sleeping function called from cpufreq
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
Quoted 19 lines · 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
Quoted 1 line
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.
include/linux/cpu.h renamed from include/linux/cpu.h.old +5 -1
@@ -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;