Add support for expanding GNU long links from NetBSD. I've had this

in my tree for ages but didn't have a proper test case.  Thanks to
otto@ for providing one.

5 files changed, 71 insertions, 58 deletions

diff --git a/bin/pax/ar_subs.c b/bin/pax/ar_subs.c
index d2d92c4c31c..10add9e75a2 100644
--- a/bin/pax/ar_subs.c
+++ b/bin/pax/ar_subs.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: ar_subs.c,v 1.26 2004/03/15 22:18:15 djm Exp $	*/
+/*	$OpenBSD: ar_subs.c,v 1.27 2004/03/30 16:14:22 millert Exp $	*/
 /*	$NetBSD: ar_subs.c,v 1.5 1995/03/21 09:07:06 cgd Exp $	*/
 
 /*-
@@ -38,7 +38,7 @@
 #if 0
 static const char sccsid[] = "@(#)ar_subs.c	8.2 (Berkeley) 4/18/94";
 #else
-static const char rcsid[] = "$OpenBSD: ar_subs.c,v 1.26 2004/03/15 22:18:15 djm Exp $";
+static const char rcsid[] = "$OpenBSD: ar_subs.c,v 1.27 2004/03/30 16:14:22 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -109,7 +109,8 @@ list(void)
 			 * we need to read, to get the real filename
 			 */
 			off_t cnt;
-			if (!(*frmt->rd_data)(arcn, -1, &cnt))
+			if (!(*frmt->rd_data)(arcn, arcn->type == PAX_GLF
+			    ? -1 : -2, &cnt))
 				(void)rd_skip(cnt + arcn->pad);
 			continue;
 		}
@@ -201,7 +202,8 @@ extract(void)
 			/*
 			 * we need to read, to get the real filename
 			 */
-			if (!(*frmt->rd_data)(arcn, -1, &cnt))
+			if (!(*frmt->rd_data)(arcn, arcn->type == PAX_GLF
+			    ? -1 : -2, &cnt))
 				(void)rd_skip(cnt + arcn->pad);
 			continue;
 		}
diff --git a/bin/pax/buf_subs.c b/bin/pax/buf_subs.c
index ab8794e9d96..930566a9e6b 100644
--- a/bin/pax/buf_subs.c
+++ b/bin/pax/buf_subs.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: buf_subs.c,v 1.18 2003/10/20 06:22:27 jmc Exp $	*/
+/*	$OpenBSD: buf_subs.c,v 1.19 2004/03/30 16:14:22 millert Exp $	*/
 /*	$NetBSD: buf_subs.c,v 1.5 1995/03/21 09:07:08 cgd Exp $	*/
 
 /*-
@@ -38,7 +38,7 @@
 #if 0
 static const char sccsid[] = "@(#)buf_subs.c	8.2 (Berkeley) 4/18/94";
 #else
-static const char rcsid[] = "$OpenBSD: buf_subs.c,v 1.18 2003/10/20 06:22:27 jmc Exp $";
+static const char rcsid[] = "$OpenBSD: buf_subs.c,v 1.19 2004/03/30 16:14:22 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -683,7 +683,7 @@ rd_wrfile(ARCHD *arcn, int ofd, off_t *left)
 	 * pass the blocksize of the file being written to the write routine,
 	 * if the size is zero, use the default MINFBSZ
 	 */
-	if (ofd == -1)
+	if (ofd < 0)
 		sz = PAXPATHLEN + 1;		/* GNU tar long link/file */
 	else if (fstat(ofd, &sb) == 0) {
 		if (sb.st_blksize > 0)
diff --git a/bin/pax/extern.h b/bin/pax/extern.h
index f8632177e07..a05050a9cf1 100644
--- a/bin/pax/extern.h
+++ b/bin/pax/extern.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: extern.h,v 1.27 2003/06/26 00:10:17 deraadt Exp $	*/
+/*	$OpenBSD: extern.h,v 1.28 2004/03/30 16:14:22 millert Exp $	*/
 /*	$NetBSD: extern.h,v 1.5 1996/03/26 23:54:16 mrg Exp $	*/
 
