Daniel Berlin [Fri, 29 Oct 2004 02:13:03 +0000 (02:13 +0000)]
tree-into-ssa.c (rewrite_virtual_phi_arguments): New function.
2004-10-28 Daniel Berlin <dberlin@dberlin.org>
* tree-into-ssa.c (rewrite_virtual_phi_arguments): New function.
(rewrite_blocks): Modify argument, and use it to decide which
function to call.
(rewrite_def_def_chains): Call rewrite_blocks with true here.
(rewrite_into_ssa): and call it with false here.
Andrew Pinski [Thu, 28 Oct 2004 23:38:47 +0000 (23:38 +0000)]
visibility-[1-9a].c: Change to use scan-hidden instead of scan-assembler.
2004-10-28 Andrew Pinski <pinskia@physics.uc.edu>
* gcc.dg/visibility-[1-9a].c: Change to use scan-hidden instead of
scan-assembler.
* lib/scanasm.exp (hidden-scan-for): New proc.
(scan-hidden): Use hidden-scan-for instead of constructing the scan
string.
(scan-not-hidden): Likewise.
* lib/target-supports.exp (check_visibility_available): mach-o is
support too.
Added pedantic_min_int to gfc_integer_info Added ARITH_ASYMMETRIC to arith...
Added pedantic_min_int to gfc_integer_info
Added ARITH_ASYMMETRIC to arith
Added support for an "asymmetric integer" warning when compiling with pedantic
Set minimum integer values to reflect realities of two's complement signed integers
Frank Ch. Eigler [Thu, 28 Oct 2004 21:24:33 +0000 (21:24 +0000)]
tree-mudflap.c (mudflap_init): Mark cache-lookup decls to prevent their later unnecessary registration.
2004-10-28 Frank Ch. Eigler <fche@redhat.com>
* tree-mudflap.c (mudflap_init): Mark cache-lookup decls to
prevent their later unnecessary registration.
(mf_build_check_statement_for): Remove "addr" argument, and
related "__mf_value" generated code. Update callers.
(mf_decl_eligible_p): New function, factored from mx_register_decls.
(mx_register_decls): Call it.
(mf_xform_derefs_1): Rewrite COMPONENT_REF and ARRAY_REF cases
to restore support for complex nested expressions.
(mudflap_enqueue_decl): Simplify. Move registration call emission
out ...
(mudflap_finish_file): ... to here.
(mudflap_enqueue_constant): Remove dump code.
Zack Weinberg [Thu, 28 Oct 2004 21:00:59 +0000 (21:00 +0000)]
re PR debug/18199 (AIX XCOFF debugging broken)
PR 18199
* dbxout.c (dbxout_begin_complex_stabs_noforcetext): New function.
(DBX_FINISH_STABS): Add (unused by default) SYM parameter.
(dbxout_finish_complex_stabs): Update to match.
(dbxout_symbol_location): Call emit_pending_bincls_if_required
and FORCE_TEXT before DBX_STATIC_BLOCK_START. Use
dbxout_begin_complex_stabs_noforcetext.
* xcoffout.h (DBX_FINISH_STABS): Restore special case for
N_GSYM, using new SYM parameter. Correct logic for special
cases for N_FUN.
Adam Nemet [Thu, 28 Oct 2004 19:31:25 +0000 (19:31 +0000)]
re PR middle-end/18160 (ICE on taking register variable address)
PR middle-end/18160
* gcc.dg/register-var-1.c: New test.
* gcc.dg/register-var-2.c: New test.
* gcc.dg/asm-7.c: Taking the address of a register variable is an
error now.
alpha.md (movqicc_internal): Allow exactly one zero in the conditional.
* config/alpha/alpha.md (movqicc_internal): Allow exactly one
zero in the conditional.
(movhicc_internal, movsicc_internal, movdicc_internal): Likewise.
* lambda-code.c (nestify_update_pending_stmts): Remove.
(perfect_nestify): Use flush_pending_stmts instead of
nestify_update_pending_stmts.
* tree-cfg.c (tree_make_forwarder_block): Use
flush_pending_stmts.
* tree-flow.h: Add a prototype for flush_pending_stmts.
* tree-ssa-loop-manip.c (tree_duplicate_loop_to_header_edge):
Use flush_pending_stmts.
