Received-SPF: pass (google.com: domain of linux-kernel+bounces-71680-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Mon, 19 Feb 2024 11:29:26 -0500
From: Steven Rostedt <rostedt@goodmis.org>
To: Oleg Nesterov <oleg@redhat.com>
Cc: wenyang.linux@foxmail.com, Masami Hiramatsu <mhiramat@kernel.org>, Ingo
 Molnar <mingo@kernel.org>, Mathieu Desnoyers
 <mathieu.desnoyers@efficios.com>, Mel Gorman <mgorman@techsingularity.net>,
 Peter Zijlstra <peterz@infradead.org>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] coredump debugging: add a tracepoint to report the
 coredumping
Message-ID: <20240219112926.77ac16f8@gandalf.local.home>
In-Reply-To: <20240217104924.GB10393@redhat.com>
References: <tencent_5CD40341EC9384E9B7CC127EA5CF2655B408@qq.com>
	<20240217104924.GB10393@redhat.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Sat, 17 Feb 2024 11:49:24 +0100
Oleg Nesterov <oleg@redhat.com> wrote:

> On 02/17, wenyang.linux@foxmail.com wrote:
> >
> > From: Wen Yang <wenyang.linux@foxmail.com>
> >
> > Currently coredump_task_exit() takes some time to wait for the generation
> > of the dump file. But if the user-space wants to receive a notification
> > as soon as possible it maybe inconvenient.
> >
> > Add the new trace_sched_process_coredump() into coredump_task_exit(),
> > this way a user-space monitor could easily wait for the exits and
> > potentially make some preparations in advance.  
> 
> Can't comment, I never know when the new tracepoint will make sense.
> 
> Stupid question. Can we simply shift trace_sched_process_exit() up
> before coredump_task_exit() ?

Reading the rest of the thread and looking at the code, we do have this:

void __noreturn do_exit(long code)
{
	struct task_struct *tsk = current;
	int group_dead;

[...]
	acct_collect(code, group_dead);
	if (group_dead)
		tty_audit_exit();
	audit_free(tsk);

	tsk->exit_code = code;
	taskstats_exit(tsk, group_dead);

	exit_mm();

	if (group_dead)
		acct_process();
	trace_sched_process_exit(tsk);

There's a lot that happens before we trigger the above event. I could
imagine that there are users expecting those actions to have taken place by
the time the event triggered. Like the "exit_mm()" call, as well as many
others.

I would be leery of moving that tracepoint.

-- Steve