 /*-
@@ -135,6 +135,7 @@ int bcpio_wr(ARCHD *);
 /*
  * file_subs.c
  */
+extern char *gnu_name_string, *gnu_link_string;
 int file_creat(ARCHD *);
 void file_close(ARCHD *, int);
 int lnk_creat(ARCHD *);
diff --git a/bin/pax/file_subs.c b/bin/pax/file_subs.c
index f1765f0f5c2..75314e6e913 100644
--- a/bin/pax/file_subs.c
+++ b/bin/pax/file_subs.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: file_subs.c,v 1.25 2003/10/20 06:22:27 jmc Exp $	*/
+/*	$OpenBSD: file_subs.c,v 1.26 2004/03/30 16:14:22 millert Exp $	*/
 /*	$NetBSD: file_subs.c,v 1.4 1995/03/21 09:07:18 cgd Exp $	*/
 
 /*-
@@ -38,7 +38,7 @@
 #if 0
 static const char sccsid[] = "@(#)file_subs.c	8.1 (Berkeley) 5/31/93";
 #else
-static const char rcsid[] = "$OpenBSD: file_subs.c,v 1.25 2003/10/20 06:22:27 jmc Exp $";
+static const char rcsid[] = "$OpenBSD: file_subs.c,v 1.26 2004/03/30 16:14:22 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -814,6 +814,7 @@ file_write(int fd, char *str, int cnt, int *rem, int *isempt, int sz,
 	char *end;
 	int wcnt;
 	char *st = str;
+	char **strp;
 
 	/*
 	 * while we have data to process
@@ -872,17 +873,28 @@ file_write(int fd, char *str, int cnt, int *rem, int *isempt, int sz,
 		/*
 		 * have non-zero data in this file system block, have to write
 		 */
-		if (fd == -1) {
-			/* GNU hack */
-			if (gnu_hack_string)
+		switch (fd) {
+		case -1:
+			strp = &gnu_name_string;
+			break;
+		case -2:
+			strp = &gnu_link_string;
+			break;
+		default:
+			strp = NULL;
+			break;
+		}
+		if (strp) {
+			if (*strp)
 				err(1, "WARNING! Major Internal Error! GNU hack Failing!");
-			gnu_hack_string = malloc(wcnt + 1);
-			if (gnu_hack_string == NULL) {
+			*strp = malloc(wcnt + 1);
+			if (*strp == NULL) {
 				paxwarn(1, "Out of memory");
 				return(-1);
 			}
-			memcpy(gnu_hack_string, st, wcnt);
-			gnu_hack_string[wcnt] = '\0';
+			memcpy(*strp, st, wcnt);
+			(*strp)[wcnt] = '\0';
+			break;
 		} else if (write(fd, st, wcnt) != wcnt) {
 			syswarn(1, errno, "Failed write to file %s", name);
 			return(-1);
diff --git a/bin/pax/tar.c b/bin/pax/tar.c
index b8b4eb3662c..e4005788794 100644
--- a/bin/pax/tar.c
+++ b/bin/pax/tar.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: tar.c,v 1.31 2003/06/26 00:10:18 deraadt Exp $	*/
+/*	$OpenBSD: tar.c,v 1.32 2004/03/30 16:14:23 millert Exp $	*/
 /*	$NetBSD: tar.c,v 1.5 1995/03/21 09:07:49 cgd Exp $	*/
 
 /*-
@@ -38,7 +38,7 @@
 #if 0
 static const char sccsid[] = "@(#)tar.c	8.2 (Berkeley) 4/18/94";
 #else
-static const char rcsid[] = "$OpenBSD: tar.c,v 1.31 2003/06/26 00:10:18 deraadt Exp $";
+static const char rcsid[] = "$OpenBSD: tar.c,v 1.32 2004/03/30 16:14:23 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -58,6 +58,7 @@ static const char rcsid[] = "$OpenBSD: tar.c,v 1.31 2003/06/26 00:10:18 deraadt
  * Routines for reading, writing and header identify of various versions of tar
  */
 
