Messages in this thread | | | From | Reiji Watanabe <> | Date | Wed, 28 Apr 2021 22:38:39 -0700 | Subject | Re: [v3 1/1] x86/cpufeatures: Implement Predictive Store Forwarding control. |
| |
> + if (!strcmp(str, "off")) { > + set_cpu_cap(&boot_cpu_data, X86_FEATURE_MSR_SPEC_CTRL); > + x86_spec_ctrl_base |= SPEC_CTRL_PSFD; > + wrmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base);
My previous suggestion about updating MSR_IA32_SPEC_CTRL meant something like:
rdmsrl(MSR_IA32_SPEC_CTRL, current); wrmsrl(MSR_IA32_SPEC_CTRL, current | SPEC_CTRL_PSFD);
And this is to keep the behavior of code below check_bugs(). (Or do you intentionally change it due to some reason ?) BTW, x86_spec_ctrl_base, which is updated in psf_cmdline(), would be overwritten by check_bugs() anyway as follows. --- void __init check_bugs(void) { <...> /* * Read the SPEC_CTRL MSR to account for reserved bits which may * have unknown values. AMD64_LS_CFG MSR is cached in the early AMD * init code as it is not enumerated and depends on the family. */ if (boot_cpu_has(X86_FEATURE_MSR_SPEC_CTRL)) rdmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base); <...> ---
> + setup_clear_cpu_cap(X86_FEATURE_PSFD);
Does X86_FEATURE_PSFD need to be cleared for the 'off' case ? Do you want to remove "psfd" from /proc/cpuinfo when PSFD is enabled ? (not when PSFD is disabled ?)
Thanks, Reiji
| |