[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <57e19865-6552-f8f1-a443-890c24d1dff4@huawei.com>
Date: Thu, 15 Sep 2022 11:57:41 +0800
From: Yicong Yang <yangyicong@...wei.com>
To: John Garry <john.garry@...wei.com>, Leo Yan <leo.yan@...aro.org>
CC: <yangyicong@...ilicon.com>, <acme@...nel.org>,
<peterz@...radead.org>, <alexander.shishkin@...ux.intel.com>,
<james.clark@....com>, <will@...nel.org>,
<mathieu.poirier@...aro.org>, <mark.rutland@....com>,
<suzuki.poulose@....com>, <jonathan.cameron@...wei.com>,
<mike.leach@...aro.org>, <gregkh@...uxfoundation.org>,
<helgaas@...nel.org>, <lorenzo.pieralisi@....com>,
<shameerali.kolothum.thodi@...wei.com>, <mingo@...hat.com>,
<linux-kernel@...r.kernel.org>,
<linux-arm-kernel@...ts.infradead.org>,
<linux-pci@...r.kernel.org>, <linux-perf-users@...r.kernel.org>,
<prime.zeng@...wei.com>, <zhangshaokun@...ilicon.com>,
<linuxarm@...wei.com>, <liuqi6124@...il.com>
Subject: Re: [RESEND PATCH v12 1/3] perf tool: arm: Refactor event list
iteration in auxtrace_record__init()
On 2022/9/14 22:27, Leo Yan wrote:
> On Wed, Sep 14, 2022 at 02:47:43PM +0100, John Garry wrote:
>
> [...]
>
>>> struct auxtrace_record
>>> *auxtrace_record__init(struct evlist *evlist, int *err)
>>> {
>>> - struct perf_pmu *cs_etm_pmu;
>>> + struct perf_pmu *cs_etm_pmu = NULL;
>>> + struct perf_pmu **arm_spe_pmus = NULL;
>>> struct evsel *evsel;
>>> - bool found_etm = false;
>>> + struct perf_pmu *found_etm = NULL;
>>> struct perf_pmu *found_spe = NULL;
>>> - struct perf_pmu **arm_spe_pmus = NULL;
>>> + int auxtrace_event_cnt = 0;
>>> int nr_spes = 0;
>>> - int i = 0;
>>> if (!evlist)
>>> return NULL;
>>> @@ -68,24 +84,23 @@ struct auxtrace_record
>>> arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
>>> evlist__for_each_entry(evlist, evsel) {
>>> - if (cs_etm_pmu &&
>>> - evsel->core.attr.type == cs_etm_pmu->type)
>>> - found_etm = true;
>>> -
>>> - if (!nr_spes || found_spe)
>>> - continue;
>>> -
>>> - for (i = 0; i < nr_spes; i++) {
>>> - if (evsel->core.attr.type == arm_spe_pmus[i]->type) {
>>> - found_spe = arm_spe_pmus[i];
>>> - break;
>>> - }
>>> - }
>>> + if (cs_etm_pmu && !found_etm) + found_etm =
>>> find_pmu_for_event(&cs_etm_pmu, 1, evsel);
>>> +
>>> + if (arm_spe_pmus && !found_spe)
>>> + found_spe = find_pmu_for_event(arm_spe_pmus, nr_spes, evsel);
>>
>> should you break if found_etm and found_spe are set? Or, indeed, error and
>> return directly as we do below? Indeed, I am not sure why you even require
>> auxtrace_event_cnt
>
> I think this was my suggestion :)
>
yes. thanks :). It's dicussed in v7 and for more information:
https://lore.kernel.org/lkml/20220430073411.GA657977@leoy-ThinkPad-X240s/
> We can check if both 'found_etm' and 'found_spe' are set and directly
> break (and bail out) for this case. But it would introduce more complex
> checking if we connect with patch 2 with new flag 'found_ptt', something
> like:
>
> if ((found_etm && found_spe) ||
> (found_etm && found_ptt) ||
> (found_spe && found_ptt))
> break;
>
> This is hard for later's extension if we need to support a new auxtrace
> event, so using auxtrace_event_cnt would be easier to extend more
> auxtrace event on Arm platforms.
>
> Thanks,
> Leo
>
>>> }
>>> +
>>> free(arm_spe_pmus);
>>> - if (found_etm && found_spe) {
>>> - pr_err("Concurrent ARM Coresight ETM and SPE operation not currently supported\n");
>>> + if (found_etm)
>>> + auxtrace_event_cnt++;
>>> +
>>> + if (found_spe)
>>> + auxtrace_event_cnt++;
>>> +
>>> + if (auxtrace_event_cnt > 1) {
>>> + pr_err("Concurrent AUX trace operation not currently supported\n");
>>> *err = -EOPNOTSUPP;
>>> return NULL;
>>> }
>>
> .
>
Powered by blists - more mailing lists