On 7/9/2025 9:58 AM, Dave Hansen wrote:

>> +	 * Avoid using memcpy() here. Instead, open code it.
>> +	 */
>> +	asm volatile("rep movsb"
>> +		     : "+D" (dst), "+S" (src), "+c" (len) : : "memory");
>> +
>> +	lass_clac();
>>  }
> 
> This didn't turn out great. At the _very_ least, we could have a:
> 
> 	inline_memcpy_i_really_mean_it()
> 

It looks like we should go back to __inline_memcpy()/_memset()
implementation that PeterZ had initially proposed. It seems to fit all
the requirements, right? Patch attached.

https://lore.kernel.org/lkml/20241028160917.1380714-3-alexander.shishkin@linux.intel.com/

> with the rep mov. Or even a #define if we were super paranoid the
> compiler is out to get us.
> 
> But _actually_ open-coding inline assembly is far too ugly to live.
>