linux-kernel - Re: [PATCH] x86/speculation: Add support for STIBP always-on preferred mode

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]

Date:   Tue, 11 Dec 2018 22:45:41 +0000
From:   "Lendacky, Thomas" <Thomas.Lendacky@....com>
To:     Borislav Petkov <bp@...en8.de>
CC:     "x86@...nel.org" <x86@...nel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Andrea Arcangeli <aarcange@...hat.com>,
        Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>,
        Jiri Kosina <jkosina@...e.cz>, Ingo Molnar <mingo@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Tim Chen <tim.c.chen@...ux.intel.com>,
        David Woodhouse <dwmw@...zon.co.uk>
Subject: Re: [PATCH] x86/speculation: Add support for STIBP always-on
 preferred mode

On 12/11/2018 04:22 PM, Borislav Petkov wrote:
> On Tue, Dec 11, 2018 at 10:17:30PM +0000, Lendacky, Thomas wrote:
>> If using just SPECTRE_V2_USER_STRICT then the code in stibp_state() would
>> have to be able differentiate between the case where the mode was switched
>> because of X86_FEATURE_AMD_STIBP_ALWAYS_ON vs a kernel command line with
>> "spectre_v2_user=on". I could always set and use a static variable in the
>> file just for the stibp_state() case.
> 
> Does it matter on X86_FEATURE_AMD_STIBP_ALWAYS_ON CPUs?
> 
> I mean, we want STIBP to be always on there so you can do:
> 
> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
> index 25e914f77bb8..d14860d1cf9c 100644
> --- a/arch/x86/kernel/cpu/bugs.c
> +++ b/arch/x86/kernel/cpu/bugs.c
> @@ -1089,7 +1089,10 @@ static char *stibp_state(void)
>  	case SPECTRE_V2_USER_NONE:
>  		return ", STIBP: disabled";
>  	case SPECTRE_V2_USER_STRICT:
> -		return ", STIBP: forced";
> +		if (boot_cpu_has(X86_FEATURE_AMD_STIBP_ALWAYS_ON))
> +			return ", STIBP: always on";
> +		else
> +			return ", STIBP: forced";
>  	case SPECTRE_V2_USER_PRCTL:
>  	case SPECTRE_V2_USER_SECCOMP:
>  		if (static_key_enabled(&switch_to_cond_stibp))
> 
> so if user has booted with spectre_v2_user=on, we say, "oh well, it is always
> enabled anyway"... or?

I'm ok with that. Having said that, we can probably just leave it as is
and return "forced" even for the ALWAYS_ON case, then.

But if we want to differentiate, though, a simple static bool that is set
when the mode is switched works just as well.

> 
>> I'll give it a bit of time and see if there's any other discussion and
>> re-submit without the new SPECTRE_V2_USER_STRICT_PREFERRED value.
> 
> Sure.

Eh, I'll just submit v2 with the changes now and we can decide between the
above or the bool.

Thanks,
Tom

>