Re: [PATCH] powerpc64: Add optimized strcpy and stpcpy for POWER10

[Paul E Murphy <murphyp@linux.ibm.com>]

On 6/16/24 6:49 AM, bmahi496@linux.ibm.com wrote:

> +/* Implements the function
> +
> +   char * [r3] strcpy (char *dest [r3], const char *src [r4])
> +
> +   or
> +
> +   char * [r3] stpcpy (char *dest [r3], const char *src [r4])
> +
> +   if USE_AS_STPCPY is defined.
> +
> +   The implementation can load bytes past a NULL terminator, but only
> +   up to the next 16B/64B boundary, so it never crosses a page.  */

I think this comment is confusing.  Maybe it suffices to say that it 
never reads across a page boundary, but may read beyond the NUL terminator.


> +
> +
> +#define LXVP(xtp,dq,ra)	    \
> +	.long(((6)<<(32-6))		\
> +	      | ((((xtp)-32)>>1)<<(32-10)) \
> +	      | ((1)<<(32-11))	     \
> +	      | ((ra)<<(32-16))	    \
> +	      | dq)
> +
> +/* Load 4 quadwords, merge into one VR for speed and check for NULLs
> +   and branch to label if NULL is found.  */
> +#define CHECK_64B(offset,addr,label)	  \
> +	LXVP(v4+32,offset,addr);	    \
> +	LXVP(v6+32,offset+32,addr);	 \

Have you benchmarked using lxv vs lxvp here?  If the performance 
difference is marginal, I think this implementation could be used for P9 
too.


> +	vminub    v14,v4,v5;		\
> +	vminub    v15,v6,v7;		\
> +	vminub    v16,v14,v15;	      \
> +	vcmpequb. v0,v16,v18;	       \
> +	beq     cr6,$+12;		   \
> +	li      r7,offset;		  \
> +	b       L(label);		   \
> +	stxv    32+v5,(offset+0)(r11);      \
> +	stxv    32+v4,(offset+16)(r11);     \
> +	stxv    32+v7,(offset+32)(r11);     \
> +	stxv    32+v6,(offset+48)(r11)

Otherwise, this LGTM with trivial comments addressed.