kern/53557: [patch]: kernel panic: fix for a divide by zero processor exception

[Christian S.J. Peron]

>Number:         53557
>Category:       kern
>Synopsis:       [patch]: kernel panic: fix for a divide by zero processor exception
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Fri Jun 20 08:10:14 PDT 2003
>Closed-Date:
>Last-Modified:
>Originator:     Christian S.J. Peron
>Release:        FreeBSD 5.1-RELEASE i386
>Organization:
>Environment:
System: FreeBSD h24-79-64-13 5.1-RELEASE FreeBSD 5.1-RELEASE #5: Fri Jun 20 09:49:44 CDT 2003 maneo at movl:/usr/src/sys/i386/compile/RAID0 i386


	
>Description:
	If root sets the kern.log_wakeups_per_second variable to 0
	the next time the callback for the syslog wakeup gets scheduled
	it will panic due to divide by zero machine exception.

	
>How-To-Repeat:
	sysctl kern.log_wakeups_per_second=0

	
>Fix:

I have enclosed a patch that appears to fix the problem:

--- /usr/src/sys/kern/subr_log.c.bak	Fri Jun 20 09:14:12 2003
+++ /usr/src/sys/kern/subr_log.c	Fri Jun 20 09:45:04 2003
@@ -99,6 +99,17 @@
 	log_open = 1;
 	callout_init(&logsoftc.sc_callout, 0);
 	fsetown(td->td_proc->p_pid, &logsoftc.sc_sigio);	/* signal process only */
+
+	/*
+	 * Check to make sure the `log_wakeups_per_second' sysctl variable
+	 * is sane so that we avoid a divide by zero machine exception.
+	 */
+	if (!log_wakeups_per_second) {
+		printf("syslog wakeup is less then one. Adjusting to "
+			"one wakeup per second.\n");
+		log_wakeups_per_second = 1;
+	}
+
 	callout_reset(&logsoftc.sc_callout, hz / log_wakeups_per_second,
 	    logtimeout, NULL);
 	return (0);
@@ -181,6 +192,17 @@
 
 	if (!log_open)
 		return;
+
+	/*
+	 * Check to make sure the `log_wakeups_per_second' sysctl variable
+	 * is sane so that we avoid a divide by zero machine exception.
+	 */
+	if (!log_wakeups_per_second) {
+		printf("syslog wakeup is less then one. Adjusting to "
+			"one wakeup per second.\n");
+		log_wakeups_per_second = 1;
+	}
+
 	if (msgbuftrigger == 0) {
 		callout_reset(&logsoftc.sc_callout,
 		    hz / log_wakeups_per_second, logtimeout, NULL);
>Release-Note:
>Audit-Trail:
>Unformatted: