs_attrmulti_by_handle(
 
 	error = 0;
 	for (i = 0; i < am_hreq.opcount; i++) {
-		ops[i].am_error = strncpy_from_user(attr_name,
+		ops[i].am_error = strncpy_from_user((char *)attr_name,
 				ops[i].am_attrname, MAXNAMELEN);
 		if (ops[i].am_error == 0 || ops[i].am_error == MAXNAMELEN)
 			error = -ERANGE;
fs/xfs/linux-2.6/xfs_ioctl.h +6 -6
index 7bd7c6a..d56173b 100644
@@ -45,23 +45,23 @@ xfs_readlink_by_handle(
 extern int
 xfs_attrmulti_attr_get(
 	struct inode		*inode,
-	char			*name,
-	char			__user *ubuf,
+	unsigned char		*name,
+	unsigned char		__user *ubuf,
 	__uint32_t		*len,
 	__uint32_t		flags);
 
 extern int
-	xfs_attrmulti_attr_set(
+xfs_attrmulti_attr_set(
 	struct inode		*inode,
-	char			*name,
-	const char		__user *ubuf,
+	unsigned char		*name,
+	const unsigned char	__user *ubuf,
 	__uint32_t		len,
 	__uint32_t		flags);
 
 extern int
 xfs_attrmulti_attr_remove(
 	struct inode		*inode,
-	char			*name,
+	unsigned char		*name,
 	__uint32_t		flags);
 
 extern struct dentry *
fs/xfs/linux-2.6/xfs_ioctl32.c +2 -2
index be1527b..0bf6d61 100644
@@ -411,7 +411,7 @@ xfs_compat_attrmulti_by_handle(
 	compat_xfs_fsop_attrmulti_handlereq_t	am_hreq;
 	struct dentry				*dentry;
 	unsigned int				i, size;
-	char					*attr_name;
+	unsigned char				*attr_name;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -XFS_ERROR(EPERM);
@@ -440,7 +440,7 @@ xfs_compat_attrmulti_by_handle(
 
 	error = 0;
 	for (i = 0; i < am_hreq.opcount; i++) {
-		ops[i].am_error = strncpy_from_user(attr_name,
+		ops[i].am_error = strncpy_from_user((char *)attr_name,
 				compat_ptr(ops[i].am_attrname),
 				MAXNAMELEN);
 		if (ops[i].am_error == 0 || ops[i].am_error == MAXNAMELEN)
fs/xfs/linux-2.6/xfs_iops.c +2 -2
index 2259460..e8566bb 100644
@@ -140,10 +140,10 @@ xfs_init_security(
 	struct xfs_inode *ip = XFS_I(inode);
 	size_t		length;
 	void		*value;
-	char		*name;
+	unsigned char	*name;
 	int		error;
 
-	error = security_inode_init_security(inode, dir, &name,
+	error = security_inode_init_security(inode, dir, (char **)&name,
 					     &value, &length);
 	if (error) {
 		if (error == -EOPNOTSUPP)
fs/xfs/linux-2.6/xfs_xattr.c +19 -8
index 0b18788..fa01b9d 100644
@@ -45,7 +45,7 @@ xfs_xattr_get(struct dentry *dentry, const char *name,
 		value = NULL;
 	}
 
-	error = -xfs_attr_get(ip, name, value, &asize, xflags);
+	error = -xfs_attr_get(ip, (unsigned char *)name, value, &asize, xflags);
 	if (error)
 		return error;
 	return asize;
@@ -67,8 +67,9 @@ xfs_xattr_set(struct dentry *dentry, const char *name, const void *value,
 		xflags |= ATTR_REPLACE;
 
 	if (!value)
-		return -xfs_attr_remove(ip, name, xflags);
-	return -xfs_attr_set(ip, name, (void *)value, size, xflags);
+		return -xfs_attr_remove(ip, (unsigned char *)name, xflags);
+	return -xfs_attr_set(ip, (unsigned char *)name,
+				(void *)value, size, xflags);
 }
 
