public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r13-3686] Do not use subword paradoxical subregs in emit_group_store
@ 2022-11-04 14:15 Eric Botcazou
  0 siblings, 0 replies; only message in thread
From: Eric Botcazou @ 2022-11-04 14:15 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:3e2bdf2460a34a2389dee813a2ba8ecf976f2ec9

commit r13-3686-g3e2bdf2460a34a2389dee813a2ba8ecf976f2ec9
Author: Eric Botcazou <ebotcazou@adacore.com>
Date:   Fri Nov 4 11:15:57 2022 +0100

    Do not use subword paradoxical subregs in emit_group_store
    
    The goal of the trick is to make life easier for the combiner, but subword
    paradoxical subregs make it harder for the register allocator instead.
    
    gcc/
            * expr.cc (emit_group_store): Do not use subword paradoxical subregs

Diff:
---
 gcc/expr.cc | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/gcc/expr.cc b/gcc/expr.cc
index 9145193c2c1..c6917fbf7bd 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -2901,19 +2901,19 @@ emit_group_store (rtx orig_dst, rtx src, tree type ATTRIBUTE_UNUSED,
       if (!REG_P (dst) || REGNO (dst) < FIRST_PSEUDO_REGISTER)
 	dst = gen_reg_rtx (outer);
 
-      /* Make life a bit easier for combine.  */
-      /* If the first element of the vector is the low part
-	 of the destination mode, use a paradoxical subreg to
-	 initialize the destination.  */
+      /* Make life a bit easier for combine: if the first element of the
+	 vector is the word (or larger) low part of the destination mode,
+	 use a paradoxical subreg to initialize the destination.  */
       if (start < finish)
 	{
 	  inner = GET_MODE (tmps[start]);
 	  bytepos = subreg_lowpart_offset (inner, outer);
-	  if (known_eq (rtx_to_poly_int64 (XEXP (XVECEXP (src, 0, start), 1)),
-			bytepos))
+	  if (known_ge (GET_MODE_BITSIZE (inner), BITS_PER_WORD)
+	      && known_eq (rtx_to_poly_int64 (XEXP (XVECEXP (src, 0,
+							     start), 1)),
+			   bytepos))
 	    {
-	      temp = simplify_gen_subreg (outer, tmps[start],
-					  inner, 0);
+	      temp = simplify_gen_subreg (outer, tmps[start], inner, 0);
 	      if (temp)
 		{
 		  emit_move_insn (dst, temp);
@@ -2929,12 +2929,12 @@ emit_group_store (rtx orig_dst, rtx src, tree type ATTRIBUTE_UNUSED,
 	{
 	  inner = GET_MODE (tmps[finish - 1]);
 	  bytepos = subreg_lowpart_offset (inner, outer);
-	  if (known_eq (rtx_to_poly_int64 (XEXP (XVECEXP (src, 0,
-							  finish - 1), 1)),
-			bytepos))
+	  if (known_ge (GET_MODE_BITSIZE (inner), BITS_PER_WORD)
+	      && known_eq (rtx_to_poly_int64 (XEXP (XVECEXP (src, 0,
+							     finish - 1), 1)),
+			   bytepos))
 	    {
-	      temp = simplify_gen_subreg (outer, tmps[finish - 1],
-					  inner, 0);
+	      temp = simplify_gen_subreg (outer, tmps[finish - 1], inner, 0);
 	      if (temp)
 		{
 		  emit_move_insn (dst, temp);

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2022-11-04 14:15 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-04 14:15 [gcc r13-3686] Do not use subword paradoxical subregs in emit_group_store Eric Botcazou

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).