[PATCH] device_shutdown can loop if the driver frees itself

This patch changes device_shutdown() to use the newly introduced safe
reverse list traversal.  We experienced loops on system reboot if we had
removed and re-inserted our device from the device list.

We noticed this problem on PPC405. Our PCI IDE device comes and goes a lot.

Our hypothesis was that there was a loop caused by the driver->shutdown
freeing memory.  It is possible that we do something wrong as well, but
being unable to reboot is kind of nasty.

Signed-off-by: Michael Richardson <mcr@marajade.sandelman.ca>
Cc: Patrick Mochel <mochel@digitalimplant.org>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

diff --git a/drivers/base/power/shutdown.c b/drivers/base/power/shutdown.c
index a47bb74..c2475f3 100644 (file)
--- a/drivers/base/power/shutdown.c
+++ b/drivers/base/power/shutdown.c
@@ -35,10 +35,10 @@ extern int sysdev_shutdown(void);
  */
 void device_shutdown(void)
 {
-       struct device * dev;
+       struct device * dev, *devn;
 
        down_write(&devices_subsys.rwsem);
-       list_for_each_entry_reverse(dev, &devices_subsys.kset.list,
+       list_for_each_entry_safe_reverse(dev, devn, &devices_subsys.kset.list,
                                kobj.entry) {
                if (dev->bus && dev->bus->shutdown) {
                        dev_dbg(dev, "shutdown\n");