tech@mandoc.bsd.lv
 help / color / mirror / Atom feed
From: Kristaps Dzonsons <kristaps@bsd.lv>
To: tech@mdocml.bsd.lv
Subject: Valgrind Error
Date: Mon, 18 Aug 2014 02:13:14 +0200	[thread overview]
Message-ID: <53F1451A.5080106@bsd.lv> (raw)

[-- Attachment #1: Type: text/plain, Size: 1888 bytes --]

Hi,

So to test HEAD (accidentally--I was supposed to be testing 1.12.4-rc1 
but ran from the wrong directory), I wrote a little script to run 
through all the manuals on my system with mandoc via valgrind.  It has 
only found one nit on ALL of them.  On a binary file, at that!

Invalid read of size 1
    at 0x100017CC4: term_word (term.c:421)
    by 0x10000F4A1: print_man_node (man_term.c:974)
    by 0x10000F38C: print_man_nodelist (man_term.c:1042)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
    by 0x10000F3BC: print_man_nodelist (man_term.c:1045)
  Address 0x100082e12 is 0 bytes after a block of size 306 alloc'd
    at 0xC658: malloc (vg_replace_malloc.c:295)
    by 0x28E358: strdup (in /usr/lib/system/libsystem_c.dylib)
    by 0x100046164: mandoc_strdup (mandoc_aux.c:102)
    by 0x10003D89B: roff_strdup (roff.c:2233)
    by 0x100025636: man_word_alloc (man.c:308)
    by 0x100024E26: man_ptext (man.c:449)
    by 0x100024316: man_parseln (man.c:126)
    by 0x100047F1D: mparse_buf_r (read.c:562)
    by 0x10004665C: mparse_parse_buffer (read.c:718)
    by 0x1000467A3: mparse_readfd (read.c:764)
    by 0x10001E2C4: parse (main.c:308)
    by 0x10001DA08: main (main.c:244)

I tracked this down to passing "\H<nil>" to mandoc_escape().  It seems 
we're not very careful in this function to receiving \0 after the 
initial marker, so enclosed are some check.  I may have missed some, so 
please pass a critical eye over this as well!

Thoughts?

Best,

Kristaps

[-- Attachment #2: mandoc_term.diff --]
[-- Type: text/plain, Size: 1488 bytes --]

? Makefile.depend.patch
? Makefile.local
? apropos
? article-template.xml
? article1.html
? article1.xml
? cgi.h
? config.h
? config.log
? demandoc
? foo.1
? foo.1.html
? foo.man
? foo.ps
? foo.sh
? gluPerspective.3
? gluPerspective.html
? hspaces.diff
? html5.diff
? html5_cgi.diff
? html5_test2.diff
? itcrash.diff
? makewhatis
? mandoc
? mandoc.dSYM
? mandoc.html
? mandoc_term.diff
? mandocdb
? patch
? preconv
? querystring.diff
? roff_res_charwidth.patch
? scale.diff
? test.1
? test.1.html
? test.1.ps
? test.2
? test.2.ps
? test.ps
? testm.ps
? testn.ps
? unit_charwidth.patch
Index: mandoc.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mandoc.c,v
retrieving revision 1.85
diff -u -p -r1.85 mandoc.c
--- mandoc.c	16 Aug 2014 19:00:01 -0000	1.85
+++ mandoc.c	17 Aug 2014 23:56:04 -0000
@@ -150,6 +150,8 @@ mandoc_escape(const char **end, const ch
 			*start = ++*end;
 			term = ']';
 			break;
+		case '\0':
+			return(ESCAPE_ERROR);
 		default:
 			*sz = 1;
 			break;
@@ -199,7 +201,8 @@ mandoc_escape(const char **end, const ch
 		/* FALLTHROUGH */
 	case 'x':
 		if (strchr(" %&()*+-./0123456789:<=>", **start)) {
-			++*end;
+			if ('\0' != **start)
+				++*end;
 			return(ESCAPE_ERROR);
 		}
 		gly = ESCAPE_IGNORE;
@@ -250,6 +253,8 @@ mandoc_escape(const char **end, const ch
 			*start = ++*end;
 			term = '\'';
 			break;
+		case '\0':
+			return(ESCAPE_ERROR);
 		default:
 			*sz = 1;
 			break;

             reply	other threads:[~2014-08-18  0:13 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-08-18  0:13 Kristaps Dzonsons [this message]
2014-08-18  5:49 ` Ingo Schwarze
2014-08-18 10:21   ` More Valgrind Errors Kristaps Dzonsons
2014-08-18 16:39     ` Ingo Schwarze

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=53F1451A.5080106@bsd.lv \
    --to=kristaps@bsd.lv \
    --cc=tech@mdocml.bsd.lv \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).