Re: [PATCH] perf, x86: Optimize intel_pmu_pebs_fixup_ip()

From: Peter Zijlstra <peterz@infradead.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Don Zickus <dzickus@redhat.com>, Andi Kleen <ak@linux.intel.com>,
	dave.hansen@linux.intel.com,
	Stephane Eranian <eranian@google.com>,
	jmario@redhat.com,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Arnaldo Carvalho de Melo <acme@infradead.org>,
	Ingo Molnar <mingo@kernel.org>
Subject: Re: [PATCH] perf, x86: Optimize intel_pmu_pebs_fixup_ip()
Date: Wed, 23 Oct 2013 22:48:38 +0200	[thread overview]
Message-ID: <20131023204838.GB19466@laptop.lan> (raw)
In-Reply-To: <CA+55aFwfsyofuZtNyUGKPWdvqc=4gOW=HTFPxsiM-3GS1G8KKQ@mail.gmail.com>

On Wed, Oct 23, 2013 at 08:09:53AM +0100, Linus Torvalds wrote:
> On Tue, Oct 22, 2013 at 10:12 PM, Peter Zijlstra <peterz@infradead.org> wrote:
> >>
> >> Careful! There is one magic piece of state that you need to
> >> save-and-restore if you do this, namely %cr2. Taking a page fault
> >> always writes to %cr2, and we must *not* corrupt it in the NMI
> >> handler.
> >
> > It looks like this is already dealt with (a similar thing is done for
> > i386).
> 
> Oh, ok then, we should be good to go. I wonder why we needed that
> special "_nmi()" version, then..

Ah, the whole fault from nmi trickery from Steve is from after we did
the copy_from_user_nmi() thing. We're only just catching up :-)

> Please do check that NMI increment the irq-counts etc.. Otherwise
> you'll need to add the explicit "pagefault_disable/enable()" pair
> around the __copy_from_user_inatomic()..

Yeah, we add NMI_OFFSET to preempt_count on nmi_enter.

I'll also make sure to test we actually hit the fault path
by concurrently running something like:

 while :; echo 1 > /proc/sys/vm/drop_caches ; done

while doing perf top or so.. 