diff mbox

Handling the 7 reserved characters when mounted to Mac

Message ID CAH2r5mtjm3dbxtfn=ke_n9uj4NCS1tK0i3Ch2iUvkVvEovjb5w@mail.gmail.com (mailing list archive)
State New, archived
Headers show

Commit Message

Steve French Sept. 18, 2014, 1:06 a.m. UTC
One minor clarification

On Mac files can be created with the 7 reserved characters (but they
get remapped automatically on the wire).  On Windows you would have to
have SFU (services for unix), or SUA or equivalent in order to create
such files (and those characters get remapped when stored), and of
course CIFS/SMB2/SMB3 treats them as reserved as well (thus the need
to remap them).

On Wed, Sep 17, 2014 at 8:03 PM, Steve French <smfrench@gmail.com> wrote:
> There are seven reserved characters that need to be remapped when
> mounting to Windows, Mac (or any server without Unix Extensions) which
> are valid in POSIX but not in the other OS.
>
> : \ < > ? * |
>
> We use the normal UCS-2 remap range for this in order to convert this
> to/from UTF8 as did Windows Services for Unix (basically add 0xF000 to
> any of the 7 reserved characters).
>
> Mac used a very slightly different "Services for Mac" remap range
> 0xF021 through 0xF027.  The attached patch allows cifs.ko (the kernel
> client) to read directories on macs containing files with these
> characters and display their names properly.  In theory this even
> might be useful on mounts to Samba when the vfs_catia module is loaded
> and we are mounted with smb3.
>
> Currently the 7 reserved characters look very strange in directory
> listings from cifs.ko to Mac server.  This patch allows these file
> name characters to be read (requires specifying mapchars on mount).
>
> Two additional changes are needed:
> 1) Make it more automatic: a way of detecting that we are mounted to a
> Mac so we know to try to always remap these characters (at least for
> smb2.1 mounts to them and eventually smb3 when they support that)
>
> 2) A deterministic way of deciding when to use the normal mapchars
> approach vs. the mac mappings when creating a file with any of the
> seven characters in its name.  It may be ok in some cases to be able
> to translate file names in both formats properly when listing
> directories - but when creating files we have to use the same
> mechanism and decide when to use which, especially on SMB3 mounts
> (which do not have an option for posix pathnames).  Will need to
> decouple turning mac character mapping on/off from mapchars mount
> option.
>
> --
> Thanks,
>
> Steve

Comments

Steve French Sept. 18, 2014, 1:08 a.m. UTC | #1
Also should note that the attached diff only allows reading the
characters in filenames, creating a file will use the normal 'SFU'
style remap range not the mac style range without an additional patch
for conversion the other direction (which I am holding off on until I
can decide for sure how to tell that the server at the other end is a
Mac)

