Date: Sun, 17 Mar 2002 13:57:24 +1100 (EST) From: Bruce Evans <bde@zeta.org.au> To: Greg Lehey <grog@FreeBSD.org> Cc: Jens Schweikhardt <schweikh@FreeBSD.org>, <cvs-committers@FreeBSD.org>, <cvs-all@FreeBSD.org> Subject: Re: cvs commit: src/share/man/man4 ng_pptpgre.4 Message-ID: <20020317134715.U35416-100000@gamplex.bde.org> In-Reply-To: <20020317130517.C35248-100000@gamplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 17 Mar 2002, I wrote:
> That may be, but ispell is not man's best friend:
>
> $ man ispell
> Segmentation fault (core dumped)
>
> The bug seems to be in libz. Upgrading to the previous version of libz
> fixes it.
The following patch (obtained from the upgrade) fixes "man ispell"
(but may break libz).
%%%
Index: infcodes.c
===================================================================
RCS file: /home/ncvs/src/lib/libz/infcodes.c,v
retrieving revision 1.3
diff -u -2 -r1.3 infcodes.c
--- infcodes.c 11 Mar 2002 22:36:26 -0000 1.3
+++ infcodes.c 17 Mar 2002 02:41:53 -0000
@@ -201,5 +201,5 @@
case COPY: /* o: copying bytes in window, waiting for space */
f = q - c->sub.copy.dist;
- while (f < s->window) /* modulo window size-"while" instead */
+ if (f < s->window) /* modulo window size-"while" instead */
f += s->end - s->window; /* of "if" handles invalid distances */
while (c->len)
%%%
The while loop caused some pointer to become invalid. I think the bug
is really in gcc. The register hildng 's' seemed to get clobbered to
(s->end - s->window) so 's' was invalid after 1 iteration. With an
"if" instead of a "while", gcc generates simpler code with line numbers
that are actually correct and without the register clobber. Compiling
the unpatched version with -O0 also unbreaks "man ispell".
Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020317134715.U35416-100000>