Date: Tue, 26 Nov 2002 01:20:24 -0800 (PST) From: "Ronald F.Guilmette" <rfg@monkeys.com> To: FreeBSD-gnats-submit@FreeBSD.org Subject: gnu/45744: /usr/bin/makeinfo sometimes segfaults Message-ID: <20021126092024.384EE42260@segfault.monkeys.com>
next in thread | raw e-mail | index | archive | help
>Number: 45744
>Category: gnu
>Synopsis: /usr/bin/makeinfo sometimes segfaults
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Tue Nov 26 01:30:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator: Ronald F. Guilmette
>Release: FreeBSD 4.7-RELEASE i386
>Organization:
Infinite Monkeys & Co.
>Environment:
System: FreeBSD segfault.monkeys.com 4.7-RELEASE FreeBSD 4.7-RELEASE #0: Fri Nov 22 02:19:43 PST 2002 root@:/usr/src/sys/compile/rfg20021102 i386
>Description:
(NOTE: This bug has also/already been reported to the GNU Project's
maintainer of the GNU texinfo package.)
I have been experiencing segmentation faults in makeinfo for quite a
long time now... over a year. Please see:
http://www.freebsd.org/cgi/query-pr.cgi?pr=45598
for more info.
Anyway, I finally got off my ass and tracked down the root cause of
these crashes.
Quite simply, there are cases where the code on line 337 of the makinfo
index.c file, i.e.:
undefindex (name_index_alist[i]->name);
gets executed when name_index_alist[i] has a NULL value. And dereferencing
a NULL is a very Bad Idea. (Doing so causes segfaults. Oh yea.)
You can tell that the code fully _expects_ that name_index_alist[i] will
have a NULL value, in some cases, at this point in the code, just by looking
at the next following code line, which reads:
if (name_index_alist[i])
Obviously, it *is* possible for name_index_alist[i] to have a NULL value,
right around this point in the code, and indeed, that condition is even
expected. The problem is that we are dereferencing that value (on line 337)
before we have checked to see if the value is NULL or not. (If it is NULL,
then we definitely DO NOT want to perform the deference.)
A suitable (and trivial, and obvious) patch for this error is provided
below. All that is needed is to move the dereferencing statement to a
point _after_ we have checked if the value of if name_index_alist[i] is
NULL or not.
(The patch below should be suitable for use against both the current
GNU version of the texinfo package, i.e. 4.3, and also the 4.2 version,
as was used in the FreeBSD 4.7-RELEASE distribution.)
>How-To-Repeat:
Please see:
http://www.freebsd.org/cgi/query-pr.cgi?pr=45598
This bug can be trivially reproduced by simply attempting to build
/usr/ports/devel/autoconf213 and/or /usr/ports/mail/nmh, which it
turn triggers a build of /usr/ports/devel/autoconf213.
>Fix:
diff -rc2 src/4.3/makeinfo/index.c build/4.3/makeinfo/index.c
*** src/4.3/makeinfo/index.c Thu Nov 7 14:16:20 2002
--- build/4.3/makeinfo/index.c Tue Nov 26 00:53:39 2002
***************
*** 335,339 ****
for (i = 0; i < defined_indices; i++)
{
- undefindex (name_index_alist[i]->name);
if (name_index_alist[i])
{ /* Suppose we're called with two input files, and the first
--- 335,338 ----
***************
*** 343,346 ****
--- 342,346 ----
here; otherwise, when we try to define the pg index again
just below, it will still point to cp. */
+ undefindex (name_index_alist[i]->name);
free (name_index_alist[i]->name);
free (name_index_alist[i]);
>Release-Note:
>Audit-Trail:
>Unformatted:
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021126092024.384EE42260>