gcc.git
19 years agoc-typeck.c (build_function_call): Call fold_buildN_initializer or fold_buildN instead...
James A. Morrison [Wed, 17 Aug 2005 04:00:21 +0000 (04:00 +0000)]
c-typeck.c (build_function_call): Call fold_buildN_initializer or fold_buildN instead of buildN then...

2005-08-16  James A. Morrison  <phython@gcc.gnu.org>

        * c-typeck.c (build_function_call): Call fold_buildN_initializer or
        fold_buildN instead of buildN then fold_initializer or fold.
        (build_unary_op): Likewise.
        (build_binary_op): Likewise.
        * fold-const.c (fold_initializer): Remove.
        (fold_build1_initializer): New function.
        (fold_build2_initializer): New function.
        (fold_build3_initializer): New function.
        * tree.h (fold_initializer): Remove.
        (fold_build1_initializer): New function.
        (fold_build2_initializer): New function.
        (fold_build3_initializer): New function.

From-SVN: r103201

19 years agofold-const.c (optimize_bit_field_compare): Remove extra fold call.
James A. Morrison [Wed, 17 Aug 2005 03:56:20 +0000 (03:56 +0000)]
fold-const.c (optimize_bit_field_compare): Remove extra fold call.

2005-08-16  James A. Morrison  <phython@gcc.gnu.org>

        * fold-const.c (optimize_bit_field_compare): Remove extra fold call.
        (try_move_mult_to_index): Call fold_build2 instead of build2.
        (fold_binary): Don't call fold after calls to try_move_mult_to_index.
        * tree-ssa-loop-niter.c (inverse): Call int_const_binop instead of
        fold_binary_to_constant.
        (infer_loop_bounds_from_undefined): Call fold_build2 instead of
        fold (build.
        * tree-data-ref.c (tree_fold_divides_p): Use tree_int_cst_equal to
        check if A == gcd (A, B).  Remove TYPE argument.
        (analyze_offset) Use fold_build2 instead of fold (build.
        (create_data_ref): Likewise.
        (analyze_siv_subscript_cst_affine): Update calls to tree_fold_divides_p.        * tree-ssa-ccp.c (widen_bitfield): Call fold_build2 instead of build2
        then fold.

From-SVN: r103200

19 years agostrverscmp.c: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 03:31:04 +0000 (03:31 +0000)]
strverscmp.c: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* strverscmp.c: Update FSF address.
* testsuite/Makefile.in: Likewise.
* testsuite/test-demangle.c: Likewise.
* testsuite/test-pexecute.c: Likewise.

From-SVN: r103199

19 years agoALL: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 03:26:45 +0000 (03:26 +0000)]
ALL: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* ALL: Update FSF address.

From-SVN: r103198

19 years agoAll files: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 03:21:42 +0000 (03:21 +0000)]
All files: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* All files: Update FSF address.

From-SVN: r103197

19 years agocrtfastmath.c, [...]: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 03:07:06 +0000 (03:07 +0000)]
crtfastmath.c, [...]: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* config/arm/unaligned-funcs.c,config/i386/crtfastmath.c,
ipa-cp.c,ipa-prop.c,ipa-prop.h: Update FSF address.

From-SVN: r103196

19 years agoFix stupidity.
James E Wilson [Wed, 17 Aug 2005 03:00:13 +0000 (20:00 -0700)]
Fix stupidity.

* gcc.dg/large-size-array.c (DIM): Use USHRT_MAX not USHORT_MAX.

From-SVN: r103195

19 years agoAll files: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 02:49:08 +0000 (02:49 +0000)]
All files: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* All files: Update FSF address.

From-SVN: r103194

19 years agoAll files: Update FSF address.
Kelley Cook [Wed, 17 Aug 2005 02:28:44 +0000 (02:28 +0000)]
All files: Update FSF address.

2005-08-17  Kelley Cook  <kcook@gcc.gnu.org>

* All files: Update FSF address.

From-SVN: r103192

19 years agore PR tree-optimization/23402 (error: statement makes a memory store, but has no...
Andrew Pinski [Wed, 17 Aug 2005 01:55:05 +0000 (01:55 +0000)]
re PR tree-optimization/23402 (error: statement makes a memory store, but has no V_MAY_DEFS nor V_MUST_DEFS)

2005-08-16  Andrew Pinski  <pinskia@physics.uc.edu>

        PR tree-opt/23402
        * gcc.c-torture/compile/zero-strct-3.c: New test.
        * gcc.c-torture/compile/zero-strct-4.c: New test.

2005-08-16  Andrew Pinski  <pinskia@physics.uc.edu>

        PR tree-opt/23402
        * gimplify.c (zero_sized_type): New function.
        (gimplify_modify_expr_rhs): If we have a zero sized type,
        replace the statement with an empty statement.

From-SVN: r103191

19 years agocrtfastmath.c (set_fast_math): Add "=m" for fxsave.
H.J. Lu [Wed, 17 Aug 2005 00:32:17 +0000 (00:32 +0000)]
crtfastmath.c (set_fast_math): Add "=m" for fxsave.

2005-08-16  H.J. Lu  <hongjiu.lu@intel.com>

* config/i386/crtfastmath.c (set_fast_math): Add "=m" for
fxsave.

From-SVN: r103189

19 years agoDaily bump.
GCC Administrator [Wed, 17 Aug 2005 00:17:02 +0000 (00:17 +0000)]
Daily bump.

[[Split portion of a mixed commit.]]

From-SVN: r103186.2

19 years ago* doc/tm.texi (Label Output): Correct typo.
Ian Lance Taylor [Tue, 16 Aug 2005 22:29:09 +0000 (22:29 +0000)]
* doc/tm.texi (Label Output): Correct typo.

From-SVN: r103182

19 years agore PR target/23376 (ICE on GCC 4.x with -O1 -funroll-loops -fvariable-expansion-in...
Steven Bosscher [Tue, 16 Aug 2005 22:24:30 +0000 (22:24 +0000)]
re PR target/23376 (ICE on GCC 4.x with -O1 -funroll-loops -fvariable-expansion-in-unroller)

PR target/23376
* loop-unroll.c (analyze_insn_to_expand_var): Make sure that
force_operand will work later on using have_insn_for.

From-SVN: r103181

19 years ago* fold-const.c (tree_expr_nonnegative_p): Regroup cases.
Volker Reichelt [Tue, 16 Aug 2005 22:13:37 +0000 (22:13 +0000)]
* fold-const.c (tree_expr_nonnegative_p): Regroup cases.

