Merge commit 'acb089b983171667467adc66f56a723b609ed22e' into kbsd/vis

Highlights:
- Some style fixes
- Bumped mbbuf in istrsenvisx() to MB_LEN_MAX to avoid VLAs
- mbslength cannot go negative, so make it unsigned
- Further bounds checking & fix an additional overrun, with dlen == 0
- Avoid duplicate call to wcslen(start)

5 files changed, 76 insertions, 30 deletions

diff --git a/contrib/libc-vis/unvis.3 b/contrib/libc-vis/unvis.3
index 7f219916de16..fba3c9489e24 100644
--- a/contrib/libc-vis/unvis.3
+++ b/contrib/libc-vis/unvis.3
@@ -1,5 +1,4 @@
-.\"	$NetBSD: unvis.3,v 1.29 2017/10/24 19:14:55 abhinav Exp $
-.\"	$FreeBSD$
+.\"	$NetBSD: unvis.3,v 1.30 2019/05/08 15:37:41 bad Exp $
 .\"
 .\" Copyright (c) 1989, 1991, 1993
 .\"	The Regents of the University of California.  All rights reserved.
@@ -30,7 +29,7 @@
 .\"
 .\"     @(#)unvis.3	8.2 (Berkeley) 12/11/93
 .\"
-.Dd March 12, 2011
+.Dd May 8, 2019
 .Dt UNVIS 3
 .Os
 .Sh NAME
@@ -100,16 +99,18 @@ should be equal to the size of
 .Pp
 The
 .Fn strunvisx
-function does the same as the
+and
+.Fn strnunvisx
+functions do the same as the
 .Fn strunvis
-function,
-but it allows you to add a flag that specifies the style the string
+and
+.Fn strnunvis
+functions,
+but take a flag that specifies the style the string
 .Ar src
 is encoded with.
-Currently, the supported flags are:
-.Dv VIS_HTTPSTYLE
-and
-.Dv VIS_MIMESTYLE .
+The meaning of the flag is the same as explained below for
+.Fn unvis .
 .Pp
 The
 .Fn unvis
@@ -158,6 +159,10 @@ The
 .Fa flag
 argument is also used to specify the encoding style of the source.
 If set to
+.Dv VIS_NOESCAPE
+.Fn unvis
+will not decode backslash escapes.
+If set to
 .Dv VIS_HTTPSTYLE
 or
 .Dv VIS_HTTP1808 ,
diff --git a/contrib/libc-vis/unvis.c b/contrib/libc-vis/unvis.c
index d96d231be96e..f290d8db271a 100644
--- a/contrib/libc-vis/unvis.c
+++ b/contrib/libc-vis/unvis.c
@@ -1,4 +1,4 @@
-/*	$NetBSD: unvis.c,v 1.44 2014/09/26 15:43:36 roy Exp $	*/
+/*	$NetBSD: unvis.c,v 1.45 2022/04/19 20:32:15 rillig Exp $	*/
 
 /*-
  * Copyright (c) 1989, 1993
@@ -34,7 +34,7 @@
 #if 0
 static char sccsid[] = "@(#)unvis.c	8.1 (Berkeley) 6/4/93";
 #else
-__RCSID("$NetBSD: unvis.c,v 1.44 2014/09/26 15:43:36 roy Exp $");
+__RCSID("$NetBSD: unvis.c,v 1.45 2022/04/19 20:32:15 rillig Exp $");
 #endif
 #endif /* LIBC_SCCS and not lint */
 __FBSDID("$FreeBSD$");
@@ -514,7 +514,7 @@ strnunvisx(char *dst, size_t dlen, const char *src, int flag)
 			errno = ENOSPC; \
 			return -1; \
 		} \
-	} while (/*CONSTCOND*/0)
+	} while (0)
 
 	while ((c = *src++) != '\0') {
  again:
diff --git a/contrib/libc-vis/vis.3 b/contrib/libc-vis/vis.3
index 35dcffea5c78..18329589d940 100644
--- a/contrib/libc-vis/vis.3
+++ b/contrib/libc-vis/vis.3
@@ -1,5 +1,4 @@
-.\"	$NetBSD: vis.3,v 1.49 2017/08/05 20:22:29 wiz Exp $
-.\"	$FreeBSD$
+.\"	$NetBSD: vis.3,v 1.50 2022/12/04 11:25:08 uwe Exp $
 .\"
 .\" Copyright (c) 1989, 1991, 1993
 .\"	The Regents of the University of California.  All rights reserved.
@@ -180,7 +179,7 @@ and
 functions return \-1 and set
 .Va errno
 to
-.Dv ENOSPC .
+.Er ENOSPC .
 The
 .Fn strenvisx
 function takes an additional argument,
diff --git a/contrib/libc-vis/vis.c b/contrib/libc-vis/vis.c
index fe3939448087..a9f60c576277 100644
--- a/contrib/libc-vis/vis.c
+++ b/contrib/libc-vis/vis.c
@@ -1,4 +1,4 @@
-/*	$NetBSD: vis.c,v 1.74 2017/11/27 16:37:21 christos Exp $	*/
+/*	$NetBSD: vis.c,v 1.83 2023/08/12 12:48:52 riastradh Exp $	*/
 
 /*-
  * Copyright (c) 1989, 1993
@@ -57,7 +57,7 @@
 
 #include <sys/cdefs.h>
 #if defined(LIBC_SCCS) && !defined(lint)
-__RCSID("$NetBSD: vis.c,v 1.74 2017/11/27 16:37:21 christos Exp $");
+__RCSID("$NetBSD: vis.c,v 1.83 2023/08/12 12:48:52 riastradh Exp $");
 #endif /* LIBC_SCCS and not lint */
 #ifdef __FBSDID
 __FBSDID("$FreeBSD$");
@@ -360,7 +360,9 @@ makeextralist(int flags, const char *src)
 	if ((dst = calloc(len + MAXEXTRAS, sizeof(*dst))) == NULL)
 		return NULL;
 
-	if ((flags & VIS_NOLOCALE) || mbsrtowcs(dst, &src, len, &mbstate) == (size_t)-1) {
+	memset(&mbstate, 0, sizeof(mbstate));
+	if ((flags & VIS_NOLOCALE)
+	    || mbsrtowcs(dst, &src, len, &mbstate) == (size_t)-1) {
 		size_t i;
 		for (i = 0; i < len; i++)
 			dst[i] = (wchar_t)(u_char)src[i];
@@ -395,7 +397,7 @@ static int
 istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
     int flags, const char *mbextra, int *cerr_ptr)
 {
-	char mbbuf[MB_CUR_MAX];
+	char mbbuf[MB_LEN_MAX];
 	wchar_t *dst, *src, *pdst, *psrc, *start, *extra;
 	size_t len, olen;
 	uint64_t bmsk, wmsk;
@@ -403,7 +405,7 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	visfun_t f;
 	int clen = 0, cerr, error = -1, i, shft;
 	char *mbdst, *mbwrite, *mdst;
-	ssize_t mbslength;
+	size_t mbslength;
 	size_t maxolen;
 	mbstate_t mbstate;
 
@@ -411,7 +413,7 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	_DIAGASSERT(mbsrc != NULL || mblength == 0);
 	_DIAGASSERT(mbextra != NULL);
 
-	mbslength = (ssize_t)mblength;
+	mbslength = mblength;
 	/*
 	 * When inputing a single character, must also read in the
 	 * next character for nextc, the look-ahead character.
@@ -432,6 +434,14 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	 * return to the caller.
 	 */
 
