[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87h80wxsze.fsf@nanos.tec.linutronix.de>
Date: Wed, 15 Jan 2020 23:19:49 +0100
From: Thomas Gleixner <tglx@...utronix.de>
To: Tony W Wang-oc <TonyWWang-oc@...oxin.com>, mingo@...hat.com,
bp@...en8.de, hpa@...or.com, x86@...nel.org,
linux-kernel@...r.kernel.org
Cc: DavidWang@...oxin.com, CooperYan@...oxin.com,
QiyuanWang@...oxin.com, HerryYang@...oxin.com
Subject: Re: [PATCH] x86/cpu: clear X86_BUG_SPECTRE_V2 on Zhaoxin family 7 CPUs
Tony W Wang-oc <TonyWWang-oc@...oxin.com> writes:
> These CPUs are not affected by spectre_v2, so clear spectre_v2 bug flag
> in their specific initialization code.
>
> if (cpu_has(c, X86_FEATURE_VMX))
> centaur_detect_vmx_virtcap(c);
> +
> + if (c->x86 == 7) {
> + setup_clear_cpu_cap(X86_BUG_SPECTRE_V2);
> + clear_bit(X86_BUG_SPECTRE_V2, (unsigned long *)cpu_caps_set);
No. Please use cpu_vuln_whitelist. It exists for exactly this
purpose. You just need to extend it with a NO_SPECTRE_V2 bit.
Thanks,
tglx
Powered by blists - more mailing lists