Date: Sat, 31 Aug 2002 22:21:48 +0100 From: David Malone <dwmalone@maths.tcd.ie> To: Anders Nordby <anders@FreeBSD.org> Cc: FreeBSD-gnats-submit@FreeBSD.org, current@FreeBSD.org, mdodd@FreeBSD.org Subject: Re: bin/42255: Truss segfaults when tracing sshd Message-ID: <20020831212147.GA34455@walton.maths.tcd.ie> In-Reply-To: <20020831154526.B4E1A20273@totem.fix.no> References: <20020831154526.B4E1A20273@totem.fix.no>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Aug 31, 2002 at 05:45:26PM +0200, Anders Nordby wrote:
> # truss -p `sockstat -l | egrep 'sshd.*tcp4' | awk '{print $3}'`
>
> Log into the system with sshd, and truss will segfault:
There is an even easier way to reproduce this:
gonzo 9% sleep 10 &
[2] 35245
gonzo 10% truss -p 35245
*segfaults*
It is actually just strcmping a NULL syscall name, which can happen
if you truss a process which is waiting for a syscall to return
when you first attach to the process.
The patch below seems to fix the problem, but I Matthew would like
a more complex fix.
David.
ndex: syscalls.c
===================================================================
RCS file: /cvs/FreeBSD-CVS/src/usr.bin/truss/syscalls.c,v
retrieving revision 1.25
diff -u -r1.25 syscalls.c
--- syscalls.c 7 Aug 2002 11:35:18 -0000 1.25
+++ syscalls.c 31 Aug 2002 21:10:51 -0000
@@ -411,7 +411,7 @@
if (trussinfo->flags & FOLLOWFORKS)
len += fprintf(trussinfo->outfile, "%5d: ", trussinfo->pid);
- if (!strcmp(name, "execve") || !strcmp(name, "exit")) {
+ if (name != NULL && (!strcmp(name, "execve") || !strcmp(name, "exit"))) {
clock_gettime(CLOCK_REALTIME, &trussinfo->after);
}
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020831212147.GA34455>