This is the mail archive of the gcc-regression@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

1 new GCC HEAD@176906 regression


With your recent patch, GCC HEAD revision 176906 had problems on:
native: regress-22 (previously the build failed)
The previous build was of revision 176899.

Regressions for native:
old   g++.sum g++.dg/eh/simd-3.C
old   g++.sum g++.dg/eh/simd-4.C
old   gcc.sum gcc.c-torture/compile/pr34856.c
old   gcc.sum gcc.c-torture/execute/20040811-1.c
old   gcc.sum gcc.c-torture/execute/pr22061-4.c
old   gcc.sum gcc.dg/20020103-1.c
NEW   gcc.sum gcc.dg/fold-compare-3.c
old   gcc.sum gcc.dg/pr34856.c
old   gcc.sum gcc.dg/sms-2.c
old   gcc.sum gcc.dg/sms-4.c
old   gcc.sum gcc.dg/sms-5.c
old   gcc.sum gcc.dg/sms-6.c
old   gcc.sum gcc.dg/sms-7.c
old   gcc.sum gcc.dg/vmx/gcc-bug-i.c
old   gcc.sum gcc.target/powerpc/405-dlmzb-strlen-1.c
old   gcc.sum gcc.target/powerpc/440-dlmzb-strlen-1.c
old   gcc.sum gcc.target/powerpc/ppc-vector-memcpy.c
old   gfortran.sum gfortran.dg/debug/pr35154-dwarf2.f
old   gfortran.sum gfortran.dg/debug/pr37738.f
old   libjava.sum Thread_Sleep_2
old   libstdc++.sum ext/mt_allocator/deallocate_global_thread-1.cc
old   libstdc++.sum ext/mt_allocator/deallocate_global_thread-3.cc

Log information for changes since the last build:
------------------------------------------------------------------------
r176900 | spop | 2011-07-28 14:57:00 -0700 (Thu, 28 Jul 2011) | 16 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/sese.c

Fix PR49876: Continue code generation with integer_zero_node on gloog_error

When setting gloog_error, graphite should continue code generation
without early returns, as otherwise the SSA representation would not
be complete.  So set the new expression to integer_zero_node, that
would not require more SSA updates, and continue code generation as
nothing happened.

Regstrapped on amd64-linux.

2011-07-28  Sebastian Pop  <sebastian.pop@amd.com>

	PR tree-optimization/49876
	* sese.c (rename_uses): Do not return false on gloog_error: set
	the new_expr to integer_zero_node and continue code generation.
	(graphite_copy_stmts_from_block): Remove early exit on gloog_error.
------------------------------------------------------------------------
r176901 | spop | 2011-07-28 14:57:10 -0700 (Thu, 28 Jul 2011) | 31 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/graphite-clast-to-gimple.c
   M /trunk/gcc/testsuite/ChangeLog
   A /trunk/gcc/testsuite/gcc.dg/graphite/id-pr48648.c

Fix PR48648: Handle CLAST assignments.

The CLAST produced by CLooG-ISL contains an assignment and GCC chokes
on it.  The exact CLAST contains an assignment followed by an if:

scat_1 = max(0,ceild(T_4-7,8));
if (scat_1 <= min(1,floord(T_4-1,8))) {
  S7(scat_1);
}

This is equivalent to a loop that iterates only once, and so CLooG
generates an assignment followed by an if instead of a loop.  This is
an important optimization that was improved in ISL, that allows
if-conversion: imagine GCC having to figure out that a loop like the
following actually iterates only once, and can be converted to an if:

for (scat_1 = max(0,ceild(T_4-7,8)); scat_1 <= min(1,floord(T_4-1,8)); scat_1++)
  S7(scat_1);

This patch implements the translation of CLAST assignments.
Bootstrapped and tested on amd64-linux.

2011-07-28  Sebastian Pop  <sebastian.pop@amd.com>

	PR middle-end/48648
	* graphite-clast-to-gimple.c (clast_get_body_of_loop): Handle
	CLAST assignments.
	(translate_clast): Same.
	(translate_clast_assignment): New.

	* gcc.dg/graphite/id-pr48648.c: New.
------------------------------------------------------------------------
r176902 | spop | 2011-07-28 14:57:20 -0700 (Thu, 28 Jul 2011) | 6 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/graphite-clast-to-gimple.c

Replace v1, v2, lb, ub with bound_one, bound_two

2011-07-28  Sebastian Pop  <sebastian.pop@amd.com>

	* graphite-clast-to-gimple.c: Replace v1, v2, lb, ub with
	bound_one, bound_two.
------------------------------------------------------------------------
r176903 | dj | 2011-07-28 15:07:39 -0700 (Thu, 28 Jul 2011) | 3 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/config/m32c/m32c.c

* config/m32c/m32c.c (m32c_secondary_reload_class): Allow PSI
registers to be reloaded in HI classes when the target is HI.

------------------------------------------------------------------------
r176904 | dj | 2011-07-28 15:26:23 -0700 (Thu, 28 Jul 2011) | 4 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/expr.c

* expr.c (expand_expr_addr_expr_1): Detect a user request for a
local frame in a naked function, and produce a suitable error for
that specific case.

------------------------------------------------------------------------
r176905 | bernds | 2011-07-28 15:47:21 -0700 (Thu, 28 Jul 2011) | 4 lines
Changed paths:
   M /trunk/gcc/ChangeLog
   M /trunk/gcc/cfgrtl.c

	PR rtl-optimization/49891
	* cfgrtl.c (force_nonfallthru_and_redirect): Set JUMP_LABEL for
	newly created returnjumps.

------------------------------------------------------------------------

For more information, see <http://gcc.gnu.org/regtest/HEAD/>.

-- 
Geoffrey Keating <geoffk@apple.com> 
(via an automated GCC regression-testing script.)

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]