Zdenek Dvorak [Thu, 17 May 2007 08:10:24 +0000 (10:10 +0200)]
tree-vrp.c (finalize_jump_threads): Do not care about dominance info.
* tree-vrp.c (finalize_jump_threads): Do not care about dominance info.
(execute_vrp): Preserve loops through jump threading.
* tree-ssa-threadupdate.c (thread_single_edge,
dbds_continue_enumeration_p, determine_bb_domination_status,
thread_through_loop_header): New functions.
(create_edge_and_update_destination_phis,
create_edge_and_update_destination_phis): Set loops for the new blocks.
(prune_undesirable_thread_requests): Removed.
(redirect_edges): Do not pretend that redirect_edge_and_branch can
create new blocks.
(thread_block): Do not call prune_undesirable_thread_requests.
Update loops.
(mark_threaded_blocks): Select edges to thread here.
(thread_through_all_blocks): Take may_peel_loop_headers argument.
Thread edges through loop headers independently.
* cfgloopmanip.c (create_preheader, mfb_keep_just): Export.
* tree-pass.h (TODO_mark_first_instance): New.
(first_pass_instance): Declare.
* cfghooks.c (duplicate_block): Put the block to the original loop
if copy is not specified.
* tree-ssa-dom.c (tree_ssa_dominator_optimize): Preserve loops through
jump threading. Pass may_peel_loop_headers to
thread_through_all_blocks according to first_pass_instance.
* cfgloop.h (create_preheader): Declare.
* tree-flow.h (thread_through_all_blocks): Declaration changed.
* basic-block.h (mfb_keep_just, mfb_kj_edge): Declare.
* passes.c (first_pass_instance): New variable.
(next_pass_1): Set TODO_mark_first_instance.
(execute_todo): Set first_pass_instance.
* gcc.dg/tree-ssa/ssa-dom-thread-2.c: New test.
* gcc.dg/vect/vect-102.c, gcc.dg/vect/vect-103.c,
gcc.dg/vect/vect-104.c: Use more complex construction to prevent vectorizing.
* gcc.dg/tree-ssa/pr21559.c: Update outcome.
PR tree-optimization/24659
* gcc.dg/vect/vect-floatint-conversion-2.c: New test.
* gcc.dg/vect/vect-intfloat-conversion-1.c: Require vect_float,
not vect_int target.
* gcc.dg/vect/vect-intfloat-conversion-2.c: Require vect_float,
not vect_int target. Loop is vectorized for vect_intfloat_cvt
targets.
* gcc.dg/vect/vect-intfloat-conversion-3.c: New test.
* gcc.dg/vect/vect-intfloat-conversion-4a.c: New test.
* gcc.dg/vect/vect-intfloat-conversion-4b.c: New test.
Keith Seitz [Wed, 16 May 2007 20:13:52 +0000 (20:13 +0000)]
java-interp.h (breakpoint_at): Declare.
* include/java-interp.h (breakpoint_at): Declare.
* interpret.cc (breakpoint_at): New function.
* gnu/classpath/jdwp/VMVirtualMachine.java (_event_list):
New member.
* gnu/classpath/jdwp/natVMVirtualMachine.cc (initialize):
Initialize _event_list.
(handle_single_step): If there is a breakpoint at the
location at which we are stopping, do not send the notification.
Instead add the event to a list of events that occur at this
location.
(jdwpBreakpointCB): If the event list is not empty, send
whatever events are in it and the breakpoint event in a single
notification.
Mark parameter jni_env as MAYBE_UNUSED.
* classpath/lib/gnu/classpath/jdwp/VMVirtualMachine.class:
Regenerated.
* gnu/classpath/jdwp/VMVirtualMachine.h: Regenerated.
Keith Seitz [Wed, 16 May 2007 20:12:33 +0000 (20:12 +0000)]
java-interp.h (breakpoint_at): Declare.
* include/java-interp.h (breakpoint_at): Declare.
* interpret.cc (breakpoint_at): New function.
* gnu/classpath/jdwp/VMVirtualMachine.java (_event_list):
New member.
* gnu/classpath/jdwp/natVMVirtualMachine.cc (initialize):
Initialize _event_list.
(handle_single_step): If there is a breakpoint at the
location at which we are stopping, do not send the notification.
Instead add the event to a list of events that occur at this
location.
(jdwpBreakpointCB): If the event list is not empty, send
whatever events are in it and the breakpoint event in a single
notification.
Mark parameter jni_env as MAYBE_UNUSED.