+	/*
+	 * Guarantee the arithmetic on input to calloc won't overflow.
+	 */
+	if (mbslength > (SIZE_MAX - 1)/16) {
+		errno = ENOMEM;
+		return -1;
+	}
+
 	/* Allocate space for the wide char strings */
 	psrc = pdst = extra = NULL;
 	mdst = NULL;
@@ -463,12 +473,18 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	 * stop at NULs because we may be processing a block of data
 	 * that includes NULs.
 	 */
-	bzero(&mbstate, sizeof(mbstate));
+	memset(&mbstate, 0, sizeof(mbstate));
 	while (mbslength > 0) {
 		/* Convert one multibyte character to wchar_t. */
-		if (!cerr)
-			clen = mbrtowc(src, mbsrc, MIN(mbslength, MB_LEN_MAX),
+		if (!cerr) {
+			clen = mbrtowc(src, mbsrc,
+			    (mbslength < MB_LEN_MAX
+				? mbslength
+				: MB_LEN_MAX),
 			    &mbstate);
+			assert(clen < 0 || (size_t)clen <= mbslength);
+			assert(clen <= MB_LEN_MAX);
+		}
 		if (cerr || clen < 0) {
 			/* Conversion error, process as a byte instead. */
 			*src = (wint_t)(u_char)*mbsrc;
@@ -482,6 +498,20 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 			 */
 			clen = 1;
 		}
+		/*
+		 * Let n := MIN(mbslength, MB_LEN_MAX).  We have:
+		 *
+		 *	mbslength >= 1
+		 *	mbrtowc(..., n, &mbstate) <= n,
+		 *		by the contract of mbrtowc
+		 *
+		 *  clen is either
+		 *  (a) mbrtowc(..., n, &mbstate), in which case
+		 *      clen <= n <= mbslength; or
+		 *  (b) 1, in which case clen = 1 <= mbslength.
+		 */
+		assert(clen > 0);
+		assert((size_t)clen <= mbslength);
 		/* Advance buffer character pointer. */
 		src++;
 		/* Advance input pointer by number of bytes read. */
@@ -539,9 +569,21 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	 * output byte-by-byte here.  Else use wctomb().
 	 */
 	len = wcslen(start);
-	maxolen = dlen ? *dlen : (wcslen(start) * MB_LEN_MAX + 1);
+	if (dlen) {
+		maxolen = *dlen;
+		if (maxolen == 0) {
+			errno = ENOSPC;
+			goto out;
+		}
+	} else {
+		if (len > (SIZE_MAX - 1)/MB_LEN_MAX) {
+			errno = ENOSPC;
+			goto out;
+		}
+		maxolen = len*MB_LEN_MAX + 1;
+	}
 	olen = 0;
-	bzero(&mbstate, sizeof(mbstate));
+	memset(&mbstate, 0, sizeof(mbstate));
 	for (dst = start; len > 0; len--) {
 		if (!cerr) {
 			/*
@@ -614,6 +656,7 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
 	}
 
 	/* Terminate the output string. */
+	assert(olen < maxolen);
 	*mbdst = '\0';
 
 	if (flags & VIS_NOLOCALE) {
diff --git a/contrib/libc-vis/vis.h b/contrib/libc-vis/vis.h
index b024b8e8d2af..6c514758a26b 100644
--- a/contrib/libc-vis/vis.h
+++ b/contrib/libc-vis/vis.h
@@ -1,5 +1,4 @@
-/*	$NetBSD: vis.h,v 1.25 2017/04/23 01:57:36 christos Exp $	*/
-/*	$FreeBSD$	*/
+/*	$NetBSD: vis.h,v 1.26 2022/05/20 21:31:24 andvar Exp $	*/
 
 /*-
  * Copyright (c) 1990, 1993
@@ -41,7 +40,7 @@
  * to select alternate encoding format
  */
 #define	VIS_OCTAL	0x0001	/* use octal \ddd format */
-#define	VIS_CSTYLE	0x0002	/* use \[nrft0..] where appropiate */
+#define	VIS_CSTYLE	0x0002	/* use \[nrft0..] where appropriate */
 
 /*
  * to alter set of characters encoded (default is to encode all