 static struct xattr_handler xfs_xattr_user_handler = {
@@ -124,8 +125,13 @@ static const char *xfs_xattr_prefix(int flags)
 }
 
 static int
-xfs_xattr_put_listent(struct xfs_attr_list_context *context, int flags,
-		char *name, int namelen, int valuelen, char *value)
+xfs_xattr_put_listent(
+	struct xfs_attr_list_context *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
 {
 	unsigned int prefix_len = xfs_xattr_prefix_len(flags);
 	char *offset;
@@ -148,7 +154,7 @@ xfs_xattr_put_listent(struct xfs_attr_list_context *context, int flags,
 	offset = (char *)context->alist + context->count;
 	strncpy(offset, xfs_xattr_prefix(flags), prefix_len);
 	offset += prefix_len;
-	strncpy(offset, name, namelen);			/* real name */
+	strncpy(offset, (char *)name, namelen);			/* real name */
 	offset += namelen;
 	*offset = '\0';
 	context->count += prefix_len + namelen + 1;
@@ -156,8 +162,13 @@ xfs_xattr_put_listent(struct xfs_attr_list_context *context, int flags,
 }
 
 static int
-xfs_xattr_put_listent_sizes(struct xfs_attr_list_context *context, int flags,
-		char *name, int namelen, int valuelen, char *value)
+xfs_xattr_put_listent_sizes(
+	struct xfs_attr_list_context *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
 {
 	context->count += xfs_xattr_prefix_len(flags) + namelen + 1;
 	return 0;
fs/xfs/xfs_acl.h +2 -2
index 00fd357..d13eeba 100644
@@ -36,8 +36,8 @@ struct xfs_acl {
 };
 
 /* On-disk XFS extended attribute names */
-#define SGI_ACL_FILE		"SGI_ACL_FILE"
-#define SGI_ACL_DEFAULT		"SGI_ACL_DEFAULT"
+#define SGI_ACL_FILE		(unsigned char *)"SGI_ACL_FILE"
+#define SGI_ACL_DEFAULT		(unsigned char *)"SGI_ACL_DEFAULT"
 #define SGI_ACL_FILE_SIZE	(sizeof(SGI_ACL_FILE)-1)
 #define SGI_ACL_DEFAULT_SIZE	(sizeof(SGI_ACL_DEFAULT)-1)
 
fs/xfs/xfs_attr.c +23 -15
index 9d11eba..f7b426a 100644
@@ -93,12 +93,12 @@ STATIC int xfs_attr_rmtval_remove(xfs_da_args_t *args);
 STATIC int
 xfs_attr_name_to_xname(
 	struct xfs_name	*xname,
-	const char	*aname)
+	const unsigned char *aname)
 {
 	if (!aname)
 		return EINVAL;
 	xname->name = aname;
-	xname->len = strlen(aname);
+	xname->len = strlen((char *)aname);
 	if (xname->len >= MAXNAMELEN)
 		return EFAULT;		/* match IRIX behaviour */
 
@@ -124,7 +124,7 @@ STATIC int
 xfs_attr_get_int(
 	struct xfs_inode	*ip,
 	struct xfs_name		*name,
-	char			*value,
+	unsigned char		*value,
 	int			*valuelenp,
 	int			flags)
 {
@@ -171,8 +171,8 @@ xfs_attr_get_int(
 int
 xfs_attr_get(
 	xfs_inode_t	*ip,
-	const char	*name,
-	char		*value,
+	const unsigned char *name,
+	unsigned char	*value,
 	int		*valuelenp,
 	int		flags)
 {
@@ -235,8 +235,12 @@ xfs_attr_calc_size(
 }
 
