Received-SPF: pass (google.com: domain of linux-kernel+bounces-138692-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Wed, 10 Apr 2024 15:58:40 +0200
From: Frederic Weisbecker <frederic@kernel.org>
To: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org,
	Adrian Hunter <adrian.hunter@intel.com>,
	Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	Arnaldo Carvalho de Melo <acme@kernel.org>,
	Ian Rogers <irogers@google.com>, Ingo Molnar <mingo@redhat.com>,
	Jiri Olsa <jolsa@kernel.org>, Marco Elver <elver@google.com>,
	Mark Rutland <mark.rutland@arm.com>,
	Namhyung Kim <namhyung@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: Re: [PATCH v3 3/4] perf: Remove perf_swevent_get_recursion_context()
 from perf_pending_task().
Message-ID: <ZhabEORWtSsdSq9x@localhost.localdomain>
References: <20240322065208.60456-1-bigeasy@linutronix.de>
 <20240322065208.60456-4-bigeasy@linutronix.de>
 <ZhRqSEbyd1rqVwfN@pavilion.home>
 <20240409062501.h4rA_ck4@linutronix.de>
 <ZhUaAjhQXN6ahtpS@localhost.localdomain>
 <20240409105405.TXUU--_W@linutronix.de>
 <ZhUt8XMndGSwNuwx@localhost.localdomain>
 <20240409133336.Y4Io-16-@linutronix.de>
 <ZhZsOvM3uTP6nTnZ@localhost.localdomain>
 <20240410125126.X26tR8tM@linutronix.de>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20240410125126.X26tR8tM@linutronix.de>

Le Wed, Apr 10, 2024 at 02:51:26PM +0200, Sebastian Andrzej Siewior a ?crit :
> On 2024-04-10 12:38:50 [+0200], Frederic Weisbecker wrote:
> > Some alternatives:
> > 
> > _ Clear event->pending_work = 0 after perf_sigtrap(), preventing an
> > event in there from adding a new task work. We may miss a signal though...
> > 
> > _ Make the recursion context per task on -RT...
> 
> The per-task counter would be indeed the easiest thing to do. But then
> only for task context, right?

It should work for CPU context as well. A context switch shouldn't be
considered as recursion. Hopefully...

> 
> But why would we miss a signal if we clean event->pending_work late?
> Isn't cleaning late same as clearing in
> perf_swevent_put_recursion_context()?

Not exactly. perf_swevent_get_recursion_context() avoids a new software event
altogether from being recorded. It is completely ignored. Whereas clearing late
event->pending_work records new software events but ignores the signal.

> If clearing pending_work late works, I would prefer to avoid yet another
> per-task counter if possible.

Yeah I know :-/

Thanks.

> 
> > > > Thanks.
> 
> Sebastian