+static size_t expandname(char *, size_t, char **, const char *);
 static u_long tar_chksm(char *, int);
 static char *name_split(char *, int);
 static int ul_oct(u_long, char *, int, int);
@@ -70,7 +71,8 @@ static int uqd_oct(u_quad_t, char *, int, int);
  */
 
 static int tar_nodir;			/* do not write dirs under old tar */
-char *gnu_hack_string;			/* GNU ././@LongLink hackery */
+char *gnu_name_string;			/* GNU ././@LongLink hackery name */
+char *gnu_link_string;			/* GNU ././@LongLink hackery link */
 
 /*
  * tar_endwr()
@@ -385,21 +387,19 @@ tar_rd(ARCHD *arcn, char *buf)
 	 */
 	if (tar_id(buf, BLKMULT) < 0)
 		return(-1);
+	memset(arcn, 0, sizeof(*arcn));
 	arcn->org_name = arcn->name;
 	arcn->sb.st_nlink = 1;
-	arcn->pat = NULL;
 
 	/*
 	 * copy out the name and values in the stat buffer
 	 */
 	hd = (HD_TAR *)buf;
-	if (gnu_hack_string) {
-		arcn->nlen = strlcpy(arcn->name, gnu_hack_string,
-		    sizeof(arcn->name));
-		free(gnu_hack_string);
-		gnu_hack_string = NULL;
-	} else {
-		arcn->nlen = strlcpy(arcn->name, hd->name, sizeof(arcn->name));
+	if (hd->linkflag != LONGLINKTYPE && hd->linkflag != LONGNAMETYPE) {
+		arcn->nlen = expandname(arcn->name, sizeof(arcn->name),
+		    &gnu_name_string, hd->name);
+		arcn->ln_nlen = expandname(arcn->ln_name, sizeof(arcn->ln_name),
+		    &gnu_link_string, hd->linkname);
 	}
 	arcn->sb.st_mode = (mode_t)(asc_ul(hd->mode,sizeof(hd->mode),OCT) &
 	    0xfff);
@@ -427,8 +427,6 @@ tar_rd(ARCHD *arcn, char *buf)
 		 * the st_mode so -v printing will look correct.
 		 */
 		arcn->type = PAX_SLK;
-		arcn->ln_nlen = strlcpy(arcn->ln_name, hd->linkname,
-			sizeof(arcn->ln_name));
 		arcn->sb.st_mode |= S_IFLNK;
 		break;
 	case LNKTYPE:
@@ -438,8 +436,6 @@ tar_rd(ARCHD *arcn, char *buf)
 		 */
 		arcn->type = PAX_HLK;
 		arcn->sb.st_nlink = 2;
-		arcn->ln_nlen = strlcpy(arcn->ln_name, hd->linkname,
-			sizeof(arcn->ln_name));
 
