Re: [PATCH v2 11/12] perf/x86: make HT bug workaround conditioned on HT enabled

From: Stephane Eranian
Date: Wed Oct 22 2014 - 12:42:22 EST


On Wed, Oct 22, 2014 at 4:58 PM, Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
> On Thu, Oct 09, 2014 at 06:34:45PM +0200, Stephane Eranian wrote:
>
> SNIP
>
>> --- a/arch/x86/kernel/cpu/perf_event_intel.c
>> +++ b/arch/x86/kernel/cpu/perf_event_intel.c
>> @@ -12,6 +12,7 @@
>> #include <linux/init.h>
>> #include <linux/slab.h>
>> #include <linux/export.h>
>> +#include <linux/watchdog.h>
>>
>> #include <asm/cpufeature.h>
>> #include <asm/hardirq.h>
>> @@ -1811,7 +1812,7 @@ intel_start_scheduling(struct cpu_hw_events *cpuc)
>> /*
>> * nothing needed if in group validation mode
>> */
>> - if (cpuc->is_fake)
>> + if (cpuc->is_fake || !is_ht_workaround_enabled())
>> return;
>> /*
>> * no exclusion needed
>> @@ -1849,8 +1850,9 @@ intel_stop_scheduling(struct cpu_hw_events *cpuc)
>> /*
>> * nothing needed if in group validation mode
>> */
>> - if (cpuc->is_fake)
>> + if (cpuc->is_fake || !is_ht_workaround_enabled())
>> return;
>
>
> Is it possible to change is_ht_workaround_enabled() result within
> intel_start_scheduling and intel_stop_scheduling time frame? like
> following would happen:
>
>
> is_ht_workaround_enabled == 1
>
> intel_start_scheduling -> spin_lock(&excl_cntrs->lock);
>
> # echo 0 > /sys/devices/cpu/ht_bug_workaround
>
> is_ht_workaround_enabled == 0
>
> intel_stop_scheduling -> wont call spin_unlock(&excl_cntrs->lock)
>
>
> for some reason I can't get HT enabled on my HSW server, so I cannot try ;-)
>
Yes, that would be possible!
I can fix this by grabbing the lock in the sysctl handler.
Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/