Merge lp:~rsandifo/gcc-linaro/lp721531-4.6 into lp:gcc-linaro/4.6

[Richard Sandiford]

This version includes some branch-local changes to account
for differences in the handling of PR 35705. In mainline,
get_pointer_and_residue reuses get_object_alignment,
but that improvement isn't in 4.6and earlier. This backport
just reuses get_object_alignment for the simple and safe*
case of DECLs.

(PR 35705 was about optimising away checks for bit 1 (rather than bit 0)
in a HPPA function address. This patch subsumes that one.

[Linaro Toolchain Builder (cbuild)]

cbuild has taken a snapshot of this branch at r106754 and queued it for build.

The snapshot is available at:
 http://ex.seabright.co.nz/snapshots/gcc-linaro-4.6+bzr106754~rsandifo~lp721531-4.6.tar.xdelta3.xz

and will be built on the following builders:
 a9-builder armv5-builder i686 x86_64

You can track the build queue at:
 http://ex.seabright.co.nz/helpers/scheduler

cbuild-snapshot: gcc-linaro-4.6+bzr106754~rsandifo~lp721531-4.6
cbuild-ancestor: lp:gcc-linaro/4.6+bzr106753
cbuild-state: check

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on i686-lucid-cbuild136-scorpius-i686r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106754~rsandifo~lp721531-4.6/logs/i686-lucid-cbuild136-scorpius-i686r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +PASS: gcc.dg/torture/pr49169.c -O0 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O1 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -g (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -Os (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 -PASS: g++.dg/cpp0x/constexpr-ptrmem.C (test for excess errors)
 +FAIL: g++.dg/cpp0x/constexpr-ptrmem.C (test for excess errors)
 -PASS: g++.dg/tree-ssa/fwprop-align.C scan-tree-dump-times forwprop2 "& 1" 0
 +FAIL: g++.dg/tree-ssa/fwprop-align.C scan-tree-dump-times forwprop2 "& 1" 0

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106754~rsandifo~lp721531-4.6/logs/i686-lucid-cbuild136-scorpius-i686r1/gcc-testsuite.txt

cbuild-checked: i686-lucid-cbuild136-scorpius-i686r1

[Linaro Toolchain Builder (cbuild)]

cbuild has taken a snapshot of this branch at r106755 and queued it for build.

The snapshot is available at:
 http://ex.seabright.co.nz/snapshots/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6.tar.xdelta3.xz

and will be built on the following builders:
 a9-builder armv5-builder i686 x86_64

You can track the build queue at:
 http://ex.seabright.co.nz/helpers/scheduler

cbuild-snapshot: gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6
cbuild-ancestor: lp:gcc-linaro/4.6+bzr106753
cbuild-state: check

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on i686-lucid-cbuild140-scorpius-i686r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/i686-lucid-cbuild140-scorpius-i686r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O0
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O1
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -g
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -Os

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/i686-lucid-cbuild140-scorpius-i686r1/gcc-testsuite.txt

cbuild-checked: i686-lucid-cbuild140-scorpius-i686r1

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on armv7l-natty-cbuild142-ursa2-cortexa9r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild142-ursa2-cortexa9r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +PASS: gcc.dg/torture/pr49169.c -O0 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O1 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -g (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -Os (test for excess errors)
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +FAIL: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild142-ursa2-cortexa9r1/gcc-testsuite.txt

cbuild-checked: armv7l-natty-cbuild142-ursa2-cortexa9r1

[Michael Hope (michaelh1)]

Does the test case assume the compiler is built for a certain mode?

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on x86_64-natty-cbuild144-oort3-x86_64r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/x86_64-natty-cbuild144-oort3-x86_64r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O0
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O1
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -g
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -Os

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/x86_64-natty-cbuild144-oort3-x86_64r1/gcc-testsuite.txt

cbuild-checked: x86_64-natty-cbuild144-oort3-x86_64r1

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on armv7l-natty-cbuild144-ursa1-armv5r2.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild144-ursa1-armv5r2

The test suite was not checked as the branch point lp:gcc-linaro/4.6+bzr106753 has nothing to compare against.

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106755~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild144-ursa1-armv5r2/gcc-testsuite.txt

cbuild-checked: armv7l-natty-cbuild144-ursa1-armv5r2

[Richard Sandiford (rsandifo)]

Michael Hope <email address hidden> writes:
> Does the test case assume the compiler is built for a certain mode?

It isn't supposed to. Investigating...

[Linaro Toolchain Builder (cbuild)]

cbuild has taken a snapshot of this branch at r106756 and queued it for build.

