Re: [RFC PATCH 2/5] entry: Add calls for save/restore auxiliary pt_regs

From: Ira Weiny <ira.weiny@intel.com>
To: Borislav Petkov <bp@alien8.de>
Cc: Rik van Riel <riel@surriel.com>,
	Dave Hansen <dave.hansen@intel.com>,
	Dave Hansen <dave.hansen@linux.intel.com>, <x86@kernel.org>,
	<linux-kernel@vger.kernel.org>, <kernel-team@fb.com>
Subject: Re: [RFC PATCH 2/5] entry: Add calls for save/restore auxiliary pt_regs
Date: Tue, 9 Aug 2022 14:49:19 -0700	[thread overview]
Message-ID: <YvLWX4NiqReB5CHB@iweiny-desk3> (raw)
In-Reply-To: <YvKsS3QuOu4JVzZU@zn.tnic>

On Tue, Aug 09, 2022 at 08:49:47PM +0200, Borislav Petkov wrote:
> On Tue, Aug 09, 2022 at 11:38:03AM -0700, Ira Weiny wrote:
> > Thomas did a lot of work to make the entry code generic. So I was
> > keeping that work consistent. This also helps to ensure I did not miss
> > any places.
> 
> How about you worry about the other arches when you actually cross that
> bridge?

I guess I miss understood the reasoning behind Thomas' work.  No one mentioned
trying to isolate this to x86 during the PKS review.

> 
> > I don't believe this is correct because instrumentation is not enabled
> > here.
> 
> Why do you have to run
> 
> 	arch_save_aux_pt_regs()
> 
> with instrumentation enabled?

Sorry, that was carried over from the PKS series where I did need to call it.

I pulled this series over quickly to show basically what needed to be done.
But I did not review it for this detail.

> 
> Patch 5 does
> 
> +       struct pt_regs_auxiliary *aux_pt_regs = &to_extended_pt_regs(regs)->aux;
> +
> +       aux_pt_regs->cpu = raw_smp_processor_id();
> 
> only?
> 
> Why does that need to run with instrumentation enabled?

This does not.

Am I wrong that instrumentation begin/end are 0 overhead in production builds?

Ira

> 
> -- 
> Regards/Gruss,
>     Boris.
> 
> https://people.kernel.org/tglx/notes-about-netiquette