From-SVN: r103178

19 years agore PR c++/23337 (ICE initializing array of vectors in C++)
Ian Lance Taylor [Tue, 16 Aug 2005 22:07:44 +0000 (22:07 +0000)]
re PR c++/23337 (ICE initializing array of vectors in C++)

./ PR c++/23337
* gimplify.c (gimplify_init_ctor_eval): If we see an element of
vector type, don't try to construct it element by element.  Add an
assertion that we use a FIELD_DECL when building a COMPONENT_REF.
testsuite/
PR c++/23337
* g++.dg/ext/vector2.C: New.

From-SVN: r103177

19 years ago* fold-const.c (fold_ternary): Simplify folding of a CALL_EXPR.
Volker Reichelt [Tue, 16 Aug 2005 22:05:14 +0000 (22:05 +0000)]
* fold-const.c (fold_ternary): Simplify folding of a CALL_EXPR.

From-SVN: r103176

19 years agore PR libfortran/23428 (inquire(iolength= ) for complex give incorrect value)
Thomas Koenig [Tue, 16 Aug 2005 21:30:32 +0000 (21:30 +0000)]
re PR libfortran/23428 (inquire(iolength= ) for complex give incorrect value)

2005-08-16  Thomas Koenig  <Thomas.Koenig@online.de>

PR libfortran/23428
* gfortran.dg/inquire-complex.f90:  Correct mangled testcase.

From-SVN: r103173

19 years agore PR libfortran/23428 (inquire(iolength= ) for complex give incorrect value)
Thomas Koenig [Tue, 16 Aug 2005 20:26:04 +0000 (20:26 +0000)]
re PR libfortran/23428 (inquire(iolength= ) for complex give incorrect value)

2005-08-16  Thomas Koenig  <Thomas.Koenig@online.de>

PR libfortran/23428
* io/transfer.c (iolength_transfer):  Remove __attribute__ ((unused))
from type.  Return correct length for inquire(iolength=)
for complex variables.

2005-08-16  Thomas Koenig  <Thomas.Koenig@online.de>

PR libfortran/23428
* gfortran.dg/inquire-complex.f90:  New test case.

From-SVN: r103168

19 years agore PR java/22113 (Buffer overflow in the lexical analyser while reading FP literals)
Ranjit Mathew [Tue, 16 Aug 2005 18:46:18 +0000 (18:46 +0000)]
re PR java/22113 (Buffer overflow in the lexical analyser while reading FP literals)

PR java/22113
* lex.c (do_java_lex): Define MAX_TOKEN_LEN. Avoid overflowing
`literal_token' for large numeric input tokens.

From-SVN: r103167

19 years agoinvoke.texi (Precompiled Headers): Document some more options which are known to...
Geoffrey Keating [Tue, 16 Aug 2005 18:43:10 +0000 (18:43 +0000)]
invoke.texi (Precompiled Headers): Document some more options which are known to be safe.

* doc/invoke.texi (Precompiled Headers): Document some more options
which are known to be safe.

From-SVN: r103166

19 years agoMore testsuite adjustments for PR java/19870.
Ranjit Mathew [Tue, 16 Aug 2005 18:32:14 +0000 (18:32 +0000)]
More testsuite adjustments for PR java/19870.

* testsuite/libjava.lang/PR19870_2.java: New testcase.
* testsuite/libjava.lang/PR19870_2.out: Expected output for the
new testcase.
* testsuite/libjava.jacks/jacks.xfail: Remove
8.5.2-non-static-member-usage-2 and add 15.12.3-explicit-constructor-9.

From-SVN: r103165

19 years agoEmit an error for too large arrays instead of an ICE.
James E Wilson [Tue, 16 Aug 2005 18:23:58 +0000 (11:23 -0700)]
Emit an error for too large arrays instead of an ICE.

PR tree-optimization/21105
* c-decl.c (grokdeclarator): Use TYPE_SIZE_UNIT not TYPE_SIZE in
TREE_OVERFLOW check.
* gcc.dg/large-size-array.c: New.

From-SVN: r103164

19 years agore PR java/19870 (gcj -C doesn't generate accessors for private members across nested...
Ranjit Mathew [Tue, 16 Aug 2005 18:22:31 +0000 (18:22 +0000)]
re PR java/19870 (gcj -C doesn't generate accessors for private members across nested class boundaries)

PR java/19870
* parse.y (nested_field_access_p): Rename to nested_member_access_p
and expand to handle method accesses across nested classes.
(build_outer_method_access_method): Rename to
build_nested_method_access_method.  Minor adjustments to comments.
(resolve_expression_name): Use the newly-renamed
nested_member_access_p method.
(resolve_qualified_expression_name): Likewise.
(patch_method_invocation): Also consider static methods for access
method generation.  Minor adjustments to comments.
(maybe_use_access_method): Use the more general
nested_memeber_access_p to determine access across nested class
boundaries.  Allow THIS_ARG to be NULL (for static methods).

From-SVN: r103163

19 years ago* gnu/xml/xpath/XPathParser.y: Removed.
Tom Tromey [Tue, 16 Aug 2005 17:22:57 +0000 (17:22 +0000)]
* gnu/xml/xpath/XPathParser.y: Removed.

From-SVN: r103162

19 years agore PR java/23300 (DECL_FIELD_OFFSET == 0 versus build_field_ref)
Tom Tromey [Tue, 16 Aug 2005 16:06:43 +0000 (16:06 +0000)]
re PR java/23300 (DECL_FIELD_OFFSET == 0 versus build_field_ref)

PR java/23300.
* expr.c (build_field_ref): Don't generate otable reference when
DECL_FIELD_OFFSET is 0.
* class.c (maybe_layout_super_class): Pass outer class to
do_resolve_class.

From-SVN: r103160

19 years agors6000.md (ltu<mode>): Convert to mode macro.
David Edelsohn [Tue, 16 Aug 2005 15:13:40 +0000 (15:13 +0000)]
rs6000.md (ltu<mode>): Convert to mode macro.

        * config/rs6000/rs6000.md (ltu<mode>): Convert to mode macro.
        (neg_ltu<mode>): Same.
        (gtu<mode>): Same.
        (neg_gtu<mode>): Same.

From-SVN: r103159

19 years agovect-40: Use aligned arrays instead of arrays to aligned type.
Dorit Nuzman [Tue, 16 Aug 2005 15:02:39 +0000 (15:02 +0000)]
vect-40: Use aligned arrays instead of arrays to aligned type.

