On Thu, Jul 30, 2009 at 11:38:22AM -0500, Langsdorf, Mark wrote:
> I'll look into it.
>
> First time I've had this bug reported, though.
>
It's happening because the suspend code runs with interrupts disabled,
and the powerpc workaround we do in the cpufreq suspend hook
calls the drivers ->get method.
powernow-k8's ->get does an smp_call_function_single
which needs interrupts enabled
If we just make the suspend code conditional on ppc, this whole problem
goes away.
Something like the patch below..
Dave
diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 5cc77fb..8d3b3d1 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -1288,6 +1288,14 @@ static int cpufreq_suspend(struct sys_device *sysdev, pm_message_t pmsg)
dprintk("suspending cpu %u\n", cpu);
+ /*
+ * This whole bogosity is here because Powerbooks are made of fail.
+ * No sane platform should need any of the code below to be run.
+ * (it's entirely the wrong thing to do, as driver->get may
+ * reenable interrupts on some architectures).
+ */
+
+#ifdef __powerpc__
if (!cpu_online(cpu))
return 0;
@@ -1346,6 +1354,7 @@ static int cpufreq_suspend(struct sys_device *sysdev, pm_message_t pmsg)
out:
cpufreq_cpu_put(cpu_policy);
+#endif /* __powerpc__ */
return ret;
}
@@ -1365,6 +1374,11 @@ static int cpufreq_resume(struct sys_device *sysdev)
dprintk("resuming cpu %u\n", cpu);
+ /* As with the ->suspend method, all the code below is
+ * only necessary because Powerbooks suck.
+ * See commit 42d4dc3f4e1e for jokes. */
+#ifdef __powerpc__
+
if (!cpu_online(cpu))
return 0;
@@ -1428,6 +1442,7 @@ out:
schedule_work(&cpu_policy->update);
fail:
cpufreq_cpu_put(cpu_policy);
+#endif /* __powerpc__ */
return ret;
}
--
To unsubscribe from this list: send the line "unsubscribe cpufreq" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
