linux-kernel - Re: [PATCH v3 2/3] x86/smp native_play_dead: Prefer cpuidle_play_dead() over mwait_play

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: <89ebe611-a9b8-43f5-bb52-e8a299f79188@linux.intel.com>
Date: Tue, 12 Nov 2024 11:55:56 +0100
From: Patryk Wlazlyn <patryk.wlazlyn@...ux.intel.com>
To: Dave Hansen <dave.hansen@...el.com>, x86@...nel.org
Cc: linux-kernel@...r.kernel.org, linux-pm@...r.kernel.org,
 rafael.j.wysocki@...el.com, len.brown@...el.com,
 artem.bityutskiy@...ux.intel.com, dave.hansen@...ux.intel.com
Subject: Re: [PATCH v3 2/3] x86/smp native_play_dead: Prefer
 cpuidle_play_dead() over mwait_play_dead()

> I don't think the bug has anything to do with this patch, really.
> There's no need to rehash it here.
>
> The issue here is that the only way to call mwait today is via
> mwait_play_dead() directly, using its internally-calculated hint.
>
> What you want is for a cpuidle-driver-calculated hint to be used.  So,
> you're using the new hint function via the cpuidle driver.  But you just
> need the cpuidle driver to be called first, not the old
> mwait_play_dead()-calculated hint.  The new code will still do mwait,
> just via a different path and with a different hint.

Ok. I'll just say that we change the order because idle driver may know better.

> The thing this doesn't mention is what the impact on everyone else is.
> I _think_ the ACPI cpuidle driver is the only worry.  Today, if there's
> a system that supports mwait and ACPI cpuidle, it'll use mwait.  After
> this patch, it'll use ACPI cpuidle.
>
> The changelog doesn't mention that behavior change or why that's OK.

True, but I think the mwait_play_dead() is exclusively for Intel. Other target
platforms get an early return. I'll include that in the commit message.

> Also, looking at this:
>
>> -    mwait_play_dead();
>>      if (cpuidle_play_dead())
>> -        hlt_play_dead();
>> +        mwait_play_dead();
>> +    hlt_play_dead();
>
> None of those return on success, right?
>
> Is there any reason this couldn't just be:
>
>     /* The first successful play_dead() will not return: */
>     cpuidle_play_dead();
>     mwait_play_dead();
>     hlt_play_dead();
>
> That has the added bonus of not needing to return anything from
> mwait_play_dead() and the resulting churn from the last patch.

mwait_play_dead may return if mwait_play_dead_with_hint returns and it only does
on non-smp builds. That being said, we do ignore the return value right now,
because in either case we want to enter hlt_play_dead() as a fallback, so I
guess we can make mwait_play_dead return void, but leave
mwait_play_dead_with_hint returning int or add ifdef CONFIG_SMP guards in
intel_idle.

When going with the return types proposed in this patch set, on non-smp builds
intel_idle would call mwait_play_dead_with_hint() which would "return 1"; and
propagate through cpuidle_play_dead().