linux-kernel - Re: [RFC PATCH] x86 / hibernate: Eliminate the redundant smp_ops.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] [day] [month] [year] [list]

Message-ID: <CAJZ5v0jqEPMQ1rLpeH0ZV8DcsJMOXdyoYPwwRkr=2vUBCADjVQ@mail.gmail.com>
Date: Wed, 26 Feb 2025 19:18:28 +0100
From: "Rafael J. Wysocki" <rafael@...nel.org>
To: Wentao Guan <guanwentao@...ontech.com>
Cc: "Rafael J. Wysocki" <rafael@...nel.org>, 王昱力 <wangyuli@...ontech.com>, 
	pavel <pavel@...nel.org>, tglx <tglx@...utronix.de>, mingo <mingo@...hat.com>, 
	bp <bp@...en8.de>, "dave.hansen" <dave.hansen@...ux.intel.com>, x86 <x86@...nel.org>, 
	hpa <hpa@...or.com>, linux-pm <linux-pm@...r.kernel.org>, 
	linux-kernel <linux-kernel@...r.kernel.org>, 占俊 <zhanjun@...ontech.com>, 
	聂诚 <niecheng1@...ontech.com>, 
	陈麟轩 <chenlinxuan@...ontech.com>, 
	Huacai Chen <chenhuacai@...ngson.cn>
Subject: Re: [RFC PATCH] x86 / hibernate: Eliminate the redundant
 smp_ops.play_dead assignment

On Wed, Feb 26, 2025 at 7:08 PM Wentao Guan <guanwentao@...ontech.com> wrote:
>
> Hello,
>
> Thanks for your reply.
>
> In my opinion, the only logic different before the patch is delete smp_ops.play_dead
> save and restore, as the comment "the resumed kernel will decide itself" and same
> logic as which in arch/arm64/kernel/hibernate.c, the ok path will work as expect.

Yes, the OK path will work as expected so long as smp_ops.play_dead is
switched over to resume_play_dead before calling
freeze_secondary_cpus().

> When discussing the error path and ret value that we not restore play_dead,
> I will try to analyze the difference between native_play_dead and resume_play_dead,
> and sev_es_play_dead [the all possiable three value], and I see some mwait and hlt
> way difference.[maybe it happens as disable the cpu failed and goes to Enable_cpus
> path in func:resume_target_kernel in hibernate.c? ] Is that it desgin to do and we can
> move it to a common place in hibernate.c and left some comments ?

Why not leave it as is?