public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] i386: Fix up ix86_expand_vector_init_general [PR105123]
@ 2022-04-02  9:25 Jakub Jelinek
  2022-04-03 19:31 ` Uros Bizjak
  0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2022-04-02  9:25 UTC (permalink / raw)
  To: Uros Bizjak; +Cc: gcc-patches

Hi!

The following testcase is miscompiled on ia32.
The problem is that at -O0 we end up with:
  vector(4) short unsigned int _1;
  short unsigned int u.0_3;
...
  _1 = {u.0_3, u.0_3, u.0_3, u.0_3};
statement (dead) which is wrongly expanded.
elt is (subreg:HI (reg:SI 83 [ u.0_3 ]) 0), tmp_mode SImode,
so after convert_mode we start with word (reg:SI 83 [ u.0_3 ]).
The intent is to manually broadcast that value to 2 SImode parts,
but because we pass word as target to expand_simple_binop, it will
overwrite (reg:SI 83 [ u.0_3 ]) and we end up with 0:
   10: {r83:SI=r83:SI<<0x10;clobber flags:CC;}
   11: {r83:SI=r83:SI|r83:SI;clobber flags:CC;}
   12: {r83:SI=r83:SI<<0x10;clobber flags:CC;}
   13: {r83:SI=r83:SI|r83:SI;clobber flags:CC;}
   14: clobber r110:V4HI
   15: r110:V4HI#0=r83:SI
   16: r110:V4HI#4=r83:SI
as the two ors do nothing and two shifts each by 16 left shift it all
away.
The following patch fixes that by using NULL_RTX target, so we expand it as
   10: {r110:SI=r83:SI<<0x10;clobber flags:CC;}
   11: {r111:SI=r110:SI|r83:SI;clobber flags:CC;}
   12: {r112:SI=r83:SI<<0x10;clobber flags:CC;}
   13: {r113:SI=r112:SI|r83:SI;clobber flags:CC;}
   14: clobber r114:V4HI
   15: r114:V4HI#0=r111:SI
   16: r114:V4HI#4=r113:SI
instead.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

Another possibility would be to pass NULL_RTX only when word == elt
and word otherwise, where word would necessarily be a pseudo from the first
shift after passing NULL_RTX there once or pass NULL_RTX for the shift and
word for ior.

2022-04-02  Jakub Jelinek  <jakub@redhat.com>

	PR target/105123
	* config/i386/i386-expand.cc (ix86_expand_vector_init_general): Avoid
	using word as target for expand_simple_binop when doing ASHIFT and
	IOR.

	* gcc.target/i386/pr105123.c: New test.

--- gcc/config/i386/i386-expand.cc.jj	2022-03-19 13:52:53.000000000 +0100
+++ gcc/config/i386/i386-expand.cc	2022-04-01 16:51:27.253154191 +0200
@@ -15830,9 +15830,9 @@ quarter:
 	      else
 		{
 		  word = expand_simple_binop (tmp_mode, ASHIFT, word, shift,
-					      word, 1, OPTAB_LIB_WIDEN);
+					      NULL_RTX, 1, OPTAB_LIB_WIDEN);
 		  word = expand_simple_binop (tmp_mode, IOR, word, elt,
-					      word, 1, OPTAB_LIB_WIDEN);
+					      NULL_RTX, 1, OPTAB_LIB_WIDEN);
 		}
 	    }
 
--- gcc/testsuite/gcc.target/i386/pr105123.c.jj	2022-04-01 16:56:44.549625810 +0200
+++ gcc/testsuite/gcc.target/i386/pr105123.c	2022-04-01 16:56:33.569782511 +0200
@@ -0,0 +1,22 @@
+/* PR target/105123 */
+/* { dg-do run { target sse2_runtime } } */
+/* { dg-options "-msse2" } */
+/* { dg-additional-options "-mtune=i686" { target ia32 } } */
+
+typedef unsigned short __attribute__((__vector_size__ (4 * sizeof (unsigned short)))) V;
+
+V
+foo (unsigned short u, V v)
+{
+  return __builtin_shuffle (u * v, v);
+}
+
+int
+main ()
+{
+  V x = foo (1, (V) { 0, 1, 2, 3 });
+  for (unsigned i = 0; i < 4; i++)
+    if (x[i] != i)
+      __builtin_abort ();
+  return 0;
+}

	Jakub


^ permalink raw reply	[flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-04-03 19:31 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-02  9:25 [PATCH] i386: Fix up ix86_expand_vector_init_general [PR105123] Jakub Jelinek
2022-04-03 19:31 ` Uros Bizjak

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).