Re: [PATCH] turbostat, servers do not support uncore power register

[Rafael J. Wysocki]

On Thursday, December 19, 2013 07:27:22 PM Prarit Bhargava wrote:
> From the Intel Software Developer's Manual section 14.7.4,
> 
> "For server platforms, PP1 domain is not supported, but its PP0 domain
> supports the MSR_PP0_PERF_STATUS interface."
> 
> When I run 'turbostat ls' I see the following error:
> 
> /dev/cpu/0/msr offset 0x641 read failed
> 
> because the server doesn't support the MSR_PP1_ENERGY_STATUS
> register (0x641).  This patch implements a get_msr_nowarn() which suppresses the
> above warning and then sets !RAPL_GFX.
> 
> Signed-off-by: Prarit Bhargava <prarit@xxxxxxxxxx>
> Cc: Len Brown <len.brown@xxxxxxxxx>
> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@xxxxxxxxx>
> Cc: Josh Triplett <josh@xxxxxxxxxxxxxxxx>
> Cc: Kristen Carlson Accardi <kristen@xxxxxxxxxxxxxxx>

Queued up for 3.14, thanks!

> ---
>  tools/power/x86/turbostat/turbostat.c |   23 ++++++++++++++++++-----
>  1 file changed, 18 insertions(+), 5 deletions(-)
> 
> diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
> index 9d77f13..ff505c2 100644
> --- a/tools/power/x86/turbostat/turbostat.c
> +++ b/tools/power/x86/turbostat/turbostat.c
> @@ -219,7 +219,7 @@ int cpu_migrate(int cpu)
>  		return 0;
>  }
>  
> -int get_msr(int cpu, off_t offset, unsigned long long *msr)
> +int _get_msr(int cpu, off_t offset, unsigned long long *msr, int warn)
>  {
>  	ssize_t retval;
>  	char pathname[32];
> @@ -234,13 +234,25 @@ int get_msr(int cpu, off_t offset, unsigned long long *msr)
>  	close(fd);
>  
>  	if (retval != sizeof *msr) {
> -		fprintf(stderr, "%s offset 0x%zx read failed\n", pathname, offset);
> +		if (warn)
> +			fprintf(stderr, "%s offset 0x%zx read failed\n",
> +				pathname, offset);
>  		return -1;
>  	}
>  
>  	return 0;
>  }
>  
> +int get_msr(int cpu, off_t offset, unsigned long long *msr)
> +{
> +	return _get_msr(cpu, offset, msr, 1);
> +}
> +
> +int get_msr_nowarn(int cpu, off_t offset, unsigned long long *msr)
> +{
> +	return _get_msr(cpu, offset, msr, 0);
> +}
> +
>  void print_header(void)
>  {
>  	if (show_pkg)
> @@ -953,9 +965,10 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)
>  		p->energy_dram = msr & 0xFFFFFFFF;
>  	}
>  	if (do_rapl & RAPL_GFX) {
> -		if (get_msr(cpu, MSR_PP1_ENERGY_STATUS, &msr))
> -			return -16;
> -		p->energy_gfx = msr & 0xFFFFFFFF;
> +		if (get_msr_nowarn(cpu, MSR_PP1_ENERGY_STATUS, &msr))
> +			do_rapl &= ~RAPL_GFX; /* unsupported on servers */
> +		else
> +			p->energy_gfx = msr & 0xFFFFFFFF;
>  	}
>  	if (do_rapl & RAPL_PKG_PERF_STATUS) {
>  		if (get_msr(cpu, MSR_PKG_PERF_STATUS, &msr))
> 

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
--
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/