Re: [PATCH v15 5/8] arm64: mte: Enable TCO in functions that can read beyond buffer limits

[Vincenzo Frascino]

On 3/12/21 3:13 PM, Catalin Marinas wrote:
> On Fri, Mar 12, 2021 at 02:22:07PM +0000, Vincenzo Frascino wrote:
>> diff --git a/arch/arm64/include/asm/mte.h b/arch/arm64/include/asm/mte.h
>> index 9b557a457f24..8603c6636a7d 100644
>> --- a/arch/arm64/include/asm/mte.h
>> +++ b/arch/arm64/include/asm/mte.h
>> @@ -90,5 +90,20 @@ static inline void mte_assign_mem_tag_range(void *addr, size_t size)
>>  
>>  #endif /* CONFIG_ARM64_MTE */
>>  
>> +#ifdef CONFIG_KASAN_HW_TAGS
>> +/* Whether the MTE asynchronous mode is enabled. */
>> +DECLARE_STATIC_KEY_FALSE(mte_async_mode);
>> +
>> +static inline bool system_uses_mte_async_mode(void)
>> +{
>> +	return static_branch_unlikely(&mte_async_mode);
>> +}
>> +#else
>> +static inline bool system_uses_mte_async_mode(void)
>> +{
>> +	return false;
>> +}
>> +#endif /* CONFIG_KASAN_HW_TAGS */
> 
> You can write this with fewer lines:
> 
> DECLARE_STATIC_KEY_FALSE(mte_async_mode);
> 
> static inline bool system_uses_mte_async_mode(void)
> {
> 	return IS_ENABLED(CONFIG_KASAN_HW_TAGS) &&
> 		static_branch_unlikely(&mte_async_mode);
> }
> 
> The compiler will ensure that mte_async_mode is not referred when
> !CONFIG_KASAN_HW_TAGS and therefore doesn't need to be defined.
>

Yes, I agree, but I introduce "#ifdef CONFIG_KASAN_HW_TAGS" in the successive
patch anyway, according to me the overall code looks more uniform like this. But
I do not have a strong opinion or preference on this.

>> diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
>> index fa755cf94e01..9362928ba0d5 100644
>> --- a/arch/arm64/kernel/mte.c
>> +++ b/arch/arm64/kernel/mte.c
>> @@ -26,6 +26,10 @@ u64 gcr_kernel_excl __ro_after_init;
>>  
>>  static bool report_fault_once = true;
>>  
>> +/* Whether the MTE asynchronous mode is enabled. */
>> +DEFINE_STATIC_KEY_FALSE(mte_async_mode);
>> +EXPORT_SYMBOL_GPL(mte_async_mode);
> 
> Maybe keep these bracketed by #ifdef CONFIG_KASAN_HW_TAGS. I think the
> mte_enable_kernel_*() aren't needed either if KASAN_HW is disabled (you
> can do it with an additional patch).
>

Makes sense, I will add it in the next version.

> With these, you can add:
> 
> Reviewed-by: Catalin Marinas <catalin.marinas@xxxxxxx>
> 

-- 
Regards,
Vincenzo