On 10/17/2018 01:59 PM, Tim Chen wrote:
> With enhanced IBRS in use, the application running on sibling CPU will not
> be able to launch Spectre v2 attack to the application on current CPU.
> There is no need to use STIBP for this case. Disable the STIBP code
> when enhanced IBRS is used.
>
> Signed-off-by: Tim Chen <tim.c.c...@linux.intel.com>
> ---
> arch/x86/kernel/cpu/bugs.c | 12 +++++++++++-
> 1 file changed, 11 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
> index 2fc7b4e..6ed82ea 100644
> --- a/arch/x86/kernel/cpu/bugs.c
> +++ b/arch/x86/kernel/cpu/bugs.c
> @@ -327,6 +327,14 @@ static enum spectre_v2_mitigation_cmd __init
> spectre_v2_parse_cmdline(void)
>
> static bool stibp_needed(void)
> {
> + /*
> + * Determine if we want to leave STIBP always on.
> + * Using enhanced IBRS makes using STIBP unnecessary.
> + */
> +
> + if (static_branch_unlikely(&spectre_v2_enhanced_ibrs))
> + return false;
> +
> if (spectre_v2_enabled == SPECTRE_V2_NONE)
> return false;
>
> @@ -881,7 +889,9 @@ static ssize_t cpu_show_common(struct device *dev, struct
> device_attribute *attr
> ", IBPB" : "",
> boot_cpu_has(X86_FEATURE_USE_IBRS_FW) ?
> ", IBRS_FW" : "",
> - (x86_spec_ctrl_base & SPEC_CTRL_STIBP) ?
> + spectre_v2_enabled == SPECTRE_V2_IBRS_ENHANCED ?
> + ", Enhanced IBRS" :
> + (x86_spec_ctrl_base & SPEC_CTRL_STIBP) ?
> ", STIBP" : "",
> boot_cpu_has(X86_FEATURE_RSB_CTXSW) ?
> ", RSB filling" : "",
The "Enhanced IBRS" is one of the states of spectre_v2_enabled. So you
don't need to print that out one more time.
Cheers,
Longman