        * gcc.dg/vect/vect-40: Use aligned arrays instead of arrays to aligned
        type.
        * gcc.dg/vect/vect-41: Likewise.
        * gcc.dg/vect/vect-42: Likewise.
        * gcc.dg/vect/vect-43: Likewise.
        * gcc.dg/vect/vect-44: Likewise.
        * gcc.dg/vect/vect-46: Likewise.
        * gcc.dg/vect/vect-47: Likewise.
        * gcc.dg/vect/vect-48: Likewise.
        * gcc.dg/vect/vect-52: Likewise.
        * gcc.dg/vect/vect-53: Likewise.
        * gcc.dg/vect/vect-54: Likewise.
        * gcc.dg/vect/vect-55: Likewise.
        * gcc.dg/vect/vect-56: Likewise.
        * gcc.dg/vect/vect-57: Likewise.
        * gcc.dg/vect/vect-58: Likewise.
        * gcc.dg/vect/vect-59: Likewise.
        * gcc.dg/vect/vect-60: Likewise.
        * gcc.dg/vect/vect-61: Likewise.
        * gcc.dg/vect/vect-85: Likewise.
        * gcc.dg/vect/vect-87: Likewise.
        * gcc.dg/vect/vect-88: Likewise.
        * gcc.dg/vect/vect-93.c: Likewise.
        * gcc.dg/vect/vect-74: Likewise, and also added documentation.
        * gcc.dg/vect/vect-75: Likewise.
        * gcc.dg/vect/vect-76: Likewise.
        * gcc.dg/vect/vect-77: Likewise.
        * gcc.dg/vect/vect-78: Likewise.
        * gcc.dg/vect/vect-80: Likewise.
        * gcc.dg/vect/vect-35: Likewise, and also check that the test wasn't
        vectorized for the expected reason, rather than checking how alignment
        was handled.
        * gcc.dg/vect/vect-92.c: Use aligned arrays instead of arrays to aligned
        type and also instead of aligned pointers. Added documentation.

From-SVN: r103158

19 years ago[multiple changes]
Feng Wang [Tue, 16 Aug 2005 12:58:46 +0000 (12:58 +0000)]
[multiple changes]

2004-08-16  Huang Chun  <chunhuang73@hotmail.com>

* trans-expr.c (gfc_conv_power_op): Evaluate the expression before
expand.

2005-08-16  Feng Wang  <fengwang@nudt.edu.cn>

* gfortran.dg/power.f90: New test.

From-SVN: r103155

19 years agobuiltins.c (expand_builtin_strcat): Remove superfluous call to fold.
Volker Reichelt [Tue, 16 Aug 2005 12:14:10 +0000 (12:14 +0000)]
builtins.c (expand_builtin_strcat): Remove superfluous call to fold.

* builtins.c (expand_builtin_strcat): Remove superfluous call to fold.
(fold_builtin_isdigit): Use fold_buildN.
(build_function_call_expr): Likewise.
* c-typeck.c (c_finish_loop): Likewise.

From-SVN: r103154

19 years agore PR middle-end/20396 (TRULY_NOOP_TRUNCATION ignored)
J"orn Rennecke [Tue, 16 Aug 2005 11:57:07 +0000 (11:57 +0000)]
re PR middle-end/20396 (TRULY_NOOP_TRUNCATION ignored)

2005-08-16  J"orn Rennecke <joern.rennecke@st.com>
            Richard Shann <rshann@superh.com>

PR middle-end/20396:
* optabs.c (expand_binop): Take TRULY_NOOP_TRUNCATION into account.

Co-Authored-By: Richard Shann <rshann@superh.com>
From-SVN: r103153

19 years ago* tree-ssa-loop-niter.c (scev_probably_wraps_p): Reword a comment.
Sebastian Pop [Tue, 16 Aug 2005 09:41:33 +0000 (11:41 +0200)]
* tree-ssa-loop-niter.c (scev_probably_wraps_p): Reword a comment.

From-SVN: r103152

19 years agore PR target/23355 (size optimizer did not eliminateing useless Push and pop instruct...
Richard Earnshaw [Tue, 16 Aug 2005 09:27:52 +0000 (09:27 +0000)]
re PR target/23355 (size optimizer did not eliminateing useless Push and pop instructions at ARM/Thumb machine)

PR target/23355
* arm.c (thumb_compute_save_reg_mask): Use similar logic to
arm_compure_save_reg0_reg12_mask to determine when the PIC register
must be saved.

From-SVN: r103151

19 years agore PR rtl-optimization/23299 ("classic" GCSE caused miscompilation)
Jakub Jelinek [Tue, 16 Aug 2005 08:38:15 +0000 (10:38 +0200)]
re PR rtl-optimization/23299 ("classic" GCSE caused miscompilation)

PR rtl-optimization/23299
* g++.dg/opt/pr23299.C: New test.

From-SVN: r103148

19 years agore PR middle-end/23369 (build_range_check generates wrong code for funcptr comparison)
John David Anglin [Tue, 16 Aug 2005 02:20:40 +0000 (02:20 +0000)]
re PR middle-end/23369 (build_range_check generates wrong code for funcptr comparison)

PR middle-end/23369
* fold-const.c (build_range_check): Disable optimization for function
pointer expressions on targets that require function pointer
canonicalization.

From-SVN: r103145

19 years agosimplify-rtx.c (simplify_const_relational_operation): When extracting arguments of...
Ulrich Weigand [Tue, 16 Aug 2005 02:01:27 +0000 (02:01 +0000)]
simplify-rtx.c (simplify_const_relational_operation): When extracting arguments of a COMPARE, recompute the mode as well.

* simplify-rtx.c (simplify_const_relational_operation): When
extracting arguments of a COMPARE, recompute the mode as well.

From-SVN: r103143

19 years agojava-tree.h (LABEL_IN_SUBR): Removed.
Tom Tromey [Tue, 16 Aug 2005 00:50:36 +0000 (00:50 +0000)]
java-tree.h (LABEL_IN_SUBR): Removed.

* java-tree.h (LABEL_IN_SUBR): Removed.
(LABEL_IN_SUBR): Likewise.
(LABEL_IS_SUBR_START): Likewise.
(LABEL_SUBR_START): Likewise.
(LABEL_SUBR_CONTEXT): Likewise.
(LABEL_CHANGED): Likewise.
(LABEL_RETURN_LABEL): Likewise.
(LABEL_RETURN_TYPE_STATE): Likewise.
(LABEL_RETURN_LABELS): Likewise.
(RETURN_MAP_ADJUSTED): Likewise.
(LABEL_PENDING_CHAIN): Likewise.

From-SVN: r103142

19 years ago* tree.c (build_string): Mark tree CONSTANT and INVARIANT.
Ian Lance Taylor [Tue, 16 Aug 2005 00:35:50 +0000 (00:35 +0000)]
* tree.c (build_string): Mark tree CONSTANT and INVARIANT.

From-SVN: r103141

19 years agomov.md (movqi_op): Immediates can't be moved to the stack.
DJ Delorie [Tue, 16 Aug 2005 00:31:39 +0000 (20:31 -0400)]
mov.md (movqi_op): Immediates can't be moved to the stack.

* config/m32c/mov.md (movqi_op): Immediates can't be moved to
the stack.
(movsi_splittable): Allow, but split, moves to the stack.
* config/m32c/m32c.c (m32c_split_move): Always split moves to the
stack.

From-SVN: r103140

19 years agoDaily bump.
GCC Administrator [Tue, 16 Aug 2005 00:16:37 +0000 (00:16 +0000)]
Daily bump.

[[Split portion of a mixed commit.]]

From-SVN: r103137.2

19 years agoaclocal.m4 (gcc_AC_FUNC_PRINTF_PTR): Delete.
Kaveh R. Ghazi [Tue, 16 Aug 2005 00:13:53 +0000 (00:13 +0000)]
aclocal.m4 (gcc_AC_FUNC_PRINTF_PTR): Delete.

* aclocal.m4 (gcc_AC_FUNC_PRINTF_PTR): Delete.
* configure.ac: Don't call gcc_AC_FUNC_PRINTF_PTR.
* system.h (HOST_PTR_PRINTF): Don't define, poison it.

* bitmap.c, c-decl.c, config/i386/i386-interix.h,
config/iq2000/iq2000.c, mips-tfile.c, print-rtl.c, print-tree.c:
Delete HOST_PTR_PRINTF.

* configure, config.in: Regenerate.

cp:
* name-lookup.c, ptree.c: Delete HOST_PTR_PRINTF.

From-SVN: r103135

19 years ago* gcc.dg/special/weak-1.c: xfail 32-bit hppa*-*-hpux*.
John David Anglin [Mon, 15 Aug 2005 23:42:33 +0000 (23:42 +0000)]
* gcc.dg/special/weak-1.c: xfail 32-bit hppa*-*-hpux*.

From-SVN: r103133

19 years agoFix typo in alpha port found while grepping for ia64 info.
James E Wilson [Mon, 15 Aug 2005 23:38:44 +0000 (16:38 -0700)]
Fix typo in alpha port found while grepping for ia64 info.

* system-linux-alpha.ads: Change ia64 to alpha.

From-SVN: r103132

19 years agoMake-lang.in (JAVA_OBJS): Removed verify.o
Tom Tromey [Mon, 15 Aug 2005 21:26:35 +0000 (21:26 +0000)]
Make-lang.in (JAVA_OBJS): Removed verify.o

* Make-lang.in (JAVA_OBJS): Removed verify.o
(java/verify.o): Removed.
* verify.c: Removed.
* lang.c (flag_new_verifier): Removed.
(java_post_options): Updated.
* java-tree.h (flag_new_verifier): Removed.
(verify_jvm_instructions): Removed.
* expr.c (pop_type_0): Assume flag_new_verifier is true.
(build_java_check_indexed_type): Likewise.
(expand_java_arraystore): Likewise.
(expand_java_arrayload): Likewise.
(pop_arguments): Likewise.
(expand_byte_code): Likewise.
(process_jvm_instruction): Likewise.

From-SVN: r103126

19 years agoFix Infinite Stack Recursion Regression.
Fariborz Jahanian [Mon, 15 Aug 2005 20:38:23 +0000 (20:38 +0000)]
Fix Infinite Stack Recursion Regression.

        * cp-tree.h (can_convert_arg, fn_type_unification): New argument.
        * call.c (add_template_candidate_real): Pass down 'flags' to
        fn_type_unification.
        (can_convert_arg): New 'flags' argument. Pass it to call to
        implicit_conversion instead of LOOKUP_NORMAL.
        (can_convert): Add LOOKUP_NORMAL to call to can_convert_arg.
        * class.c (resolve_address_of_overloaded_function): Ditto.
        (resolve_address_of_overloaded_function): Ditto.
        * decl.c (reshape_init, check_default_argument): Ditto.
        * typeck.c (build_ptrmemfunc): Ditto.
        * pt.c (type_unification_real): Add 'flags' argument.
        (fn_type_unification): Pass 'flags' to type_unification_real.
        (type_unification_real): Pass new 'flags' argument to call to
        can_convert_arg.

Oked by Geoff Keating.

From-SVN: r103120

19 years agomf-hooks3.c (main_seen_p): Remove.
Ulrich Weigand [Mon, 15 Aug 2005 20:26:46 +0000 (20:26 +0000)]
mf-hooks3.c (main_seen_p): Remove.

* mf-hooks3.c (main_seen_p): Remove.
(__mf_get_state): Remove attempt to recognize the main thread.

From-SVN: r103119

19 years agors6000.md (QHSI): New mode macro.
David Edelsohn [Mon, 15 Aug 2005 17:58:58 +0000 (17:58 +0000)]
rs6000.md (QHSI): New mode macro.

* config/rs6000/rs6000.md (QHSI): New mode macro.
(wd): Extend mode attr for QImode and HImode
(dbits): New mode attr.
(zero_extend<mode>di2): Convert to mode macro.

From-SVN: r103118

19 years agoconfigure.ac: Test for the name of the symbol used for the entry point...
Maciej W. Rozycki [Mon, 15 Aug 2005 16:15:16 +0000 (16:15 +0000)]
configure.ac: Test for the name of the symbol used for the entry point...

* configure.ac: Test for the name of the symbol used for the entry
point; define ENTRY_POINT to the result.
* configure: Regenerate.
* config.h.in: Regenerate.
* mf-heuristics.c: Replace _start with ENTRY_POINT throughout.

From-SVN: r103113

19 years agore PR target/21841 (Documentation should say -mhp-ld/-mgnu-ld are only for hppa64...
Steve Ellcey [Mon, 15 Aug 2005 15:23:38 +0000 (15:23 +0000)]
re PR target/21841 (Documentation should say -mhp-ld/-mgnu-ld are only for hppa64-*-hpux*)

PR target/21841
* doc/invoke.texi (-mgnu-ld): Update description.
(-mhp-ld): Ditto.

From-SVN: r103112

19 years agoAdded missing ChangeLog entry.
Andreas Krebbel [Mon, 15 Aug 2005 14:41:21 +0000 (14:41 +0000)]
Added missing ChangeLog entry.

From-SVN: r103111

19 years agoMAINTAINERS (write after approval): Added myself.
Andreas Krebbel [Mon, 15 Aug 2005 14:38:41 +0000 (14:38 +0000)]
MAINTAINERS (write after approval): Added myself.

2005-08-15  Andreas Krebbel  <krebbel1@de.ibm.com>

* MAINTAINERS (write after approval): Added myself.

