Add unicode support to msdosfs and smbfs; original pathes from imura,

bug fixes by Kuan-Chung Chiu <buganini at gmail dot com>.

Tested by me in production for several days at work.

4 files changed, 84 insertions, 22 deletions

diff --git a/sys/netsmb/smb_conn.c b/sys/netsmb/smb_conn.c
index 2095eb19ef55..2ee851bd4aec 100644
--- a/sys/netsmb/smb_conn.c
+++ b/sys/netsmb/smb_conn.c
@@ -444,13 +444,29 @@ smb_vc_create(struct smb_vcspec *vcspec,
 		goto fail;
 	if (vcspec->servercs[0]) {
 		error = (int)iconv_open(vcspec->servercs, vcspec->localcs,
-		    &vcp->vc_toserver);
+		    &vcp->vc_cp_toserver);
 		if (error)
 			goto fail;
 		error = (int)iconv_open(vcspec->localcs, vcspec->servercs,
-		    &vcp->vc_tolocal);
+		    &vcp->vc_cp_tolocal);
 		if (error)
 			goto fail;
+		vcp->vc_toserver = vcp->vc_cp_toserver;
+		vcp->vc_tolocal = vcp->vc_cp_tolocal;
+		iconv_add(ENCODING_UNICODE, ENCODING_UNICODE, SMB_UNICODE_NAME);
+		iconv_add(ENCODING_UNICODE, SMB_UNICODE_NAME, ENCODING_UNICODE);
+		error = (int)iconv_open(SMB_UNICODE_NAME, vcspec->localcs,
+		    &vcp->vc_ucs_toserver);
+		if (!error) {
+			error = (int)iconv_open(vcspec->localcs, SMB_UNICODE_NAME,
+			    &vcp->vc_ucs_tolocal);
+		}
+		if (error) {
+			if (vcp->vc_ucs_toserver)
+				iconv_close(vcp->vc_ucs_toserver);
+			vcp->vc_ucs_toserver = NULL;
+			vcp->vc_ucs_tolocal = NULL;
+		}
 	}
 	error = (int)smb_iod_create(vcp);
 	if (error)
@@ -486,9 +502,17 @@ smb_vc_free(struct smb_connobj *cp)
 	if (vcp->vc_toupper)
 		iconv_close(vcp->vc_toupper);
 	if (vcp->vc_tolocal)
-		iconv_close(vcp->vc_tolocal);
+		vcp->vc_tolocal = NULL;
 	if (vcp->vc_toserver)
-		iconv_close(vcp->vc_toserver);
+		vcp->vc_toserver = NULL;
+	if (vcp->vc_cp_tolocal)
+		iconv_close(vcp->vc_cp_tolocal);
+	if (vcp->vc_cp_toserver)
+		iconv_close(vcp->vc_cp_toserver);
+	if (vcp->vc_ucs_tolocal)
+		iconv_close(vcp->vc_ucs_tolocal);
+	if (vcp->vc_ucs_toserver)
+		iconv_close(vcp->vc_ucs_toserver);
 	smb_co_done(VCTOCP(vcp));
 	smb_sl_destroy(&vcp->vc_stlock);
 	free(vcp, M_SMBCONN);
diff --git a/sys/netsmb/smb_conn.h b/sys/netsmb/smb_conn.h
index 370f6cfaa7f0..94a7a820ed4e 100644
--- a/sys/netsmb/smb_conn.h
+++ b/sys/netsmb/smb_conn.h
@@ -242,6 +242,10 @@ struct smb_vc {
 	void *		vc_toupper;	/* local charset */
 	void *		vc_toserver;	/* local charset to server one */
 	void *		vc_tolocal;	/* server charset to local one */
+	void *		vc_cp_toserver;	/* local charset to server one (using CodePage) */
+	void *		vc_cp_tolocal;	/* server charset to local one (using CodePage) */
+	void *		vc_ucs_toserver; /* local charset to server one (using UCS-2) */
+	void *		vc_ucs_tolocal;	/* server charset to local one (using UCS-2) */
 	int		vc_number;	/* number of this VC from the client side */
 	int		vc_genid;
 	uid_t		vc_uid;		/* user id of connection */
@@ -272,6 +276,8 @@ struct smb_vc {
 
