public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
From: Max Filippov <jcmvbkbc@gcc.gnu.org>
To: gcc-cvs@gcc.gnu.org
Subject: [gcc r13-5333] xtensa: Revise complex hard register clobber elimination
Date: Tue, 24 Jan 2023 21:23:24 +0000 (GMT)	[thread overview]
Message-ID: <20230124212324.E8AE23858D1E@sourceware.org> (raw)

https://gcc.gnu.org/g:1c407dc088231ba5f2cc63d9278f4b797db48de1

commit r13-5333-g1c407dc088231ba5f2cc63d9278f4b797db48de1
Author: Takayuki 'January June' Suwa <jjsuwa_sys3175@yahoo.co.jp>
Date:   Fri Jan 20 12:33:37 2023 +0900

    xtensa: Revise complex hard register clobber elimination
    
    In the previously posted patch
    "xtensa: Make complex hard register clobber elimination more robust and accurate",
    the check code for insns that refer to the [DS]Cmode hard register before
    it is overwritten after it is clobbered is incomplete.  Fortunately such
    insns are seldom emitted, so it didn't matter.
    
    This patch fixes that for the sake of completeness.
    
    gcc/ChangeLog:
    
            * config/xtensa/xtensa.md:
            Fix exit from loops detecting references before overwriting in the
            split pattern.

Diff:
---
 gcc/config/xtensa/xtensa.md | 72 +++++++++++++++++++++++----------------------
 1 file changed, 37 insertions(+), 35 deletions(-)

diff --git a/gcc/config/xtensa/xtensa.md b/gcc/config/xtensa/xtensa.md
index dd3fc37353b..d3996b26cb5 100644
--- a/gcc/config/xtensa/xtensa.md
+++ b/gcc/config/xtensa/xtensa.md
@@ -2973,45 +2973,47 @@
 {
   auto_sbitmap bmp (FIRST_PSEUDO_REGISTER);
   rtx_insn *insn;
-  rtx reg = gen_rtx_REG (SImode, 0);
+  rtx reg = gen_rtx_REG (SImode, 0), dest;
+  unsigned int regno;
+  sbitmap_iterator iter;
   bitmap_set_range (bmp, REGNO (operands[0]), REG_NREGS (operands[0]));
   for (insn = next_nonnote_nondebug_insn_bb (curr_insn);
        insn; insn = next_nonnote_nondebug_insn_bb (insn))
-    {
-      sbitmap_iterator iter;
-      unsigned int regno;
-      if (NONJUMP_INSN_P (insn))
-	{
-	  EXECUTE_IF_SET_IN_BITMAP (bmp, 2, regno, iter)
-	    {
-	      set_regno_raw (reg, regno, REG_NREGS (reg));
-	      if (reg_overlap_mentioned_p (reg, PATTERN (insn)))
-		break;
-	    }
-	  if (GET_CODE (PATTERN (insn)) == SET)
-	    {
-	      rtx x = SET_DEST (PATTERN (insn));
-	      if (REG_P (x) && HARD_REGISTER_P (x))
-		bitmap_clear_range (bmp, REGNO (x), REG_NREGS (x));
-	      else if (SUBREG_P (x) && HARD_REGISTER_P (SUBREG_REG (x)))
-		{
-		  struct subreg_info info;
-		  subreg_get_info (regno = REGNO (SUBREG_REG (x)),
-				   GET_MODE (SUBREG_REG (x)),
-				   SUBREG_BYTE (x), GET_MODE (x), &info);
-		  if (!info.representable_p)
-		    break;
-		  bitmap_clear_range (bmp, regno + info.offset, info.nregs);
-		}
-	    }
-	  if (bitmap_empty_p (bmp))
-	    goto FALLTHRU;
-	}
-      else if (CALL_P (insn))
+    if (NONJUMP_INSN_P (insn))
+      {
 	EXECUTE_IF_SET_IN_BITMAP (bmp, 2, regno, iter)
-	 if (call_used_or_fixed_reg_p (regno))
-	   break;
-    }
+	  {
+	    set_regno_raw (reg, regno, REG_NREGS (reg));
+	    if (reg_referenced_p (reg, PATTERN (insn)))
+	      goto ABORT;
+	  }
+	if (GET_CODE (PATTERN (insn)) == SET
+	    || GET_CODE (PATTERN (insn)) == CLOBBER)
+	  {
+	    dest = SET_DEST (PATTERN (insn));
+	    if (REG_P (dest) && HARD_REGISTER_P (dest))
+	      bitmap_clear_range (bmp, REGNO (dest), REG_NREGS (dest));
+	    else if (SUBREG_P (dest)
+		     && HARD_REGISTER_P (SUBREG_REG (dest)))
+	      {
+		struct subreg_info info;
+		subreg_get_info (regno = REGNO (SUBREG_REG (dest)),
+				 GET_MODE (SUBREG_REG (dest)),
+				 SUBREG_BYTE (dest), GET_MODE (dest),
+				 &info);
+		if (!info.representable_p)
+		  break;
+		bitmap_clear_range (bmp, regno + info.offset, info.nregs);
+	      }
+	  }
+	if (bitmap_empty_p (bmp))
+	  goto FALLTHRU;
+      }
+    else if (CALL_P (insn))
+      EXECUTE_IF_SET_IN_BITMAP (bmp, 2, regno, iter)
+	if (call_used_or_fixed_reg_p (regno))
+	  goto ABORT;
+ABORT:
   FAIL;
 FALLTHRU:;
 })

                 reply	other threads:[~2023-01-24 21:23 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230124212324.E8AE23858D1E@sourceware.org \
    --to=jcmvbkbc@gcc.gnu.org \
    --cc=gcc-cvs@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).