Re: [PATCH v2 08/32] perf/x86/intel/cqm: add pmu sysfs attribute

[Thomas Gleixner]

On Wed, 11 May 2016, David Carrillo-Cisneros wrote:

> Expose max_recycle_threshold pmu attribute to user-space.
> 
> Reviewed-by: Stephane Eranian <eranian@xxxxxxxxxx>
> Signed-off-by: David Carrillo-Cisneros <davidcc@xxxxxxxxxx>
> ---
>  arch/x86/events/intel/cqm.c | 48 +++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 48 insertions(+)
> 
> diff --git a/arch/x86/events/intel/cqm.c b/arch/x86/events/intel/cqm.c
> index 54f219f..225b0c8 100644
> --- a/arch/x86/events/intel/cqm.c
> +++ b/arch/x86/events/intel/cqm.c
> @@ -393,9 +393,57 @@ static struct attribute_group intel_cqm_format_group = {
>  	.attrs = intel_cqm_formats_attr,
>  };
>  
> +static ssize_t
> +max_recycle_threshold_show(
> +	struct device *dev, struct device_attribute *attr, char *page)
> +{
> +	ssize_t rv;
> +
> +	monr_hrchy_acquire_mutexes();
> +	rv = snprintf(page, PAGE_SIZE - 1, "%u\n",
> +		      __intel_cqm_max_threshold);
> +	monr_hrchy_release_mutexes();

So we acquire a gazillion of mutexes to read a single variable?

> +
> +	return rv;
> +}
> +
> +static ssize_t
> +max_recycle_threshold_store(struct device *dev,
> +			    struct device_attribute *attr,
> +			    const char *buf, size_t count)
> +{
> +	unsigned int bytes;
> +	int ret;
> +
> +	ret = kstrtouint(buf, 0, &bytes);

That number is not limited by any means. So 0 ... UINT_MAX is valid, correct?

> +	if (ret)
> +		return ret;
> +
> +	/* Mutex waits for rotation logic in all packages to complete. */

What's wrong with just setting the variable and let it take effect on the next
rotation? That locking here is just pointless. It does not protect anything.

> +	monr_hrchy_acquire_mutexes();
> +
> +	__intel_cqm_max_threshold = bytes;
> +
> +	monr_hrchy_release_mutexes();

Thanks,

	tglx