Jan Hubicka [Sat, 14 May 2005 14:24:49 +0000 (16:24 +0200)]
tree-cfg.c (execute_warn_function_return): Break out noreturn warning too..
* tree-cfg.c (execute_warn_function_return): Break out noreturn
warning too..
(execute_warn_function_noreturn): ... here.
(pass_warn_function_noreturn): New pass.
* tree-pass.h (pass_warn_function_noreturn): Declare
* tree-optimize.c (init_tree_optimization_passes): Move return
warnings early and add noreturn warnings at place of previous
return warnings.
Kazu Hirata [Sat, 14 May 2005 13:05:37 +0000 (13:05 +0000)]
tree-ssa-live.c (tpa_init, [...]): Use VEC instead of VARRAY.
* tree-ssa-live.c (tpa_init, tpa_delete, tpa_compact,
root_var_init, type_var_init): Use VEC instead of VARRAY.
* tree-ssa-live.h (tree_partition_associator_d): Change the
type of trees to VEC(tree,heap)*.
(tpa_tree): Use VEC instead of VARRAY.
Kazu Hirata [Sat, 14 May 2005 03:25:44 +0000 (03:25 +0000)]
tree-eh.c (leh_tf_state): Change the type of dest_array to VEC(tree,heap)*.
* tree-eh.c (leh_tf_state): Change the type of dest_array to
VEC(tree,heap)*.
(maybe_record_in_goto_queue, lower_try_finally_onedest,
lower_try_finally_copy, lower_try_finally_switch,
lower_try_finally): Use VEC instead of VARRAY.
Josh Conner [Fri, 13 May 2005 21:13:17 +0000 (21:13 +0000)]
arm.c (arm_size_return_regs): New.
2005-05-13 Josh Conner <jconner@apple.com>
* config/arm/arm.c (arm_size_return_regs): New.
(thumb_unexpanded_epilogue): replace in-line calculation
of return registers with call to arm_size_return_regs.
(use_return_insn): Include test of which registers are
being used to hold return values, to accommodate
vector return values.
* config/cris/cris.h (CRIS_ASSERT): New macro.
* config/cris/cris.md ("*movsi_internal"): Use CRIS_ASSERT and
internal_error instead of abort.
("movsi"): Ditto. Unquote preparation statement section.
* config/cris/cris.c (ASSERT_PLT_UNSPEC): Define in terms of
CRIS_ASSERT.
("call", "call_value"): Ditto, but use gcc_assert where
appropriate. Tweak comments.
(cris_op_str): Call internal_error, not abort.
(cris_print_operand): Tweak "abort" reference to "die" in comment.
(cris_initial_elimination_offset): Call gcc_unreachable, not abort.
(cris_notice_update_cc): Call internal_error, not abort.
(cris_expand_return, cris_expand_prologue, cris_gen_movem_load)
(cris_emit_movem_store, cris_asm_output_symbol_ref)
(cris_output_addr_const_extra): Use CRIS_ASSERT, don't call abort.
(cris_symbol, cris_got_symbol): Use CRIS_ASSERT instead of
ENABLE_CHECKING-guarded abort call.
(cris_split_movdx): Use CRIS_ASSERT instead of abort with test.
Use internal_error at end of test-chain.
Magnus Fromreide [Fri, 13 May 2005 15:24:30 +0000 (17:24 +0200)]
1.cc: Use initialization instead of copying as the string is used only once.
2005-05-13 Magnus Fromreide <magfr@lysator.liu.se>
* testsuite/27_io/basic_streambuf/sgetn/char/1.cc: Use
initialization instead of copying as the string is used only once.
* testsuite/27_io/basic_streambuf/sgetn/wchar_t/1.cc: Likewise.
* testsuite/27_io/basic_streambuf/sputn/char/1.cc: Likewise.
* testsuite/27_io/basic_streambuf/sputn/wchar_t/1.cc: Likewise.
Joern Rennecke [Fri, 13 May 2005 13:34:17 +0000 (14:34 +0100)]
re PR rtl-optimization/20756 (Clobbers of deleted insns are processed, but left out)
PR rtl-optimization/20756:
* basic-block.h (PROP_DEAD_INSN): Define.
* flow.c (propagate_one_insn): Set during mark_set_regs call
for a dead insn.
(mark_set_1): Emit a clobber when we are about to delete a dead insn
that kills a live register.
J"orn Rennecke [Fri, 13 May 2005 12:25:28 +0000 (12:25 +0000)]
re PR rtl-optimization/20756 (Clobbers of deleted insns are processed, but left out)
PR rtl-optimization/20756:
* basic-block.h (PROP_DEAD_INSN): Define.
* flow.c (propagate_one_insn): Set during mark_set_regs call
for a dead insn.