From-SVN: r103110

19 years agore PR tree-optimization/23391 (Tree checking failure due to scev)
Sebastian Pop [Mon, 15 Aug 2005 12:26:12 +0000 (14:26 +0200)]
re PR tree-optimization/23391 (Tree checking failure due to scev)

PR 23391
* Makefile.in (tree-chrec.o): Depends on real.h.
* tree-chrec.c: Include real.h.
(chrec_fold_plus_poly_poly, chrec_fold_multiply_poly_poly,
chrec_fold_plus_1): Use build_real for SCALAR_FLOAT_TYPE_P.
* tree-scalar-evolution.c (add_to_evolution_1,
interpret_rhs_modify_expr): Ditto.

From-SVN: r103109

19 years agore PR tree-optimization/23386 (bitmap.c is being miscompiled (VRP))
Sebastian Pop [Mon, 15 Aug 2005 07:51:42 +0000 (09:51 +0200)]
re PR tree-optimization/23386 (bitmap.c is being miscompiled (VRP))

PR 23386
  * tree-data-ref.c (estimate_niter_from_size_of_data): When
  step is negative compute the estimation from init downwards to
  zero.

* testsuite/gcc.dg/tree-ssa/pr23386.c: New.

From-SVN: r103106

19 years agofinally-1.m: Check that the catch and finally are being called.
Andrew Pinski [Mon, 15 Aug 2005 01:13:39 +0000 (01:13 +0000)]
finally-1.m: Check that the catch and finally are being called.

2005-08-14  Andrew Pinski  <pinskia@physics.uc.edu>

        * execute/exceptions/finally-1.m: Check that the catch and finally are
        being called.  Also add a "return 0;" at the end of main.

From-SVN: r103098

19 years agogcc_release: Update with FSF address.
Kelley Cook [Mon, 15 Aug 2005 01:01:02 +0000 (01:01 +0000)]
gcc_release: Update with FSF address.

2005-08-14  Kelley Cook  <kcook@gcc.gnu.org>

* gcc_release: Update with FSF address.

From-SVN: r103097

19 years agoAll files: Update with new FSF address.
Kelley Cook [Mon, 15 Aug 2005 00:50:43 +0000 (00:50 +0000)]
All files: Update with new FSF address.

2005-08-14  Kelley Cook  <kcook@gcc.gnu.org>

* All files: Update with new FSF address.

From-SVN: r103096

19 years agoAll files: Update with new FSF address.
Kelley Cook [Mon, 15 Aug 2005 00:41:31 +0000 (00:41 +0000)]
All files: Update with new FSF address.

2005-08-14  Kelley Cook  <kcook@gcc.gnu.org>

* All files: Update with new FSF address.

From-SVN: r103095

19 years agoDaily bump.
GCC Administrator [Mon, 15 Aug 2005 00:16:18 +0000 (00:16 +0000)]
Daily bump.

[[Split portion of a mixed commit.]]

From-SVN: r103092.2

19 years agoCommitted forgotten changelog:
Andrew Pinski [Sun, 14 Aug 2005 23:49:29 +0000 (16:49 -0700)]
Committed forgotten changelog:

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        * objc/execute/exceptions/exceptions.exp: New exp for the exceptions
        testsuite.
        * objc/execute/exceptions/trivial.m: New test.
        * objc.dg/try-catch-11.m: Move to ...
        * objc/execute/exceptions/foward-1.m: Here.
        * objc.dg/try-catch-2.m: Move to ...
        * objc/execute/exceptions/catchall-1.m: Here.
        * objc.dg/try-catch-8.m: Move to ...
        * objc/execute/exceptions/local-variables-1.m: Here.

From-SVN: r103090

19 years agofold-const (fold_binary): Call fold_build2 instead of fold (build.
James A. Morrison [Sun, 14 Aug 2005 23:19:05 +0000 (23:19 +0000)]
fold-const (fold_binary): Call fold_build2 instead of fold (build.

2005-08-14  James A. Morrison  <phython@gcc.gnu.org>

        * fold-const (fold_binary): Call fold_build2 instead of fold (build.

From-SVN: r103089

19 years agos390.c (s390_const_ok_for_constraint_p): Add 'P' constraint.
Ulrich Weigand [Sun, 14 Aug 2005 23:13:58 +0000 (23:13 +0000)]
s390.c (s390_const_ok_for_constraint_p): Add 'P' constraint.

* config/s390/s390.c (s390_const_ok_for_constraint_p): Add 'P'
constraint.
(legitimate_reload_constant_p): Fix handling of lliXX operands.
Accept double-word constants that can be split.
* config/s390/s390.md ("movti"): Use 'P' constraint.
("*movdi_31", "*movdf_31"): Likewise.

From-SVN: r103088

19 years agoparse.c (match): Enclosed macro in do...while(0) and braces.
Tobias Schlüter [Sun, 14 Aug 2005 21:45:03 +0000 (23:45 +0200)]
parse.c (match): Enclosed macro in do...while(0) and braces.

2005-08-14  Asher Langton  <langton2@llnl.gov>

        * parse.c (match): Enclosed macro in do...while(0) and braces.

Also fix formatting of a few previous ChangeLog entries

From-SVN: r103086

19 years agore PR fortran/21432 (gfortran does not support printing of namelists)
Paul Thomas [Sun, 14 Aug 2005 21:43:12 +0000 (21:43 +0000)]
re PR fortran/21432 (gfortran does not support printing of namelists)

2005-08-14 Paul Thomas  <pault@gcc.gnu.org>

PR fortran/21432.
* gfortran.texi: Document PRINT namelist.

From-SVN: r103085

19 years agomf-runtime.c (__mf_state_1): Initialize to reentrant.
Ulrich Weigand [Sun, 14 Aug 2005 19:38:34 +0000 (19:38 +0000)]
mf-runtime.c (__mf_state_1): Initialize to reentrant.

* mf-runtime.c (__mf_state_1): Initialize to reentrant.
(__mf_init): Set thread state active.
* mf-hooks3.c (__mf_pthread_spawner): Always set thread
state active.
(pthread_create wrapper): Always use thread spawner.

* testsuite/libmudflap.cth/pass37-frag.c: Increase timeout.
* testsuite/libmudflap.cth/pass39-frag.c: Likewise.

From-SVN: r103084

19 years agore PR tree-optimization/22615 (ICE in first_vi_for_offset, at tree-ssa-structalias...
Daniel Berlin [Sun, 14 Aug 2005 19:23:56 +0000 (19:23 +0000)]
re PR tree-optimization/22615 (ICE in first_vi_for_offset, at tree-ssa-structalias.c:2858)

2005-08-14  Daniel Berlin  <dberlin@dberlin.org>

Fix PR tree-optimization/22615

* tree-ssa-structalias.c (solution_set_add): Handle
first_vi_for_offset returning NULL.
(do_da_constraint): Ditto.
(do_sd_constraint): Ditto.
(do_ds_constraint): Ditto
(find_func_aliases): Ditto.
(build_constraint_graph): RHS is allowed be ANYTHING.
(first_vi_for_offset): Return NULL if we couldn't find anything at
the offset.