On Wed, Sep 17, 2014 at 8:06 PM, Steve French <smfrench@gmail.com> wrote:
> One minor clarification
>
> On Mac files can be created with the 7 reserved characters (but they
> get remapped automatically on the wire).  On Windows you would have to
> have SFU (services for unix), or SUA or equivalent in order to create
> such files (and those characters get remapped when stored), and of
> course CIFS/SMB2/SMB3 treats them as reserved as well (thus the need
> to remap them).
>
> On Wed, Sep 17, 2014 at 8:03 PM, Steve French <smfrench@gmail.com> wrote:
>> There are seven reserved characters that need to be remapped when
>> mounting to Windows, Mac (or any server without Unix Extensions) which
>> are valid in POSIX but not in the other OS.
>>
>> : \ < > ? * |
>>
>> We use the normal UCS-2 remap range for this in order to convert this
>> to/from UTF8 as did Windows Services for Unix (basically add 0xF000 to
>> any of the 7 reserved characters).
>>
>> Mac used a very slightly different "Services for Mac" remap range
>> 0xF021 through 0xF027.  The attached patch allows cifs.ko (the kernel
>> client) to read directories on macs containing files with these
>> characters and display their names properly.  In theory this even
>> might be useful on mounts to Samba when the vfs_catia module is loaded
>> and we are mounted with smb3.
>>
>> Currently the 7 reserved characters look very strange in directory
>> listings from cifs.ko to Mac server.  This patch allows these file
>> name characters to be read (requires specifying mapchars on mount).
>>
>> Two additional changes are needed:
>> 1) Make it more automatic: a way of detecting that we are mounted to a
>> Mac so we know to try to always remap these characters (at least for
>> smb2.1 mounts to them and eventually smb3 when they support that)
>>
>> 2) A deterministic way of deciding when to use the normal mapchars
>> approach vs. the mac mappings when creating a file with any of the
>> seven characters in its name.  It may be ok in some cases to be able
>> to translate file names in both formats properly when listing
>> directories - but when creating files we have to use the same
>> mechanism and decide when to use which, especially on SMB3 mounts
>> (which do not have an option for posix pathnames).  Will need to
>> decouple turning mac character mapping on/off from mapchars mount
>> option.
>>
>> --
>> Thanks,
>>
>> Steve
>
>
>
> --
> Thanks,
>
> Steve
diff mbox

Patch

diff --git a/fs/cifs/cifs_fs_sb.h b/fs/cifs/cifs_fs_sb.h
index 9409fa1..8c0b3ff 100644
--- a/fs/cifs/cifs_fs_sb.h
+++ b/fs/cifs/cifs_fs_sb.h
@@ -45,6 +45,7 @@ 
 #define CIFS_MOUNT_POSIXACL	0x100000 /* mirror of MS_POSIXACL in mnt_cifs_flags */
 #define CIFS_MOUNT_CIFS_BACKUPUID 0x200000 /* backup intent bit for a user */
 #define CIFS_MOUNT_CIFS_BACKUPGID 0x400000 /* backup intent bit for a group */
+#define CIFS_MOUNT_MAP_MAC_CHR	0x800000 /* remap illegal chars :*?<> ala SFM */
 
 struct cifs_sb_info {
 	struct rb_root tlink_tree;
diff --git a/fs/cifs/cifs_unicode.c b/fs/cifs/cifs_unicode.c
index 15e9505..bda9c70 100644
--- a/fs/cifs/cifs_unicode.c
+++ b/fs/cifs/cifs_unicode.c
@@ -61,26 +61,10 @@  cifs_utf16_bytes(const __le16 *from, int maxbytes,
 	return outlen;
 }
 
-/*
- * cifs_mapchar - convert a host-endian char to proper char in codepage
- * @target - where converted character should be copied
- * @src_char - 2 byte host-endian source character
- * @cp - codepage to which character should be converted
- * @mapchar - should character be mapped according to mapchars mount option?
- *
- * This function handles the conversion of a single character. It is the
- * responsibility of the caller to ensure that the target buffer is large
- * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
- */
-static int
-cifs_mapchar(char *target, const __u16 src_char, const struct nls_table *cp,
-	     bool mapchar)
+/* Convert character using the SFU - "Services for Unix" remapping range */
+static bool
+convert_sfu_char(const __u16 src_char, char *target)
 {
-	int len = 1;
-
-	if (!mapchar)
-		goto cp_convert;
-
 	/*
 	 * BB: Cannot handle remapping UNI_SLASH until all the calls to
 	 *     build_path_from_dentry are modified, as they use slash as
@@ -106,19 +90,73 @@  cifs_mapchar(char *target, const __u16 src_char, const struct nls_table *cp,
 		*target = '<';
 		break;
 	default:
-		goto cp_convert;
+		return false;
 	}
+	return true;
+}
 