 STATIC int
-xfs_attr_set_int(xfs_inode_t *dp, struct xfs_name *name,
-		char *value, int valuelen, int flags)
+xfs_attr_set_int(
+	struct xfs_inode *dp,
+	struct xfs_name	*name,
+	unsigned char	*value,
+	int		valuelen,
+	int		flags)
 {
 	xfs_da_args_t	args;
 	xfs_fsblock_t	firstblock;
@@ -452,8 +456,8 @@ out:
 int
 xfs_attr_set(
 	xfs_inode_t	*dp,
-	const char	*name,
-	char		*value,
+	const unsigned char *name,
+	unsigned char	*value,
 	int		valuelen,
 	int		flags)
 {
@@ -600,7 +604,7 @@ out:
 int
 xfs_attr_remove(
 	xfs_inode_t	*dp,
-	const char	*name,
+	const unsigned char *name,
 	int		flags)
 {
 	int		error;
@@ -669,9 +673,13 @@ xfs_attr_list_int(xfs_attr_list_context_t *context)
  */
 /*ARGSUSED*/
 STATIC int
-xfs_attr_put_listent(xfs_attr_list_context_t *context, int flags,
-		     char *name, int namelen,
-		     int valuelen, char *value)
+xfs_attr_put_listent(
+	xfs_attr_list_context_t *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
 {
 	struct attrlist *alist = (struct attrlist *)context->alist;
 	attrlist_ent_t *aep;
@@ -1980,7 +1988,7 @@ xfs_attr_rmtval_get(xfs_da_args_t *args)
 	xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];
 	xfs_mount_t *mp;
 	xfs_daddr_t dblkno;
-	xfs_caddr_t dst;
+	void *dst;
 	xfs_buf_t *bp;
 	int nmap, error, tmp, valuelen, blkcnt, i;
 	xfs_dablk_t lblkno;
@@ -2039,7 +2047,7 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 	xfs_inode_t *dp;
 	xfs_bmbt_irec_t map;
 	xfs_daddr_t dblkno;
-	xfs_caddr_t src;
+	void *src;
 	xfs_buf_t *bp;
 	xfs_dablk_t lblkno;
 	int blkcnt, valuelen, nmap, error, tmp, committed;
fs/xfs/xfs_attr.h +1 -1
index 9c3a243..e920d68 100644
@@ -113,7 +113,7 @@ typedef struct attrlist_cursor_kern {
 
 
 typedef int (*put_listent_func_t)(struct xfs_attr_list_context *, int,
-				      char *, int, int, char *);
+			      unsigned char *, int, int, unsigned char *);
 
 typedef struct xfs_attr_list_context {
 	struct xfs_inode		*dp;		/* inode */
fs/xfs/xfs_attr_leaf.c +14 -14
index baf41b5..52519a2 100644
@@ -521,11 +521,11 @@ xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
 
 	sfe = &sf->list[0];
 	for (i = 0; i < sf->hdr.count; i++) {
-		nargs.name = (char *)sfe->nameval;
+		nargs.name = sfe->nameval;
 		nargs.namelen = sfe->namelen;
-		nargs.value = (char *)&sfe->nameval[nargs.namelen];
+		nargs.value = &sfe->nameval[nargs.namelen];
 		nargs.valuelen = sfe->valuelen;
-		nargs.hashval = xfs_da_hashname((char *)sfe->nameval,
+		nargs.hashval = xfs_da_hashname(sfe->nameval,
 						sfe->namelen);
 		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(sfe->flags);
 		error = xfs_attr_leaf_lookup_int(bp, &nargs); /* set a->index */
@@ -612,10 +612,10 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 		for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
 			error = context->put_listent(context,
 					   sfe->flags,
-					   (char *)sfe->nameval,
+					   sfe->nameval,
 					   (int)sfe->namelen,
 					   (int)sfe->valuelen,
-					   (char*)&sfe->nameval[sfe->namelen]);
+					   &sfe->nameval[sfe->namelen]);
 
