Re: [PATCH V6 3/8] x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()

From: Borislav Petkov <bp@alien8.de>
To: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: linux-kernel@vger.kernel.org,
	Peter Zijlstra <peterz@infradead.org>,
	Josh Poimboeuf <jpoimboe@redhat.com>,
	Andy Lutomirski <luto@kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Juergen Gross <jgross@suse.com>,
	x86@kernel.org, Lai Jiangshan <jiangshan.ljs@antgroup.com>,
	Ingo Molnar <mingo@redhat.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	"H. Peter Anvin" <hpa@zytor.com>
Subject: Re: [PATCH V6 3/8] x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()
Date: Wed, 27 Apr 2022 19:45:00 +0200	[thread overview]
Message-ID: <YmmBHABKMk7Ctx46@zn.tnic> (raw)
In-Reply-To: <20220421141055.316239-4-jiangshanlai@gmail.com>

On Thu, Apr 21, 2022 at 10:10:50PM +0800, Lai Jiangshan wrote:
> From: Lai Jiangshan <jiangshan.ljs@antgroup.com>
> 
> The macro idtentry calls error_entry() unconditionally even on XENPV.
> But the code XENPV needs in error_entry() is PUSH_AND_CLEAR_REGS only.
> And error_entry() also calls sync_regs() which has to deal with the
> case of XENPV via an extra branch so that it doesn't copy the pt_regs.

What extra branch?

Do you mean the

	if (regs != eregs)

test in sync_regs()?

I'm confused. Are you, per chance, aiming to optimize XENPV here or
what's up?

> And PUSH_AND_CLEAR_REGS in error_entry() makes the stack not return to
> its original place when the function returns, which means it is not
> possible to convert it to a C function.
> 
> Move PUSH_AND_CLEAR_REGS out of error_entry(), add a function to wrap
> PUSH_AND_CLEAR_REGS and call it before error_entry().
> 
> The new function call adds two instructions (CALL and RET) for every
> interrupt or exception.

Not only - it pushes all the regs in PUSH_AND_CLEAR_REGS too. I don't
understand why that matters here? It was done in error_entry anyway.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette