similar named 8.3.3-REL crashes on two different machines....

Greg A. Woods woods at weird.com
Fri Aug 16 07:03:17 UTC 2002


[ On Friday, August 16, 2002 at 06:39:42 (+0000), Paul Vixie wrote: ]
> Subject: Re: similar named 8.3.3-REL crashes on two different machines.... 
>
> indeed, that is worrisome.  i don't see that comment in the master sources,
> so perhaps it showed up in an internal netbsd audit.  your line numbers are
> a little different from mine, too.  but: the comment seems accurate.

OI!  My apologies!  That seems to be my own comment.  I must have noted
it in my own reading at some point and then forgot about it -- good job
I made the comment in my source at the time!  ;-)

(the other line number changes are due to other comments I added
elsewhere too, though they seem far less meaningful to me reading them
now -- just questions regarding the EFAULT at line 717...)

> just sq_remove() as far as i'm concerned.  here's what i'm considering:
> 
> Index: ns_main.c
> ===================================================================
> RCS file: /proj/cvs/isc/bind8/src/bin/named/ns_main.c,v
> retrieving revision 8.160
> diff -u -r8.160 ns_main.c
> --- ns_main.c   24 Jun 2002 07:06:55 -0000      8.160
> +++ ns_main.c   16 Aug 2002 06:34:23 -0000
> @@ -825,8 +825,13 @@
>         sp->s_ifp = ifp;
>         INSIST(sizeof sp->s_temp >= INT16SZ);
>         iov = evConsIovec(sp->s_temp, INT16SZ);
> -       INSIST_ERR(evRead(lev, rfd, &iov, 1, stream_getlen, sp, &sp->evID_r)
> -                  != -1);
> +       if (evRead(lev, rfd, &iov, 1, stream_getlen, sp, &sp->evID_r) == -1) {
> +               ns_info(ns_log_default,
> +                       "stream_accept: evRead() failed: %s",
> +                       strerror(errno));
> +               sq_remove(sp);
> +               return;
> +       }
>         sp->flags |= STREAM_READ_EV;
>         ns_debug(ns_log_default, 1, "IP/TCP connection from %s (fd %d)",
>                  sin_ntoa(sp->s_from), rfd);
> 

That looks good to me.  I'll give it a try, though I've no idea at the
moment how to trigger the problem on demand....  :-(

-- 
								Greg A. Woods

+1 416 218-0098;            <g.a.woods at ieee.org>;           <woods at robohack.ca>
Planix, Inc. <woods at planix.com>; VE3TCP; Secrets of the Weird <woods at weird.com>


More information about the bind-workers mailing list