linux-kernel - Re: [PATCH 1/4] perf, x86: Avoid checkpointed counters causing excessive TSX aborts v4

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20130831150755.GY31370@twins.programming.kicks-ass.net>
Date:	Sat, 31 Aug 2013 17:07:55 +0200
From:	Peter Zijlstra <peterz@...radead.org>
To:	Andi Kleen <ak@...ux.intel.com>
Cc:	Andi Kleen <andi@...stfloor.org>, mingo@...nel.org,
	acme@...radead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] perf, x86: Avoid checkpointed counters causing
 excessive TSX aborts v4

On Fri, Aug 30, 2013 at 01:44:45PM -0700, Andi Kleen wrote:
> On Fri, Aug 30, 2013 at 06:02:15PM +0200, Peter Zijlstra wrote:
> > On Wed, Aug 21, 2013 at 04:47:23PM -0700, Andi Kleen wrote:
> > > @@ -1224,6 +1240,15 @@ again:
> > >  		x86_pmu.drain_pebs(regs);
> > >  	}
> > >  
> > > +	/*
> > > +	 * To avoid spurious interrupts with perf stat always reset checkpointed
> > > +	 * counters.
> > > +	 *
> > > +	 * XXX move somewhere else.
> > > +	 */
> > > +	if (cpuc->events[2] && event_is_checkpointed(cpuc->events[2]))
> > > +		status |= (1ULL << 2);
> > > +
> > >  	for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) {
> > >  		struct perf_event *event = cpuc->events[bit];
> > >  

> So can just drop the XXX comment. Ok?

How about hiding the entire thing in a hsw function. I'm fairly sure
that eventually we'll need to check all counters for this nonsense.

Something like so perhaps?

---
 arch/x86/kernel/cpu/perf_event_intel.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index a45d8d4..2a400b7 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1170,6 +1170,20 @@ static void intel_pmu_reset(void)
 	local_irq_restore(flags);
 }
 
+static void intel_pmu_hsw_tsx_status(struct cpu_hw_event *cpuc, u64 *status)
+{
+	const int idx = 2; /* only cnt2 supports TSX for now */
+	struct perf_event *event = cpuc->event[idx];
+
+	if (event_is_checkpoint(event)) {
+		/*
+		 * In order to avoid spurious interrupts always reset
+		 * checkpointed counters.
+		 */
+		*status |= (1ULL << idx);
+	}
+}
+
 /*
  * This handler is triggered by the local APIC, so the APIC IRQ handling
  * rules apply:
@@ -1224,6 +1238,8 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
 		x86_pmu.drain_pebs(regs);
 	}
 
+	intel_pmu_hsw_tsx_status(cpuc, &status);
+
 	for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) {
 		struct perf_event *event = cpuc->events[bit];
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/