From-SVN: r103083

19 years agos390.c (s390_canonicalize_comparison): Prefer register over memory as first operand.
Ulrich Weigand [Sun, 14 Aug 2005 19:06:56 +0000 (19:06 +0000)]
s390.c (s390_canonicalize_comparison): Prefer register over memory as first operand.

* config/s390/s390.c (s390_canonicalize_comparison): Prefer register
over memory as first operand.

From-SVN: r103082

19 years agore PR fortran/21432 (gfortran does not support printing of namelists)
Paul Thomas [Sun, 14 Aug 2005 16:15:40 +0000 (16:15 +0000)]
re PR fortran/21432 (gfortran does not support printing of namelists)

2005-08-14 Paul Thomas  <pault@gcc.gnu.org>

PR fortran/21432.
* io.c (match_io): Add code to implement PRINT namelist.

2005-08-14 Paul Thomas  <pault@gcc.gnu.org>

PR fortran/21432.
* gfortran.dg/namelist_print_1.f: New test of functionality of
PRINT namelist.
* gfortran.dg/namelist_print_2.f: New test to check that PRINT
namelist generates error with -std=f95.

From-SVN: r103079

19 years agore PR target/23360 (-ffast-math startup broken on i686 (maybe Athlon-xp))
H.J. Lu [Sun, 14 Aug 2005 14:26:51 +0000 (14:26 +0000)]
re PR target/23360 (-ffast-math startup broken on i686 (maybe Athlon-xp))

2005-08-14  H.J. Lu  <hongjiu.lu@intel.com>

PR target/23360
* config/i386/crtfastmath.c (set_fast_math): Check if DAZ is
available for setting it.

From-SVN: r103078

19 years agore PR tree-optimization/23320 (ICE in in base_addr_differ_p, at tree-data-ref.c:430)
Ira Rosen [Sun, 14 Aug 2005 06:28:03 +0000 (06:28 +0000)]
re PR tree-optimization/23320 (ICE in in base_addr_differ_p, at tree-data-ref.c:430)

        PR tree-optimization/23320
        * tree-data-ref.c (base_addr_differ_p): Add comment. Check
        data-refs' types instead of base object nullness. Add check for
        pointer type data-refs before first location comparison. Remove
        assert.

From-SVN: r103076

19 years agore PR tree-optimization/23119 (gcc.dg/vect/vect-105.c scan-tree-dump-times vectorized...
Ira Rosen [Sun, 14 Aug 2005 06:13:30 +0000 (06:13 +0000)]
re PR tree-optimization/23119 (gcc.dg/vect/vect-105.c scan-tree-dump-times vectorized 1 loops 1 fails)

        PR tree-optimization/23119
        * gcc.dg/vect/vect-105.c: Add xfail for no_align targets.

From-SVN: r103075

19 years agotrans-stmt.c (gfc_trans_arithmetic_if): Optimized in case of equal labels.
Canqun Yang [Sun, 14 Aug 2005 02:06:06 +0000 (10:06 +0800)]
trans-stmt.c (gfc_trans_arithmetic_if): Optimized in case of equal labels.

* trans-stmt.c (gfc_trans_arithmetic_if): Optimized in case of equal
labels.

From-SVN: r103074

19 years agore PR libobjc/22492 ([PATCH] Abort after @finally: libobjc passing exception value...
Marcin Koziej [Sun, 14 Aug 2005 02:01:52 +0000 (04:01 +0200)]
re PR libobjc/22492 ([PATCH] Abort after @finally: libobjc passing exception value instead of exception.)

libobjc/ChangeLog:

2005-08-13  Marcin Koziej  <creep@desk.pl>
            Andrew Pinski  <pinskia@physics.uc.edu>

PR libobjc/22492
* exception.c (PERSONALITY_FUNCTION): Fix the PC with finally.

testsuite/ChangeLog:
2005-08-13  Marcin Koziej  <creep@desk.pl>
            Andrew Pinski  <pinskia@physics.uc.edu>

PR libobjc/22492
* execute/exceptions/finally-1.m: New test.

Co-Authored-By: Andrew Pinski <pinskia@physics.uc.edu>
From-SVN: r103073

19 years agonested-func-1.m: Move to ...
Andrew Pinski [Sun, 14 Aug 2005 01:33:13 +0000 (01:33 +0000)]
nested-func-1.m: Move to ...

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        * objc.dg/nested-func-1.m: Move to ...
        * objc/execute/nested-func-1.m: Here.

From-SVN: r103072

19 years agoexceptions.exp: New exp for the exceptions testsuite.
Andrew Pinski [Sun, 14 Aug 2005 00:58:26 +0000 (17:58 -0700)]
exceptions.exp: New exp for the exceptions testsuite.

* objc/execute/exceptions/exceptions.exp: New exp for the exceptions
testsuite.
* objc/execute/exceptions/trivial.m: New test.
* objc.dg/try-catch-11.m: Move to ...
* objc/execute/exceptions/foward-1.m: Here.
* objc.dg/try-catch-2.m: Move to ...
* objc/execute/exceptions/catchall-1.m: Here.
* objc.dg/try-catch-8.m: Move to ...
* objc/execute/exceptions/local-variables-1.m: Here.

From-SVN: r103071

19 years agoDaily bump.
GCC Administrator [Sun, 14 Aug 2005 00:16:14 +0000 (00:16 +0000)]
Daily bump.

[[Split portion of a mixed commit.]]

From-SVN: r103068.2

19 years ago* doc/md.texi (Machine Constraints): Fix misplaced @end table.
Andreas Schwab [Sat, 13 Aug 2005 23:54:11 +0000 (23:54 +0000)]
* doc/md.texi (Machine Constraints): Fix misplaced @end table.

From-SVN: r103065

19 years agoMakefile.in (extra_ldflags_libobjc): New.
Andrew Pinski [Sat, 13 Aug 2005 23:44:14 +0000 (23:44 +0000)]
Makefile.in (extra_ldflags_libobjc): New.

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        * Makefile.in (extra_ldflags_libobjc): New.
        (libobjc$(libext).la): Add extra_ldflags_libobjc to the link line.
        (libobjc_gc$(libext).la): Likewise.
        * configure.ac (extra_ldflags_libgfortran): Set for *-darwin* to
        "-Wl,-single_module".
        * configure: Regenerate.
        * linking.m (_objcInit): Remove.

From-SVN: r103064

19 years agoobjc.exp (objc_target_compile): Check for -fgnu-runtime in ${options} instead of...
Andrew Pinski [Sat, 13 Aug 2005 23:36:21 +0000 (23:36 +0000)]
objc.exp (objc_target_compile): Check for -fgnu-runtime in ${options} instead of [target_info name].

