Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Hello, On Wed, Oct 10, 2012 at 9:48 PM, Andreas Herrmann wrote: > Please ignore my patch. It was insufficiently tested -- sorry for this. > (powernowk8_target_fn might sleep). > > Have to take a closer look how to avoid below issue. Probably the only thing which can be done is always bouncing to the percpu work item. Thanks. -- tejun -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Rafael,
Please ignore my patch. It was insufficiently tested -- sorry for this.
(powernowk8_target_fn might sleep).
Have to take a closer look how to avoid below issue.
Regards,
Andreas
On Tue, Oct 09, 2012 at 09:38:44PM +0200, Andreas Herrmann wrote:
>
> Commit 6889125b8b4e09c5e53e6ecab3433bed1ce198c9
> (cpufreq/powernow-k8: workqueue user shouldn't migrate the kworker to another
> CPU)
> causes powernow-k8 to trigger a preempt warning, e.g.:
>
> BUG: using smp_processor_id() in preemptible [] code: cpufreq/3776
> caller is powernowk8_target+0x20/0x49
> Pid: 3776, comm: cpufreq Not tainted 3.6.0 #9
> Call Trace:
>[] debug_smp_processor_id+0xc7/0xe0
>[] powernowk8_target+0x20/0x49
>[] __cpufreq_driver_target+0x82/0x8a
>[] cpufreq_governor_performance+0x4e/0x54
>[] __cpufreq_governor+0x8c/0xc9
>[] __cpufreq_set_policy+0x1a9/0x21e
>[] store_scaling_governor+0x16f/0x19b
>[] ? cpufreq_update_policy+0x124/0x124
>[] ? _raw_spin_unlock_irqrestore+0x2c/0x49
>[] store+0x60/0x88
>[] sysfs_write_file+0xf4/0x130
>[] vfs_write+0xb5/0x151
>[] sys_write+0x4a/0x71
>[] system_call_fastpath+0x16/0x1b
>
> Fix this by using get_cpu/put_cpu in powernowk8_target().
>
> Cc: Tejun Heo
> Cc: sta...@vger.kernel.org
> Signed-off-by: Andreas Herrmann
> ---
> drivers/cpufreq/powernow-k8.c | 19 +++
> 1 file changed, 11 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/cpufreq/powernow-k8.c b/drivers/cpufreq/powernow-k8.c
> index 1a40935..3d1df2a 100644
> --- a/drivers/cpufreq/powernow-k8.c
> +++ b/drivers/cpufreq/powernow-k8.c
> @@ -1220,17 +1220,20 @@ static long powernowk8_target_fn(void *arg)
> static int powernowk8_target(struct cpufreq_policy *pol,
> unsigned targfreq, unsigned relation)
> {
> + int this_cpu, ret;
> struct powernowk8_target_arg pta = { .pol = pol, .targfreq = targfreq,
>.relation = relation };
>
> - /*
> - * Must run on @pol->cpu. cpufreq core is responsible for ensuring
> - * that we're bound to the current CPU and pol->cpu stays online.
> - */
> - if (smp_processor_id() == pol->cpu)
> - return powernowk8_target_fn();
> - else
> - return work_on_cpu(pol->cpu, powernowk8_target_fn, );
> + this_cpu = get_cpu();
> + if (this_cpu == pol->cpu) {
> + ret = powernowk8_target_fn();
> + put_cpu();
> + } else {
> + put_cpu();
> + ret = work_on_cpu(pol->cpu, powernowk8_target_fn, );
> + }
> +
> + return ret;
> }
>
> /* Driver entry point to verify the policy and range of frequencies */
> --
> 1.7.12
>
--
Operating | Advanced Micro Devices GmbH
System | Einsteinring 24, 85609 Dornach/Aschheim, Germany
Research | Geschäftsführer: Alberto Bozzo
Center | Sitz: Dornach, Gemeinde Aschheim, Landkreis München
(OSRC) | Registergericht München, HRB Nr. 43632, WEEE-Reg-Nr: DE 12919551
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Rafael,
Please ignore my patch. It was insufficiently tested -- sorry for this.
(powernowk8_target_fn might sleep).
Have to take a closer look how to avoid below issue.
