x86/hwmon: avoid deadlock on CPU removal in pkgtemp
authorJan Beulich <jbeulich@novell.com>
Mon, 13 Sep 2010 10:11:05 +0000 (10:11 +0000)
committerGuenter Roeck <guenter.roeck@ericsson.com>
Fri, 24 Sep 2010 18:44:18 +0000 (11:44 -0700)
pkgtemp_device_remove(), holding the list protecting mutex, calls
pkgtemp_device_add(), which itself wants to acquire the same mutex.
Holding the mutex over the entire loop body in pkgtemp_device_remove()
isn't really necessary, as long as the loop gets exited after
processing the matched CPU.

Once exiting the loop after removing an eventual match, there's no
need for using the "safe" list iterator anymore.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
drivers/hwmon/pkgtemp.c

index 74157fcda6edf4bc569469db4ae1de9efa5b882d..844957333fe511f2716d2d9ddb2a43652844dd39 100644 (file)
@@ -339,17 +339,18 @@ exit:
 #ifdef CONFIG_HOTPLUG_CPU
 static void pkgtemp_device_remove(unsigned int cpu)
 {
-       struct pdev_entry *p, *n;
+       struct pdev_entry *p;
        unsigned int i;
        int err;
 
        mutex_lock(&pdev_list_mutex);
-       list_for_each_entry_safe(p, n, &pdev_list, list) {
+       list_for_each_entry(p, &pdev_list, list) {
                if (p->cpu != cpu)
                        continue;
 
                platform_device_unregister(p->pdev);
                list_del(&p->list);
+               mutex_unlock(&pdev_list_mutex);
                kfree(p);
                for_each_cpu(i, cpu_core_mask(cpu)) {
                        if (i != cpu) {
@@ -358,7 +359,7 @@ static void pkgtemp_device_remove(unsigned int cpu)
                                        break;
                        }
                }
-               break;
+               return;
        }
        mutex_unlock(&pdev_list_mutex);
 }