Paolo Carlini [Sun, 13 Dec 2009 23:45:12 +0000 (23:45 +0000)]
re PR libstdc++/40133 (exception propagation support not enabled in libstdc++ 4.4 on {armeabi,hppa}-linux)
2009-12-11 Paolo Carlini <paolo.carlini@oracle.com>
Matthias Klose <doko@ubuntu.com>
PR libstdc++/40133
* acinclude.m4 ([GLIBCXX_ENABLE_ATOMIC_BUILTINS]): On *-*-linux*,
*-*-uclinux*, *-*-kfreebsd*-gnu | *-*-gnu* targets do link tests when
possible.
* configure: Regenerate.
Michael Matz [Sun, 13 Dec 2009 21:51:34 +0000 (21:51 +0000)]
re PR tree-optimization/42027 (Performance regression in convolution loop optimization)
PR tree-optimization/42027
* dojump.c (do_jump <TRUTH_AND_EXPR, TRUTH_OR_EXPR>): Go to
TRUTH_ANDIF_EXPR resp. TRUTH_ORIF_EXPR expander, instead of
falling through.
Richard Guenther [Sat, 12 Dec 2009 01:12:55 +0000 (01:12 +0000)]
tree.c (need_assembler_name_p): Abstract decls do not need an assembler name.
2009-12-12 Richard Guenther <rguenther@suse.de>
* tree.c (need_assembler_name_p): Abstract decls do not need
an assembler name.
* dwarf2out.c (dwarf2out_abstract_function): Preserve
DECL_ABSTRACT properly.
Paolo Carlini [Fri, 11 Dec 2009 22:04:56 +0000 (22:04 +0000)]
re PR libstdc++/42352 (-std=c++0x reference binding problem)
2009-12-11 Paolo Carlini <paolo.carlini@oracle.com>
PR libstdc++/42352
* include/bits/list.tcc (sort): Use _GLIBCXX_MOVE for list::splice
and list::merge calls.
* testsuite/23_containers/list/operations/42352.cc: New.
Daniel Franke [Fri, 11 Dec 2009 21:08:39 +0000 (16:08 -0500)]
re PR fortran/40290 (Spurious warning on REAL*COMPLEX with -Wconversion)
2009-12-11 Daniel Franke <franke.daniel@gmail.com>
PR fortran/40290
* expr.c (gfc_type_convert_binary): Added warn-on-conversion flag,
passed on to gfc_convert_type_warn() instead of gfc_convert_type();
enabled warnings on all callers but ...
* arith.c (eval_intrinsic): Disabled warnings on implicit type
conversion.
* gfortran.h gfc_type_convert_binary): Adjusted prototype.
Richard Guenther [Fri, 11 Dec 2009 18:00:24 +0000 (18:00 +0000)]
re PR lto/41658 (Execution testsuite fails with -O2 -fwhopr)
2009-12-11 Richard Guenther <rguenther@suse.de>
PR lto/41658
PR lto/41657
* lto-streamer-out.c (write_global_stream): Do not make decls
weak but resort to TREE_ASM_WRITTEN to avoid multiple definitions.
Make sure to mark all decls that we have written.
Paolo Carlini [Fri, 11 Dec 2009 17:54:37 +0000 (17:54 +0000)]
PR libstdc++/22634, DR 539 [Ready]
2009-12-11 Paolo Carlini <paolo.carlini@oracle.com>
PR libstdc++/22634, DR 539 [Ready]
* include/bits/stl_numeric.h (adjacent_difference): Use std::move
at the end of the loop body, per the Ready resolution.
* include/std/numeric: Do not include unnecessarily <cstddef>.
* doc/xml/manual/intro.xml: Add an entry for DR 539.
Jakub Jelinek [Thu, 10 Dec 2009 21:58:49 +0000 (22:58 +0100)]
re PR c++/42317 (Issues with comdat virtual dtors)
PR c++/42317
* cgraph.h (struct cgraph_node): Add same_comdat_group field.
* cgraph.c (cgraph_remove_node): Unchain node from same_comdat_group
circular list.
(cgraph_node_can_be_local_p): Return false for DECL_COMDAT with
node->same_comdat_group.
* ipa.c (cgraph_remove_unreachable_nodes): For any reachable node
mark all its same_comdat_group nodes as also reachable.
