On May 9, 4:51pm, [EMAIL PROTECTED] (Dave Dykstra) wrote:
-- Subject: Re: [PATCH] rsync kills all user processes on fork failure
>
> What's the best fix against the current CVS? Should we back out the
> previous fix because that one only solved half the problem? Somebody
> please provide an updated patch.
>
-- End of excerpt from Dave Dykstra
I've appended an updated patch that can be applied to the current CVS.
Description:
-- Be extra careful: reject any negative pid in do_fork(),
not just -1 (since other negative values would also be
harmful if they were ever passed to kill, even though
such values should never occur).
-- Expand all_pids[] dynamically as needed in do_fork(), to
avoid any possibility of overflow.
-- Prevent kill(-1, SIGUSR2) in main.c; this is independent of
the interaction between do_fork() and kill_all(), which
involves SIGUSR1.
Bob Byrnes e-mail: [EMAIL PROTECTED]
Curl Corporation phone: 617-761-1200
400 Technology Square, 8th Floor fax: 617-761-1201
Cambridge, MA 02139
--- util.c.orig Thu May 9 22:12:37 2002
+++ util.c Thu May 9 22:15:20 2002
@@ -373,15 +373,25 @@
}
-static pid_t all_pids[10];
+static pid_t *all_pids;
static int num_pids;
+static int max_pids;
+
+#define MIN_PIDS 16
/** Fork and record the pid of the child. **/
pid_t do_fork(void)
{
pid_t newpid = fork();
- if (newpid != 0 && newpid != -1) {
+ /* we must ensure that newpid == -1 is not recorded in all_pids
+ if fork() fails, otherwise kill_all() could nuke the world! */
+ if (newpid > 0) {
+ if (num_pids >= max_pids) {
+ max_pids = (max_pids ? 2*max_pids : MIN_PIDS);
+ all_pids = (pid_t *)Realloc(all_pids, sizeof(pid_t)*max_pids);
+ if (!all_pids) out_of_memory("do_fork");
+ }
all_pids[num_pids++] = newpid;
}
return newpid;
--- main.c.orig Thu May 9 22:12:31 2002
+++ main.c Thu May 9 22:15:28 2002
@@ -408,6 +408,13 @@
;
}
+ /* if fork fails, we must avoid doing kill(-1, SIGUSR2) later,
+ since that could nuke the world! */
+ if (pid < 0) {
+ rprintf(FERROR,"could not fork in do_recv: %s\n",strerror(errno));
+ exit_cleanup(RERR_IPC);
+ }
+
close(recv_pipe[1]);
close(error_pipe[1]);
if (f_in != f_out) close(f_in);
--
To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/rsync
Before posting, read: http://www.tuxedo.org/~esr/faqs/smart-questions.html