* classpath/lib/gnu/classpath/jdwp/VMVirtualMachine.class:
Regenerated.
* gnu/classpath/jdwp/VMVirtualMachine.h: Regenerated.
Uros Bizjak [Wed, 16 May 2007 17:07:07 +0000 (19:07 +0200)]
README: Update for new files.
* soft-fp/README: Update for new files.
* soft-fp/floattisf.c: New file.
* soft-fp/floattidf.c: New file.
* soft-fp/floattitf.c: New file.
* soft-fp/floatuntisf.c: New file.
* soft-fp/floatuntidf.c: New file.
* soft-fp/floatuntitf.c: New file.
* soft-fp/fixsfti.c: New file.
* soft-fp/fixdfti.c: New file.
* soft-fp/fixtfti.c: New file.
* soft-fp/fixunssfti.c: New file.
* soft-fp/fixunsdfti.c: New file.
* soft-fp/fixunstfti.c: New file.
* soft-fp/extendxftf.c: New file.
* soft-fp/trunctfxf.c: New file.
* libgcc-std.ver (__extendxftf2): Added to GCC_4.3.0 section.
(__trunctfxf2): Ditto.
* config/i386/t-fprules-softfp64: New file.
* config/i386/sfp-machine.h: New file.
* config.gcc (x86_64-*-linux*, x86_64-*-kfreebsd*-gnu,
x86_64-*-knetbsd*-gnu): Add i386/t-fprules-softfp64
and soft-fp/t-softfp to tmake_file.
(i[34567]86-*-linux*, i[34567]86-*-kfreebsd*-gnu,
i[34567]86-*-knetbsd*-gnu): Ditto for --enable-targets=all.
* config/i386/t-linux64 (softfp_wrap_start): New.
(softfp_wrap_end): New.
* config/i386/i386.c (ix86_scalar_mode_supported): TFmode is
supported for TARGET_64BIT.
testsuite/ChangeLog:
* gcc.dg/torture/fp-int-convert-float128.c: Do not xfail for i?86-*-*
and x86_64-*-* targets.
* gcc.dg/torture/fp-int-convert-float128-timode.c: Ditto.
Benjamin Kosnik [Wed, 16 May 2007 17:00:58 +0000 (17:00 +0000)]
[multiple changes]
2007-05-16 Benjamin Kosnik <bkoz@redhat.com>
* acinclude.m4 (GLIBCXX_CHECK_COMPILER_FEATURES): Add a function
to the test code compiled with -ffunction-sections -fdata-sections.
* configure: Regenerate.
* testsuite/22_locale/ctype/scan/wchar_t/1.cc: Remove
_GLIBCXX_USE_WCHAR_T guards as superfluous.
Benjamin Kosnik [Wed, 16 May 2007 16:54:48 +0000 (16:54 +0000)]
acinclude.m4 (GLIBCXX_CHECK_COMPILER_FEATURES): Add a function to the test code compiled with -ffunction-sections -fdata-sections.
2007-05-11 Benjamin Kosnik <bkoz@redhat.com>
* acinclude.m4 (GLIBCXX_CHECK_COMPILER_FEATURES): Add a function
to the test code compiled with -ffunction-sections -fdata-sections.
* configure: Regenerate.
* testsuite/22_locale/ctype/scan/wchar_t/1.cc: Remove
_GLIBCXX_USE_WCHAR_T guards as superfluous.
Benjamin Kosnik [Wed, 16 May 2007 16:47:47 +0000 (16:47 +0000)]
acinclude.m4 (GLIBCXX_CHECK_COMPILER_FEATURES): Add a function to the test code compiled with -ffunction-sections -fdata-sections.
2007-05-11 Benjamin Kosnik <bkoz@redhat.com>
* acinclude.m4 (GLIBCXX_CHECK_COMPILER_FEATURES): Add a function
to the test code compiled with -ffunction-sections -fdata-sections.
* configure: Regenerate.
* testsuite/22_locale/ctype/scan/wchar_t/1.cc: Remove
_GLIBCXX_USE_WCHAR_T guards as surpufulous.
revert: configure.ac (gcc_gxx_include_dir): Use $(libsubdir_to_prefix).
gcc/
Revert:
2007-05-12 Richard Sandiford <richard@codesourcery.com>
* configure.ac (gcc_gxx_include_dir): Use $(libsubdir_to_prefix).
(gcc_tooldir): Likewise.
* configure: Regenerate.