(lv_update_pending_stmts): Remove.
(tree_ssa_loop_version): Use flush_pending_stmts instead of
lv_update_pending_stmts.
* tree-ssa.c (flush_pending_stmts): New.
Ulrich Weigand [Thu, 28 Oct 2004 12:47:21 +0000 (12:47 +0000)]
re PR target/15286 (ICE cause by reload)
PR target/15286
* final.c (alter_subreg): Compute correct offset to use with
paradoxical SUBREGs of memory operands.
* recog.c (general_operand): Allow paradoxical SUBREGs of
memory operands after reload.
* simplify-rtx.c (simplify_gen_subreg): Fail if simplify_subreg
has failed when passed a hard register.
Mark Mitchell [Thu, 28 Oct 2004 05:17:14 +0000 (05:17 +0000)]
re PR c++/17435 (Binding a temporary of derived type to reference of base)
PR c++/17435
* call.c (convert_like_real): Fix formatting.
(initialize_reference): When binding a temporary to a base class,
ensure that the nominal copy made is to the derived class, not the
base class.
PR c++/18140
* parser.c (cp_parser_next_token_ends_template_argument_p): Do not
include ">>".
Geoffrey Keating [Thu, 28 Oct 2004 00:16:57 +0000 (00:16 +0000)]
typeof-2.c: Needs aliases as well as weak.
* gcc.dg/weak/typeof-2.c: Needs aliases as well as weak.
* gcc.dg/weak/weak-3.c: Likewise.
* gcc.dg/weak/weak-5.c: Likewise.
* gcc.dg/weak/weak-9.c: Likewise.
* lib/file-format.exp (gcc_target_object_format): Darwin has Mach-O.
* lib/target-supports.exp (check_weak_available): Mach-O does weak.
Zdenek Dvorak [Wed, 27 Oct 2004 20:27:20 +0000 (22:27 +0200)]
re PR tree-optimization/18048 (mgrid loop performance regression with ivopts (register pressure))
PR tree-optimization/18048
* fold-const.c (try_move_mult_to_index): New function.
(fold): Use try_move_mult_to_index.
* tree-ssa-loop-ivopts.c (try_add_cand_for): Prefer common candidates.
* tree-ssa-loop-niter.c (number_of_iterations_cond): Produce
an all-ones unsigned constant without extra bits.
* tree.c (build_low_bits_mask): New function.
* tree.h (build_low_bits_mask): Declare.
Daniel Berlin [Wed, 27 Oct 2004 17:45:21 +0000 (17:45 +0000)]
re PR tree-optimization/17133 (wrong code with -ftree-lim)
2004-10-27 Daniel Berlin <dberlin@dberlin.org>
Fix PR tree-optimization/17133
* tree-cfg.c (rewrite_to_new_ssa_names_bb): Also rewrite must
def kill operand.
* tree-flow-inline.h: V_MUST_DEF_OP became V_MUST_DEF_RESULT.
(get_v_must_def_result_ptr): Modify for new structure of
v_must_defs array.
(get_v_must_def_kill_ptr): New.
(op_iter_next_use): Add support for the kill that occurs in V_MUST_DEFs.
(op_iter_next_tree): Ditto. Also V_MAY_DEF_OP became V_MAY_DEF_RESULT.
(op_iter_next_def): V_MAY_DEF_OP became V_MAY_DEF_RESULT.
(op_iter_init): Initialize new mustu members.
(op_iter_next_mustdef): New function.
(op_iter_init_mustdef): Ditto.
* tree-flow.h (rewrite_def_def_chains): New function.
* tree-into-ssa.c (mark_def_sites): Handle mustdefkill operands.
(ssa_mark_def_sites): Ditto.
(rewrite_stmt): Ditto.
(ssa_rewrite_stmt): Ditto.
(rewrite_blocks): Factor out from rewrite_into_ssa.
(mark_def_block_sites): Ditto.
(rewrite_def_def_chains): New function, just rewrites def-def
chains without phi node insertion.
* tree-pass.h (TODO_fix_def_def_chains): New todo flag.
* tree-pretty-print.c (dump_vops): Print out MUST_DEF's so that
they include the rhs now.
* tree-ssa-ccp.c (visit_assignment): V_MUST_DEF_OP became
V_MUST_DEF_RESULT.