(cgraph_externally_visible_p): Return true even if any of
same_comdat_group nodes has address taken.
* lto-cgraph.c (lto_output_node): Stream out same_comdat_group.
(output_cgraph): Ensure other same_comdat_group nodes are also
included.
(input_node): Stream in same_comdat_group.
(input_cgraph_1): Fix up same_comdat_group fields from references
to pointers.
* cgraphunit.c (cgraph_analyze_functions): Mark all other
same_comdat_group nodes as reachable.
(cgraph_mark_functions_to_output): For each node->process process
also other same_comdat_group nodes.
* ipa-inline.c (cgraph_clone_inlined_nodes): Don't reuse nodes
with same_comdat_group non-NULL.
(cgraph_mark_inline_edge): Likewise.
* decl2.c (cp_write_global_declarations): Clear DECL_EXTERNAL
also on all other functions in the same comdat group.
* optimize.c (maybe_clone_body): Also optimize virtual implicit
dtors. For virtual comdat dtors tell cgraph that base and deleting
dtor are in the same comdat group.
* config/abi/pre/gnu.ver: Don't export certain base dtors that
weren't previously exported.
* g++.dg/opt/dtor2.C: New test.
* g++.dg/opt/dtor2.h: New file.
* g++.dg/opt/dtor2-aux.cc: New file.
Bernd Schmidt [Thu, 10 Dec 2009 18:03:05 +0000 (18:03 +0000)]
re PR rtl-optimization/42216 (changes in scheduling regress 464.h264ref 20%)
PR rtl-opt/42216
* regrename.c: Error out if MAX_RECOG_OPERANDS is larger than
HOST_BITS_PER_WIDE_INT.
(verify_reg_in_set): New function, broken out of verify_reg_tracked.
(verify_reg_tracked): Use it.
(scan_rtx_reg): When seeing a use involving a superset of the
registers in an existing chain, enlarge that chain. Otherwise,
allow subsets and set fail_current_block for all other kinds of
overlap.
(hide_operands): New argument UNTRACKED_OPERANDS; callers changed.
Do not modify operands when the bit with the corresponding number
is set in that bitmap.
(build_def_use): When we see matching operands with different
modes, don't set fail_current_block, but keep track of such
operands in a bitmap if their registers aren't already tracked
in an open chain. Pass this bitmap to all hide_operands calls.
* include/pa-signal.h: Do not include ucontext.h.
(SIGNAL_HANDLER): Add __attribute__ ((unused)) to signal handler
arguments.
(MAKE_THROW_FRAME): Do not adjust program counter.
* include/hppa-signal.h: Do not include sys/types.h, sys/syscall.h and
unistd.h.
(MAKE_THROW_FRAME): Do not adjust program counter.
(HANDLE_FPE): Define.
(INIT_SEGV, INIT_FPE): Revise indentation.
Nathan Froyd [Wed, 9 Dec 2009 16:46:57 +0000 (16:46 +0000)]
vector.md (absv2sf2, [...]): New expanders.
* config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
mulv2sf3, divv2sf3): New expanders.
* config/rs6000/spe.md (spe_evabs, spe_evand, spe_evaddw, spe_evsubfw,
spe_evdivws): Rename to use standard GCC names.
* config/rs6000/paired.md (negv2sf, absv2sf2, addv2sf3, subv2sf3,
mulv2sf3, divv2sf3): Rename to avoid conflict with the new expanders.
* config/rs6000/rs6000.c (bdesc_2arg, bdesc_1arg): Use new CODE_FOR_
names for renamed patterns.
Sebastian Pop [Wed, 9 Dec 2009 15:18:22 +0000 (15:18 +0000)]
re PR tree-optimization/42285 (ICE in Graphite's scan_tree_for_params for 416.gamess)
Fix PR42285.
2009-12-08 Sebastian Pop <sebpop@gmail.com>
PR middle-end/42285
* graphite-scop-detection.c (graphite_can_represent_init): Also
handle more complex MULT_EXPRs containing parameters by recursion
on the structure.
Jakub Jelinek [Wed, 9 Dec 2009 09:13:43 +0000 (10:13 +0100)]
configure.ac: Compute ld_ver...