* Makefile.in (libsubdir_to_prefix): New variable, based on the
old configure.ac gcc_tooldir setting.
(prefix_to_exec_prefix): New variable.
(DRIVER_DEFINES): Use $(libsubdir_to_prefix)$(prefix_to_exec_prefix)
rather than $(unlibsubdir)/../ to derive TOOLDIR_BASE_PREFIX.
Zdenek Dvorak [Mon, 14 May 2007 21:58:42 +0000 (23:58 +0200)]
tree-ssa-loop-niter.c (record_estimate): Use GGC_NEW to allocate struct nb_iter_bound.
* tree-ssa-loop-niter.c (record_estimate): Use GGC_NEW to allocate
struct nb_iter_bound.
(free_numbers_of_iterations_estimates_loop): Use ggc_free.
* gengtype.c (open_base_files): Add cfhloop.h to the list of includes.
* cfgloopmanip.c (place_new_loop): Vector larray is gc-allocated.
* tree-scalar-evolution.c: Include gt-tree-scalar-evolution.h.
(struct scev_info_str, scalar_evolution_info): Add GTY markers.
(new_scev_info_str): Use GGC_NEW to allocate struct scev_info_str.
(del_scev_info): Use ggc_free.
(scev_initialize): Allocate scalar_evolution_info in gc memory.
* loop-init.c: Include ggc.h.
(loop_optimizer_init): Use GGC_CNEW to allocate struct loops.
(loop_optimizer_finalize): Use ggc_free.
* tree-ssa-loop.c (pass_tree_unswitch, pass_vectorize,
pass_linear_transfom, pass_empty_loop, pass_complete_unroll,
pass_iv_optimize): Add TODO_ggc_collect.
* function.h (struct function): Remove skip marker from x_current_loops.
* cfgloop.c: Include ggc.h.
(flow_loops_free, flow_loop_free): Free the loop descriptions in gc
memory.
(establish_preds): Vector superloops is gc allocated.
(alloc_loop): Allocate loop using GGC_CNEW. Allocate head of
loop->exits list.
(flow_loops_find): Vector larray is gc allocated.
(loop_exit_free): Use ggc_free.
(rescan_loop_exit): Use GGC_NEW to allocate struct loop_exit. Reflect
that head of exits list is now not a part of struct loop.
(record_loop_exits): Allocate exits table in gc memory.
(get_loop_exit_edges, verify_loop_structure, single_exit): Reflect that
head of exits list is now not a part of struct loop.
* cfgloop.h (struct lpt_decision, struct nb_iter_bound,
struct loop_exit): Add GTY marker.
(struct loop): Add GTY marker. Make superloops vector gc allocated.
Add skip marker to aux field. Make head of exits list a separate
object.
(struct loops): Add GTY marker. Make larray vector gc allocated.
Add param marker to exits table.
(get_loops): Type changed.
* Makefile.in (tree-scalar-evolution.o): Add gt-tree-scalar-evolution.h
dependency.
(cfgloop.o, loop-init.o): Add ggc.h dependency.
(GTFILES): Add cfgloop.h and tree-scalar-evolution.c.
* basic-block.h (struct basic_block_def): Remove skip marker from
loop_father field.
Uros Bizjak [Mon, 14 May 2007 19:42:47 +0000 (21:42 +0200)]
builtins.c (expand_builtin_mathfn): Use EXPAND_NORMAL instead of 0 in the call to expand_expr().
* builtins.c (expand_builtin_mathfn): Use EXPAND_NORMAL instead
of 0 in the call to expand_expr().
(expand_builtin_mathfn_3): Ditto.
(expand_builtin_interclass_mathfn): Ditto.
(expand_builtin_cexpi): Ditto.
(expand_builtin_int_roundingfn): Ditto.
(expand_builtin_int_roundingfn_2): Ditto.
(expand_builtin_pow): Ditto.
(expand_builtin_powi): Ditto.
(expand_builtin_bswap): Ditto.
(expand_builtin_unop): Ditto.
(expand_builtin_fabs): Ditto.
(get_builtin_sync_mem): Use NULL_RTX instead of NULL in
the call to expand_expr().
(expand_builtin_sync_operation): Ditto.
(expand_builtin_compare_and_swap): Ditto.
(expand_builtin_lock_test_and_set): Ditto.
* except.c (expand_builtin_eh_return_data_regno): Use EXPAND_NORMAL
instead of 0 in the call to expand_expr().
(expand_builtin_extract_return_addr): Ditto.
(expand_builtin_eh_return): Ditto.
