On Tue, Jan 22, 2019 at 08:54:33AM +0000, Zhang, Lei wrote:
> diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
> index efb7b2c..37e4f18 100644
> --- a/arch/arm64/mm/fault.c
> +++ b/arch/arm64/mm/fault.c
> @@ -666,6 +666,28 @@ static int do_sea(unsigned long addr, unsigned int esr,
> struct pt_regs *regs)
> return 0;
> }
>
> +static int do_bad_unknown_63(unsigned long addr, unsigned int esr, struct
> pt_regs *regs)
> +{
> + /*
> + * On some variants of the Fujitsu-A64FX cores ver(1.0, 1.1),
> + * memory accesses may spuriously trigger data aborts with
> + * DFSC=0b111111.
> + */
> + if (IS_ENABLED(CONFIG_FUJITSU_ERRATUM_010001)) {
> + if (cpus_have_cap(ARM64_WORKAROUND_FUJITSU_A64FX_0100001)) {
> + return 0;
> + } else { /* cpu capabilities maybe not ready*/
> + unsigned int current_cpu_midr = read_cpuid_id();
> + const struct midr_range fujitsu_a64fx_midr_range = {
> + MIDR_FUJITSU_A64FX, MIDR_CPU_VAR_REV(0, 0),
> MIDR_CPU_VAR_REV(1, 0)
> + };
> + if (is_midr_in_range(current_cpu_midr,
> &fujitsu_a64fx_midr_range) == TRUE)
> + return 0;
> + }
> + }
> + return do_bad(addr, esr, regs);
> +}
IIUC, this can happen very early when the errata framework isn't yet
ready. Given that this is not on a fast path (you already took a fault),
I don't think it's worth optimising for cpus_have_cap() (and
ARM64_WORKAROUND_FUJITSU_A64FX_0100001). I've seen Mark's comments on
why checking MIDR in a preemptible context is not a good idea but I
suspect your platform is homogeneous (i.e. not big.LITTLE).
--
Catalin
