linux-kernel - Re: [PATCH v2] perf/core: Add support for PMUs that can be read from more than 1 CPU

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: <5A9DBE64.3060209@codeaurora.org>
Date:   Mon, 05 Mar 2018 14:02:12 -0800
From:   Saravana Kannan <skannan@...eaurora.org>
To:     Mark Rutland <mark.rutland@....com>
CC:     suzuki.poulose@....com, Peter Zijlstra <peterz@...radead.org>,
        rananta@...eaurora.org, Arnaldo Carvalho de Melo <acme@...nel.org>,
        linux-kernel@...r.kernel.org,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Ingo Molnar <mingo@...hat.com>,
        Namhyung Kim <namhyung@...nel.org>, avilaj@...eaurora.org,
        Jiri Olsa <jolsa@...hat.com>,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH v2] perf/core: Add support for PMUs that can be read from
 more than 1 CPU

On 03/05/2018 04:21 AM, Mark Rutland wrote:
> On Mon, Mar 05, 2018 at 12:17:02PM +0000, Mark Rutland wrote:
>> On Fri, Mar 02, 2018 at 05:14:53PM -0800, Saravana Kannan wrote:
>
>>> @@ -629,6 +629,7 @@ struct perf_event {
>>>
>>>   	int				oncpu;
>>>   	int				cpu;
>>> +	cpumask_t			readable_on_cpus;
>>
>> For most PMUs, this will be emptry, and it's potentially *very* large
>> (e.g. on systems where NR_CPUS is 4096). Please use a poitner to a mask,
>> as I suggested in [1], e.g.
>
>> [1] https://lkml.kernel.org/r/20171128124534.3jvuala525wvn64r@wfg-t540p.sh.intel.com
>
> Whoops, that should've been:
>
> [1] https://lkml.kernel.org/r/20180225143802.denbkubqjg2dc7af@salmiak
>

I didn't notice you mentioned the use of pointers, but I was planning on 
doing that anyway. But then I realize people will complain about 
cacheline bouncing across 4096 CPUs if I use a cpu mask pointer.

Thanks,
Saravana

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project