From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753199AbbEKK0S (ORCPT ); Mon, 11 May 2015 06:26:18 -0400 Received: from ozlabs.org ([103.22.144.67]:39870 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752814AbbEKK0R (ORCPT ); Mon, 11 May 2015 06:26:17 -0400 In-Reply-To: <1430363091-27722-1-git-send-email-sukadev@linux.vnet.ibm.com> To: Sukadev Bhattiprolu , Paul Mackerras From: Michael Ellerman Cc: gusld@br.ibm.com, linuxppc-dev@lists.ozlabs.org, dev@codyps.com, linux-kernel@vger.kernel.org Subject: Re: [1/1] powerpc/hv-24x7: Check support before registering PMU Message-Id: <20150511102616.2004E140187@ozlabs.org> Date: Mon, 11 May 2015 20:26:15 +1000 (AEST) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, 2015-30-04 at 03:04:51 UTC, Sukadev Bhattiprolu wrote: > We currently try to register the 24x7 PMU unconditionally. Not all > Power systems support 24x7 counters (eg: Power7). On these systems > we get a backtrace during boot when trying to register the 24x7 PMU. > > Check if the hypervisor supports 24x7 counters before attempting to > register the 24x7 PMU. The concept is good, but .. > diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c > index ec2eb20..19a59a3 100644 > --- a/arch/powerpc/perf/hv-24x7.c > +++ b/arch/powerpc/perf/hv-24x7.c > @@ -1268,12 +1268,44 @@ static struct pmu h_24x7_pmu = { > .read = h_24x7_event_read, > }; > > +/* > + * Return 1 if we can access the 24x7 counter catalog from the hypervisor. > + * Return 0 otherwise. > + */ > +static int hv_has_24x7(void) bool please > +{ > + int page_size = 4096; > + unsigned long hret; > + void *page; > + struct hv_24x7_catalog_page_0 *page_0; > + > + page_0 = kmalloc(page_size*2, GFP_KERNEL); > + if (!page_0) > + return 0; > + > + /* buffer for catalog needs to be 4K-aligned */ > + page = (void *)((long)page_0 & (~page_size-1)) + page_size; > + hret = h_get_24x7_catalog_page(page, 0, 0); > + kfree(page_0); It shouldn't need to be this complicated. You should just be able to do the hcall with bogus args and if you get H_FUNCTION then 24x7 is not supported. cheers