 			/*
 			 * Either search callback finished early or
@@ -659,8 +659,8 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 		}
 
 		sbp->entno = i;
-		sbp->hash = xfs_da_hashname((char *)sfe->nameval, sfe->namelen);
-		sbp->name = (char *)sfe->nameval;
+		sbp->hash = xfs_da_hashname(sfe->nameval, sfe->namelen);
+		sbp->name = sfe->nameval;
 		sbp->namelen = sfe->namelen;
 		/* These are bytes, and both on-disk, don't endian-flip */
 		sbp->valuelen = sfe->valuelen;
@@ -818,9 +818,9 @@ xfs_attr_leaf_to_shortform(xfs_dabuf_t *bp, xfs_da_args_t *args, int forkoff)
 			continue;
 		ASSERT(entry->flags & XFS_ATTR_LOCAL);
 		name_loc = xfs_attr_leaf_name_local(leaf, i);
-		nargs.name = (char *)name_loc->nameval;
+		nargs.name = name_loc->nameval;
 		nargs.namelen = name_loc->namelen;
-		nargs.value = (char *)&name_loc->nameval[nargs.namelen];
+		nargs.value = &name_loc->nameval[nargs.namelen];
 		nargs.valuelen = be16_to_cpu(name_loc->valuelen);
 		nargs.hashval = be32_to_cpu(entry->hashval);
 		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(entry->flags);
@@ -2370,10 +2370,10 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 
 			retval = context->put_listent(context,
 						entry->flags,
-						(char *)name_loc->nameval,
+						name_loc->nameval,
 						(int)name_loc->namelen,
 						be16_to_cpu(name_loc->valuelen),
-						(char *)&name_loc->nameval[name_loc->namelen]);
+						&name_loc->nameval[name_loc->namelen]);
 			if (retval)
 				return retval;
 		} else {
@@ -2397,15 +2397,15 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 					return retval;
 				retval = context->put_listent(context,
 						entry->flags,
-						(char *)name_rmt->name,
+						name_rmt->name,
 						(int)name_rmt->namelen,
 						valuelen,
-						(char*)args.value);
+						args.value);
 				kmem_free(args.value);
 			} else {
 				retval = context->put_listent(context,
 						entry->flags,
-						(char *)name_rmt->name,
+						name_rmt->name,
 						(int)name_rmt->namelen,
 						valuelen,
 						NULL);
fs/xfs/xfs_attr_sf.h +1 -1
index 76ab7b0..919756e 100644
@@ -52,7 +52,7 @@ typedef struct xfs_attr_sf_sort {
 	__uint8_t	valuelen;	/* length of value */
 	__uint8_t	flags;		/* flags bits (see xfs_attr_leaf.h) */
 	xfs_dahash_t	hash;		/* this entry's hash value */
-	char		*name;		/* name value, pointer into buffer */
+	unsigned char	*name;		/* name value, pointer into buffer */
 } xfs_attr_sf_sort_t;
 
 #define XFS_ATTR_SF_ENTSIZE_BYNAME(nlen,vlen)	/* space name/value uses */ \
fs/xfs/xfs_vnodeops.h +5 -5
index 167a467..774f407 100644
@@ -43,11 +43,11 @@ int xfs_change_file_space(struct xfs_inode *ip, int cmd,
 int xfs_rename(struct xfs_inode *src_dp, struct xfs_name *src_name,
 		struct xfs_inode *src_ip, struct xfs_inode *target_dp,
 		struct xfs_name *target_name, struct xfs_inode *target_ip);
-int xfs_attr_get(struct xfs_inode *ip, const char *name, char *value,
-		int *valuelenp, int flags);
-int xfs_attr_set(struct xfs_inode *dp, const char *name, char *value,
-		int valuelen, int flags);
-int xfs_attr_remove(struct xfs_inode *dp, const char *name, int flags);
+int xfs_attr_get(struct xfs_inode *ip, const unsigned char *name,
+		unsigned char *value, int *valuelenp, int flags);
+int xfs_attr_set(struct xfs_inode *dp, const unsigned char *name,
+		unsigned char *value, int valuelen, int flags);
+int xfs_attr_remove(struct xfs_inode *dp, const unsigned char *name, int flags);
 int xfs_attr_list(struct xfs_inode *dp, char *buffer, int bufsize,
 		int flags, struct attrlist_cursor_kern *cursor);
 ssize_t xfs_read(struct xfs_inode *ip, struct kiocb *iocb,
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 6/8
#12 lore raw

