From: Robin Dapp <rdapp@ventanamicro.com>
Date: Mon, 26 Jun 2023 11:30:26 +0000 (+0200)
Subject: match.pd: Use element_mode instead of TYPE_MODE.
X-Git-Tag: basepoints/gcc-15~8004
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=d915762ea9043da858d388b60b2d8093ff77eeab;p=thirdparty%2Fgcc.git

match.pd: Use element_mode instead of TYPE_MODE.

This patch changes TYPE_MODE into element_mode in a match.pd
simplification.  As the simplification can be also called with vector
types real_can_shorten_arithmetic would ICE in REAL_MODE_FORMAT which
expects a scalar mode.  Therefore, use element_mode instead of
TYPE_MODE.

Additionally, check if the target supports the resulting operation.  One
target that supports e.g. a float addition but not a _Float16 addition
is the RISC-V vector extension Zvfhmin.

gcc/ChangeLog:

	* match.pd: Use element_mode and check if target supports
	operation with new type.
---

diff --git a/gcc/match.pd b/gcc/match.pd
index 33ccda3e7b6b..83bcefa914bf 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -7454,10 +7454,12 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
 		  values representable in the TYPE to be within the
 		  range of normal values of ITYPE.  */
 	      (if (element_precision (newtype) < element_precision (itype)
+		   && (!VECTOR_MODE_P (TYPE_MODE (newtype))
+		       || target_supports_op_p (newtype, op, optab_default))
 		   && (flag_unsafe_math_optimizations
 		       || (element_precision (newtype) == element_precision (type)
-			   && real_can_shorten_arithmetic (TYPE_MODE (itype),
-							   TYPE_MODE (type))
+			   && real_can_shorten_arithmetic (element_mode (itype),
+							   element_mode (type))
 			   && !excess_precision_type (newtype)))
 		   && !types_match (itype, newtype))
 		 (convert:type (op (convert:newtype @1)