Stole unicode translation table from mount_msdos. Add kernel code

to support this translation.

MFC after:	2 weeks

6 files changed, 137 insertions, 18 deletions

diff --git a/sys/fs/ntfs/ntfs.h b/sys/fs/ntfs/ntfs.h
index d74fe6f..f5e9125 100644
--- a/sys/fs/ntfs/ntfs.h
+++ b/sys/fs/ntfs/ntfs.h
@@ -258,6 +258,8 @@ struct ntfsmount {
 #if !defined(__FreeBSD__)
 	struct netexport ntm_export;	/* export information */
 #endif
+ 	wchar *		ntm_82u;	/* 8bit to Unicode */
+ 	char **		ntm_u28;	/* Unicode to 8 bit */
 };
 
 #define ntm_mftcn	ntm_bootfile.bf_mftcn
diff --git a/sys/fs/ntfs/ntfs_subr.c b/sys/fs/ntfs/ntfs_subr.c
index 7b9ae76..e1e6006 100644
--- a/sys/fs/ntfs/ntfs_subr.c
+++ b/sys/fs/ntfs/ntfs_subr.c
@@ -63,21 +63,20 @@ MALLOC_DEFINE(M_NTFSDECOMP, "NTFS decomp", "NTFS decompression temporary");
 
 static int ntfs_ntlookupattr __P((struct ntfsmount *, const char *, int, int *, char **));
 static int ntfs_findvattr __P((struct ntfsmount *, struct ntnode *, struct ntvattr **, struct ntvattr **, u_int32_t, const char *, size_t, cn_t));
-static int ntfs_uastricmp __P((const wchar *, size_t, const char *, size_t));
-static int ntfs_uastrcmp __P((const wchar *, size_t, const char *, size_t));
+static int ntfs_uastricmp __P((struct ntfsmount *, const wchar *, size_t, const char *, size_t));
+static int ntfs_uastrcmp __P((struct ntfsmount *, const wchar *, size_t, const char *, size_t));
 
 /* table for mapping Unicode chars into uppercase; it's filled upon first
  * ntfs mount, freed upon last ntfs umount */
 static wchar *ntfs_toupper_tab;
-#define NTFS_U28(ch)		((((ch) & 0xFF) == 0) ? '_' : (ch) & 0xFF)
-#define NTFS_TOUPPER(ch)	(ntfs_toupper_tab[(unsigned char)(ch)])
+#define NTFS_TOUPPER(ch)	(ntfs_toupper_tab[(ch)])
 static struct lock ntfs_toupper_lock;
 static signed int ntfs_toupper_usecount;
 
 /* support macro for ntfs_ntvattrget() */
 #define NTFS_AALPCMP(aalp,type,name,namelen) (				\
   (aalp->al_type == type) && (aalp->al_namelen == namelen) &&		\
-  !ntfs_uastrcmp(aalp->al_name,aalp->al_namelen,name,namelen) )
+  !NTFS_UASTRCMP(aalp->al_name,aalp->al_namelen,name,namelen) )
 
 /*
  * 
@@ -662,7 +661,8 @@ ntfs_runtovrun(
  * Compare unicode and ascii string case insens.
  */
 static int
-ntfs_uastricmp(ustr, ustrlen, astr, astrlen)
+ntfs_uastricmp(ntmp, ustr, ustrlen, astr, astrlen)
+	struct ntfsmount *ntmp;
 	const wchar *ustr;
 	size_t ustrlen;
 	const char *astr;
@@ -671,9 +671,13 @@ ntfs_uastricmp(ustr, ustrlen, astr, astrlen)
 	size_t             i;
 	int             res;
 
+	/*
+	 * XXX We use NTFS_82U(NTFS_U28(c)) to get rid of unicode
+	 * symbols not covered by translation table
+	 */
 	for (i = 0; i < ustrlen && i < astrlen; i++) {
-		res = ((int) NTFS_TOUPPER(NTFS_U28(ustr[i]))) -
-			((int)NTFS_TOUPPER(astr[i]));
+		res = ((int) NTFS_TOUPPER(NTFS_82U(NTFS_U28(ustr[i])))) -
+			((int)NTFS_TOUPPER(NTFS_82U(astr[i])));
 		if (res)
 			return res;
 	}
@@ -684,7 +688,8 @@ ntfs_uastricmp(ustr, ustrlen, astr, astrlen)
  * Compare unicode and ascii string case sens.
  */
 static int
-ntfs_uastrcmp(ustr, ustrlen, astr, astrlen)
+ntfs_uastrcmp(ntmp, ustr, ustrlen, astr, astrlen)
+	struct ntfsmount *ntmp;
 	const wchar *ustr;
 	size_t ustrlen;
 	const char *astr;
@@ -918,7 +923,7 @@ ntfs_ntlookupfile(
 			/* check the name - the case-insensitible check
 			 * has to come first, to break from this for loop
 			 * if needed, so we can dive correctly */
-			res = ntfs_uastricmp(iep->ie_fname, iep->ie_fnamelen,
+			res = NTFS_UASTRICMP(iep->ie_fname, iep->ie_fnamelen,
 				fname, fnamelen);
 			if (res > 0) break;
 			if (res < 0) continue;
@@ -926,7 +931,7 @@ ntfs_ntlookupfile(
 			if (iep->ie_fnametype == 0 ||
 			    !(ntmp->ntm_flag & NTFS_MFLAG_CASEINS))
 			{
-				res = ntfs_uastrcmp(iep->ie_fname,
+				res = NTFS_UASTRCMP(iep->ie_fname,
 					iep->ie_fnamelen, fname, fnamelen);
 				if (res != 0) continue;
 			}
@@ -1956,13 +1961,13 @@ ntfs_toupper_use(mp, ntmp)
 	 * XXX for now, just the first 256 entries are used anyway,
 	 * so don't bother reading more
 	 */
-	MALLOC(ntfs_toupper_tab, wchar *, 256 * sizeof(wchar),
+	MALLOC(ntfs_toupper_tab, wchar *, 65536 * sizeof(wchar),
 		M_NTFSRDATA, M_WAITOK);
 