The snapshot is available at:
 http://ex.seabright.co.nz/snapshots/gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6.tar.xdelta3.xz

and will be built on the following builders:
 a9-builder armv5-builder i686 x86_64

You can track the build queue at:
 http://ex.seabright.co.nz/helpers/scheduler

cbuild-snapshot: gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6
cbuild-ancestor: lp:gcc-linaro/4.6+bzr106753
cbuild-state: check

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on x86_64-natty-cbuild148-oort1-x86_64r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6/logs/x86_64-natty-cbuild148-oort1-x86_64r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O0
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O1
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -O3 -g
 +UNSUPPORTED: gcc.dg/torture/pr49169.c -Os

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6/logs/x86_64-natty-cbuild148-oort1-x86_64r1/gcc-testsuite.txt

cbuild-checked: x86_64-natty-cbuild148-oort1-x86_64r1

[Linaro Toolchain Builder (cbuild)]

cbuild successfully built this on armv7l-natty-cbuild148-ursa2-cortexa9r1.

The build results are available at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild148-ursa2-cortexa9r1

The test suite results changed compared to the branch point lp:gcc-linaro/4.6+bzr106753:
 +PASS: gcc.dg/torture/pr49169.c -O0 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O1 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto -flto-partition=none (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 -flto (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O2 (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -fomit-frame-pointer (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -O3 -g (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c -Os (test for excess errors)
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 +PASS: gcc.dg/torture/pr49169.c scan-assembler abort
 -WARNING: program timed out.

The full testsuite results are at:
 http://ex.seabright.co.nz/build/gcc-linaro-4.6+bzr106756~rsandifo~lp721531-4.6/logs/armv7l-natty-cbuild148-ursa2-cortexa9r1/gcc-testsuite.txt

cbuild-checked: armv7l-natty-cbuild148-ursa2-cortexa9r1

[Ramana Radhakrishnan (ramana)]

This is Ok.

Ramana

review: Approve

=== modified file 'ChangeLog.linaro'
--- ChangeLog.linaro	2011-07-03 09:08:14 +0000
+++ ChangeLog.linaro	2011-07-04 10:08:02 +0000
@@ -1,3 +1,4 @@
+<<<<<<< TREE
 2011-07-03  Ira Rosen  <[e-mail]>
 
 	Backport from FSF:
@@ -293,6 +294,58 @@
 	gcc/testsuite/
 	* gcc.target/arm/mla-2.c: New test.
 
+=======
[phone]  Richard Sandiford  <[e-mail]>
+
+	gcc/
+	* builtins.c (get_object_alignment): Fix comment.
+	* fold-const.c (get_pointer_modulus_and_residue): Remove
+	allow_func_align.  Use get_object_alignment.
+	(fold_binary_loc): Update caller.
+
[phone]  Richard Sandiford  <[e-mail]>
+
+	gcc/
+	Backport from mainline:
+
+	2011-06-29  Richard Sandiford  <[e-mail]>
+
+	PR tree-optimization/49545
+	* builtins.c (get_object_alignment_1): Update function comment.
+	Do not use DECL_ALIGN for functions, but test
+	TARGET_PTRMEMFUNC_VBIT_LOCATION instead.
+	* fold-const.c (get_pointer_modulus_and_residue): Don't check
+	for functions here.
+	* tree-ssa-ccp.c (get_value_from_alignment): Likewise.
+
+	gcc/testsuite/
+	Backport from mainline:
+
+	2011-06-29  Richard Sandiford  <[e-mail]>
+
+	* gcc.dg/torture/pr49169.c: Restrict to ARM and MIPS targets.
+
[phone]  Richard Sandiford  <[e-mail]>
+
+	gcc/
+	Backport from mainline:
+
+	2011-07-27  Richard Guenther  <[e-mail]>
+
+	PR tree-optimization/49169
+	* fold-const.c (get_pointer_modulus_and_residue): Don't rely on
+	the alignment of function decls.
+
+	gcc/testsuite/
+	Backport from mainline:
+
+	2011-07-27  Michael Hope  <[e-mail]>
+		    Richard Sandiford  <[e-mail]>
+
+	PR tree-optimization/49169
+	* gcc.dg/torture/pr49169.c: New test.
+
+>>>>>>> MERGE-SOURCE
 2001-06-02  Richard Sandiford  <[e-mail]>
 
 	gcc/

=== modified file 'gcc/builtins.c'
--- gcc/builtins.c	2011-03-03 21:56:58 +0000
+++ gcc/builtins.c	2011-07-04 10:08:02 +0000
@@ -264,7 +264,14 @@
 }
 
