ipmi/watchdog: fix watchdog timeout set on reboot
authorValentin Vidic <Valentin.Vidic@CARNet.hr>
Fri, 5 May 2017 19:07:33 +0000 (21:07 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 7 Aug 2017 02:19:42 +0000 (19:19 -0700)
commit 860f01e96981a68553f3ca49f574ff14fe955e72 upstream.

systemd by default starts watchdog on reboot and sets the timer to
ShutdownWatchdogSec=10min.  Reboot handler in ipmi_watchdog than reduces
the timer to 120s which is not enough time to boot a Xen machine with
a lot of RAM.  As a result the machine is rebooted the second time
during the long run of (XEN) Scrubbing Free RAM.....

Fix this by setting the timer to 120s only if it was previously
set to a low value.

Signed-off-by: Valentin Vidic <Valentin.Vidic@CARNet.hr>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/char/ipmi/ipmi_watchdog.c

index 096f0cef4da14152653ea61daa010b367174759d..40d400fe5bb74d2b7817adba7a330a2c9ce5c2e0 100644 (file)
@@ -1162,10 +1162,11 @@ static int wdog_reboot_handler(struct notifier_block *this,
                        ipmi_watchdog_state = WDOG_TIMEOUT_NONE;
                        ipmi_set_timeout(IPMI_SET_TIMEOUT_NO_HB);
                } else if (ipmi_watchdog_state != WDOG_TIMEOUT_NONE) {
-                       /* Set a long timer to let the reboot happens, but
-                          reboot if it hangs, but only if the watchdog
+                       /* Set a long timer to let the reboot happen or
+                          reset if it hangs, but only if the watchdog
                           timer was already running. */
-                       timeout = 120;
+                       if (timeout < 120)
+                               timeout = 120;
                        pretimeout = 0;
                        ipmi_watchdog_state = WDOG_TIMEOUT_RESET;
                        ipmi_set_timeout(IPMI_SET_TIMEOUT_NO_HB);