 	if ((error = VFS_VGET(mp, NTFS_UPCASEINO, &vp)))
 		goto out;
 	error = ntfs_readattr(ntmp, VTONT(vp), NTFS_A_DATA, NULL,
-			0, 256*sizeof(wchar), (char *) ntfs_toupper_tab, NULL);
+			0, 65536*sizeof(wchar), (char *) ntfs_toupper_tab, NULL);
 	vput(vp);
 
     out:
@@ -1997,6 +2002,86 @@ ntfs_toupper_unuse()
 	LOCKMGR(&ntfs_toupper_lock, LK_RELEASE, NULL);
 } 
 
+int
+ntfs_u28_init(
+	struct ntfsmount *ntmp,
+	wchar *u2w)
+{
+	char ** u28;
+	int i, j, h, l;
+
+	MALLOC(u28, char **, 256 * sizeof(char*), M_TEMP, M_WAITOK | M_ZERO);
+
+	for (i=0; i<256; i++) {
+		h = (u2w[i] >> 8) & 0xFF;
+		l = (u2w[i]) &0xFF;
+
+		if (u28[h] == NULL) {
+			MALLOC(u28[h], char *, 256 * sizeof(char), M_TEMP, M_WAITOK);
+			for (j=0; j<256; j++)
+				u28[h][j] = '_';
+		}
+
+		u28[h][l] = i & 0xFF;
+	}
+
+	ntmp->ntm_u28 = u28;
+
+	return (0);
+}
+
+int
+ntfs_u28_uninit(struct ntfsmount *ntmp)
+{
+	char ** u28;
+	int i;
+
+	if (ntmp->ntm_u28 == NULL)
+		return (0);
+
+	u28 = ntmp->ntm_u28;
+
+	for (i=0; i<256; i++)
+		if (u28[i] != NULL)
+			FREE(u28[i], M_TEMP);
+
+	FREE(u28, M_TEMP);
+
+	return (0);
+}
+
+int
+ntfs_82u_init(
+	struct ntfsmount *ntmp,
+	u_int16_t *u2w)
+{
+	wchar * _82u;
+	int i;
+
+	MALLOC(_82u, wchar *, 256 * sizeof(wchar), M_TEMP, M_WAITOK);
+
+	if (u2w == NULL) {
+		for (i=0; i<256; i++)
+			_82u[i] = i;
+	} else {
+		for (i=0; i<128; i++)
+			_82u[i] = i;
+		for (i=0; i<128; i++)
+			_82u[i+128] = u2w[i];
+	}
+
+	ntmp->ntm_82u = _82u;
+
+	return (0);
+}
+
+int
+ntfs_82u_uninit(struct ntfsmount *ntmp)
+{
+	FREE(ntmp->ntm_82u, M_TEMP);
+	return (0);
+}
+
 /*
  * maps the Unicode char to 8bit equivalent
  * XXX currently only gets lower 8bit from the Unicode char
@@ -2004,9 +2089,15 @@ ntfs_toupper_unuse()
  * something better has to be definitely though out
  */
 char
-ntfs_u28(unichar)
-  wchar unichar;
+ntfs_u28(
+	struct ntfsmount *ntmp, 
+	wchar wc)
 {
-	return (char) NTFS_U28(unichar);
+	char * p;
+
+	p = ntmp->ntm_u28[(wc>>8)&0xFF];
+	if (p == NULL)
+		return ('_');
+	return (p[wc&0xFF]);
 }
 
diff --git a/sys/fs/ntfs/ntfs_subr.h b/sys/fs/ntfs/ntfs_subr.h
index 8f1480a..41a2dc7 100644
--- a/sys/fs/ntfs/ntfs_subr.h
+++ b/sys/fs/ntfs/ntfs_subr.h
@@ -84,7 +84,6 @@ int ntfs_filesize __P(( struct ntfsmount *, struct fnode *, u_int64_t *, u_int64
 int ntfs_times __P(( struct ntfsmount *, struct ntnode *, ntfs_times_t *));
 struct timespec	ntfs_nttimetounix __P(( u_int64_t ));
 int ntfs_ntreaddir __P(( struct ntfsmount *, struct fnode *, u_int32_t, struct attr_indexentry **));
-char ntfs_u28 __P((wchar));
 int ntfs_runtovrun __P(( cn_t **, cn_t **, u_long *, u_int8_t *));
 int ntfs_attrtontvattr __P(( struct ntfsmount *, struct ntvattr **, struct attr * ));
 void ntfs_freentvattr __P(( struct ntvattr * ));
@@ -108,3 +107,15 @@ int ntfs_toupper_use __P((struct mount *, struct ntfsmount *));
 void ntfs_toupper_unuse __P((void));
 int ntfs_fget __P((struct ntfsmount *, struct ntnode *, int, char *, struct fnode **));
 void ntfs_frele __P((struct fnode *));
+
+int ntfs_u28_init(struct ntfsmount *ntmp, wchar *u2w);
+int ntfs_u28_uninit(struct ntfsmount *ntmp);
+int ntfs_82u_init(struct ntfsmount *ntmp, u_int16_t *u2w);
+int ntfs_82u_uninit(struct ntfsmount *ntmp);
+char ntfs_u28(struct ntfsmount *ntmp, wchar wc);
+#define NTFS_U28(ch)		ntfs_u28(ntmp, (ch))
+#define NTFS_82U(ch)		(ntmp->ntm_82u[(ch)&0xFF])
+#define	NTFS_UASTRCMP(ustr, ustrlen, astr, astrlen)	\
+	ntfs_uastrcmp(ntmp, (ustr), (ustrlen), (astr), (astrlen))
+#define	NTFS_UASTRICMP(ustr, ustrlen, astr, astrlen)	\
+	ntfs_uastricmp(ntmp, (ustr), (ustrlen), (astr), (astrlen))
diff --git a/sys/fs/ntfs/ntfs_vfsops.c b/sys/fs/ntfs/ntfs_vfsops.c
index 94af95b..26a09b8 100644
--- a/sys/fs/ntfs/ntfs_vfsops.c
+++ b/sys/fs/ntfs/ntfs_vfsops.c
@@ -501,6 +501,17 @@ ntfs_mountfs(devvp, mp, argsp, p)
 	ntmp->ntm_gid = argsp->gid;
 	ntmp->ntm_mode = argsp->mode;
 	ntmp->ntm_flag = argsp->flag;
+
+	/* Copy in the 8-bit to Unicode conversion table */
+	if (argsp->flag & NTFSMNT_U2WTABLE) {
+		ntfs_82u_init(ntmp, argsp->u2w);
+	} else {
+		ntfs_82u_init(ntmp, NULL);
+	}
+
+	/* Initialize Unicode to 8-bit table from 8toU table */
+	ntfs_u28_init(ntmp, ntmp->ntm_82u);
+
 	mp->mnt_data = (qaddr_t)ntmp;
 
 	dprintf(("ntfs_mountfs(): case-%s,%s uid: %d, gid: %d, mode: %o\n",
@@ -697,6 +708,8 @@ ntfs_unmount(
 	ntfs_toupper_unuse();
 
 	dprintf(("ntfs_umount: freeing memory...\n"));
+	ntfs_u28_uninit(ntmp);
+	ntfs_82u_uninit(ntmp);
 	mp->mnt_data = (qaddr_t)0;
 	mp->mnt_flag &= ~MNT_LOCAL;
 	FREE(ntmp->ntm_ad, M_NTFSMNT);
diff --git a/sys/fs/ntfs/ntfs_vnops.c b/sys/fs/ntfs/ntfs_vnops.c
index be192a1..5c2297f 100644
--- a/sys/fs/ntfs/ntfs_vnops.c
+++ b/sys/fs/ntfs/ntfs_vnops.c
@@ -586,7 +586,7 @@ ntfs_readdir(ap)
 				continue;
 
 			for(i=0; i<iep->ie_fnamelen; i++) {
-				cde.d_name[i] = ntfs_u28(iep->ie_fname[i]);
+				cde.d_name[i] = NTFS_U28(iep->ie_fname[i]);
 			}
 			cde.d_name[i] = '\0';
 			dprintf(("ntfs_readdir: elem: %d, fname:[%s] type: %d, flag: %d, ",
diff --git a/sys/fs/ntfs/ntfsmount.h b/sys/fs/ntfs/ntfsmount.h
index f3a6d25..0c143ed 100644
--- a/sys/fs/ntfs/ntfsmount.h
+++ b/sys/fs/ntfs/ntfsmount.h
@@ -30,6 +30,7 @@
 
 #define	NTFS_MFLAG_CASEINS	0x00000001
 #define	NTFS_MFLAG_ALLNAMES	0x00000002
+#define	NTFSMNT_U2WTABLE	0x00000004
 
 struct ntfs_args {
 	char	*fspec;			/* block special device to mount */
@@ -38,4 +39,5 @@ struct ntfs_args {
 	gid_t	gid;			/* gid that owns ntfs files */
 	mode_t	mode;			/* mask to be applied for ntfs perms */
 	u_long	flag;			/* additional flags */
+	u_int16_t u2w[256];		/* Unix to Wchar */
 };