* configure.ac: Compute ld_ver, ld_vers, ld_date, ld_vers_major,
ld_vers_minor and ld_vers_path always, not just when gcc_cv_ld_hidden
isn't cached.
* configure: Regenerated.
Eric Botcazou [Tue, 8 Dec 2009 18:38:23 +0000 (18:38 +0000)]
configure.host (sparc*-sun-solaris2.*): New case.
* configure.host (sparc*-sun-solaris2.*): New case.
* include/dwarf2-signal.h (MAKE_THROW_FRAME, SPARC case): Make dummy.
* include/sparc-signal.h: Do not include ucontext.h.
(SIGNAL_HANDLER): Rename symbol.
(FLUSH_REGISTER_WINDOWS): Delete.
(MAKE_THROW_FRAME): Make dummy.
(INIT_SEGV): Adjust for above renaming.
(INIT_FPE): Likewise.
Michael Matz [Tue, 8 Dec 2009 13:56:06 +0000 (13:56 +0000)]
re PR middle-end/38474 (compile time explosion in dataflow_set_preserve_mem_locs at -O3)
PR middle-end/38474
* function.c (free_temp_slots): Only walk the temp slot
addresses and combine slots if we actually changes something.
(pop_temp_slots): Ditto.
Daniel Kraft [Tue, 8 Dec 2009 11:39:20 +0000 (12:39 +0100)]
re PR fortran/41177 (Wrong base-object checks for type-bound procedures)
2008-12-08 Daniel Kraft <d@domob.eu>
PR fortran/41177
* gfortran.dg/typebound_proc_4.f03: Remove check for wrong error.
* gfortran.dg/typebound_proc_13.f03: New test.
2008-12-08 Daniel Kraft <d@domob.eu>
PR fortran/41177
* gfortran.h (struct symbol_attribute): New flag `class_pointer'.
* symbol.c (gfc_build_class_symbol): Set the new flag.
* resolve.c (update_compcall_arglist): Remove wrong check for
non-scalar base-object.
(check_typebound_baseobject): Add the correct version here as well
as some 'not implemented' message check in the old case.
(resolve_typebound_procedure): Check that the passed-object dummy
argument is scalar, non-pointer and non-allocatable as it should be.
Sebastian Pop [Mon, 7 Dec 2009 22:22:45 +0000 (22:22 +0000)]
Remove XOP splitters.
* config/i386/i386-protos.h (ix86_expand_fma4_multiple_memory):
Removed.
* config/i386/i386.c (ix86_expand_fma4_multiple_memory): Removed.
* config/i386/sse.md: Remove all XOP splitters.
Allow the second and fourth operands of XOP multiply-add insns
to be nonimmediate.
* config.gcc (cpu_is_64bit): Add new core e500mc64.
(powerpc*-*-*): Add new core e500mc64.
* config/rs6000/e500mc64.md: New file.
* config/rs6000/rs6000.c (processor_costs): Add new costs for
e500mc64.
(rs6000_override_options): Add e500mc64 case to
processor_target_table. Altivec and Spe options not allowed with
e500mc64. Disable string instructions for e500mc64. Enable branch
targets alignment for both e500mc and e500mc64. Initialize
rs6000_cost for e500mc64.
(rs6000_emit_sISEL): New function.
(rs6000_emit_sCOND): Call rs6000_emit_sISEL for isel targets.
(rs6000_emit_int_cmove): Fix mode of 64 bit isel pattern
generation.
(rs6000_issue_rate): Set issue rate for e500mc64.
(rs6000_rtx_costs): Set more accurate cost for mfcr instruction
on architectures with isel.
* config/rs6000/rs6000-protos.h (rs6000_emit_sISEL): Declare.
* config/rs6000/rs6000.h (processor_type): Add
PROCESSOR_PPCE500MC64.
(ASM_CPU_SPEC): Add e500mc64.
* config/rs6000/rs6000.md (define_attr "cpu"): Add ppce500mc64.
Include e500mc64.md.
(abssi2_isel): Expand pattern to handle DImode.
(nabs<mode>2_isel): New pattern.
(absdi2): Change pattern to handle 64 bit isel targets.
(absdi2_internal): Exclude ISEL targets.
(nabsdi2): Exclude ISEL targets.
* doc/invoke.texi: Add e500mc64 to list of cpus.
git://gcc.gnu.org / gcc.git / log