 		/*
 		 * no idea of what type this thing really points at, but
@@ -448,19 +444,15 @@ tar_rd(ARCHD *arcn, char *buf)
 		arcn->sb.st_mode |= S_IFREG;
 		break;
 	case LONGLINKTYPE:
-		arcn->type = PAX_GLL;
-		/* FALLTHROUGH */
 	case LONGNAMETYPE:
 		/*
 		 * GNU long link/file; we tag these here and let the
 		 * pax internals deal with it -- too ugly otherwise.
 		 */
-		if (hd->linkflag != LONGLINKTYPE)
-			arcn->type = PAX_GLF;
+		arcn->type =
+		    hd->linkflag == LONGLINKTYPE ? PAX_GLL : PAX_GLF;
 		arcn->pad = TAR_PAD(arcn->sb.st_size);
 		arcn->skip = arcn->sb.st_size;
-		arcn->ln_name[0] = '\0';
-		arcn->ln_nlen = 0;
 		break;
 	case DIRTYPE:
 		/*
@@ -469,8 +461,6 @@ tar_rd(ARCHD *arcn, char *buf)
 		arcn->type = PAX_DIR;
 		arcn->sb.st_mode |= S_IFDIR;
 		arcn->sb.st_nlink = 2;
-		arcn->ln_name[0] = '\0';
-		arcn->ln_nlen = 0;
 		break;
 	case AREGTYPE:
 	case REGTYPE:
@@ -757,10 +747,9 @@ ustar_rd(ARCHD *arcn, char *buf)
 	 */
 	if (ustar_id(buf, BLKMULT) < 0)
 		return(-1);
+	memset(arcn, 0, sizeof(*arcn));
 	arcn->org_name = arcn->name;
 	arcn->sb.st_nlink = 1;
-	arcn->pat = NULL;
-	arcn->nlen = 0;
 	hd = (HD_USTAR *)buf;
 
 	/*
@@ -773,14 +762,15 @@ ustar_rd(ARCHD *arcn, char *buf)
 		dest += cnt;
 		*dest++ = '/';
 		cnt++;
-	}
-	if (gnu_hack_string) {
-		arcn->nlen = strlcpy(dest, gnu_hack_string,
-		    sizeof(arcn->name) - cnt);
-		free(gnu_hack_string);
-		gnu_hack_string = NULL;
 	} else {
-		arcn->nlen = strlcpy(dest, hd->name, sizeof(arcn->name) - cnt);
+		cnt = 0;
+	}
+
+	if (hd->typeflag != LONGLINKTYPE && hd->typeflag != LONGNAMETYPE) {
+		arcn->nlen = expandname(dest, sizeof(arcn->name) - cnt,
+		    &gnu_name_string, hd->name);
+		arcn->ln_nlen = expandname(arcn->ln_name, sizeof(arcn->ln_name),
+		    &gnu_link_string, hd->linkname);
 	}
 
 	/*
@@ -813,8 +803,6 @@ ustar_rd(ARCHD *arcn, char *buf)
 	/*
 	 * set the defaults, these may be changed depending on the file type
 	 */
-	arcn->ln_name[0] = '\0';
-	arcn->ln_nlen = 0;
 	arcn->pad = 0;
 	arcn->skip = 0;
 	arcn->sb.st_rdev = (dev_t)0;
@@ -869,11 +857,6 @@ ustar_rd(ARCHD *arcn, char *buf)
 			arcn->sb.st_mode |= S_IFREG;
 			arcn->sb.st_nlink = 2;
 		}
-		/*
-		 * copy the link name
-		 */
-		arcn->ln_nlen = strlcpy(arcn->ln_name, hd->linkname,
-			sizeof(arcn->ln_name));
 		break;
 	case LONGLINKTYPE:
 	case LONGNAMETYPE:
@@ -885,8 +868,6 @@ ustar_rd(ARCHD *arcn, char *buf)
 		    hd->typeflag == LONGLINKTYPE ? PAX_GLL : PAX_GLF;
 		arcn->pad = TAR_PAD(arcn->sb.st_size);
 		arcn->skip = arcn->sb.st_size;
-		arcn->ln_name[0] = '\0';
-		arcn->ln_nlen = 0;
 		break;
 	case CONTTYPE:
 	case AREGTYPE:
@@ -1135,3 +1116,20 @@ name_split(char *name, int len)
 	 */
 	return(start);
 }
+
+static size_t
+expandname(char *buf, size_t len, char **gnu_name, const char *name)
+{
+	size_t nlen;
+
+	if (*gnu_name) {
+		if ((nlen = strlcpy(buf, *gnu_name, len)) >= len)
+			nlen = len - 1;
+		free(*gnu_name);
+		*gnu_name = NULL;
+	} else {
+		if ((nlen = strlcpy(buf, name, len)) >= len)
+			nlen = len - 1;
+	}
+	return(nlen);
+}