Re: [PATCH 6/8] xfs: convert attr to use unsigned names

fs/xfs/linux-2.6/xfs_acl.c

Quoted 22 lines · Dave Chinner

To be consistent with the directory code, the attr code should use unsigned names. Convert the names from the vfs at the highest level to unsigned, and ??nsure they are consistenly used as unsigned down to disk.

---
 fs/xfs/linux-2.6/xfs_acl.c     |   11 ++++++-----
 fs/xfs/linux-2.6/xfs_ioctl.c   |   18 +++++++++---------
 fs/xfs/linux-2.6/xfs_ioctl.h   |   12 ++++++------
 fs/xfs/linux-2.6/xfs_ioctl32.c |    4 ++--
 fs/xfs/linux-2.6/xfs_iops.c    |    4 ++--
 fs/xfs/linux-2.6/xfs_xattr.c   |   27 +++++++++++++++++++--------
 fs/xfs/xfs_acl.h               |    4 ++--
 fs/xfs/xfs_attr.c              |   38 +++++++++++++++++++++++---------------
 fs/xfs/xfs_attr.h              |    2 +-
 fs/xfs/xfs_attr_leaf.c         |   28 ++++++++++++++--------------
 fs/xfs/xfs_attr_sf.h           |    2 +-
 fs/xfs/xfs_vnodeops.h          |   10 +++++-----
 12 files changed, 90 insertions(+), 70 deletions(-)
fs/xfs/linux-2.6/xfs_acl.c +0 -0

It's a bit unfortunate that the highlevel xattr code gets char froms the VFS, but better convert them at the highest level and keep the lower ones clean, so:

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#13 lore raw

PATCH 7/8 xfs: clean up sign warnings in dir2 code

In reply to Dave Chinner

We are now consistently using unsigned char strings for names so fix up the remaining warnings in the dir2 code to complete the cleanup.

---
 fs/xfs/xfs_dir2.c       |    2 +-
 fs/xfs/xfs_dir2_block.c |    9 +++++----
 fs/xfs/xfs_dir2_leaf.c  |    2 +-
 fs/xfs/xfs_dir2_sf.c    |    2 +-
 4 files changed, 8 insertions(+), 7 deletions(-)
fs/xfs/xfs_dir2.c +1 -1
index 3a8c6ba..42520f0 100644
@@ -44,7 +44,7 @@
 #include "xfs_vnodeops.h"
 #include "xfs_trace.h"
 
-struct xfs_name xfs_name_dotdot = {"..", 2};
+struct xfs_name xfs_name_dotdot = { (unsigned char *)"..", 2};
 
 /*
  * ASCII case-insensitive (ie. A-Z) support for directories that was
fs/xfs/xfs_dir2_block.c +5 -4
index ddc4ecc..779a267 100644
@@ -57,8 +57,8 @@ static xfs_dahash_t xfs_dir_hash_dot, xfs_dir_hash_dotdot;
 void
 xfs_dir_startup(void)
 {
-	xfs_dir_hash_dot = xfs_da_hashname(".", 1);
-	xfs_dir_hash_dotdot = xfs_da_hashname("..", 2);
+	xfs_dir_hash_dot = xfs_da_hashname((unsigned char *)".", 1);
+	xfs_dir_hash_dotdot = xfs_da_hashname((unsigned char *)"..", 2);
 }
 