Regards,
Andreas
On Tue, Oct 09, 2012 at 09:38:44PM +0200, Andreas Herrmann wrote:
Commit 6889125b8b4e09c5e53e6ecab3433bed1ce198c9
(cpufreq/powernow-k8: workqueue user shouldn't migrate the kworker to another
CPU)
causes powernow-k8 to trigger a preempt warning, e.g.:
BUG: using smp_processor_id() in preemptible [] code: cpufreq/3776
caller is powernowk8_target+0x20/0x49
Pid: 3776, comm: cpufreq Not tainted 3.6.0 #9
Call Trace:
[8125b447] debug_smp_processor_id+0xc7/0xe0
[814877e7] powernowk8_target+0x20/0x49
[81482b02] __cpufreq_driver_target+0x82/0x8a
[81484fc6] cpufreq_governor_performance+0x4e/0x54
[81482c50] __cpufreq_governor+0x8c/0xc9
[81482e6f] __cpufreq_set_policy+0x1a9/0x21e
[814839af] store_scaling_governor+0x16f/0x19b
[81484f16] ? cpufreq_update_policy+0x124/0x124
[8162b4a5] ? _raw_spin_unlock_irqrestore+0x2c/0x49
[81483640] store+0x60/0x88
[811708c0] sysfs_write_file+0xf4/0x130
[8111243b] vfs_write+0xb5/0x151
[811126e0] sys_write+0x4a/0x71
[816319a9] system_call_fastpath+0x16/0x1b
Fix this by using get_cpu/put_cpu in powernowk8_target().
Cc: Tejun Heo t...@kernel.org
Cc: sta...@vger.kernel.org
Signed-off-by: Andreas Herrmann andreas.herrma...@amd.com
---
drivers/cpufreq/powernow-k8.c | 19 +++
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/drivers/cpufreq/powernow-k8.c b/drivers/cpufreq/powernow-k8.c
index 1a40935..3d1df2a 100644
--- a/drivers/cpufreq/powernow-k8.c
+++ b/drivers/cpufreq/powernow-k8.c
@@ -1220,17 +1220,20 @@ static long powernowk8_target_fn(void *arg)
static int powernowk8_target(struct cpufreq_policy *pol,
unsigned targfreq, unsigned relation)
{
+ int this_cpu, ret;
struct powernowk8_target_arg pta = { .pol = pol, .targfreq = targfreq,
.relation = relation };
- /*
- * Must run on @pol-cpu. cpufreq core is responsible for ensuring
- * that we're bound to the current CPU and pol-cpu stays online.
- */
- if (smp_processor_id() == pol-cpu)
- return powernowk8_target_fn(pta);
- else
- return work_on_cpu(pol-cpu, powernowk8_target_fn, pta);
+ this_cpu = get_cpu();
+ if (this_cpu == pol-cpu) {
+ ret = powernowk8_target_fn(pta);
+ put_cpu();
+ } else {
+ put_cpu();
+ ret = work_on_cpu(pol-cpu, powernowk8_target_fn, pta);
+ }
+
+ return ret;
}
/* Driver entry point to verify the policy and range of frequencies */
--
1.7.12
--
Operating | Advanced Micro Devices GmbH
System | Einsteinring 24, 85609 Dornach/Aschheim, Germany
Research | Geschäftsführer: Alberto Bozzo
Center | Sitz: Dornach, Gemeinde Aschheim, Landkreis München
(OSRC) | Registergericht München, HRB Nr. 43632, WEEE-Reg-Nr: DE 12919551
--
To unsubscribe from this list: send the line unsubscribe linux-kernel in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Hello, On Wed, Oct 10, 2012 at 9:48 PM, Andreas Herrmann andreas.herrma...@amd.com wrote: Please ignore my patch. It was insufficiently tested -- sorry for this. (powernowk8_target_fn might sleep). Have to take a closer look how to avoid below issue. Probably the only thing which can be done is always bouncing to the percpu work item. Thanks. -- tejun -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Hello,
On Tue, Oct 09, 2012 at 09:38:44PM +0200, Andreas Herrmann wrote:
>
> Commit 6889125b8b4e09c5e53e6ecab3433bed1ce198c9
> (cpufreq/powernow-k8: workqueue user shouldn't migrate the kworker to another
> CPU)
> causes powernow-k8 to trigger a preempt warning, e.g.:
>
> BUG: using smp_processor_id() in preemptible [] code: cpufreq/3776
> caller is powernowk8_target+0x20/0x49
> Pid: 3776, comm: cpufreq Not tainted 3.6.0 #9
> Call Trace:
>[] debug_smp_processor_id+0xc7/0xe0
>[] powernowk8_target+0x20/0x49
>[] __cpufreq_driver_target+0x82/0x8a
>[] cpufreq_governor_performance+0x4e/0x54
>[] __cpufreq_governor+0x8c/0xc9
>[] __cpufreq_set_policy+0x1a9/0x21e
>[] store_scaling_governor+0x16f/0x19b
>[] ? cpufreq_update_policy+0x124/0x124
>[] ? _raw_spin_unlock_irqrestore+0x2c/0x49
>[] store+0x60/0x88
>[] sysfs_write_file+0xf4/0x130
>[] vfs_write+0xb5/0x151
>[] sys_write+0x4a/0x71
>[] system_call_fastpath+0x16/0x1b
...