 /* Return the alignment in bits of EXP, an object.
-   Don't return more than MAX_ALIGN no matter what.  */
+   Don't return more than MAX_ALIGN no matter what.
+
+   Note that the address (and thus the alignment) computed here is based
+   on the address to which a symbol resolves, whereas DECL_ALIGN is based
+   on the address at which an object is actually located.  These two
+   addresses are not always the same.  For example, on ARM targets,
+   the address &foo of a Thumb function foo() has the lowest bit set,
+   whereas foo() itself starts on an even address.  */
 
 unsigned int
 get_object_alignment (tree exp, unsigned int max_align)
@@ -286,7 +293,21 @@
     exp = DECL_INITIAL (exp);
   if (DECL_P (exp)
       && TREE_CODE (exp) != LABEL_DECL)
-    align = DECL_ALIGN (exp);
+    {
+      if (TREE_CODE (exp) == FUNCTION_DECL)
+	{
+	  /* Function addresses can encode extra information besides their
+	     alignment.  However, if TARGET_PTRMEMFUNC_VBIT_LOCATION
+	     allows the low bit to be used as a virtual bit, we know
+	     that the address itself must be 2-byte aligned.  */
+	  if (TARGET_PTRMEMFUNC_VBIT_LOCATION == ptrmemfunc_vbit_in_pfn)
+	    align = 2 * BITS_PER_UNIT;
+	  else
+	    align = BITS_PER_UNIT;
+	}
+      else
+	align = DECL_ALIGN (exp);
+    }
   else if (CONSTANT_CLASS_P (exp))
     {
       align = TYPE_ALIGN (TREE_TYPE (exp));

=== modified file 'gcc/fold-const.c'
--- gcc/fold-const.c	2011-05-23 20:37:18 +0000
+++ gcc/fold-const.c	2011-07-04 10:08:02 +0000
@@ -9232,15 +9232,10 @@
    0 <= N < M as is common.  In general, the precise value of P is unknown.
    M is chosen as large as possible such that constant N can be determined.
 
-   Returns M and sets *RESIDUE to N.
-
-   If ALLOW_FUNC_ALIGN is true, do take functions' DECL_ALIGN_UNIT into
-   account.  This is not always possible due to PR 35705.
- */
+   Returns M and sets *RESIDUE to N.  */
 
 static unsigned HOST_WIDE_INT
-get_pointer_modulus_and_residue (tree expr, unsigned HOST_WIDE_INT *residue,
-				 bool allow_func_align)
+get_pointer_modulus_and_residue (tree expr, unsigned HOST_WIDE_INT *residue)
 {
   enum tree_code code;
 
@@ -9270,9 +9265,8 @@
 	    }
 	}
 
-      if (DECL_P (expr)
-	  && (allow_func_align || TREE_CODE (expr) != FUNCTION_DECL))
-	return DECL_ALIGN_UNIT (expr);
+      if (DECL_P (expr))
+	return get_object_alignment (expr, ~0U) / BITS_PER_UNIT;
     }
   else if (code == POINTER_PLUS_EXPR)
     {
@@ -9282,8 +9276,7 @@
 
       op0 = TREE_OPERAND (expr, 0);
       STRIP_NOPS (op0);
-      modulus = get_pointer_modulus_and_residue (op0, residue,
-						 allow_func_align);
+      modulus = get_pointer_modulus_and_residue (op0, residue);
 
       op1 = TREE_OPERAND (expr, 1);
       STRIP_NOPS (op1);
@@ -11163,8 +11156,7 @@
 	  unsigned HOST_WIDE_INT modulus, residue;
 	  unsigned HOST_WIDE_INT low = TREE_INT_CST_LOW (arg1);
 
-	  modulus = get_pointer_modulus_and_residue (arg0, &residue,
-						     integer_onep (arg1));
+	  modulus = get_pointer_modulus_and_residue (arg0, &residue);
 
 	  /* This works because modulus is a power of 2.  If this weren't the
 	     case, we'd have to replace it by its greatest power-of-2

=== added file 'gcc/testsuite/gcc.dg/torture/pr49169.c'
--- gcc/testsuite/gcc.dg/torture/pr49169.c	1970-01-01 00:00:00 +0000
+++ gcc/testsuite/gcc.dg/torture/pr49169.c	2011-07-04 10:08:02 +0000
@@ -0,0 +1,15 @@
+/* { dg-do compile { target { arm*-*-* || mips*-*-* } } } */
+
+#include <stdlib.h>
+#include <stdint.h>
+
+int
+main (void)
+{
+  void *p = main;
+  if ((intptr_t) p & 1)
+    abort ();
+  return 0;
+}
+
+/* { dg-final { scan-assembler "abort" } } */

=== modified file 'gcc/tree-ssa-ccp.c'
--- gcc/tree-ssa-ccp.c	2011-05-05 15:42:22 +0000
+++ gcc/tree-ssa-ccp.c	2011-07-04 10:08:02 +0000
@@ -522,10 +522,6 @@
     val = bit_value_binop (PLUS_EXPR, TREE_TYPE (expr),
 			   TREE_OPERAND (base, 0), TREE_OPERAND (base, 1));
   else if (base
-	   /* ???  While function decls have DECL_ALIGN their addresses
-	      may encode extra information in the lower bits on some
-	      targets (PR47239).  Simply punt for function decls for now.  */
-	   && TREE_CODE (base) != FUNCTION_DECL
 	   && ((align = get_object_alignment (base, BIGGEST_ALIGNMENT))
 		> BITS_PER_UNIT))
     {