* tree-ssa-dce.c (mark_operand_necessary): Add phionly argument.
Update callers.
(mark_really_necessary_kill_operand_phis): New function.
(perform_tree_ssa_dce): Call it.
(pass_dce): Add TODO_fix_def_def_chains.
(pass_cd_dce): Ditto.
* tree-ssa-loop-im.c (determine_max_movement): Look at kills as
well.
(rewrite_mem_refs): Ditto.
* tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Look at kills
as well.
* tree-ssa-operands.c (allocate_v_may_def_optype):
v_may_def_operand_type_t became v_def_use_operand_type_t.
(allocate_v_must_def_optype) Ditto.
(finalize_ssa_v_must_defs): Update for new operand type, as well
as setting the use portion as well.
(copy_virtual_operands): Copy the kill operand as well.
(create_ssa_artficial_load_stmt): V_MUST_DEF_OP became
V_MUST_DEF_RESULT.
* tree-ssa-operands.h (v_may_def_operand_type): Renamed to
v_def_use_operand_type.
(v_must_def_optype_d): Use v_def_use_operand_type.
(V_MUST_DEF_OP_*): Renamed to V_MUST_DEF_RESULT_*
(V_MUST_DEF_KILL_*): New macros.
(struct ssa_operand_iterator_d): Add num_v_mustu and v_mustu_i
members.
Rename existing must_i and num_v_must members to mustd_i and
num_v_mustd.
(SSA_OP_VMUSTDEFKILL): New flag.
(SSA_OP_VIRTUAL_KILLS): New flag.
(SSA_OP_ALL_OPERANDS): Add in SSA_OP_ALL_KILLS.
(SSA_OP_ALL_KILLS): New flag.
(FOR_EACH_SSA_MUSTDEF_OPERAND): New macro.
* tree-ssa.c (verify_ssa): Verify virtual kills as well.
* tree-vectorizer.c (vect_create_data_ref_ptr): V_MUST_DEF_OP
became V_MUST_DEF_RESULT.
(rename_variables_in_bb): Rename kill pointer as well.
* tree-dfa.c (compute_immediate_uses_for_stmt): Add kills into the
immediate uses.
Steven Bosscher [Wed, 27 Oct 2004 17:14:19 +0000 (17:14 +0000)]
tree-phinodes.c (remove_phi_arg_num): Don't remove PHIs without any PHI arguments left.
* tree-phinodes.c (remove_phi_arg_num): Don't remove PHIs
without any PHI arguments left. Make sure the argument that
we're supposed to remove exists at all.
Richard Earnshaw [Wed, 27 Oct 2004 15:10:22 +0000 (15:10 +0000)]
ffi.c (ffi_prep_cif_machdep): Handle functions that return long long values.
* src/arm/ffi.c (ffi_prep_cif_machdep): Handle functions that return
long long values. Round stack allocation to a multiple of 8 bytes
for ATPCS compatibility.
* src/arm/sysv.S (ffi_call_SYSV): Rework to avoid use of APCS register
names. Handle returning long long types. Add Thumb and interworking
support. Improve soft-float code.
Cyrille Comar [Wed, 27 Oct 2004 13:56:19 +0000 (15:56 +0200)]
gnat_ugn.texi: Generalize "finding memory problems" section into a "memory management issues"...
2004-10-26 Cyrille Comar <comar@act-europe.fr>
Vasiliy Fofanov <fofanov@act-europe.fr>
Vincent Celier <celier@gnat.com>
* gnat_ugn.texi: Generalize "finding memory problems" section into a
"memory management issues" section and document some of the useful
memory pools provided as part of the GNAT library.
Remove "virtual" from declaration of A::method2 in
the simple example of Ada/C++ mixed system.
Library Projects may be virtually extended: their virtual extensions
are not Library Projects.
Added section on extending project hierarchies.
Eric Botcazou [Wed, 27 Oct 2004 13:55:45 +0000 (15:55 +0200)]
trans.c (call_to_gnu): For an (in-)out parameter passed by reference whose type is a constructed...
2004-10-26 Eric Botcazou <ebotcazou@act-europe.fr>
* trans.c (call_to_gnu): For an (in-)out parameter passed by reference
whose type is a constructed subtype of an aliased object with an
unconstrained nominal subtype, convert the actual to the constructed
subtype before taking its address.
