From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753320AbdESMY3 (ORCPT ); Fri, 19 May 2017 08:24:29 -0400 Received: from mga06.intel.com ([134.134.136.31]:51542 "EHLO mga06.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750783AbdESMY1 (ORCPT ); Fri, 19 May 2017 08:24:27 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.38,364,1491289200"; d="scan'208";a="104160313" Subject: Re: [PATCH] perf/x86/intel: Drop kernel samples even though :u is specified To: Peter Zijlstra Cc: acme@kernel.org, jolsa@kernel.org, mingo@redhat.com, alexander.shishkin@linux.intel.com, Linux-kernel@vger.kernel.org, ak@linux.intel.com, kan.liang@intel.com, yao.jin@intel.com, Will Deacon , Mark Rutland References: <1495189152-5766-1-git-send-email-yao.jin@linux.intel.com> <20170519092905.yivnfjyhblm5wiap@hirez.programming.kicks-ass.net> <20170519121056.c45jbsfgcdjfuoow@hirez.programming.kicks-ass.net> From: "Jin, Yao" Message-ID: Date: Fri, 19 May 2017 20:24:19 +0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.1.1 MIME-Version: 1.0 In-Reply-To: <20170519121056.c45jbsfgcdjfuoow@hirez.programming.kicks-ass.net> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 5/19/2017 8:10 PM, Peter Zijlstra wrote: > On Fri, May 19, 2017 at 08:06:09PM +0800, Jin, Yao wrote: >> SNIP >> >>> I would much rather see this in generic code, somewhere around >>> __perf_event_overflow() I suppose. That would retain proper accounting >>> for the interrupt rate etc.. >>> >>> Also it would work for all architectures. Because I'm thinking more than >>> just x86 will suffer from skid. >> Yes, moving to generic code is better. Thanks for the suggestion! I will do >> that. >> >>> If you're really worried, I suppose you can put it behind a PERF_PMU_CAP >>> flag or something. >> I guess what you are suggesting is to add checking like: >> >> if (is_sampling_event(event)) { >> if (event->pmu->capabilities & PERF_PMU_CAP_NO_INTERRUPT) { >> return; >> } >> } > Ah, I was more thinking of something like PERF_PMU_CAP_NO_SKID or > something that would skip the test and preserve current behaviour. OK, I understand now. For example, for PEBS event, its capabilities should be set with PERF_PMU_CAP_NO_SKID. If the event's capabilities is set with PERF_PMU_CAP_NO_SKID, it should skip the checking and keep current behavior. Thanks Jin Yao