sleeping function called from cpufreq

sleeping function called from cpufreq

[Pierre Ossman]

As of lately I've been getting tonnes of these:

[  610.185635] Debug: sleeping function called from invalid context at 
include/linux/rwsem.h:43
[  610.185647] in_atomic():1, irqs_disabled():0
[  610.185653]  [<c01041be>] dump_stack+0x1e/0x20
[  610.185667]  [<c0119b62>] __might_sleep+0xa2/0xc0
[  610.185678]  [<c029de86>] cpufreq_notify_transition+0x46/0x220
[  610.185690]  [<e09d08fc>] centrino_target+0xfc/0x130 [speedstep_centrino]
[  610.185708]  [<c029f17f>] __cpufreq_driver_target+0x5f/0x70
[  610.185718]  [<c02a029d>] cpufreq_set+0x7d/0xa0
[  610.185728]  [<c02a0339>] store_speed+0x49/0x50
[  610.185737]  [<c029e6c6>] store+0x46/0x60
[  610.185745]  [<c01a5f27>] flush_write_buffer+0x37/0x40
[  610.185754]  [<c01a5f98>] sysfs_write_file+0x68/0x90
[  610.185763]  [<c01639b8>] vfs_write+0xa8/0x190
[  610.185773]  [<c0163b57>] sys_write+0x47/0x70
[  610.185781]  [<c01032bb>] sysenter_past_esp+0x54/0x75

Ideas on solving it?

Rgds
Pierre

Re: sleeping function called from cpufreq

[Zwane Mwaikambo]

On Mon, 7 Nov 2005, Pierre Ossman wrote:

> As of lately I've been getting tonnes of these:
> 
> [  610.185635] Debug: sleeping function called from invalid context at
> include/linux/rwsem.h:43
> [  610.185647] in_atomic():1, irqs_disabled():0
> [  610.185653]  [<c01041be>] dump_stack+0x1e/0x20
> [  610.185667]  [<c0119b62>] __might_sleep+0xa2/0xc0
> [  610.185678]  [<c029de86>] cpufreq_notify_transition+0x46/0x220
> [  610.185690]  [<e09d08fc>] centrino_target+0xfc/0x130 [speedstep_centrino]
> [  610.185708]  [<c029f17f>] __cpufreq_driver_target+0x5f/0x70
> [  610.185718]  [<c02a029d>] cpufreq_set+0x7d/0xa0
> [  610.185728]  [<c02a0339>] store_speed+0x49/0x50
> [  610.185737]  [<c029e6c6>] store+0x46/0x60
> [  610.185745]  [<c01a5f27>] flush_write_buffer+0x37/0x40
> [  610.185754]  [<c01a5f98>] sysfs_write_file+0x68/0x90
> [  610.185763]  [<c01639b8>] vfs_write+0xa8/0x190
> [  610.185773]  [<c0163b57>] sys_write+0x47/0x70
> [  610.185781]  [<c01032bb>] sysenter_past_esp+0x54/0x75
> 
> Ideas on solving it?

There is a fix for it in 2.6.14-mm1 but i think it also requires the 
following patch in order to link. I'm not actually advocating the hotplug 
in progress flag and would prefer using a previously posted 
preempt_disable/enable method.

--- linux-mm/include/linux/cpu.h.old	2005-11-06 22:08:39.000000000 -0500
+++ linux-mm/include/linux/cpu.h	2005-11-06 22:41:17.000000000 -0500
@@ -33,7 +33,6 @@
 
 extern int register_cpu(struct cpu *, int, struct node *);
 extern struct sys_device *get_cpu_sysdev(int cpu);
-extern int current_in_cpu_hotplug(void);
 #ifdef CONFIG_HOTPLUG_CPU
 extern void unregister_cpu(struct cpu *, struct node *);
 #endif
@@ -43,6 +42,7 @@
 /* Need to know about CPUs going up/down? */
 extern int register_cpu_notifier(struct notifier_block *nb);
 extern void unregister_cpu_notifier(struct notifier_block *nb);
+extern int current_in_cpu_hotplug(void);
 
 int cpu_up(unsigned int cpu);
 
@@ -55,6 +55,10 @@
 static inline void unregister_cpu_notifier(struct notifier_block *nb)
 {
 }
+static inline int current_in_cpu_hotplug(void)
+{
+	return 0;
+}
 
 #endif /* CONFIG_SMP */
 extern struct sysdev_class cpu_sysdev_class;