Pascal Obry [Wed, 27 Oct 2004 13:55:31 +0000 (15:55 +0200)]
tracebak.c (IS_BAD_PTR): Use IsBadCodePtr on Win32 to check for ptr validity (process must have read access).
2004-10-26 Pascal Obry <obry@gnat.com>
* tracebak.c (IS_BAD_PTR): Use IsBadCodePtr on Win32 to check for ptr
validity (process must have read access). Set to 0 in all other cases.
(STOP_FRAME): Now check for ptr validity to avoid a segmentation
violation on Win32.
(VALID_STACK_FRAME): Check for ptr validity on Win32 to avoid a
segmentation violation.
Thomas Quinot [Wed, 27 Oct 2004 13:54:52 +0000 (15:54 +0200)]
sem_util.adb (Is_Aliased_View): Defend against the case where this subprogram is called with a parameter that...
2004-10-26 Thomas Quinot <quinot@act-europe.fr>
Ed Schonberg <schonberg@gnat.com>
* sem_util.adb (Is_Aliased_View): Defend against the case where this
subprogram is called with a parameter that is not an object name. This
situation arises for some cases of illegal code, which is diagnosed
later, and in this case it is wrong to call Is_Aliased, as that might
cause a compiler crash.
(Explain_Limited_Type): Refine previous fix to include
inherited components of derived types, to provide complete information.
* exp_ch9.adb (Set_Privals): Set the Ekind of the actual object that
is the prival for a protected object.
It is necessary to mark this entity as a variable, in addition to
flagging it as Aliased, because Sem_Util.Is_Aliased_View has been
modified to avoid checking the Aliased flag on entities that are not
objects. (Checking that flag for non-objects is erroneous and could
lead to a compiler crash).
Ed Schonberg [Wed, 27 Oct 2004 13:54:38 +0000 (15:54 +0200)]
sem_elab.adb (Check_Elab_Call): Do not check a call that does not appear in the code for the main unit.
2004-10-26 Ed Schonberg <schonberg@gnat.com>
* sem_elab.adb (Check_Elab_Call): Do not check a call that does not
appear in the code for the main unit. Dependencies among units in the
context of the main unit are established when those other units are
compiled. Otherwise spurious elaboration constraints can generate
incorrect elaboration circularities.
Ed Schonberg [Wed, 27 Oct 2004 13:42:11 +0000 (15:42 +0200)]
sem_ch6.adb (Analyze_Subprogram_Body): If body is a subunit for a different kind of stub (possibly wrong name for file)...
2004-10-26 Ed Schonberg <schonberg@gnat.com>
* sem_ch6.adb (Analyze_Subprogram_Body): If body is a subunit for a
different kind of stub (possibly wrong name for file), do not check
for conformance.
(Uses_Secondary_Stack): New subsidiary to Build_Body_To_Inline. If body
includes call to some function that returns an unconstrained type, do
not inline.
Ed Schonberg [Wed, 27 Oct 2004 13:41:55 +0000 (15:41 +0200)]
sem_ch3.adb (Complete_Private_Subtype): If the full view is a task or protected type with discriminants...
2004-10-26 Ed Schonberg <schonberg@gnat.com>
* sem_ch3.adb (Complete_Private_Subtype): If the full view is a task
or protected type with discriminants, do not constrain the
corresponding record type if the subtype declaration has no
discriminant constraints. This can be the case in source code, or in
the subtype declaration created to rename an actual type within an
instantiation.
Ed Schonberg [Wed, 27 Oct 2004 13:40:55 +0000 (15:40 +0200)]
sem_ch12.adb (In_Main_Context): Predicate to determine whether the current instance appears within a unit that...
2004-10-26 Ed Schonberg <schonberg@gnat.com>
Javier Miranda <miranda@gnat.com>
* sem_ch12.adb (In_Main_Context): Predicate to determine whether the
current instance appears within a unit that is directly in the context
of the main unit.
Used to determine whether the body of the instance should be analyzed
immediately after its spec, to make its subprogram bodies available
for front-end inlining.
(Analyze_Formal_Array_Type): Cleanup condition that checks that range
constraint is not allowed on the component type (AARM 12.5.3(3))