> - /*
> - * Must run on @pol->cpu. cpufreq core is responsible for ensuring
> - * that we're bound to the current CPU and pol->cpu stays online.
> - */
Urgh... so this wasn't true? Well, the perils of the last minute
changes.
> - if (smp_processor_id() == pol->cpu)
> - return powernowk8_target_fn();
> - else
> - return work_on_cpu(pol->cpu, powernowk8_target_fn, );
> + this_cpu = get_cpu();
> + if (this_cpu == pol->cpu) {
> + ret = powernowk8_target_fn();
> + put_cpu();
> + } else {
> + put_cpu();
> + ret = work_on_cpu(pol->cpu, powernowk8_target_fn, );
> + }
> +
> + return ret;
Looking at the code, yes, I think the above is correct. Rafael, can
you please confirm?
Thanks.
--
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] cpufreq, powernow-k8: Fix usage of smp_processor_id() in preemptible code
Hello,
On Tue, Oct 09, 2012 at 09:38:44PM +0200, Andreas Herrmann wrote:
Commit 6889125b8b4e09c5e53e6ecab3433bed1ce198c9
(cpufreq/powernow-k8: workqueue user shouldn't migrate the kworker to another
CPU)
causes powernow-k8 to trigger a preempt warning, e.g.:
BUG: using smp_processor_id() in preemptible [] code: cpufreq/3776
caller is powernowk8_target+0x20/0x49
Pid: 3776, comm: cpufreq Not tainted 3.6.0 #9
Call Trace:
[8125b447] debug_smp_processor_id+0xc7/0xe0
[814877e7] powernowk8_target+0x20/0x49
[81482b02] __cpufreq_driver_target+0x82/0x8a
[81484fc6] cpufreq_governor_performance+0x4e/0x54
[81482c50] __cpufreq_governor+0x8c/0xc9
[81482e6f] __cpufreq_set_policy+0x1a9/0x21e
[814839af] store_scaling_governor+0x16f/0x19b
[81484f16] ? cpufreq_update_policy+0x124/0x124
[8162b4a5] ? _raw_spin_unlock_irqrestore+0x2c/0x49
[81483640] store+0x60/0x88
[811708c0] sysfs_write_file+0xf4/0x130
[8111243b] vfs_write+0xb5/0x151
[811126e0] sys_write+0x4a/0x71
[816319a9] system_call_fastpath+0x16/0x1b
...
- /*
- * Must run on @pol-cpu. cpufreq core is responsible for ensuring
- * that we're bound to the current CPU and pol-cpu stays online.
- */
Urgh... so this wasn't true? Well, the perils of the last minute
changes.
- if (smp_processor_id() == pol-cpu)
- return powernowk8_target_fn(pta);
- else
- return work_on_cpu(pol-cpu, powernowk8_target_fn, pta);
+ this_cpu = get_cpu();
+ if (this_cpu == pol-cpu) {
+ ret = powernowk8_target_fn(pta);
+ put_cpu();
+ } else {
+ put_cpu();
+ ret = work_on_cpu(pol-cpu, powernowk8_target_fn, pta);
+ }
+
+ return ret;
Looking at the code, yes, I think the above is correct. Rafael, can
you please confirm?
Thanks.
--
tejun
--
To unsubscribe from this list: send the line unsubscribe linux-kernel in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