-out:
-	return len;
+/* Convert character using the SFM - "Services for Mac" remapping range */
+static bool
+convert_sfm_char(const __u16 src_char, char *target)
+{
+	switch (src_char) {
+	case SFM_COLON:
+		*target = ':';
+		break;
+	case SFM_ASTERISK:
+		*target = '*';
+		break;
+	case SFM_QUESTION:
+		*target = '?';
+		break;
+	case SFM_PIPE:
+		*target = '|';
+		break;
+	case SFM_GRTRTHAN:
+		*target = '>';
+		break;
+	case SFM_LESSTHAN:
+		*target = '<';
+		break;
+	case SFM_SLASH:
+		*target = '\\';
+		break;
+	default:
+		return false;
+	}
+	return true;
+}
+
+
+/*
+ * cifs_mapchar - convert a host-endian char to proper char in codepage
+ * @target - where converted character should be copied
+ * @src_char - 2 byte host-endian source character
+ * @cp - codepage to which character should be converted
+ * @mapchar - should character be mapped according to mapchars mount option?
+ *
+ * This function handles the conversion of a single character. It is the
+ * responsibility of the caller to ensure that the target buffer is large
+ * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
+ */
+static int
+cifs_mapchar(char *target, const __u16 src_char, const struct nls_table *cp,
+	     bool mapchar, bool map_mac_char)
+{
+	int len = 1;
+
+	if (mapchar && convert_sfu_char(src_char, target))
+		return len;
+
+	if (map_mac_char && convert_sfm_char(src_char, target))
+		return len;
 
-cp_convert:
 	len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
 	if (len <= 0) {
 		*target = '?';
 		len = 1;
 	}
-	goto out;
+	return len;
 }
 
 /*
@@ -145,7 +183,7 @@  cp_convert:
  */
 int
 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
-		 const struct nls_table *codepage, bool mapchar)
+		const struct nls_table *codepage, bool mapchar, bool mapmacchar)
 {
 	int i, charlen, safelen;
 	int outlen = 0;
@@ -172,13 +210,15 @@  cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
 		 * conversion bleed into the null terminator
 		 */
 		if (outlen >= safelen) {
-			charlen = cifs_mapchar(tmp, ftmp, codepage, mapchar);
+			charlen = cifs_mapchar(tmp, ftmp, codepage, mapchar,
+					       mapmacchar);
 			if ((outlen + charlen) > (tolen - nullsize))
 				break;
 		}
 
 		/* put converted char into 'to' buffer */
-		charlen = cifs_mapchar(&to[outlen], ftmp, codepage, mapchar);
+		charlen = cifs_mapchar(&to[outlen], ftmp, codepage, mapchar,
+					mapmacchar);
 		outlen += charlen;
 	}
 
@@ -267,7 +307,7 @@  cifs_strndup_from_utf16(const char *src, const int maxlen,
 		if (!dst)
 			return NULL;
 		cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
-			       false);
+			       false, false);
 	} else {
 		len = strnlen(src, maxlen);
 		len++;
diff --git a/fs/cifs/cifs_unicode.h b/fs/cifs/cifs_unicode.h
index d8eac3b..1270077 100644
--- a/fs/cifs/cifs_unicode.h
+++ b/fs/cifs/cifs_unicode.h
@@ -52,6 +52,19 @@ 
 #define UNI_PIPE        (__u16) ('|' + 0xF000)
 #define UNI_SLASH       (__u16) ('\\' + 0xF000)
 
+/*
+ * Macs use an older "SFM" mapping of the symbols above. Fortunately it does
+ * not conflict (although almost does) with the mapping above.
+ */
+
+#define SFM_ASTERISK    ((__u16) 0xF021)
+#define SFM_QUESTION    ((__u16) 0xF025)
+#define SFM_COLON       ((__u16) 0xF022)
+#define SFM_GRTRTHAN    ((__u16) 0xF024)
+#define SFM_LESSTHAN    ((__u16) 0xF023)
+#define SFM_PIPE        ((__u16) 0xF027)
+#define SFM_SLASH       ((__u16) 0xF026)
+
 /* Just define what we want from uniupr.h.  We don't want to define the tables
  * in each source file.
  */
@@ -75,7 +88,7 @@  extern const struct UniCaseRange CifsUniLowerRange[];
 