 #define SMB_UNICODE_STRINGS(vcp)	((vcp)->vc_hflags2 & SMB_FLAGS2_UNICODE)
 
+#define	SMB_UNICODE_NAME	"UCS-2LE"
+
 /*
  * smb_share structure describes connection to the given SMB share (tree).
  * Connection to share is always built on top of the VC.
diff --git a/sys/netsmb/smb_smb.c b/sys/netsmb/smb_smb.c
index d3fb94aececf..6c732f9d4043 100644
--- a/sys/netsmb/smb_smb.c
+++ b/sys/netsmb/smb_smb.c
@@ -121,9 +121,17 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 	u_int8_t wc, stime[8], sblen;
 	u_int16_t dindex, tw, tw1, swlen, bc;
 	int error, maxqsz;
+	int unicode = SMB_UNICODE_STRINGS(vcp);
+	void * servercharset = vcp->vc_toserver;
+	void * localcharset = vcp->vc_tolocal;
 
 	if (smb_smb_nomux(vcp, scred, __func__) != 0)
 		return EINVAL;
+	/* Disable Unicode for SMB_COM_NEGOTIATE requests */ 
+	if (unicode) {
+		vcp->vc_toserver = vcp->vc_cp_toserver;
+		vcp->vc_tolocal  = vcp->vc_cp_tolocal;
+	}
 	vcp->vc_hflags = 0;
 	vcp->vc_hflags2 = 0;
 	vcp->obj.co_flags &= ~(SMBV_ENCRYPT);
@@ -180,7 +188,7 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 					SMBERROR("Unexpected length of security blob (%d)\n", sblen);
 					break;
 				}
-				error = md_get_uint16(mdp, &bc);
+				error = md_get_uint16le(mdp, &bc);
 				if (error)
 					break;
 				if (sp->sv_caps & SMB_CAP_EXT_SECURITY)
@@ -193,6 +201,13 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 			}
 			if (sp->sv_sm & SMB_SM_SIGS_REQUIRE)
 				vcp->vc_hflags2 |= SMB_FLAGS2_SECURITY_SIGNATURE;
+			if (vcp->vc_ucs_toserver &&
+				sp->sv_caps & SMB_CAP_UNICODE) {
+				/*
+				* They do Unicode.
+				*/
+				vcp->obj.co_flags |= SMBV_UNICODE;
+			}
 			vcp->vc_hflags2 |= SMB_FLAGS2_KNOWS_LONG_NAMES;
 			if (dp->d_id == SMB_DIALECT_NTLM0_12 &&
 			    sp->sv_maxtx < 4096 &&
@@ -200,7 +215,13 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 				vcp->obj.co_flags |= SMBV_WIN95;
 				SMBSDEBUG("Win95 detected\n");
 			}
-		} else if (dp->d_id > SMB_DIALECT_CORE) {
+			error = 0;
+			break;
+		}
+		vcp->vc_hflags2 &= ~(SMB_FLAGS2_EXT_SEC|SMB_FLAGS2_DFS|
+				     SMB_FLAGS2_ERR_STATUS|SMB_FLAGS2_UNICODE);
+		unicode = 0;
+		if (dp->d_id > SMB_DIALECT_CORE) {
 			md_get_uint16le(mdp, &tw);
 			sp->sv_sm = tw;
 			md_get_uint16le(mdp, &tw);
@@ -217,7 +238,7 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 				if (swlen > SMB_MAXCHALLENGELEN)
 					break;
 				md_get_uint16(mdp, NULL);	/* mbz */
-				if (md_get_uint16(mdp, &bc) != 0)
+				if (md_get_uint16le(mdp, &bc) != 0)
 					break;
 				if (bc < swlen)
 					break;
@@ -259,6 +280,12 @@ smb_smb_negotiate(struct smb_vc *vcp, struct smb_cred *scred)
 		SMBSDEBUG("MAXTX = %d\n", sp->sv_maxtx);
 	}
 bad:
+	/* Restore Unicode conversion state */
+	if (unicode) {
+		vcp->vc_toserver = servercharset;
+		vcp->vc_tolocal  = localcharset;
+		vcp->vc_hflags2 |= SMB_FLAGS2_UNICODE;
+	}
 	smb_rq_done(rqp);
 	return error;
 }
@@ -273,9 +300,13 @@ smb_smb_ssnsetup(struct smb_vc *vcp, struct smb_cred *scred)
 	smb_uniptr unipp, ntencpass = NULL;
 	char *pp, *up, *pbuf, *encpass;
 	int error, plen, uniplen, ulen, upper;
+	u_int32_t caps = 0;
 
 	upper = 0;
 
