Re: [PATCH v5 08/10] arm64: Always enable ssb vulnerability detection

[Jeremy Linton <jeremy.linton@arm.com>]

On 3/1/19 1:02 AM, Andre Przywara wrote:
> Hi,
> 
> On 2/26/19 7:05 PM, Jeremy Linton wrote:
>> The ssb detection logic is necessary regardless of whether
>> the vulnerability mitigation code is built into the kernel.
>> Break it out so that the CONFIG option only controls the
>> mitigation logic and not the vulnerability detection.
>>
>> Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
>> ---
>>   arch/arm64/include/asm/cpufeature.h |  4 ----
>>   arch/arm64/kernel/cpu_errata.c      | 11 +++++++----
>>   2 files changed, 7 insertions(+), 8 deletions(-)
>>
>> diff --git a/arch/arm64/include/asm/cpufeature.h 
>> b/arch/arm64/include/asm/cpufeature.h
>> index dfcfba725d72..c2b60a021437 100644
>> --- a/arch/arm64/include/asm/cpufeature.h
>> +++ b/arch/arm64/include/asm/cpufeature.h
>> @@ -628,11 +628,7 @@ static inline int arm64_get_ssbd_state(void)
>>   #endif
>>   }
>> -#ifdef CONFIG_ARM64_SSBD
>>   void arm64_set_ssbd_mitigation(bool state);
>> -#else
>> -static inline void arm64_set_ssbd_mitigation(bool state) {}
>> -#endif
>>   extern int do_emulate_mrs(struct pt_regs *regs, u32 sys_reg, u32 rt);
>> diff --git a/arch/arm64/kernel/cpu_errata.c 
>> b/arch/arm64/kernel/cpu_errata.c
>> index 0f6e8f5d67bc..5f5611d17dc1 100644
>> --- a/arch/arm64/kernel/cpu_errata.c
>> +++ b/arch/arm64/kernel/cpu_errata.c
>> @@ -276,7 +276,6 @@ static int detect_harden_bp_fw(void)
>>       return 1;
>>   }
>> -#ifdef CONFIG_ARM64_SSBD
>>   DEFINE_PER_CPU_READ_MOSTLY(u64, arm64_ssbd_callback_required);
>>   int ssbd_state __read_mostly = ARM64_SSBD_KERNEL;
>> @@ -347,6 +346,7 @@ void __init arm64_enable_wa2_handling(struct 
>> alt_instr *alt,
>>           *updptr = cpu_to_le32(aarch64_insn_gen_nop());
>>   }
>> +#ifdef CONFIG_ARM64_SSBD
>>   void arm64_set_ssbd_mitigation(bool state)
>>   {
>>       if (this_cpu_has_cap(ARM64_SSBS)) {
>> @@ -371,6 +371,12 @@ void arm64_set_ssbd_mitigation(bool state)
>>           break;
>>       }
>>   }
>> +#else
>> +void arm64_set_ssbd_mitigation(bool state)
>> +{
>> +    pr_info_once("SSBD, disabled by kernel configuration\n");
> 
> Is there a stray comma or is the continuation of some previous printout?

This is on purpose because I didn't like the way it read if you expanded 
the acronym. I still don't, maybe a ":" is more appropriate.


> 
> Regardless of that it looks good and compiles with both 
> CONFIG_ARM64_SSBD defined or not:
> 
> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
> 
> Cheers,
> Andre.
> 
>> +}
>> +#endif    /* CONFIG_ARM64_SSBD */
>>   static bool has_ssbd_mitigation(const struct arm64_cpu_capabilities 
>> *entry,
>>                       int scope)
>> @@ -468,7 +474,6 @@ static bool has_ssbd_mitigation(const struct 
>> arm64_cpu_capabilities *entry,
>>       return required;
>>   }
>> -#endif    /* CONFIG_ARM64_SSBD */
>>   static void __maybe_unused
>>   cpu_enable_cache_maint_trap(const struct arm64_cpu_capabilities 
>> *__unused)
>> @@ -760,14 +765,12 @@ const struct arm64_cpu_capabilities 
>> arm64_errata[] = {
>>           ERRATA_MIDR_RANGE_LIST(arm64_harden_el2_vectors),
>>       },
>>   #endif
>> -#ifdef CONFIG_ARM64_SSBD
>>       {
>>           .desc = "Speculative Store Bypass Disable",
>>           .capability = ARM64_SSBD,
>>           .type = ARM64_CPUCAP_LOCAL_CPU_ERRATUM,
>>           .matches = has_ssbd_mitigation,
>>       },
>> -#endif
>>   #ifdef CONFIG_ARM64_ERRATUM_1188873
>>       {
>>           /* Cortex-A76 r0p0 to r2p0 */
>>