public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "John David Anglin" <dave@hiauly1.hia.nrc.ca>
To: gcc-patches@gcc.gnu.org
Subject: [patch committed] Fix PR target/32847 -- ICE in find_reloads
Date: Tue, 31 Jul 2007 16:44:00 -0000	[thread overview]
Message-ID: <200707311615.l6VGFesY029802@hiauly1.hia.nrc.ca> (raw)

After the dataflow merge, tmpdir-g++.dg-struct-layout-1/t028 cp_compat_y_tst.o
started to fail with an ICE in find_reloads.  Using match_scratch resolves
this problem.

The assembler code should be the same as before except for casesi32p
where the constraint register has been been changed from 'a' to 'r'
(i.e., any general register).  I believe that the 'a' constaint was
a remnant of an attempt to place the jump table in readonly data.  In
any case, this restriction isn't needed for the current assembler
code.

Tested on hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11 and
hppa64-hp-hpux11.11.  Committed to trunk.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

2007-07-31  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>

	PR target/32847
        * pa.md (casesi32): Use match_scratch.  Revise insn condition.
	(casesi32p, casesi64p): Likewise.
	(casesi): Adjust for above.

Index: config/pa/pa.md
===================================================================
--- config/pa/pa.md	(revision 127022)
+++ config/pa/pa.md	(working copy)
@@ -7626,27 +7626,11 @@
   if (TARGET_BIG_SWITCH)
     {
       if (TARGET_64BIT)
-	{
-          rtx tmp1 = gen_reg_rtx (DImode);
-          rtx tmp2 = gen_reg_rtx (DImode);
-
-          emit_jump_insn (gen_casesi64p (operands[0], operands[3],
-                                         tmp1, tmp2));
-	}
+	emit_jump_insn (gen_casesi64p (operands[0], operands[3]));
+      else if (flag_pic)
+	emit_jump_insn (gen_casesi32p (operands[0], operands[3]));
       else
-	{
-	  rtx tmp1 = gen_reg_rtx (SImode);
-
-	  if (flag_pic)
-	    {
-	      rtx tmp2 = gen_reg_rtx (SImode);
-
-	      emit_jump_insn (gen_casesi32p (operands[0], operands[3],
-					     tmp1, tmp2));
-	    }
-	  else
-	    emit_jump_insn (gen_casesi32 (operands[0], operands[3], tmp1));
-	}
+	emit_jump_insn (gen_casesi32 (operands[0], operands[3]));
     }
   else
     emit_jump_insn (gen_casesi0 (operands[0], operands[3]));
@@ -7673,8 +7657,8 @@
 		       (mult:SI (match_operand:SI 0 "register_operand" "r")
 				(const_int 4))
 		       (label_ref (match_operand 1 "" "")))))
-   (clobber (match_operand:SI 2 "register_operand" "=&r"))]
-  "!TARGET_64BIT && TARGET_BIG_SWITCH"
+   (clobber (match_scratch:SI 2 "=&r"))]
+  "!flag_pic"
   "ldil L'%l1,%2\;ldo R'%l1(%2),%2\;{ldwx|ldw},s %0(%2),%2\;bv,n %%r0(%2)"
   [(set_attr "type" "multi")
    (set_attr "length" "16")])
@@ -7685,9 +7669,9 @@
 		       (mult:SI (match_operand:SI 0 "register_operand" "r")
 				(const_int 4))
 		       (label_ref (match_operand 1 "" "")))))
-   (clobber (match_operand:SI 2 "register_operand" "=&a"))
-   (clobber (match_operand:SI 3 "register_operand" "=&r"))]
-  "!TARGET_64BIT && TARGET_BIG_SWITCH"
+   (clobber (match_scratch:SI 2 "=&r"))
+   (clobber (match_scratch:SI 3 "=&r"))]
+  "flag_pic"
   "{bl .+8,%2\;depi 0,31,2,%2|mfia %2}\;ldo {16|20}(%2),%2\;\
 {ldwx|ldw},s %0(%2),%3\;{addl|add,l} %2,%3,%3\;bv,n %%r0(%3)"
   [(set_attr "type" "multi")
@@ -7703,9 +7687,9 @@
 				  (match_operand:SI 0 "register_operand" "r"))
 				(const_int 8))
 		       (label_ref (match_operand 1 "" "")))))
-   (clobber (match_operand:DI 2 "register_operand" "=&r"))
-   (clobber (match_operand:DI 3 "register_operand" "=&r"))]
-  "TARGET_64BIT && TARGET_BIG_SWITCH"
+   (clobber (match_scratch:DI 2 "=&r"))
+   (clobber (match_scratch:DI 3 "=&r"))]
+  ""
   "mfia %2\;ldo 24(%2),%2\;ldw,s %0(%2),%3\;extrd,s %3,63,32,%3\;\
 add,l %2,%3,%3\;bv,n %%r0(%3)"
   [(set_attr "type" "multi")

                 reply	other threads:[~2007-07-31 16:26 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=200707311615.l6VGFesY029802@hiauly1.hia.nrc.ca \
    --to=dave@hiauly1.hia.nrc.ca \
    --cc=gcc-patches@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).