linux-kernel - Re: [PATCH v6 2/3]: perf/core: use context tstamp

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]

Date:   Thu, 3 Aug 2017 21:47:56 +0300
From:   Alexey Budankov <alexey.budankov@...ux.intel.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Andi Kleen <ak@...ux.intel.com>,
        Kan Liang <kan.liang@...el.com>,
        Dmitri Prokhorov <Dmitry.Prohorov@...el.com>,
        Valery Cherepennikov <valery.cherepennikov@...el.com>,
        Mark Rutland <mark.rutland@....com>,
        Stephane Eranian <eranian@...gle.com>,
        David Carrillo-Cisneros <davidcc@...gle.com>,
        linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v6 2/3]: perf/core: use context tstamp_data for skipped
 events on mux interrupt

On 03.08.2017 18:00, Peter Zijlstra wrote:
> On Wed, Aug 02, 2017 at 11:15:39AM +0300, Alexey Budankov wrote:
>> @@ -772,6 +780,10 @@ struct perf_event_context {
>>  	 */
>>  	u64				time;
>>  	u64				timestamp;
>> +	/*
>> +	 * Context cache for filtered out events;
>> +	 */
>> +	struct perf_event_tstamp	tstamp_data;
>>  
>>  	/*
>>  	 * These fields let us detect when two contexts have both
> 
> 
>> @@ -1379,6 +1379,9 @@ static void update_context_time(struct perf_event_context *ctx)
>>  
>>  	ctx->time += now - ctx->timestamp;
>>  	ctx->timestamp = now;
>> +
>> +	ctx->tstamp_data.running += ctx->time - ctx->tstamp_data.stopped;
>> +	ctx->tstamp_data.stopped = ctx->time;
>>  }
>>  
>>  static u64 perf_event_time(struct perf_event *event)
> 
> It appears to me we have some redundancy here.
> 
> 
>> @@ -1968,9 +1971,13 @@ event_sched_out(struct perf_event *event,
>>  	 */
>>  	if (event->state == PERF_EVENT_STATE_INACTIVE &&
>>  	    !event_filter_match(event)) {
>> +		delta = tstamp - event->tstamp->stopped;
>> +		event->tstamp->running += delta;
>> +		event->tstamp->stopped = tstamp;
>> +		if (event->tstamp != &event->tstamp_data) {
>> +			event->tstamp_data = *event->tstamp;
> 
> This,
> 
>> +			event->tstamp = &event->tstamp_data;
>> +		}
>>  	}
>>  
>>  	if (event->state != PERF_EVENT_STATE_ACTIVE)
> 
> 
>> @@ -3239,8 +3246,11 @@ ctx_pinned_sched_in(struct perf_event *event, void *data)
>>  
>>  	if (event->state <= PERF_EVENT_STATE_OFF)
>>  		return 0;
>> -	if (!event_filter_match(event))
>> +	if (!event_filter_match(event)) {
>> +		if (event->tstamp != &params->ctx->tstamp_data)
>> +			event->tstamp = &params->ctx->tstamp_data;
> 
> this and
> 
>>  		return 0;
>> +	}
>>  
>>  	/* may need to reset tstamp_enabled */
>>  	if (is_cgroup_event(event))
>> @@ -3273,8 +3283,11 @@ ctx_flexible_sched_in(struct perf_event *event, void *data)
>>  	 * Listen to the 'cpu' scheduling filter constraint
>>  	 * of events:
>>  	 */
>> -	if (!event_filter_match(event))
>> +	if (!event_filter_match(event)) {
>> +		if (event->tstamp != &params->ctx->tstamp_data)
>> +			event->tstamp = &params->ctx->tstamp_data;
> 
> this..
> 
>>  		return 0;
>> +	}
>>  
>>  	/* may need to reset tstamp_enabled */
>>  	if (is_cgroup_event(event))
> 
> 
> Are the magic spots, right? And I'm not convinced its right.
> 
> Suppose I have two events in my context, and I created them 1 minute
> apart. Then their respective tstamp_enabled are 1 minute apart as well.
> But the above doesn't seem to preserve that difference.
> 
> A similar argument can be made for running I think. That is a per event
> value and cannot be passed along to the ctx and back.

Aww, I see your point and it challenges my initial assumptions. 
Let me think thru the case more. There must be some solution. Thanks!

> 
> 
>