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        * lib/objc.exp (objc_target_compile): Check for -fgnu-runtime
        in ${options} instead of [target_info name].
        * lib/obj-c++.exp (objc_target_compile): Likewise.

From-SVN: r103063

19 years agostubify-2.m: Correct the name of the rtl dump file.
Andrew Pinski [Sat, 13 Aug 2005 21:41:53 +0000 (21:41 +0000)]
stubify-2.m: Correct the name of the rtl dump file.

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        * objc.dg/stubify-2.m: Correct the name of the rtl dump file.

From-SVN: r103062

19 years agopart of PR testsuite/23348
Andrew Pinski [Sat, 13 Aug 2005 21:16:04 +0000 (21:16 +0000)]
part of PR testsuite/23348

2005-08-13  Andrew Pinski  <pinskia@physics.uc.edu>

        part of PR testsuite/23348
        * lib/objc-torture.exp: Add -fgnu-runtime and/or -fnext-runtime to each
        of the torture options.
        * objc/compile/trivial.m: New test.

From-SVN: r103061

19 years agoDocument compatibility with earlier ProPolice implementation.
James E Wilson [Sat, 13 Aug 2005 20:58:02 +0000 (13:58 -0700)]
Document compatibility with earlier ProPolice implementation.

* c-cppbuiltin.c (c_cpp_builtins): Add comment for flag_stack_protect
macros.

From-SVN: r103058

19 years agoAdd missing docs for feature added by Richard Henderson.
James E Wilson [Sat, 13 Aug 2005 20:52:01 +0000 (13:52 -0700)]
Add missing docs for feature added by Richard Henderson.

* doc/cpp.texi (__SSP__, __SSP_ALL__): Document.
* doc/invoke.texi (-Wstack-protector, -fstack-protector,
-fstack-protector-all, --param ssp-buffer-size): Document.
(-Wvariadic-macros): Alphabetize.
(-fsched-stalled-insns-dep): Add missing 'f'.

From-SVN: r103057

19 years agors6000.h (EXTRA_CONSTRAINT): Add 'a' for indexed or indirect address operand.
David Edelsohn [Sat, 13 Aug 2005 18:47:08 +0000 (18:47 +0000)]
rs6000.h (EXTRA_CONSTRAINT): Add 'a' for indexed or indirect address operand.

        * config/rs6000/rs6000.h (EXTRA_CONSTRAINT): Add 'a' for indexed
        or indirect address operand.
        (EXTRA_ADDRESS_CONSTRAINT): New.
        * config/rs6000/rs6000.md (prefetch): Change constraint "p" to "a".

From-SVN: r103056

19 years agore PR tree-optimization/22236 (wrong code for casts and scev)
Sebastian Pop [Sat, 13 Aug 2005 17:28:43 +0000 (19:28 +0200)]
re PR tree-optimization/22236 (wrong code for casts and scev)

PR tree-optimization/22236
* tree-cfg.c (print_pred_bbs, print_succ_bbs): Correctly print
successors and predecessors.
* tree-chrec.c (chrec_convert): Before converting, check that
sequences don't wrap.
* tree-data-ref.c (compute_estimated_nb_iterations): Moved ...
(analyze_array): Extern.
(find_data_references_in_loop): Remove call to
compute_estimated_nb_iterations.
* tree-data-ref.h (analyze_array): Declared.
* tree-flow-inline.h (single_ssa_tree_operand, single_ssa_use_operand,
single_ssa_def_operand, zero_ssa_operands): Fix documentation.
* tree-flow.h (scev_probably_wraps_p): Declare with an extra parameter.
* tree-scalar-evolution.c (instantiate_parameters_1): Factor entry
condition.
* tree-ssa-loop-ivcanon.c: Fix documentation.
* tree-ssa-loop-ivopts.c (idx_find_step): Add a fixme note.
* tree-ssa-loop-niter.c (compute_estimated_nb_iterations): ... here.
(infer_loop_bounds_from_undefined): New.
(estimate_numbers_of_iterations_loop): Use
infer_loop_bounds_from_undefined.
(used_in_pointer_arithmetic_p): New.
(scev_probably_wraps_p): Pass an extra parameter.  Call
used_in_pointer_arithmetic_p.  Check that AT_STMT is not null.
(convert_step): Fix documentation.
* tree-vrp.c (adjust_range_with_scev): Call instantiate_parameters.
Use initial_condition_in_loop_num and evolution_part_in_loop_num
instead of CHREC_LEFT and CHREC_RIGHT.  Adjust the call to
scev_probably_wraps_p.

From-SVN: r103055

19 years agos390.c (s390_split_branches): Revert 2005-08-12 change.
Ulrich Weigand [Sat, 13 Aug 2005 11:08:21 +0000 (11:08 +0000)]
s390.c (s390_split_branches): Revert 2005-08-12 change.

* config/s390/s390.c (s390_split_branches): Revert 2005-08-12 change.
(s390_register_info): Ignore clobbered_regs information for fixed
registers, and only fixed registers.
(s390_init_frame_layout): Remove redundant call.

From-SVN: r103054

19 years agoDaily bump.
GCC Administrator [Sat, 13 Aug 2005 00:16:12 +0000 (00:16 +0000)]
Daily bump.

[[Split portion of a mixed commit.]]

From-SVN: r103048.2

19 years agoeoshift.f90: Correct format.
Thomas Koenig [Fri, 12 Aug 2005 20:02:43 +0000 (20:02 +0000)]
eoshift.f90: Correct format.

2005-08-12  Thomas Koenig  <Thomas.Koenig@online.de>

* gfortran.dg/eoshift.f90:  Correct format.

From-SVN: r103045

19 years agoinvoke.texi (C++ Dialect Options): Add dynamic_cast to description of -Wold-style...
Gerald Pfeifer [Fri, 12 Aug 2005 20:02:27 +0000 (20:02 +0000)]
invoke.texi (C++ Dialect Options): Add dynamic_cast to description of -Wold-style-casts.

* doc/invoke.texi (C++ Dialect Options): Add dynamic_cast to
description of -Wold-style-casts.

From-SVN: r103044

19 years ago005-08-12 Thomas Koenig <Thomas.Koenig@online.de>
Thomas Koenig [Fri, 12 Aug 2005 19:42:12 +0000 (19:42 +0000)]
005-08-12  Thomas Koenig  <Thomas.Koenig@online.de>

        * gfortran.fortran-torture/execute/character_passing.f90:
        Fix comment typo.

From-SVN: r103042

19 years ago* MAINTAINERS (Write After Approval): Add myself.
Maciej W. Rozycki [Fri, 12 Aug 2005 18:09:47 +0000 (18:09 +0000)]
* MAINTAINERS (Write After Approval): Add myself.

