2014-04-23 17:04:26

by Stephane Eranian

[permalink] [raw]
Subject: [PATCH] perf/x86: fix RAPL rdmsrl_safe() usage


This patch fixes a bug introduced by commit 24223657.

The rdmsrl_safe() function returns 0 on success.
The current code was failing to detect the RAPL PMU
on real hardware (missing /sys/devices/power) because
the return value of rdmsrl_safe() was misinterpreted.

Signed-off-by: Stephane Eranian <[email protected]>

diff --git a/arch/x86/kernel/cpu/perf_event_intel_rapl.c b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
index 7c87424..619f769 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_rapl.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
@@ -543,7 +543,8 @@ static int rapl_cpu_prepare(int cpu)
if (phys_id < 0)
return -1;

- if (!rdmsrl_safe(MSR_RAPL_POWER_UNIT, &msr_rapl_power_unit_bits))
+ /* protect rdmsrl() to handle virtualization */
+ if (rdmsrl_safe(MSR_RAPL_POWER_UNIT, &msr_rapl_power_unit_bits))
return -1;

pmu = kzalloc_node(sizeof(*pmu), GFP_KERNEL, cpu_to_node(cpu));


2014-04-23 19:00:26

by Borislav Petkov

[permalink] [raw]
Subject: Re: [PATCH] perf/x86: fix RAPL rdmsrl_safe() usage

On Wed, Apr 23, 2014 at 07:04:19PM +0200, Stephane Eranian wrote:
>
> This patch fixes a bug introduced by commit 24223657.

Maybe add

Fixes: <commit-id>

tag?

> The rdmsrl_safe() function returns 0 on success.
> The current code was failing to detect the RAPL PMU
> on real hardware (missing /sys/devices/power) because
> the return value of rdmsrl_safe() was misinterpreted.
>
> Signed-off-by: Stephane Eranian <[email protected]>

Acked-by: Borislav Petkov <[email protected]>

--
Regards/Gruss,
Boris.

Sent from a fat crate under my desk. Formatting is fine.
--

2014-04-23 19:27:29

by Venkatesh Srinivas

[permalink] [raw]
Subject: Re: [PATCH] perf/x86: fix RAPL rdmsrl_safe() usage

On Wed, Apr 23, 2014 at 12:00 PM, Borislav Petkov <[email protected]> wrote:
> On Wed, Apr 23, 2014 at 07:04:19PM +0200, Stephane Eranian wrote:
>>
>> This patch fixes a bug introduced by commit 24223657.
>
> Maybe add
>
> Fixes: <commit-id>
>
> tag?
>
>> The rdmsrl_safe() function returns 0 on success.
>> The current code was failing to detect the RAPL PMU
>> on real hardware (missing /sys/devices/power) because
>> the return value of rdmsrl_safe() was misinterpreted.
>>
>> Signed-off-by: Stephane Eranian <[email protected]>
>
> Acked-by: Borislav Petkov <[email protected]>

Brown-Paper-Bag-Worn-By: Venkatesh Srinivas <[email protected]>

-- vs;

2014-04-23 20:00:25

by Andi Kleen

[permalink] [raw]
Subject: Re: [PATCH] perf/x86: fix RAPL rdmsrl_safe() usage

> > Acked-by: Borislav Petkov <[email protected]>
>
> Brown-Paper-Bag-Worn-By: Venkatesh Srinivas <[email protected]>

Again the bug is really in KVM. We wouldn't need all this crap
if KVM didn't throw the bogus #GPs. I'm sure you missed plenty
of MSR accesses elsewhere, and it's a ticking time bomb.

If the VM lies to the kernel with the family-model you need to at
least limit the consequences. KVM is spectacularly failing
to do that by throwing bogus #GPs.

The brown paper bad needs to be employed on the KVM code
(or better the bug be fixed there, by defaulting to not #GP
on unknown MSRs)

-Andi

--
[email protected] -- Speaking for myself only

Subject: [tip:perf/urgent] perf/x86: Fix RAPL rdmsrl_safe() usage

Commit-ID: 9f7ff8931e3c5ddc8535476971ec9501e9555c05
Gitweb: http://git.kernel.org/tip/9f7ff8931e3c5ddc8535476971ec9501e9555c05
Author: Stephane Eranian <[email protected]>
AuthorDate: Wed, 23 Apr 2014 19:04:19 +0200
Committer: Ingo Molnar <[email protected]>
CommitDate: Thu, 24 Apr 2014 08:12:41 +0200

perf/x86: Fix RAPL rdmsrl_safe() usage

This patch fixes a bug introduced by:

24223657806a ("perf/x86/intel: Use rdmsrl_safe() when initializing RAPL PMU")

The rdmsrl_safe() function returns 0 on success.
The current code was failing to detect the RAPL PMU
on real hardware (missing /sys/devices/power) because
the return value of rdmsrl_safe() was misinterpreted.

Signed-off-by: Stephane Eranian <[email protected]>
Acked-by: Borislav Petkov <[email protected]>
Acked-by: Venkatesh Srinivas <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: http://lkml.kernel.org/r/20140423170418.GA12767@quad
Signed-off-by: Ingo Molnar <[email protected]>
---
arch/x86/kernel/cpu/perf_event_intel_rapl.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel_rapl.c b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
index 7c87424..619f769 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_rapl.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
@@ -543,7 +543,8 @@ static int rapl_cpu_prepare(int cpu)
if (phys_id < 0)
return -1;

- if (!rdmsrl_safe(MSR_RAPL_POWER_UNIT, &msr_rapl_power_unit_bits))
+ /* protect rdmsrl() to handle virtualization */
+ if (rdmsrl_safe(MSR_RAPL_POWER_UNIT, &msr_rapl_power_unit_bits))
return -1;

pmu = kzalloc_node(sizeof(*pmu), GFP_KERNEL, cpu_to_node(cpu));