 #ifdef __KERNEL__
 int cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
-		    const struct nls_table *codepage, bool mapchar);
+		    const struct nls_table *cp, bool mapchar, bool mapmacchars);
 int cifs_utf16_bytes(const __le16 *from, int maxbytes,
 		     const struct nls_table *codepage);
 int cifs_strtoUTF16(__le16 *, const char *, int, const struct nls_table *);
diff --git a/fs/cifs/cifsencrypt.c b/fs/cifs/cifsencrypt.c
index 4934347..cdeb80b 100644
--- a/fs/cifs/cifsencrypt.c
+++ b/fs/cifs/cifsencrypt.c
@@ -431,7 +431,7 @@  find_domain_name(struct cifs_ses *ses, const struct nls_table *nls_cp)
 						return -ENOMEM;
 				cifs_from_utf16(ses->domainName,
 					(__le16 *)blobptr, attrsize, attrsize,
-					nls_cp, false);
+					nls_cp, false, false);
 				break;
 			}
 		}
diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h
index 25b8392..53652a2 100644
--- a/fs/cifs/cifsglob.h
+++ b/fs/cifs/cifsglob.h
@@ -498,7 +498,8 @@  struct smb_vol {
 
 #define CIFS_MOUNT_MASK (CIFS_MOUNT_NO_PERM | CIFS_MOUNT_SET_UID | \
 			 CIFS_MOUNT_SERVER_INUM | CIFS_MOUNT_DIRECT_IO | \
-			 CIFS_MOUNT_NO_XATTR | CIFS_MOUNT_MAP_SPECIAL_CHR | \
+			 CIFS_MOUNT_NO_XATTR | \
+			 CIFS_MOUNT_MAP_SPECIAL_CHR | CIFS_MOUNT_MAP_MAC_CHR | \
 			 CIFS_MOUNT_UNX_EMUL | CIFS_MOUNT_NO_BRL | \
 			 CIFS_MOUNT_CIFS_ACL | CIFS_MOUNT_OVERR_UID | \
 			 CIFS_MOUNT_OVERR_GID | CIFS_MOUNT_DYNPERM | \
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 36ca204..b1755a9 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -3196,8 +3196,10 @@  void cifs_setup_cifs_sb(struct smb_vol *pvolume_info,
 		cifs_sb->mnt_cifs_flags |= CIFS_MOUNT_SET_UID;
 	if (pvolume_info->server_ino)
 		cifs_sb->mnt_cifs_flags |= CIFS_MOUNT_SERVER_INUM;
-	if (pvolume_info->remap)
+	if (pvolume_info->remap) {
 		cifs_sb->mnt_cifs_flags |= CIFS_MOUNT_MAP_SPECIAL_CHR;
+		cifs_sb->mnt_cifs_flags |= CIFS_MOUNT_MAP_MAC_CHR;
+	}
 	if (pvolume_info->no_xattr)
 		cifs_sb->mnt_cifs_flags |= CIFS_MOUNT_NO_XATTR;
 	if (pvolume_info->sfu_emul)
diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
index b334a89..3867d1a 100644
--- a/fs/cifs/readdir.c
+++ b/fs/cifs/readdir.c
@@ -716,7 +716,9 @@  static int cifs_filldir(char *find_entry, struct file *file,
 					min_t(size_t, de.namelen,
 					      (size_t)max_len), nlt,
 					cifs_sb->mnt_cifs_flags &
-						CIFS_MOUNT_MAP_SPECIAL_CHR);
+						CIFS_MOUNT_MAP_SPECIAL_CHR,
+					cifs_sb->mnt_cifs_flags &
+						CIFS_MOUNT_MAP_MAC_CHR);
 		name.len -= nls_nullsize(nlt);
 	} else {
 		name.name = de.name;