(mark_set_1): Emit a clobber when we are about to delete a dead insn
that kills a live register.
function.c (instantiate_virtual_regs_in_insn): Do not simplify addition to move unless dest is a register.
* function.c (instantiate_virtual_regs_in_insn): Do not simplify
addition to move unless dest is a register. Do not skip
instantiation in dest on addition simplification.
cris.c (cris_print_operand): Use operand_subword and HOST_WIDE_INT_PRINT_DEC for output...
* config/cris/cris.c (cris_print_operand) <case 'H'; case
CONST_INT>: Use operand_subword and HOST_WIDE_INT_PRINT_DEC for
output; don't have special case for HOST_BITS_PER_WIDE_INT == 32.
Kaz Kojima [Thu, 12 May 2005 21:45:05 +0000 (21:45 +0000)]
sh.c: Declare the prototype of sh_adjust_unroll_max only when...
* config/sh/sh.c: Declare the prototype of sh_adjust_unroll_max
only when TARGET_ADJUST_UNROLL_MAX is defined.
(general_movsrc_operand): Use the C style comment.
(ua_offset): Use prototype.
Zdenek Dvorak [Thu, 12 May 2005 19:41:10 +0000 (21:41 +0200)]
tree-ssa-loop-im.c (struct mem_ref): Add field "next".
* tree-ssa-loop-im.c (struct mem_ref): Add field "next".
(struct hmr_data, hoist_memory_reference, memref_del,
struct fmrv_data): Removed.
(hoist_memory_references, free_mem_ref, free_mem_refs): New functions.
(gather_mem_refs, gather_mem_refs_stmt): Add new references to the
list.
(find_more_ref_vops): Traverse the list of memory references.
(determine_lsm_loop): Work with the list of memory references instead
of traversing the hashtable.
Steven Bosscher [Thu, 12 May 2005 19:29:21 +0000 (19:29 +0000)]
except.c (duplicate_eh_region_1, [...]): New functions, based on old implementation removed with RTL inliner.
* except.c (duplicate_eh_region_1, duplicate_eh_region_2,
duplicate_eh_regions): New functions, based on old
implementation removed with RTL inliner.
(reachable_next_level): Modify behaviour pre-inline.
* except.h (duplicate_eh_regions_map, duplicate_eh_regions):
Declare.
* function.h (struct function): Add after_inlining flag.
* tree-optimize.c (tree_rest_of_compilation): Set it.
* tree-eh.c (add_stmt_to_eh_region_fn): Initialize hashtable
when needed.
* tree-cfg.c (label_to_block_fn): Deal with cases where
label was never insterted into code.
Co-Authored-By: Jan Hubicka <jh@suse.cz> Co-Authored-By: Stuart Hastings <stuart@apple.com>
From-SVN: r99625
Zack Weinberg [Thu, 12 May 2005 18:31:38 +0000 (18:31 +0000)]
directives.c (#sccs table entry): Mark IN_I, consistent with #ident.
libcpp:
* directives.c (#sccs table entry): Mark IN_I, consistent with #ident.
(do_sccs): Delete function definition, #define to do_ident.
(do_ident): Don't hardwire directive name.
gcc:
* doc/cpp.texi: Document that #sccs is a synonym for #ident.
Thomas Koenig [Thu, 12 May 2005 17:50:33 +0000 (17:50 +0000)]
re PR libfortran/21324 (#undef GFC_CLEAR_MEMORY causes testsuite failures)
2005-05-12 Thomas Koenig <Thomas.Koenig@online.de>
PR libfortran/21324
* runtime/memory.c: Don't define GFC_CLEAR_MEMORY (it's a
performance hog).
* io/open.c (new_unit): Zero freshly allocated memory for
unit structure.
* io/unit.c (init_units): Zero freshly allocated memory for
STDIN, STDOUT and STDERR.
* io/unix.c (open_internal): Zero freshly allocated memory
for unix_stream.
(fd_to_stream): Likewise.
Eric Botcazou [Thu, 12 May 2005 11:43:05 +0000 (13:43 +0200)]
sparc.h: Remove dead code.
* config/sparc/sparc.h: Remove dead code.
* config/sparc/sparc.c (sparc_compute_frame_size): Use
FIRST_PARM_OFFSET for the size of the register window area.
(emit_save_regs): Rename into emit_save_or_restore_regs.
Add 'action' parameter. Use 4095 as upper bound for the offset.
Pass 'action' to save_or_restore_regs.
(emit_restore_regs): Delete.
(sparc_expand_prologue): Call emit_save_or_restore_regs.
(sparc_expand_epilogue): Likewise.
* config/sparc/sparc.md (mode macro P): Move.
(movdi_insn_sp32_v9, movdi_insn_sp32): Swap.
(mov<V32:mode> expander): Move to the top of the V32 section.
(movdf_insn_sp32_v9_no_fpu, movdf_insn_sp32_v9): Swap.
(movtf_insn_sp64_hq, movtf_insn_sp64): Swap.
(sibcall_epilogue): Move.
Richard Earnshaw [Thu, 12 May 2005 08:05:11 +0000 (08:05 +0000)]
vfp.md (negsf2_vfp): Add alternative using integer registers.
* arm/vfp.md (negsf2_vfp): Add alternative using integer registers.
(negdf2_vfp): Likewise. Convert to define_insn_and_split and split
the alternatives using integer registers into the appropriate
primitives.
Eli Zaretskii [Thu, 12 May 2005 03:44:58 +0000 (03:44 +0000)]
pex-djgpp.c: Include string.h, fcntl.h, unistd.h, and sys/stat.h.
2005-05-11 Eli Zaretskii <eliz@gnu.org>
* pex-djgpp.c: Include string.h, fcntl.h, unistd.h, and
sys/stat.h.
(pex_init): Fix last argument to pex_init_common.
(pex_djgpp_exec_child): Remove leading underscore from _open,
_dup, _dup2, _close, and _spawnv/_spawnvp. Replace `program',
which is undeclared, with `executable', which was unused. Remove
unused variable `e'. Fix casting of last arg to spawnv/spawnvp.
(pex_djgpp_wait): Declare arguments with ATTRIBUTE_UNUSED.
7000.md (rm7_impy_si_mult): Just match imul and imadd.
* config/mips/7000.md (rm7_impy_si_mult): Just match imul and imadd.
Remove hilo_operand check.
(rm7_impy_si_mul): Just match imul3. Remove hilo_operand check.
* config/mips/predicates.md (hilo_operand): Delete.
Tobias Schlüter [Wed, 11 May 2005 14:52:51 +0000 (16:52 +0200)]
resolve.c (resolve_symbol): Copy 'pointer' and 'dimension' attribute from result symbol to function symbol.
fortran/
* resolve.c (resolve_symbol): Copy 'pointer' and 'dimension'
attribute from result symbol to function symbol.
testsuite/
* gfortran.dg/func_result_2.f90: New test.
J"orn Rennecke [Wed, 11 May 2005 12:24:43 +0000 (12:24 +0000)]
re PR middle-end/20371 (Some corner cases of MS bitfields don't work)
PR middle-end/20371:
* tree.h (record_layout_info_s): New member prev_packed.
* stor-layout.c (update_alignment_for_field): Fix comment about
KNOWN_ALIGN. For MS bitfields, if we start a new run, make sure
we start it properly aligned.
(place_field): At the beginning of a record, pass 0 as KNOWN_ALIGN
to update_alignment_for_field, and recompute it afterwards using
the alignment of the record.
When a packed bitfield precedes an MS bitfield, don't add padding
at the end of the packed bitfield on behalf of the base type of
the packed bit field.
Don't adjust rli->bitpos at the end
of an MS bitfield run if we already adjusted bitpos/offset for an
alignment as large or larger than the bitfield type size.
Take possible record alignment > BIGGEST_ALIGNMENT into account
when calculating actual_align.
Only put packed buit fields into rli->prev_field if they end up
suitably aligned.
Also set rli->remaining_in_alignment when we re-set rli->prev_field.
Update rli->remaining_in_alignment when we have already started a
run of bit fields and we process a packed bit field.
Sebastian Pop [Wed, 11 May 2005 11:03:31 +0000 (13:03 +0200)]
tree-data-ref.c (find_data_references_in_loop): Give up when the body of the loop contains a CALL_EXPR or an ASM_EXPR...
* tree-data-ref.c (find_data_references_in_loop): Give up when
the body of the loop contains a CALL_EXPR or an ASM_EXPR: they
may embed arbitrary side effects.
Remove the assumption that GIMPLE form contains a single array
access per statement.
When the statement contains virtual operands, fail if it is not
a MODIFY_EXPR or a CALL_EXPR.
Return after the dont know node is inserted.
Richard Guenther [Wed, 11 May 2005 08:14:44 +0000 (08:14 +0000)]
re PR middle-end/19807 (fold does not fold &a[4]-1)
2005-05-11 Richard Guenther <rguenth@gcc.gnu.org>
PR middle-end/19807
PR tree-optimization/19639
* fold-const.c (try_move_mult_to_index): Handle INTEGER_CST
and generic summands for char* as s * delta, too, folding &a[i]
CODE x to &a[i CODE x/s]. Use tree_int_cst_equal
for comparison of steps. Convert types for index addition.
(fold_binary): Adjust the callers to always dispatch to
try_move_mult_to_index.
* tree-ssa-propagate.c (set_rhs): Avoid setting rhs to
expr with non-gimple ARRAY_REF offset.
re PR libfortran/21471 ('POSITION = "APPEND"' doesn't seem to work)
PR libfortran/21471
* open.c (new_unit): Take care of the case where POSITION_APPEND
is specified (sseek to the end, and set u>-endfile).
* gfortran.dg/append-1.f90: New test.
Ian Lance Taylor [Tue, 10 May 2005 22:40:37 +0000 (22:40 +0000)]
read-rtl.c (struct macro_traverse_data): Define.
* read-rtl.c (struct macro_traverse_data): Define.
(map_attr_string): New static function, broken out of
apply_macro_to_string.
(mode_attr_index, apply_mode_maps): New static functions.
(apply_macro_to_string): Call map_attr_string.
(apply_macro_to_rtx): Add mode_maps and infile parameters. Change
all callers.
(apply_macro_traverse): Expect data to point to a struct
macro_traverse_data.
(read_rtx): Add mode_maps local variable. Use mode_traverse_data
to pass data through htab_traverse.
(read_rtx_1): Add mode_maps parameter. Change all callers.
Handle mode names which are attribute strings.
* doc/md.texi (Substitutions): Rename from String Substitutions.
Change references. Document using attributes as modes.
Tobias Schlüter [Tue, 10 May 2005 22:06:55 +0000 (00:06 +0200)]
re PR fortran/20178 (COMPLEX function returns incompatible with g77)
gcc/fortran/
PR fortran/20178
* gfortran.h (gfc_option): Add flag_f2c.
* invoke.texi: Document '-ff2c' command line option. Adapt
documentation for '-fno-second-underscore' and '-fno-underscoring'.
* lang.opt (ff2c): New entry.
* options.c (gfc-init_options): Set default calling convention
to -fno-f2c. Mark -fsecond-underscore unset.
(gfc_post_options): Set -fsecond-underscore if not explicitly set
by user.
(handle_options): Set gfc_option.flag_f2c according to requested
calling convention.
* trans-decl.c (gfc_get_extern_function_decl): Use special f2c
intrinsics where necessary.
(gfc_trans_deferred_vars): Change todo error to assertion.
* trans-expr.c (gfc_conv_variable): Dereference access
to hidden result argument.
(gfc_conv_function_call): Add hidden result argument to argument
list if f2c calling conventions requested. Slightly restructure
tests. Convert result of default REAL function to requested type
if f2c calling conventions are used. Dereference COMPLEX result
if f2c cc are used.
* trans-types.c (gfc_sym_type): Return double for default REAL
function if f2c cc are used.
(gfc_return_by_reference): Slightly restructure logic. Return
COMPLEX by reference depending on calling conventions.
(gfc_get_function_type): Correctly make hidden result argument a
pass-by-reference argument for COMPLEX. Remove old code which does
this for derived types.
libgfortran/
PR fortran/20178
* Makefile.am (gfor_specific_src): Add 'intrinsics/f2c_intrinsics.F90'
to dependencies.
* Makefile.in: Regenerate.
* intrinsics/f2c_specific.F90: New file.
gcc/testsuite/
PR fortran/20178
* gfortran.dg/f2c_1.f90, gfortran.dg/f2c_2.f90,
gfortran.dg/f2c_3.f90: New tests.
s390.c: (s390_const_double_ok_for_constraint_p): New function.
2005-05-10 Adrian Straetling <straetling@de.ibm.com>
* config/s390/s390.c: (s390_const_double_ok_for_constraint_p): New
function.
(legitimate_reload_constant_p): Add check for const double zero.
* config/s390/s390.md: Add comment for constraint letter 'G'.
("*movdf_64", "*movdf_31", "movsf"): Add constraint and proper
attributes for new case.
* config/s390/s390.h: (CONST_DOUBLE_OK_FOR_CONSTRAINT_P): Define it as
s390_const_double_ok_for_constraint_p.
* config/s390/s390-protos.h (s390_const_double_ok_for_constraint_p):
Add prototype.
Zdenek Dvorak [Tue, 10 May 2005 20:04:27 +0000 (22:04 +0200)]
tree-ssa-loop-im.c: Include hashtab.h.
* tree-ssa-loop-im.c: Include hashtab.h.
(struct mem_ref_loc): New.
(struct mem_ref): Describe the set of references with the same
shape.
(max_stmt_uid, get_stmt_uid, record_mem_ref, free_mem_refs,
maybe_queue_var, fem_single_reachable_address,
for_each_memref, single_reachable_address,
is_call_clobbered_ref, determine_lsm_reg): Removed.
(record_mem_ref_loc, free_mem_ref_locs, determine_lsm_ref,
hoist_memory_reference, memref_hash, memref_eq, memref_del,
gather_mem_refs_stmt, gather_mem_refs, find_more_ref_vops):
New functions.
(rewrite_mem_refs, schedule_sm): Use mem_ref_loc list.
(determine_lsm_loop): Rewritten.
(determine_lsm): Do not set stmt uids.
projects / thirdparty / gcc.git / log