From-SVN: r103041

19 years agore PR libgcj/23353 (clone() copies internal lock data with hash synchronization off)
Morgan Deters [Fri, 12 Aug 2005 16:40:32 +0000 (16:40 +0000)]
re PR libgcj/23353 (clone() copies internal lock data with hash synchronization off)

2005-08-12  Morgan Deters  <mdeters@morgandeters.com>

PR libgcj/23353:
* java/lang/natObject.cc (clone): Clear sync_info.

From-SVN: r103039

19 years agoconfigure.in: Replace NCN_STRICT_CHECK_TOOL with NCN_STRICT_CHECK_TOOLS...
Paolo Bonzini [Fri, 12 Aug 2005 14:18:46 +0000 (14:18 +0000)]
configure.in: Replace NCN_STRICT_CHECK_TOOL with NCN_STRICT_CHECK_TOOLS...

toplevel:
2005-08-12  Paolo Bonzini  <bonzini@gnu.org>

* configure.in: Replace NCN_STRICT_CHECK_TOOL with
NCN_STRICT_CHECK_TOOLS, and likewise for NCN_STRICT_CHECK_TARGET_TOOLS.
Look for alternate names of the target cc and c++
* configure: Regenerate.

config:
2005-08-12  Paolo Bonzini  <bonzini@gnu.org>

* config/acx.m4 (NCN_CHECK_TARGET_TOOL, NCN_STRICT_CHECK_TOOL,
NCN_STRICT_CHECK_TARGET_TOOL): Remove.
(NCN_STRICT_CHECK_TOOLS, NCN_STRICT_CHECK_TARGET_TOOLS): New,
based on the deleted macros.

From-SVN: r103036

19 years agolinux.h (TARGET_THREAD_SSP_OFFSET): Define.
Andreas Krebbel [Fri, 12 Aug 2005 13:49:18 +0000 (13:49 +0000)]
linux.h (TARGET_THREAD_SSP_OFFSET): Define.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>
    Jakub Jelinek  <jakub@redhat.com>

* config/s390/linux.h (TARGET_THREAD_SSP_OFFSET): Define.
* config/s390/s390-protos.h (s390_get_thread_pointer): Prototype added.
* config/s390/s390.c (print_operand): New output modifier 'G' added.
(get_thread_pointer): Renamed to s390_get_thread_pointer.
* config/s390/s390.md (stack_protect_set, stack_protect_test): If
TARGET_THREAD_SSP_OFFSET is defined, change operands[1] to
(MEM:P (PLUS:P (tp, TARGET_THREAD_SSP_OFFSET))).
(UNSPEC_SP_SET, UNSPEC_SP_TEST): New constants.
("stack_protect_set", "stack_protect_test"): New expanders.
("stack_protect_setsi", "stack_protect_setdi", "stack_protect_testsi",
"stack_protect_testdi"): New insn definitions.

Co-Authored-By: Jakub Jelinek <jakub@redhat.com>
From-SVN: r103034

19 years agos390.md ("*movdf_31"): Changed constraint from P to K.
Andreas Krebbel [Fri, 12 Aug 2005 13:41:28 +0000 (13:41 +0000)]
s390.md ("*movdf_31"): Changed constraint from P to K.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>

* config/s390/s390.md ("*movdf_31"): Changed constraint from P to K.

From-SVN: r103033

19 years agolib1funcs.asm: Error if __ARM_ARCH__ not set.
Paul Brook [Fri, 12 Aug 2005 13:18:36 +0000 (13:18 +0000)]
lib1funcs.asm: Error if __ARM_ARCH__ not set.

2005-08-12  Paul Brook  <paul@codesourcery.com>

* config/arm/lib1funcs.asm: Error if __ARM_ARCH__ not set.

[[Split portion of a mixed commit.]]

From-SVN: r103029.2

19 years agopredicates.md (setmem_operand): New predicate.
Andreas Krebbel [Fri, 12 Aug 2005 12:24:05 +0000 (12:24 +0000)]
predicates.md (setmem_operand): New predicate.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>

* config/s390/predicates.md (setmem_operand): New predicate.
(shift_count_operand): Accept ANDs with special constants as
operand.
* config/s390/s390.c (print_shift_count_operand): Skip ANDs
with special constants.
* config/s390/s390.md ("setmem_long", "*setmem_long"): Replaced
shift_count_operand with setmem_operand.

From-SVN: r103028

19 years agos390.c (s390_extract_part, [...]): Type cast added.
Andreas Krebbel [Fri, 12 Aug 2005 12:10:30 +0000 (12:10 +0000)]
s390.c (s390_extract_part, [...]): Type cast added.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>

* config/s390/s390.c (s390_extract_part, s390_single_part):
Type cast added.
(s390_const_ok_for_constraint_p): Added SImode to the N constraint.
(s390_output_mi_thunk): Don't use lg on 31 bit.
* config/s390/s390.md ("*movdi_31", "*movdf_31"): Added lmy and stmy.
("*llgt_sisi" and splitter): Replaced TARGET_64BIT with TARGET_ZARCH.

From-SVN: r103027

19 years agos390.c (CONST_OK_FOR_J, [...]): New macros.
Andreas Krebbel [Fri, 12 Aug 2005 12:02:21 +0000 (12:02 +0000)]
s390.c (CONST_OK_FOR_J, [...]): New macros.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>

* config/s390/s390.c (CONST_OK_FOR_J, CONST_OK_FOR_K): New macros.
(s390_select_ccmode, s390_rtx_costs, legitimate_reload_constant_p,
s390_init_frame_layout, s390_emit_prologue, s390_emit_epilogue,
s390_output_mi_thunk): Replaced uses of CONST_OK_FOR_CONSTRAINT_P
with one of the new macros.

From-SVN: r103026

19 years agos390.c (s390_split_branches, [...]): Don't set save_return_addr_p.
Andreas Krebbel [Fri, 12 Aug 2005 11:58:05 +0000 (11:58 +0000)]
s390.c (s390_split_branches, [...]): Don't set save_return_addr_p.

2005-08-12  Andreas Krebbel  <krebbel1@de.ibm.com>

* config/s390/s390.c (s390_split_branches, s390_init_frame_layout):
Don't set save_return_addr_p.
(s390_register_info): Make clobbered_regs not depending on
save_return_addr_p.

From-SVN: r103025

19 years agore PR c++/8271 (Templates and pointers to const member functions)
Giovanni Bajo [Fri, 12 Aug 2005 09:34:12 +0000 (11:34 +0200)]
re PR c++/8271 (Templates and pointers to const member functions)

PR c++/8271
* g++.dg/template/unify10.C: New.

From-SVN: r103022