+	if (vcp->obj.co_flags & SMBV_UNICODE)
+		caps |= SMB_CAP_UNICODE;
+
 again:
 
 	vcp->vc_smbuid = SMB_UID_UNKNOWN;
@@ -374,8 +405,7 @@ again:
 	} else {
 		mb_put_uint16le(mbp, uniplen);
 		mb_put_uint32le(mbp, 0);		/* reserved */
-		mb_put_uint32le(mbp, vcp->obj.co_flags & SMBV_UNICODE ?
-				     SMB_CAP_UNICODE : 0);
+		mb_put_uint32le(mbp, caps);
 		smb_rq_wend(rqp);
 		smb_rq_bstart(rqp);
 		mb_put_mem(mbp, pp, plen, MB_MSYSTEM);
@@ -477,24 +507,13 @@ smb_smb_treeconnect(struct smb_share *ssp, struct smb_cred *scred)
 	upper = 0;
 
 again:
-
-#if 0
 	/* Disable Unicode for SMB_COM_TREE_CONNECT_ANDX requests */
 	if (SSTOVC(ssp)->vc_hflags2 & SMB_FLAGS2_UNICODE) {
 		vcp = SSTOVC(ssp);
-		if (vcp->vc_toserver) {
-			iconv_close(vcp->vc_toserver);
-			/* Use NULL until UTF-8 -> ASCII works */
-			vcp->vc_toserver = NULL;
-		}
-		if (vcp->vc_tolocal) {
-			iconv_close(vcp->vc_tolocal);
-			/* Use NULL until ASCII -> UTF-8 works*/
-			vcp->vc_tolocal = NULL;
-		}
+		vcp->vc_toserver = vcp->vc_cp_toserver;
+		vcp->vc_tolocal = vcp->vc_cp_tolocal;
 		vcp->vc_hflags2 &= ~SMB_FLAGS2_UNICODE;
 	}
-#endif
 
 	ssp->ss_tid = SMB_TID_UNKNOWN;
 	error = smb_rq_alloc(SSTOCP(ssp), SMB_COM_TREE_CONNECT_ANDX, scred, &rqp);
@@ -559,6 +578,15 @@ again:
 	ssp->ss_tid = rqp->sr_rptid;
 	ssp->ss_vcgenid = vcp->vc_genid;
 	ssp->ss_flags |= SMBS_CONNECTED;
+	/*
+	 * If the server can speak Unicode then switch
+	 * our converters to do Unicode <--> Local
+	 */
+	if (vcp->obj.co_flags & SMBV_UNICODE) {
+		vcp->vc_toserver = vcp->vc_ucs_toserver;
+		vcp->vc_tolocal = vcp->vc_ucs_tolocal;
+		vcp->vc_hflags2 |= SMB_FLAGS2_UNICODE;
+	}
 bad:
 	if (encpass)
 		free(encpass, M_SMBTEMP);
diff --git a/sys/netsmb/smb_subr.c b/sys/netsmb/smb_subr.c
index c92f865fee86..3846b75f71c1 100644
--- a/sys/netsmb/smb_subr.c
+++ b/sys/netsmb/smb_subr.c
@@ -350,6 +350,8 @@ smb_put_dmem(struct mbchain *mbp, struct smb_vc *vcp, const char *src,
 	}
 	mbp->mb_copy = smb_copy_iconv;
 	mbp->mb_udata = dp;
+	if (SMB_UNICODE_STRINGS(vcp))
+		mb_put_padbyte(mbp);
 	return mb_put_mem(mbp, src, size, MB_MCUSTOM);
 }
 
@@ -362,6 +364,8 @@ smb_put_dstring(struct mbchain *mbp, struct smb_vc *vcp, const char *src,
 	error = smb_put_dmem(mbp, vcp, src, strlen(src), caseopt);
 	if (error)
 		return error;
+	if (SMB_UNICODE_STRINGS(vcp))
+		return mb_put_uint16le(mbp, 0);
 	return mb_put_uint8(mbp, 0);
 }