Re: [RFC] x86/mce: Add workaround for SKX/CLX/CPX spurious machine checks

[Luck, Tony]

On Sun, Feb 06, 2022 at 08:36:40PM -0800, Jue Wang wrote:
> +static bool quirk_skylake_repmov(void)
> +{
> +	/*
> +	 * State that represents if an SRAR MCE has already signaled on the DCU bank.
> +	 */
> +	static DEFINE_PER_CPU(bool, srar_dcu_signaled);
> +
> +	if (unlikely(!__this_cpu_read(srar_dcu_signaled))) {
> +		u64 mc1_status = mce_rdmsrl(MSR_IA32_MCx_STATUS(1));

Jue,

When I reviewed this for you off-list, I didn't notice that you
dropped the test for mcgstatus & MCG_STATUS_LMCES as part of
moving to a helper function and expanding the test for more
bits in mc1_status.

I think that test still is still important ... knowing that this is
a *local* machine check before making decision based on just what this
CPU observes makes this a bit more robust.

> +
> +		if (is_intel_srar(mc1_status)) {
> +			__this_cpu_write(srar_dcu_signaled, true);
> +			msr_clear_bit(MSR_IA32_MISC_ENABLE,
> +				      MSR_IA32_MISC_ENABLE_FAST_STRING_BIT);
> +			mce_wrmsrl(MSR_IA32_MCG_STATUS, 0);
> +			mce_wrmsrl(MSR_IA32_MCx_STATUS(1), 0);
> +			pr_err("First SRAR MCE on DCU, CPU: %d, disable fast string copy.\n",
> +			       smp_processor_id());
> +			return true;
> +		}
> +	}
> +	return false;
> +}

-Tony