 /*
@@ -513,8 +513,9 @@ xfs_dir2_block_getdents(
 		/*
 		 * If it didn't fit, set the final offset to here & return.
 		 */
-		if (filldir(dirent, dep->name, dep->namelen, cook & 0x7fffffff,
-			    be64_to_cpu(dep->inumber), DT_UNKNOWN)) {
+		if (filldir(dirent, (char *)dep->name, dep->namelen,
+			    cook & 0x7fffffff, be64_to_cpu(dep->inumber),
+			    DT_UNKNOWN)) {
 			*offset = cook & 0x7fffffff;
 			xfs_da_brelse(NULL, bp);
 			return 0;
fs/xfs/xfs_dir2_leaf.c +1 -1
index 29f484c..e2d8985 100644
@@ -1081,7 +1081,7 @@ xfs_dir2_leaf_getdents(
 		dep = (xfs_dir2_data_entry_t *)ptr;
 		length = xfs_dir2_data_entsize(dep->namelen);
 
-		if (filldir(dirent, dep->name, dep->namelen,
+		if (filldir(dirent, (char *)dep->name, dep->namelen,
 			    xfs_dir2_byte_to_dataptr(mp, curoff) & 0x7fffffff,
 			    be64_to_cpu(dep->inumber), DT_UNKNOWN))
 			break;
fs/xfs/xfs_dir2_sf.c +1 -1
index 9d4f17a..c1a5945 100644
@@ -782,7 +782,7 @@ xfs_dir2_sf_getdents(
 		}
 
 		ino = xfs_dir2_sf_get_inumber(sfp, xfs_dir2_sf_inumberp(sfep));
-		if (filldir(dirent, sfep->name, sfep->namelen,
+		if (filldir(dirent, (char *)sfep->name, sfep->namelen,
 			    off & 0x7fffffff, ino, DT_UNKNOWN)) {
 			*offset = off & 0x7fffffff;
 			return 0;
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 7/8
#14 lore raw

Re: [PATCH 7/8] xfs: clean up sign warnings in dir2 code

Quoted 3 lines · Dave Chinner

We are now consistently using unsigned char strings for names so fix up the remaining warnings in the dir2 code to complete the cleanup.

The cast in the initializers looks a bit ugly, but I don't know a better way either.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#15 lore raw

PATCH 8/8 xfs: make compile warn about char sign mismatches again

In reply to Dave Chinner

The -fno-unsigned-char directive has no effect anymore as the XFs build is clean. However, the kernel build hides pointer sign differences so turn that back on so that we can clean up all the mismatches prior to a userspace code resync.

---
 fs/xfs/Makefile       |    2 +-
 fs/xfs/xfs_vnodeops.c |    6 ++++--
 2 files changed, 5 insertions(+), 3 deletions(-)
fs/xfs/Makefile +1 -1
index 56641fe..1926701 100644
@@ -16,7 +16,7 @@
 # Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #
 
-EXTRA_CFLAGS +=	 -I$(src) -I$(src)/linux-2.6 -funsigned-char
+EXTRA_CFLAGS +=	 -I$(src) -I$(src)/linux-2.6 -Wpointer-sign
 
 XFS_LINUX := linux-2.6
 
fs/xfs/xfs_vnodeops.c +4 -2
index 6f26875..9f7c001 100644
@@ -2199,7 +2199,8 @@ xfs_symlink(
 	if (DM_EVENT_ENABLED(dp, DM_EVENT_SYMLINK)) {
 		error = XFS_SEND_NAMESP(mp, DM_EVENT_SYMLINK, dp,
 					DM_RIGHT_NULL, NULL, DM_RIGHT_NULL,
-					link_name->name, target_path, 0, 0, 0);
+					link_name->name,
+					(unsigned char *)target_path, 0, 0, 0);
 		if (error)
 			return error;
 	}
@@ -2395,7 +2396,8 @@ std_return:
 					dp, DM_RIGHT_NULL,
 					error ? NULL : ip,
 					DM_RIGHT_NULL, link_name->name,
-					target_path, 0, error, 0);
+					(unsigned char *)target_path,
+					0, error, 0);
 	}
 
