[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression between 4.6(4.5) and 4.4 for ARM

[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression between 4.6(4.5) and 4.4 for ARM

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.5.2                       |4.5.3

--- Comment #6 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-12-16 13:03:16 UTC ---
GCC 4.5.2 is being released, adjusting target milestone.

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.5.3                       |4.5.4

--- Comment #7 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-04-28 14:51:27 UTC ---
GCC 4.5.3 is being released, adjusting target milestone.

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[bmei at broadcom dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #8 from Bingfeng Mei <bmei at broadcom dot com> 2011-04-28 15:22:26 UTC ---
I am currently on vacation until 4/5/2011. I may access my mail irregularly.
Cheers,
Bingfeng Mei

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-05 20:31:17 UTC ---
We produce better code with the trunk now:
    xorl    %eax, %eax
    testl    $1024, %edi
    setne    %al
    ret

But that is still worse than before with the and ...

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #10 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-05 20:40:59 UTC ---
  /* If this is an equality or inequality test of a single bit, we can
     do this by shifting the bit being tested to the low-order bit and
     masking the result with the constant 1.  If the condition was EQ,
     we xor it with 1.  This does not require an scc insn and is faster
     than an scc insn even if we have it.

     The code to make this transformation was moved into fold_single_bit_test,
     so we just call into the folder and expand its result.  */

  if ((code == NE || code == EQ)
      && TREE_CODE (arg0) == BIT_AND_EXPR && integer_zerop (arg1)
      && integer_pow2p (TREE_OPERAND (arg0, 1))
      && (TYPE_PRECISION (ops->type) != 1 || TYPE_UNSIGNED (ops->type)))

This needs to be updated for the out-of-ssa expand.

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-05 20:51:58 UTC ---
I have a patch:
Index: expr.c
===================================================================
--- expr.c    (revision 182017)
+++ expr.c    (working copy)
@@ -10563,15 +10563,24 @@ do_store_flag (sepops ops, rtx target, e
      so we just call into the folder and expand its result.  */

   if ((code == NE || code == EQ)
-      && TREE_CODE (arg0) == BIT_AND_EXPR && integer_zerop (arg1)
-      && integer_pow2p (TREE_OPERAND (arg0, 1))
+      && TREE_CODE (arg0) == SSA_NAME && integer_zerop (arg1)
       && (TYPE_PRECISION (ops->type) != 1 || TYPE_UNSIGNED (ops->type)))
     {
-      tree type = lang_hooks.types.type_for_mode (mode, unsignedp);
-      return expand_expr (fold_single_bit_test (loc,
-                        code == NE ? NE_EXPR : EQ_EXPR,
-                        arg0, arg1, type),
-              target, VOIDmode, EXPAND_NORMAL);
+      gimple srcstmt = get_gimple_for_ssa_name (arg0);
+      if (srcstmt
+      && TREE_CODE (gimple_assign_rhs_code (srcstmt)) == BIT_AND_EXPR
+      && integer_pow2p (gimple_assign_rhs2 (srcstmt))
+    {
+      enum tree_code tcode = code == NE ? NE_EXPR : EQ_EXPR;
+      tree type = lang_hooks.types.type_for_mode (mode, unsignedp);
+      tree temp = fold_build2 (loc, BIT_AND_EXPR, type,
+                   gimple_assign_rhs1 (srcstmt),
+                   gimple_assign_rhs2 (srcstmt));
+      return expand_expr (fold_single_bit_test (loc,
+                                  tcode,
+                                  arg0, arg1, type),
+                  target, VOIDmode, EXPAND_NORMAL);
+    }
     }

   if (! get_subtarget (target)

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #12 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-05 20:56:14 UTC ---
(In reply to comment #11)
> I have a patch:
With a couple of fixes above that.

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch
             Target|arm-elf                     |
             Status|NEW                         |ASSIGNED
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2011-12/msg00466.htm
                   |                            |l
         AssignedTo|unassigned at gcc dot       |pinskia at gcc dot gnu.org
                   |gnu.org                     |

[Bug middle-end/45416] [4.5/4.6/4.7 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #13 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-07 19:23:13 UTC ---
Author: pinskia
Date: Wed Dec  7 19:23:10 2011
New Revision: 182084

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=182084
Log:
2011-12-07  Andrew Pinski  <apinski@cavium.com>

    PR middle-end/45416
    * expr.c (do_store_flag): Rewrite code that looks for BIT_AND_EXPR for
    SSA-expand.

2011-12-07  Andrew Pinski  <apinski@cavium.com>

    PR middle-end/45416
    * gcc.dg/pr45416.c: New testcase.


Added:
    trunk/gcc/testsuite/gcc.dg/pr45416.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/expr.c
    trunk/gcc/testsuite/ChangeLog

[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression between 4.6/4.7(4.5) and 4.4 for ARM

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |NEW
      Known to work|                            |4.7.0
         AssignedTo|pinskia at gcc dot gnu.org  |unassigned at gcc dot
                   |                            |gnu.org
            Summary|[4.5/4.6/4.7 Regression]    |[4.5/4.6 Regression] Code
                   |Code size regression        |size regression between
                   |between 4.6/4.7(4.5) and    |4.6/4.7(4.5) and 4.4 for
                   |4.4 for ARM                 |ARM

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-07 19:24:22 UTC ---
Fixed on the trunk.  If someone wants to backport the patch, that is ok with
me.

[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression from 4.4 for ARM

[gjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Georg-Johann Lay <gjl at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |eric.weddington at atmel
                   |                            |dot com, gjl at gcc dot
                   |                            |gnu.org

--- Comment #15 from Georg-Johann Lay <gjl at gcc dot gnu.org> 2011-12-09 13:01:39 UTC ---
Appears that the patch leads to code bloat on AVR at least.

It is general wrong to assume that it's cheap to perform x >> n without caring
for the underlying architecture.

[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression from 4.4 for ARM

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

--- Comment #16 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2012-01-23 17:59:56 UTC ---
Author: ramana
Date: Mon Jan 23 17:59:51 2012
New Revision: 183446

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=183446
Log:


2012-01-23  Ramana Radhakrishnan  <ramana.radhakrishnan@linaro.org>

       PR middle-end/45416
        * gcc.dg/pr45416.c: Skip if Thumb1. Handle ubfx.


Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/pr45416.c

[Bug middle-end/45416] [4.5/4.6 Regression] Code size regression from 4.4 for ARM

[rearnsha at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45416

Richard Earnshaw <rearnsha at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |arm
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|4.5.4                       |4.7.0

--- Comment #17 from Richard Earnshaw <rearnsha at gcc dot gnu.org> 2012-02-01 14:47:09 UTC ---
fixed.  Not a wrong-code bug, so not backporting.