(expand_eh_return): Ditto.
* explow.c (expr_size): Ditto.
* expr.c (optimize_bitfield_assignment_op): Ditto.
(expand_assignement): Ditto.
(store_expr): Ditto.
(store_field): Ditto.
(expand_expr_addr_expr_1): Use NULL_RTX instead of NULL in
the call to expand_expr().
(expand_expr_real_1) [COMPLEX_CST]: Use EXPAND_NORMAL instead of 0
in the call to expand_expr().
[CONSTRUCTOR, PLUS_EXPR, MINUS_EXPR, NEGATE_EXPR, ABS_EXPR,
BIT_NOT_EXPR, LSHIFT_EXPR, LT_EXPR, TRUTH_NOT_EXPR]: Ditto.
[VEC_UNPACK_HI_EXPR, VEC_UNPACK_LO]: Use expand_normal() instead
of expand_expr().
* optabs.c (expand_vec_shift_expr): Ditto.
(expand_vec_cond_expr): Ditto.
(vector_compare_rtx): Use EXPAND_STACK_PARM instead of 1 in the
call to expand_expr().
* stmt.c (expand_return): Use EXPAND_NORMAL instead of 0
in the call to expand_expr().
Kazu Hirata [Mon, 14 May 2007 13:52:18 +0000 (13:52 +0000)]
m68k-protos.h: Rename m68k_interrupt_function_p to m68k_get_function_kind.
gcc/
* config/m68k/m68k-protos.h: Rename m68k_interrupt_function_p
to m68k_get_function_kind. Update its prototype.
* config/m68k/m68k.c (m68k_attribute_table): Add an entry for
interrupt_thread.
(m68k_interrupt_function_p): Return enum m68k_function_type
instead of bool. Rename to m68k_get_function_kind.
(m68k_handle_fndecl_attribute): Reject interrupt_thread if the
target is not fido.
(m68k_compute_frame_layout): Don't mark any register for save
if an interrupt_thread attribute is specified.
(m68k_hard_regno_rename_ok): Update a use of
m68k_interrupt_function_p.
* config/m68k/m68k.h (EPILOGUE_USES): Update a use of
m68k_interrupt_function_p.
(m68k_function_type): New.
* config/m68k/m68k.md (*return): Output a 'sleep' instruction
for a function with an interrupt_thread attribute.
* doc/extend.texi: Document the interrupt_thread attribute.
Steven Bosscher [Sat, 12 May 2007 18:12:54 +0000 (18:12 +0000)]
re PR rtl-optimization/31848 (Invalid loop optimization causes bootstrap failure in genautomata)
PR rtl-optimization/31848
* loop-invariant.c (move_invariant_reg): If we move an insn
with a REG_EQUAL note, and that insn is not always executed,
remove the REG_EQUAL note.
configure.ac (gcc_gxx_include_dir): Use $(libsubdir_to_prefix).
gcc/
* configure.ac (gcc_gxx_include_dir): Use $(libsubdir_to_prefix).
(gcc_tooldir): Likewise.
* configure: Regenerate.
* Makefile.in (libsubdir_to_prefix): New variable, based on the
old configure.ac gcc_tooldir setting.
(prefix_to_exec_prefix): New variable.
(DRIVER_DEFINES): Use $(libsubdir_to_prefix)$(prefix_to_exec_prefix)
rather than $(unlibsubdir)/../ to derive TOOLDIR_BASE_PREFIX.
Paul Thomas [Sat, 12 May 2007 06:19:43 +0000 (06:19 +0000)]
re PR fortran/30746 (50th Anniversary Bug - Forward reference to contained function)
2007-05-12 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30746
* resolve.c (check_host_association): New function that detects
incorrect host association and corrects it.
(gfc_resolve_expr): Call the new function for variables and
functions.
* match.h : Remove prototype for gfc_match_rvalue.
* gfortran.h : Add prototype for gfc_match_rvalue.
2007-05-12 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30746
* gfortran.dg/host_assoc_function_1.f90: New test.
Jan Hubicka [Fri, 11 May 2007 12:18:17 +0000 (14:18 +0200)]
semantics.c (expand_or_defer_fn): Do not call c_record_cdtor_fn.
* semantics.c (expand_or_defer_fn): Do not call c_record_cdtor_fn.
* decl2.c (start_objects): ctors and dtors are no longer public.
(cp_write_global_declarations): Do not call c_build_cdtor_fns.
* cgraphunit.c: Include gt-cgraphunit.h
(static_ctors, static_dtors): New static vars.
(record_cdtor_fn, build_cdtor, cgraph_build_cdtor_fns): New functions,
based on implementation in c-common.c
(cgraph_finalize_function): Call record_cdtor_fn.
(cgraph_optimize): Call cgraph_build_cdtor_fns.
* decl.c (finish_function): Do not call c_record_cdtor_fn.
(c_write_global_declarations): Do not call c_build_cdtor_fns.
* c-common.c (static_ctors, static_dtors, c_record_cdtor_fn,
build_cdtor, c_build_cdtor_fns): Remove.
* c-common.h (static_ctors, static_dtors, c_record_cdtor_fn,
c_build_cdtor_fns): Remove prototype.
Paul Thomas [Fri, 11 May 2007 11:46:47 +0000 (11:46 +0000)]
re PR fortran/30876 (Array valued recursive function rejected)
2007-05-11 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30876
* trans-expr.c (gfc_conv_function_call): Reduce indirection for
direct assignments of recursive array valued functions.
* primary.c (gfc_match_rvalue): Correct error for recursive
function calls such that directly recursive calls of scalar
function without an explicit result are disallowed.
2007-05-11 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30876
* gfortran.dg/recursive_reference_1.f90: Put error at correct
line.
* gfortran.dg/recursive_reference_2.f90: New test.
Paul Thomas [Fri, 11 May 2007 11:42:56 +0000 (11:42 +0000)]
re PR fortran/30878 (Rejects function f1; namelist /nml/ f1)
2007-05-11 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30878
* resolve.c (resolve_fl_namelist): It is not an error if the
namelist element is the result variable of the enclosing
function. Search for the symbol in current and all parent
namespaces for a potential conflict.
* symbol.c (check_conflict): Remove the conflict between
'in_namelist' and 'FL_PROCEDURE' because the symbol info
is not available to exclude function result variables.
* trans-io.c (nml_get_addr_expr): Use the fake result decl
if the symbol is an implicit result variable.
2007-05-11 Paul Thomas <pault@gcc.gnu.org>
PR fortran/30878
* gfortran.dg/namelist_29.f90: New test.
PR fortran/31474
* decl.c (get_proc_name): If an entry has already been declared
as a module procedure, pick up the symbol and the symtree and
use them for the entry.
2007-05-11 Paul Thomas <pault@gcc.gnu.org>
PR fortran/31474
* gfortran.dg/entry_10.f90: New test.
Zdenek Dvorak [Thu, 10 May 2007 21:33:11 +0000 (23:33 +0200)]
re PR tree-optimization/31885 (Not removing empty loop, scev not finding the correct result)
PR tree-optimization/31885
* tree-chrec.c (chrec_contains_undetermined): Do not consider NULL_TREE
to be undetermined.
(automatically_generated_chrec_p): Return false for NULL.
gcc/
* config.gcc (sparc-wrs-vxworks): New target.
* config/sparc/vxworks.h, config/sparc/t-vxworks: New files.
* config/sparc/sparc-protos.h (sparc_emit_call_insn): Declare.
* config/sparc/sparc.h: Include vxworks-dummy.h.
(PRINT_OPERAND_ADDRESS): Extend SYMBOL_REF handling to
include LABEL_REFs too.
* config/sparc/sparc.c (sparc_expand_move): Don't assume that
_GLOBAL_OFFSET_TABLE_ - label_ref is a link-time constant on
VxWorks.
(legitimize_pic_address): Handle LABEL_REFs like SYMBOL_REFs
on VxWorks.
(load_pic_register): Use gen_vxworks_load_got for VxWorks.
(sparc_emit_call_insn): New function.
(sparc_function_ok_for_sibcall): Restrict sibcalls to locally-binding
functions when generating VxWorks PIC.
* config/sparc/sparc.md (vxworks_load_got): New pattern.
(call, call_value): Use sparc_emit_call_insn instead of
emit_call_insn.
libgcc/
* config.host (sparc-wrs-vxworks): New target.
Bob Wilson [Wed, 9 May 2007 17:05:14 +0000 (17:05 +0000)]
xtensa.c (xtensa_output_literal): Mask out high bits for floating-point values if HOST_BITS_PER_LONG > 32.
* config/xtensa/xtensa.c (xtensa_output_literal): Mask out high bits
for floating-point values if HOST_BITS_PER_LONG > 32. Use split_double
instead of operand_subword.
projects / thirdparty / gcc.git / log
