Date: Fri, 17 May 2013 11:16:51 +0200
From: Borislav Petkov <bp@alien8.de>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Robert Richter <rric@kernel.org>, Josh Boyer <jwboyer@redhat.com>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@ghostprotocols.net>, x86@kernel.org,
        linux-kernel@vger.kernel.org, gleb@redhat.com
Subject: Re: Drop WARN on AMD lack of perfctrs
Message-ID: <20130517091651.GB23035@pd.tnic>
References: <20130516151026.GB18325@hansolo.jdub.homelinux.org>
 <20130516175117.GK19669@dyad.programming.kicks-ass.net>
 <20130516175557.GC18325@hansolo.jdub.homelinux.org>
 <20130516181018.GO19669@dyad.programming.kicks-ass.net>
 <20130516205558.GE8356@rric.localhost>
 <20130516213420.GB31393@pd.tnic>
 <20130517090451.GQ19669@dyad.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20130517090451.GQ19669@dyad.programming.kicks-ass.net>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 948
Lines: 25

On Fri, May 17, 2013 at 11:04:51AM +0200, Peter Zijlstra wrote:
> Mostly so that check_hw_exists() doesn't need to know about the vendor
> specifics like where the MSRs live, how many there are etc..

Yep, but there will still be issues with perf when booted on a guest and
kvm not supporting it. And AFAIU, they're signalling this by turning off
CPUID bits so that initializing perf doesn't happen.

So, I think init_hw_perf_events should as a first step look at CPUID
bits and then do anything else. And this is done on Intel with
X86_FEATURE_ARCH_PERFMON. But Robert is fixing this on AMD too so...

Thanks.

-- 
Regards/Gruss,
    Boris.

Sent from a fat crate under my desk. Formatting is fine.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/