linux-kernel - Re: [PATCH] x86: add hintable NOPs emulation

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: <aKWfYo8lKmcHE14a@lx-t490>
Date: Wed, 20 Aug 2025 12:11:46 +0200
From: "Ahmed S. Darwish" <darwi@...utronix.de>
To: Marcos Del Sol Vives <marcos@...a.pet>
Cc: linux-kernel@...r.kernel.org, Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
	Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
	"H. Peter Anvin" <hpa@...or.com>, Brian Gerst <brgerst@...il.com>,
	Uros Bizjak <ubizjak@...il.com>, Ard Biesheuvel <ardb@...nel.org>,
	David Kaplan <david.kaplan@....com>, Kees Cook <kees@...nel.org>,
	"Peter Zijlstra (Intel)" <peterz@...radead.org>,
	Andrew Cooper <andrew.cooper3@...rix.com>,
	Oleg Nesterov <oleg@...hat.com>, "Xin Li (Intel)" <xin@...or.com>,
	Sabyrzhan Tasbolatov <snovitoll@...il.com>
Subject: Re: [PATCH] x86: add hintable NOPs emulation

On Wed, 20 Aug 2025, Marcos Del Sol Vives wrote:
>
> But I think the kernel should let the user know the binaries they're
> running are having some performance penalty due to this emulation, in case
> they want to recompile without the offending flags.
>
> Without the logging, they'd be in the dark and might get confused on why
> their programs are running slower than on other machines.
>

Not convinced; especially all the extra 'thread_struct' noise.

>
> I originally did that, but then realized it was not possible due to
> "handle_hnop" depending on the conditionally-available "hnop_warn" flag.
>

Please do:

    #ifdef CONFIG_X86_HNOP_EMU
    static bool handle_hnop(struct pt_regs *regs)
    {
	// Reference 'hnop_warn' as much as you like
    }
    #else
    static bool handle_hnop(struct pt_regs *regs)
    {
	return false;
    }
    # endif

Then this ugliness:

    static inline void handle_invalid_op(struct pt_regs *regs)
    {
    #ifdef CONFIG_X86_HNOP_EMU
     	if (user_mode(regs) && handle_hnop(regs))
     		return;
    #endif
	...
    }

can become normal code:

    static inline void handle_invalid_op(struct pt_regs *regs)
    {
     	if (user_mode(regs) && handle_hnop(regs))
     		return;
	...
    }

Good luck,

--
Ahmed S. Darwish
Linutronix GmbH