Re: Linux 4.15-rc2: Regression in resume from ACPI S3

From: Pavel Machek <pavel@ucw.cz>
To: Linus Torvalds <torvalds@linux-foundation.org>,
	luto@kernel.org, tglx@linutronix.de,
	jarkko.nikula@linux.intel.com
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>,
	Thomas Gleixner <tglx@linutronix.de>,
	Andy Lutomirski <luto@kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	the arch/x86 maintainers <x86@kernel.org>
Subject: Re: Linux 4.15-rc2: Regression in resume from ACPI S3
Date: Sun, 10 Dec 2017 19:56:38 +0100	[thread overview]
Message-ID: <20171210185638.GA10363@amd> (raw)
In-Reply-To: <CA+55aFxc9e=CuwrAzhX7BgV4Or+unTQDXW1Xk-gXOZPiAtN3Bw@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1950 bytes --]

On Sun 2017-12-10 08:37:56, Linus Torvalds wrote:
> On Sun, Dec 10, 2017 at 8:23 AM, Pavel Machek <pavel@ucw.cz> wrote:
> >
> > Can you do something about html emails? Quoting them doesn't work too well.
> 
> Yeah, and they don't show up onlkml either because of rules. I try to
> avoid them, but have been more on mobile for various reasons lately
> than usual. That should be over and done with after today, though.
> 
> >> Any chance to bisect it? This doesn't sound like the other problem
> >> we had.
> >
> > Let me try...
> >
> > v4.15-rc2: suspends/resumes ok.
> > 4ded3be: hangs on resume.
> >
> > Given that between those, there was supposed "fix" for suspend, I
> > believe I should try reverting that one first. .. if someone can tell
> > me commit id, that would help.
> 
> The fix in there should be 5b06bbcfc2c6 ("x86/power: Fix some ordering
> bugs in __restore_processor_context()") so you can certainly see if it
> works before that (or just reverting it).

Revert is easier.

> But there are also a few other x86 low-level things there, and that
> fix really looks very safe, so I'd almost expect something else to
> have triggered your problem. There's less than 500 commits in that
> range you have, so a few bisections should narrow it down a lot.

No, that commit does _look_ pretty suspect to me...

Confirmed, revert fixes it. You see how it moves fix_processor_context
around #ifdef CONFIG_X86_32 block? And how people forget 32-bit
machines exist? Aha.

Which brings me to .. various people do automated testing of
kernel. Testing 32-bit kernel for boot, and both 32-bit and 64-bit for
boot and suspend would be very nice. The last item is not hard, either:

sudo rtcwake -l -m mem -s 5

...should take 10 seconds or so. 

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]