 	if (!error)
Signature · 6 lines
1.6.5

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 8/8
#16 lore raw

Re: [PATCH 8/8] xfs: make compile warn about char sign mismatches again

Quoted 4 lines · Dave Chinner

The -fno-unsigned-char directive has no effect anymore as the XFs build is clean. However, the kernel build hides pointer sign differences so turn that back on so that we can clean up all the mismatches prior to a userspace code resync.

Not so happy about turning the warning on so we have local compiler warning differences from the rest of the kernel again, but it's not that bad, so

Quoted 14 lines

index 6f26875..9f7c001 100644

fs/xfs/xfs_vnodeops.c +2 -1
@@ -2199,7 +2199,8 @@ xfs_symlink(
 	if (DM_EVENT_ENABLED(dp, DM_EVENT_SYMLINK)) {
 		error = XFS_SEND_NAMESP(mp, DM_EVENT_SYMLINK, dp,
 					DM_RIGHT_NULL, NULL, DM_RIGHT_NULL,
-					link_name->name, target_path, 0, 0, 0);
+					link_name->name,
+					(unsigned char *)target_path, 0, 0, 0);
 		if (error)
 			return error;
 	}

fs/xfs/xfs_vnodeops.c std_return: · line 2400

Quoted 6 lines
fs/xfs/xfs_vnodeops.c +2 -1
@@ -2395,7 +2396,8 @@ std_return:
 					dp, DM_RIGHT_NULL,
 					error ? NULL : ip,
 					DM_RIGHT_NULL, link_name->name,
-					target_path, 0, error, 0);
+					(unsigned char *)target_path,
+					0, error, 0);

These changes are nowhere mentioned in the changelog. And they probably belong into the patch changing the dmapi events to take unsigned chars.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#17 lore raw

Re: [PATCH 8/8] xfs: make compile warn about char sign mismatches again

Quoted 28 lines · Christoph Hellwig
Quoted 4 lines · Dave Chinner

The -fno-unsigned-char directive has no effect anymore as the XFs build is clean. However, the kernel build hides pointer sign differences so turn that back on so that we can clean up all the mismatches prior to a userspace code resync.

Not so happy about turning the warning on so we have local compiler warning differences from the rest of the kernel again, but it's not that bad, so

Quoted 15 lines

index 6f26875..9f7c001 100644

fs/xfs/xfs_vnodeops.c +2 -1
@@ -2199,7 +2199,8 @@ xfs_symlink(
 	if (DM_EVENT_ENABLED(dp, DM_EVENT_SYMLINK)) {
 		error = XFS_SEND_NAMESP(mp, DM_EVENT_SYMLINK, dp,
 					DM_RIGHT_NULL, NULL, DM_RIGHT_NULL,
-					link_name->name, target_path, 0, 0, 0);
+					link_name->name,
+					(unsigned char *)target_path, 0, 0, 0);
 		if (error)
 			return error;
 	}
@@ -2395,7 +2396,8 @@ std_return:
 					dp, DM_RIGHT_NULL,
Quoted 8 lines

fs/xfs/xfs_vnodeops.c std_return: · line 2400

Quoted 5 lines
fs/xfs/xfs_vnodeops.c +2 -1
 					error ? NULL : ip,
 					DM_RIGHT_NULL, link_name->name,
-					target_path, 0, error, 0);
+					(unsigned char *)target_path,
+					0, error, 0);

These changes are nowhere mentioned in the changelog. And they probably belong into the patch changing the dmapi events to take unsigned chars.

I though I fixed that. Oh well, I'll fix it up again....

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs