Thomas Koenig [Sat, 26 Jan 2008 15:22:59 +0000 (15:22 +0000)]
re PR libfortran/34887 (reverse tabbing before slash descriptor (regression vs. g77))
2008-01-26 Thomas Koenig <tkoenig@gcc.gnu.org>
PR libfofortran/34887
* io/transfer.c (next_record_w): Always move to the farthest
position when completing the record (also when we are
processing a slash edit descriptor).
2008-01-26 Thomas Koenig <tkoenig@gcc.gnu.org>
PR libfofortran/34887
* gfortran.dg/x_slash_2.f: New test.
re PR rtl-optimization/34959 (23_containers/bitset/ext/15361.cc fails for MIPS n64)
gcc/
PR rtl-optimization/34959
* optabs.c (expand_unop): In libcall notes, give ffs, clz, ctz,
popcount and parity rtxes the same mode as their operand.
Truncate or extend the result to the return value's mode
if necessary.
re PR target/34981 (Lazily-bound function called twice)
gcc/
PR target/34981
* config/mips/mips-protos.h (mips_expand_call): Return an rtx.
* config/mips/mips.h (FIRST_PSEUDO_REGISTER): Rename FAKE_CALL_REGNO
to GOT_VERSION_REGNUM.
(CALL_REALLY_USED_REGISTERS): Set the GOT_VERSION_REGNUM entry to 0.
(EPILOGUE_USES): Include GOT_VERSION_REGNUM if TARGET_USE_GOT.
* config/mips/mips.c (mips_emit_call_insn): New function.
(mips_call_tls_get_addr): Call mips_expand_call directly.
(mips16_copy_fpr_return_value): Use mips_emit_call_insn rather than
emit_call_insn.
(mips16_build_call_stub): Likewise. Return the call insn or null.
(mips_expand_call): Update the call to mips16_build_call_stub
accordingly and a remove redundant condition. Assert that MIPS16
stubs do not use lazy binding. Use mips_emit_call_insn and return
the call insn.
(mips_extra_live_on_entry): Include GOT_VERSION_REGNUM if
TARGET_USE_GOT.
(mips_hard_regno_mode_ok_p): Allow SImode for GOT_VERSION_REGNUM.
(mips_avoid_hazard): Remove hazard_set handling.
* config/mips/mips.md (UNSPEC_EH_RECEIVER): Rename to...
(UNSPEC_RESTORE_GP): ...this.
(UNSPEC_SET_GOT_VERSION, UNSPEC_UPDATE_GOT_VERSION): New constants.
(FAKE_CALL_REGNO): Rename to...
(GOT_VERSION_REGNUM): ...this.
(type): Add "ghost" value. Add an associated insn reservation.
(hazard_set): Remove.
(exception_receiver): Rename to...
(restore_gp): ...this and update the unspec identifier accordingly.
(exception_receiver, nonlocal_got_receiver): New expanders.
(load_call<mode>): Use GOT_VERSION_REGNUM. Don't set
FAKE_CALL_REGNO. Remove hazard_set attribute.
(set_got_version, update_got_version): New patterns.
gcc/testsuite/
PR target/34981
* gcc.target/mips/lazy-binding-1.c: New test.
* gcc.target/mips/mips.exp (setup_mips_tests): Set
mips_forced_no_abicalls and mips_forced_no_shared.
(dg-mips-options): Avoid using -mabicalls with an implicit -mabi=eabi.
Avoid using small data with -mabicalls. Don't make -G0 force
-mn-abicalls. Skip -mabicalls and -mshared tests if the multilib
forces the opposite option.
Jerry DeLisle [Fri, 25 Jan 2008 23:34:53 +0000 (23:34 +0000)]
re PR fortran/34876 (Can't read/write array sections with negative stride not specified)
2008-01-25 Jerry DeLisle <jvdelisle@gcc.gnu.org>
PR libfortran/34876
* io/transfer.c (write_buf): Handle case of zero sized array.
(transfer_array): Set data pointer to NULL and size to zero. Then
make a data transfer and return.
Richard Guenther [Fri, 25 Jan 2008 21:20:00 +0000 (21:20 +0000)]
re PR c++/33887 (Reference to bitfield gets wrong value when optimizing)
2008-01-25 Richard Guenther <rguenther@suse.de>
PR c++/33887
* decl.c (record_builtin_java_type): Make __java_boolean
a variant of bool.
* typeck.c (structural_comptypes): Move TYPE_FOR_JAVA check
after TYPE_MAIN_VARIANT check.
Jason Merrill [Fri, 25 Jan 2008 19:45:11 +0000 (14:45 -0500)]
re PR c++/31780 (ICE with incompatible types for ?: with "complex type" conversion)
PR c++/31780
* call.c (standard_conversion): Allow conversion from integer/real
to complex.
(compare_ics): Such a conversion is worse than a normal arithmetic
conversion.
Co-Authored-By: Mark Mitchell <mark@codesourcery.com>
From-SVN: r131832
Richard Guenther [Fri, 25 Jan 2008 12:08:52 +0000 (12:08 +0000)]
re PR tree-optimization/34966 (ICE: verify_ssa fails when optimization trigonometric code)
2008-01-25 Richard Guenther <rguenther@suse.de>
PR tree-optimization/34966
* tree-ssa-math-opts.c (execute_cse_sincos_1): For all but
default defs and PHI_NODEs we have to insert after the
defining statement.
Nick Clifton [Fri, 25 Jan 2008 11:23:44 +0000 (11:23 +0000)]
stormy16-lib2.c (MIN_UNITS_PER_WORD): Provide a default definition.
* config/stormy16/stormy16-lib2.c (MIN_UNITS_PER_WORD):
Provide a default definition.
(LIBGCC2_UNITS_PER_WORD): Likewise.
* config/stormy16/stormy16.c: Include df.h for the prototype
for df_regs_ever_live_p.
(xstormy16_expand_builtin_va_start): Convert the stack offset
into a component_ref and then use POINTER_PLUS_EXPR to add it
to the incoming_virtual_args_rtx.
(xstormy16_gimplify_va_arg_expr): Rename to
xstormy16_gimplify_va_arg_expr.
Use POINTER_PLUS_EXPR when performing pointer arithmetic.
(TARGET_GIMPLIFY_VA_ARG_EXPR): Use renamed
xstormy16_gimplify_va_arg_expr.
Fix up some formatting issues.
* config/stormy16/stormy16.c: (xstormy16_carry_plus_operand):
Move to predicates.md.
(xs_hi_general_operand): Likewise.
(xs_hi_nonmemory_operand): Likewise.
* config/stormy16/predicates.md:
(xstormy16_carry_plus_operand): New predicate.
(xs_hi_general_operand): New predicate.
(xs_hi_nonmemory_operand): New predicate.
* config/stormy16/stormy16-protos.h:
(xstormy16_carry_plus_operand): Delete prototype.
(xs_hi_general_operand): Likewise.
(xs_hi_nonmemory_operand): Likewise.
* config/storm16/stormy16.md (addhi3): Remove earlyclobber
modifiers as they are no longer needed and they can trigger
reload spill failures.
* config/storm16/stormy16.md (ineqbranchsi): Replace match_dup
with a match_operand in order to help reload.
* config/storm16/stormy16.md (movhi_internal): Replace 'r'
constraint with 'e' for the 8th alternative as this version of
the mov.w instruction only accepts the lower 8 registers.
Jakub Jelinek [Fri, 25 Jan 2008 09:01:27 +0000 (10:01 +0100)]
re PR preprocessor/34692 (Internal error with pragma in macro)
PR preprocessor/34692
* macro.c (collect_args): Add pragma_buff argument. Push
CPP_PRAGMA ... CPP_PRAGMA_EOL tokens to *pragma_buff, rather
than into arguments. Reset prevent_expansion and parsing_args
state at CPP_PRAGMA_EOL/CPP_EOF.
(funlike_invocation_p): Add pragma_buff argument, pass it through
to collect_args.
(enter_macro_context): Add result argument. Adjust
funlike_invocation_p caller. Emit all deferred pragma tokens
gathered during collect_args before the expansion, add a padding
token. Return 2 instead of 1 if any pragma tokens were prepended.
(cpp_get_token): If enter_macro_context returns 2, don't return
a padding token, instead cycle to grab CPP_PRAGMA token.
* directives.c (_cpp_handle_directive): If was_parsing_args
in deferred pragma, leave parsing_args and prevent_expansion as is.
* gcc.dg/cpp/pr34692.c: New test.
* gcc.dg/gomp/pr34692.c: New test.
ipa-struct-reorg.c (remove_str_allocs_in_func, [...]): New functions.
2008-01-25 Golovanevsky Olga <olga@il.ibm.com>
* ipa-struct-reorg.c (remove_str_allocs_in_func, remove_str_allocs):
New functions.
(remove_structure): Update allocations list before removing structure.
* ipa-struct-reorg.c (is_safe_cond_expr,
create_new_stmts_for_cond_expr): Use integer_zerop function,
that recognize not only zero-pointer, but zero-integer too.
2008-01-24 Daniel Franke <franke.daniel@gmail.com>
* gfortran.texi (OpenMP): Extended existing documentation.
(contributors): Added major contributors of 2008 that were
not listed yet.
(proposed extensions): Removed implemented items.
re PR tree-optimization/34472 (gcc.dg/struct/wo_prof_malloc_size_var.c doesn't work)
gcc/
PR tree-optimization/34472
* ipa-struct-reorg.c (safe_cond_expr_check): Change the DATA
parameter to a "bool *" and set *DATA to false if there is
an unsafe access. Do not delete the structure here.
(check_cond_exprs): Delete it here instead.
(check_cond_exprs, exclude_cold_structs): Do not increase
I when removing a structure.
Paul Thomas [Thu, 24 Jan 2008 08:40:38 +0000 (08:40 +0000)]
re PR fortran/34872 (Spurious error in snapshot of 01/18/08: Statement at (1) is not a valid branch target statement for the branch statement at (2))
2008-01-24 Paul Thomas <pault@gcc.gnu.org>
PR fortran/34872
* parse.c (next_statement) : If ST_GET_FCN_CHARACTERISTICS is
seen, check for a statement label and, if present, delete it
and set the locus to the start of the statement.
2008-01-24 Paul Thomas <pault@gcc.gnu.org>
PR fortran/34872
* gfortran.dg/function_charlen_3.f: New test.
David Daney [Wed, 23 Jan 2008 23:00:39 +0000 (23:00 +0000)]
2008-01-23 David Daney <ddaney@avtrex.com>
* class.c (hide) Rename to...
(java_hide_decl) ... this throughout, and make public.
* resource.c (Jr_count): Remove.
(compile_resource_data): Call java_mangle_resource_name to generate
decl name. Make resource decl public and hidden.
* mangle.c (java_mangle_resource_name): New function.
* java-tree.h (java_hide_decl, java_mangle_resource_name): Declare
functions.
* cplus-dem.c (demangle_function_name): Changed to return value
indicating if a name was correctly demangled.
(iterate_demangle_function): Use demangle_function_name return
value.
Bernd Schmidt [Wed, 23 Jan 2008 02:17:00 +0000 (02:17 +0000)]
bfin-protos.h (WA_RETS, [...]): New macros.
* config/bfin/bfin-protos.h (WA_RETS, ENABLE_WA_RETS): New macros.
* config/bfin/bfin.c (bfin_cpus): Add WA_RETS everywhere.
(cputype_selected): New static variable.
(bfin_handle_option): Set it if -mcpu is used.
(override_option): Select default set of workarounds if no cpu type
selected on the command line.
(workaround_rts_anomaly): Only run if ENABLE_WA_RETS.
Jakub Jelinek [Wed, 23 Jan 2008 01:50:45 +0000 (02:50 +0100)]
re PR c++/33984 (bit-fields, references and overloads)
gcc/cp/ChangeLog:
PR c++/33984
* call.c (reference_binding): For bitfields use the declared bitfield
type.
(add_builtin_candidates): Likewise.
* class.c (layout_class_type): For bitfields copy over the
original type quals.
gcc/testsuite/ChangeLog:
PR c++/33984
* g++.dg/conversion/bitfield7.C: New test.
* g++.dg/cpp0x/decltype4.C: Fixed xfail.
Bernd Schmidt [Tue, 22 Jan 2008 23:55:43 +0000 (23:55 +0000)]
gcc/
From Michael Frysinger <michael.frysinger@analog.com>
* config/bfin/bfin-protos.h (enum bfin_cpu_type): Add
BFIN_CPU_BF547, BFIN_CPU_BF523, BFIN_CPU_BF524, and BFIN_CPU_BF526.
* config/bfin/elf.h (LIB_SPEC): Use proper linker script
for bf547, bf523, bf524, and bf526.
* config/bfin/bfin.c (bfin_cpus[]): Add bf547, bf523, bf524, and
bf526.
* config/bfin/bfin.h (TARGET_CPU_CPP_BUILTINS): Define
__ADSPBF523__ for bf523, __ADSPBF524__ for bf524,
__ADSPBF526__ for bf526, __ADSPBF52x__ for all three, as well as
__ADSPBF547__ and __ADSPBF54x__ for bf547.
* doc/invoke.texi (Blackfin Options): Document that
-mcpu now accept bf547, bf523, bf524, and bf526.
gcc/testsuite/
From Michael Frysinger <michael.frysinger@analog.com>
* gcc.target/bfin/mcpu-bf547.c: New test.
* gcc.target/bfin/mcpu-bf523.c: New test.
* gcc.target/bfin/mcpu-bf524.c: New test.
* gcc.target/bfin/mcpu-bf526.c: New test.
Eric Botcazou [Tue, 22 Jan 2008 22:27:47 +0000 (22:27 +0000)]
re PR rtl-optimization/34628 (problems with inlining on ARM)
PR rtl-optimization/34628
* combine.c (try_combine): Stop and undo after the first combination
if an autoincrement side-effect on the first insn has effectively
been lost.
Paul Thomas [Tue, 22 Jan 2008 21:22:13 +0000 (21:22 +0000)]
re PR fortran/34875 (read into vector-valued section doesn't transfer any values)
2008-01-22 Paul Thomas <pault@gcc.gnu.org>
PR fortran/34875
* trans-io.c (gfc_trans_transfer): If the array reference in a
read has a vector subscript, use gfc_conv_subref_array_arg to
copy back the temporary.
2008-01-22 Paul Thomas <pault@gcc.gnu.org>
PR fortran/34875
* gfortran.dg/vector_subscript_3.f90: New test.
Jason Merrill [Tue, 22 Jan 2008 19:50:37 +0000 (14:50 -0500)]
re PR c++/34912 (ICE with friend in local class)
PR c++/34912
* friend.c (do_friend): Check for prior declaration of a friend
function of a local class.
* name-lookup.c (lookup_name_innermost_nonclass_level):
No longer static.
* name-lookup.h: Declare it.
Eric Botcazou [Tue, 22 Jan 2008 18:15:49 +0000 (18:15 +0000)]
decl.c (components_to_record): Do not reuse the empty union type if there is a representation clause on the record.
* decl.c (components_to_record): Do not reuse the empty union type
if there is a representation clause on the record.
* trans.c (addressable_p): Return true for INTEGER_CST.
Jakub Jelinek [Tue, 22 Jan 2008 17:03:12 +0000 (18:03 +0100)]
re PR c++/34914 (Member pointer to vector rejected)
PR c++/34914
* c-common.c (handle_vector_size_attribute): Only allow
integral, scalar float and fixed point types. Handle OFFSET_TYPE
the same way as pointer, array etc. types.
* tree.c (reconstruct_complex_type): Handle OFFSET_TYPE.
projects / thirdparty / gcc.git / log
