Re: [PATCH, rs6000 8/9] [v2] Enable gimple folding for vec_xl, vec_xst

[Bill Schmidt <wschmidt at linux dot ibm dot com>]

On 6/12/18 11:56 AM, Will Schmidt wrote:
> Hi,
> Gimple folding for unaligned vector loads and stores.
> Regtest completed across variety of systems, P6,P7,P8,P9.
>     
> [v2] Added the type for the MEM_REF, per feedback.
> Testcases for gimple-folding of the same are currently in-tree
> as powerpc/fold-vec-load-*.c and powerpc/fold-vec-store-*.c.
> Re-tested, still looks good. :-)
>     
> Thanks
> -Will
>     
> [gcc]
>     
> 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
>     
> 	* config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
> 	vec_xst variants to the list.
> 	(rs6000_gimple_fold_builtin): Add support for folding unaligned
> 	vector loads and stores.
>
> diff --git a/gcc/config/rs6000/rs6000.c b/gcc/config/rs6000/rs6000.c
> index d62abdf..374666c 100644
> --- a/gcc/config/rs6000/rs6000.c
> +++ b/gcc/config/rs6000/rs6000.c
> @@ -15360,10 +15360,16 @@ rs6000_builtin_valid_without_lhs (enum rs6000_builtins fn_code)
>      case ALTIVEC_BUILTIN_STVX_V8HI:
>      case ALTIVEC_BUILTIN_STVX_V4SI:
>      case ALTIVEC_BUILTIN_STVX_V4SF:
>      case ALTIVEC_BUILTIN_STVX_V2DI:
>      case ALTIVEC_BUILTIN_STVX_V2DF:
> +    case VSX_BUILTIN_STXVW4X_V16QI:
> +    case VSX_BUILTIN_STXVW4X_V8HI:
> +    case VSX_BUILTIN_STXVW4X_V4SF:
> +    case VSX_BUILTIN_STXVW4X_V4SI:
> +    case VSX_BUILTIN_STXVD2X_V2DF:
> +    case VSX_BUILTIN_STXVD2X_V2DI:
>        return true;
>      default:
>        return false;
>      }
>  }
> @@ -15869,10 +15875,78 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
>  	gimple_set_location (g, loc);
>  	gsi_replace (gsi, g, true);
>  	return true;
>        }
>
> +    /* unaligned Vector loads.  */
> +    case VSX_BUILTIN_LXVW4X_V16QI:
> +    case VSX_BUILTIN_LXVW4X_V8HI:
> +    case VSX_BUILTIN_LXVW4X_V4SF:
> +    case VSX_BUILTIN_LXVW4X_V4SI:
> +    case VSX_BUILTIN_LXVD2X_V2DF:
> +    case VSX_BUILTIN_LXVD2X_V2DI:
> +      {
> +	 arg0 = gimple_call_arg (stmt, 0);  // offset
> +	 arg1 = gimple_call_arg (stmt, 1);  // address
> +	 lhs = gimple_call_lhs (stmt);
> +	 location_t loc = gimple_location (stmt);
> +	 /* Since arg1 may be cast to a different type, just use ptr_type_node
> +	    here instead of trying to enforce TBAA on pointer types.  */
> +	 tree arg1_type = ptr_type_node;
> +	 tree lhs_type = TREE_TYPE (lhs);
> +	 /* POINTER_PLUS_EXPR wants the offset to be of type 'sizetype'.  Create
> +	    the tree using the value from arg0.  The resulting type will match
> +	    the type of arg1.  */
> +	 gimple_seq stmts = NULL;
> +	 tree temp_offset = gimple_convert (&stmts, loc, sizetype, arg0);
> +	 tree temp_addr = gimple_build (&stmts, loc, POINTER_PLUS_EXPR,
> +				       arg1_type, arg1, temp_offset);
> +	 gsi_insert_seq_before (gsi, stmts, GSI_SAME_STMT);
> +	 /* Use the build2 helper to set up the mem_ref.  The MEM_REF could also
> +	    take an offset, but since we've already incorporated the offset
> +	    above, here we just pass in a zero.  */
> +	 gimple *g;
> +	 tree align_ltype = build_aligned_type (lhs_type, TYPE_ALIGN ((lhs_type)));

This alignment is too strong (assuming lhs is a vector type).  lxvd2x and lxvw4x only
require word alignment, so

	tree align_ltype = build_aligned_type (lhs_type, 4);

seems like what you need.  See my earlier discussion with Richard.  Sorry I didn't
notice this before!

> +	 g = gimple_build_assign (lhs, build2 (MEM_REF, align_ltype, temp_addr,
> +						build_int_cst (arg1_type, 0)));
> +	 gimple_set_location (g, loc);
> +	 gsi_replace (gsi, g, true);
> +	 return true;
> +      }
> +
> +    /* unaligned Vector stores.  */
> +    case VSX_BUILTIN_STXVW4X_V16QI:
> +    case VSX_BUILTIN_STXVW4X_V8HI:
> +    case VSX_BUILTIN_STXVW4X_V4SF:
> +    case VSX_BUILTIN_STXVW4X_V4SI:
> +    case VSX_BUILTIN_STXVD2X_V2DF:
> +    case VSX_BUILTIN_STXVD2X_V2DI:
> +      {
> +	 arg0 = gimple_call_arg (stmt, 0); /* Value to be stored.  */
> +	 arg1 = gimple_call_arg (stmt, 1); /* Offset.  */
> +	 tree arg2 = gimple_call_arg (stmt, 2); /* Store-to address.  */
> +	 location_t loc = gimple_location (stmt);
> +	 tree arg0_type = TREE_TYPE (arg0);
> +	 /* Use ptr_type_node (no TBAA) for the arg2_type.  */
> +	 tree arg2_type = ptr_type_node;
> +	 /* POINTER_PLUS_EXPR wants the offset to be of type 'sizetype'.  Create
> +	    the tree using the value from arg0.  The resulting type will match
> +	    the type of arg2.  */
> +	 gimple_seq stmts = NULL;
> +	 tree temp_offset = gimple_convert (&stmts, loc, sizetype, arg1);
> +	 tree temp_addr = gimple_build (&stmts, loc, POINTER_PLUS_EXPR,
> +				       arg2_type, arg2, temp_offset);
> +	 /* Mask off any lower bits from the address.  */

Spurious comment?  You're not doing that, nor do you want to...

Thanks,
Bill

> +	 gsi_insert_seq_before (gsi, stmts, GSI_SAME_STMT);
> +	 gimple *g;
> +	 g = gimple_build_assign (build2 (MEM_REF, arg0_type, temp_addr,
> +					   build_int_cst (arg2_type, 0)), arg0);
> +	 gimple_set_location (g, loc);
> +	 gsi_replace (gsi, g, true);
> +	 return true;
> +      }
> +
>      /* Vector Fused multiply-add (fma).  */
>      case ALTIVEC_BUILTIN_VMADDFP:
>      case VSX_BUILTIN_XVMADDDP:
>      case ALTIVEC_BUILTIN_VMLADDUHM:
>        {
>
>