linux-kernel - Re: [PATCH 4/8] shazptr: Avoid synchronize

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: <20250625115643.GE1613376@noisy.programming.kicks-ass.net>
Date: Wed, 25 Jun 2025 13:56:43 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Boqun Feng <boqun.feng@...il.com>
Cc: linux-kernel@...r.kernel.org, rcu@...r.kernel.org, lkmm@...ts.linux.dev,
	Ingo Molnar <mingo@...nel.org>, Will Deacon <will@...nel.org>,
	Waiman Long <longman@...hat.com>,
	Davidlohr Bueso <dave@...olabs.net>,
	"Paul E. McKenney" <paulmck@...nel.org>,
	Josh Triplett <josh@...htriplett.org>,
	Frederic Weisbecker <frederic@...nel.org>,
	Neeraj Upadhyay <neeraj.upadhyay@...nel.org>,
	Joel Fernandes <joelagnelf@...dia.com>,
	Uladzislau Rezki <urezki@...il.com>,
	Steven Rostedt <rostedt@...dmis.org>,
	Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
	Lai Jiangshan <jiangshanlai@...il.com>,
	Zqiang <qiang.zhang@...ux.dev>, Breno Leitao <leitao@...ian.org>,
	aeh@...a.com, netdev@...r.kernel.org, edumazet@...gle.com,
	jhs@...atatu.com, kernel-team@...a.com,
	Erik Lundgren <elundgren@...a.com>
Subject: Re: [PATCH 4/8] shazptr: Avoid synchronize_shaptr() busy waiting


Response is a bit weird because non-linear editing..

On Tue, Jun 24, 2025 at 08:10:57PM -0700, Boqun Feng wrote:

> +	/* Whether the scan kthread has been scheduled to scan */
> +	bool scheduled;

> +static int __noreturn shazptr_scan_kthread(void *unused)
> +{
> +	for (;;) {
> +		swait_event_idle_exclusive(shazptr_scan.wq,
> +					   READ_ONCE(shazptr_scan.scheduled));

This seems weird; why use a whole wait-queue, in exclusive mode no less,
for something that is one known thread.

Also, I think this thing needs to be FREEZABLE, otherwise suspend might
have issues.

Why not just write it like:

		for (;;) {
			set_current_state(TASK_IDLE | TASK_FREEZABLE);
			if (!list_empty(&scan->queue))
				break;
			schedule();
		}
		__set_current_state(TASK_RUNNABLE);

		for (;;) {
			scoped_guard (mutex, scan->lock) {
				if (list_empty(scan->queued) &&
				    list_empty(scan->scanning))
					break;
			}

			shazptr_do_scan(scan);
		}


> +		shazptr_do_scan(&shazptr_scan);
> +
> +		scoped_guard(mutex, &shazptr_scan.lock) {
> +			if (list_empty(&shazptr_scan.queued) &&
> +			    list_empty(&shazptr_scan.scanning))
> +				shazptr_scan.scheduled = false;

This condition, why can't we directly use this condition instead of
scheduled?

> +		}
> +	}
> +}


> +static void synchronize_shazptr_normal(void *ptr)
> +{

> +
> +	/* Found blocking slots, prepare to wait. */
> +	if (blocking_grp_mask) {
> +		struct shazptr_scan *scan = &shazptr_scan;
> +		struct shazptr_wait wait = {
> +			.blocking_grp_mask = blocking_grp_mask,
> +		};
> +
> +		INIT_LIST_HEAD(&wait.list);
> +		init_completion(&wait.done);
> +
> +		scoped_guard(mutex, &scan->lock) {
> +			list_add_tail(&wait.list, &scan->queued);
> +
> +			if (!scan->scheduled) {
> +				WRITE_ONCE(scan->scheduled, true);
> +				swake_up_one(&shazptr_scan.wq);
> +			}

Or perhaps; just write this like:

			bool was_empty = list_empty(&scan->queued);
			list_add_tail(&wait.list, &scan->queued);
			if (was_empty)
				wake_up_process(scan->thread);

> +		}
> +
> +		wait_for_completion(&wait.done);
> +	}
> +}