On 11/7/22 14:39, Kim Phillips wrote:
> I've started a version that has AUTOIBRS reuse SPECTRE_V2_EIBRS
> spectre_v2_mitigation enum, but, so far, it's change to bugs.c
> looks bigger: 58 lines changed vs. 34 (see below).
>
> Let me know if you want me to send it as a part of a v2 submission
> after I take care of the kvm CPUID review.
Thanks for putting that together. I generally like how this looks.
I think it probably goes to a _bit_ too much trouble to turn off
"eibrs,lfence/retpoline". If someone goes to the trouble of specifying
those, a warning or pr_info() is probably enough. You don't need to
actively override it.
> - } else if (boot_cpu_has(X86_FEATURE_IBRS) && !spectre_v2_in_ibrs_mode(mode) &&
> - mode != SPECTRE_V2_AUTO_IBRS) {
> + } else if ((boot_cpu_has(X86_FEATURE_IBRS) && !spectre_v2_in_ibrs_mode(mode)) ||
> + (boot_cpu_has(X86_FEATURE_AUTOIBRS) && !spectre_v2_in_ibrs_mode(mode))) {
> setup_force_cpu_cap(X86_FEATURE_USE_IBRS_FW);
> pr_info("Enabling Restricted Speculation for firmware calls\n");
Did the "mode != SPECTRE_V2_AUTO_IBRS" check get dropped accidentally?
Or is it unnecessary now?
