linux-kernel - Re: [PATCH v2] ACPI: Replace msleep() with usleep_range() in acpi_os

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: <CAJvTdKni91yBbCnUvzq1cPdYMNzbckFN4mJjn3ZEthhi=5PzhA@mail.gmail.com>
Date: Wed, 20 Nov 2024 13:35:48 -0500
From: Len Brown <lenb@...nel.org>
To: "Rafael J. Wysocki" <rafael@...nel.org>
Cc: anna-maria@...utronix.de, tglx@...utronix.de, peterz@...radead.org, 
	frederic@...nel.org, corbet@....net, akpm@...ux-foundation.org, 
	linux-acpi@...r.kernel.org, linux-pm@...r.kernel.org, 
	linux-kernel@...r.kernel.org, Len Brown <len.brown@...el.com>, 
	Arjan van de Ven <arjan@...ux.intel.com>, Todd Brandt <todd.e.brandt@...el.com>
Subject: Re: [PATCH v2] ACPI: Replace msleep() with usleep_range() in acpi_os_sleep().

On Mon, Nov 18, 2024 at 6:03 AM Rafael J. Wysocki <rafael@...nel.org> wrote:

> if (ms >= 12 * MSEC_PER_SEC / HZ) {
>         msleep(ms);
> } else {
>        u64 us = ms * USEC_PER_MSEC;
>
>       usleep_range(us, us / 8);
> }

The purpose of delaying a timer via timer slack is to afford the
opportunity to coalesce timers, when the precision wasn't actually
necessary.

I agree that  precision is not necessary here -- we are talking about
acpi_os_sleep(), which has a granularity of 1ms.

So the question becomes how much coalescing do we get for how much delay?

In the suspend/resume flows that sparked this discussion, the answer
is zero --- and so for that case, zero timer slack is justified.

I acknowledge that you refuse to apply my patch.

However, if you are going to add slack, I would like it to be clear
why users tasks, which may invoke HR timers at a dizzying rate, are
subject to a flat 50us timerslack_ns, while the entire ACPI
sub-system, which invokes a single timer, would see a 650usec delay
added for 5,000 usec sleep.
-- 
Len Brown, Intel