Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1760729AbZDIEnT (ORCPT ); Thu, 9 Apr 2009 00:43:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753549AbZDIEnI (ORCPT ); Thu, 9 Apr 2009 00:43:08 -0400 Received: from bilbo.ozlabs.org ([203.10.76.25]:37854 "EHLO bilbo.ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753244AbZDIEnH (ORCPT ); Thu, 9 Apr 2009 00:43:07 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <18909.31952.873098.336615@cargo.ozlabs.ibm.com> Date: Thu, 9 Apr 2009 14:42:56 +1000 From: Paul Mackerras To: Ingo Molnar , Peter Zijlstra CC: linux-kernel@vger.kernel.org Subject: [PATCH] perf_counter: powerpc: add nmi_enter/nmi_exit calls X-Mailer: VM 8.0.9 under Emacs 22.2.1 (i486-pc-linux-gnu) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3012 Lines: 95 Impact: powerpc bug fix Now that the core is using in_nmi() (added in e30e08f6, "perf_counter: fix NMI race in task clock"), we need the powerpc perf_counter_interrupt to call nmi_enter() and nmi_exit() in those cases where the interrupt happens when interrupts are soft-disabled. If interrupts were soft-enabled, we can treat it as a regular interrupt and do irq_enter/irq_exit around the whole routine. This lets us get rid of the test_perf_counter_pending() call at the end of perf_counter_interrupt, thus simplifying things a little. Signed-off-by: Paul Mackerras --- arch/powerpc/kernel/perf_counter.c | 31 +++++++++++++++++-------------- 1 files changed, 17 insertions(+), 14 deletions(-) diff --git a/arch/powerpc/kernel/perf_counter.c b/arch/powerpc/kernel/perf_counter.c index c9d019f..bd76d0f 100644 --- a/arch/powerpc/kernel/perf_counter.c +++ b/arch/powerpc/kernel/perf_counter.c @@ -714,7 +714,7 @@ hw_perf_counter_init(struct perf_counter *counter) * here so there is no possibility of being interrupted. */ static void record_and_restart(struct perf_counter *counter, long val, - struct pt_regs *regs) + struct pt_regs *regs, int nmi) { s64 prev, delta, left; int record = 0; @@ -749,7 +749,7 @@ static void record_and_restart(struct perf_counter *counter, long val, * Finally record data if requested. */ if (record) - perf_counter_overflow(counter, 1, regs, 0); + perf_counter_overflow(counter, nmi, regs, 0); } /* @@ -762,6 +762,17 @@ static void perf_counter_interrupt(struct pt_regs *regs) struct perf_counter *counter; long val; int found = 0; + int nmi; + + /* + * If interrupts were soft-disabled when this PMU interrupt + * occurred, treat it as an NMI. + */ + nmi = !regs->softe; + if (nmi) + nmi_enter(); + else + irq_enter(); for (i = 0; i < cpuhw->n_counters; ++i) { counter = cpuhw->counter[i]; @@ -769,7 +780,7 @@ static void perf_counter_interrupt(struct pt_regs *regs) if ((int)val < 0) { /* counter has overflowed */ found = 1; - record_and_restart(counter, val, regs); + record_and_restart(counter, val, regs, nmi); } } @@ -796,18 +807,10 @@ static void perf_counter_interrupt(struct pt_regs *regs) */ mtspr(SPRN_MMCR0, cpuhw->mmcr[0]); - /* - * If we need a wakeup, check whether interrupts were soft-enabled - * when we took the interrupt. If they were, we can wake stuff up - * immediately; otherwise we'll have do the wakeup when interrupts - * get soft-enabled. - */ - if (test_perf_counter_pending() && regs->softe) { - irq_enter(); - clear_perf_counter_pending(); - perf_counter_do_pending(); + if (nmi) + nmi_exit(); + else irq_exit(); - } } void hw_perf_counter_setup(int cpu) -- 1.5.6.3 -- 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/