Kazu Hirata [Tue, 15 Feb 2005 07:19:51 +0000 (07:19 +0000)]
flow.c (recompute_reg_usage): Remove all arguments.
* flow.c (recompute_reg_usage): Remove all arguments.
* passes.c (rest_of_compilation): Adjust the call to
recompute_reg_usage.
* rtl.h: Adjust the prototype for recompute_reg_usage.
Daniel Berlin [Mon, 14 Feb 2005 21:32:14 +0000 (21:32 +0000)]
tree-dump.c (dump_files): Update to reflect new member of dump_file_info.
2005-02-14 Daniel Berlin <dberlin@dberlin.org>
* tree-dump.c (dump_files): Update to reflect new member
of dump_file_info.
(dump_register): Add new argument.
Set glob name.
(dump_switch_p_1): Check against glob name if doglob was passed.
(dump_switch_p): Check against regular first, then glob if
we didn't have any hits on the dump name.
* tree-optimize.c (register_one_dump_file):
Pass in glob name as well.
* tree-pass.h (struct dump_file_info): Add glob
member.
Nathan Sidwell [Mon, 14 Feb 2005 17:50:58 +0000 (17:50 +0000)]
re PR c++/19608 (ICE after friend function definition in local class)
cp:
PR c++/19608
* parser.c (cp_parser_late_parsing_for_member): Use
current_function_decl as scope to push to and from.
testsuite:
PR c++/19608
* g++.dg/parse/local-class1.C: New.
Andrew Haley [Mon, 14 Feb 2005 14:58:22 +0000 (14:58 +0000)]
re PR java/19907 (Incorrect code generated for ManifestElement.java)
2005-02-14 Andrew Haley <aph@redhat.com>
PR java/19907
* expr.c (expand_byte_code): Call promote_arguments().
(promote_arguments): New function.
* decl.c (check_local_unnamed_variable): Remve special case for
new verifier.
(find_local_variable): Promote all boolean types to int
when searching for local variable decls.
Anthony Green [Mon, 14 Feb 2005 14:57:37 +0000 (14:57 +0000)]
re PR libgcj/18116 (JNI uses dot instead of slash as the package separator)
2005-02-14 Anthony Green <green@redhat.com>
PR libgcj/18116
* testsuite/libjava.jni/PR18116.c: New file.
* testsuite/libjava.jni/PR18116.java: New file.
* testsuite/libjava.jni/PR18116.out: New file.
Sebastian Pop [Mon, 14 Feb 2005 14:54:37 +0000 (15:54 +0100)]
lambda-code.c (lambda_loopnest_to_gcc_loopnest, [...]): Use standard_iv_increment_position for computing the bsi position for create_iv.
* lambda-code.c (lambda_loopnest_to_gcc_loopnest, perfect_nestify):
Use standard_iv_increment_position for computing the bsi position
for create_iv.
* tree-ssa-loop-manip.c (create_iv): Add a comment on how to compute
the bsi position.
Anthony Green [Mon, 14 Feb 2005 13:51:29 +0000 (13:51 +0000)]
jni.cc (nathash_add): Don't strdup the method signature.
2005-02-13 Anthony Green <green@redhat.com>
* jni.cc (nathash_add): Don't strdup the method signature.
(_Jv_JNI_RegisterNatives): Convert the slashes to dots in the
method signature.
Update copyright.
Nathan Sidwell [Mon, 14 Feb 2005 13:45:42 +0000 (13:45 +0000)]
re PR c++/19884 (ICE on explicit instantiation of a non-template constructor)
cp:
PR c++/19884
* pt.c (check_explicit_specialization): Make sure namespace
binding lookup found an overloaded function.
(lookup_template_function): Just assert FNS is an overloaded
function.
PR c++/19895
* decl.c (grokdeclarator): Check for error mark node in ptrmem
construction.
testsuite:
PR c++/19895
* g++.dg/parse/ptrmem3.C: New.
PR c++/19884
* g++.old-deja/g++.oliva/template6.C: Add another case.
* g++.dg/template/explicit6.C: New.
Zdenek Dvorak [Mon, 14 Feb 2005 11:37:52 +0000 (12:37 +0100)]
re PR rtl-optimization/17428 (internal compiler error: in spill_failure, at reload1.c:1880 (-fspeculative-prefetching))
PR target/17428
* cfgrtl.c (safe_insert_insn_on_edge): Avoid extending life range of hard
registers.
* value-prof.c (insn_prefetch_values_to_profile): Only scan normal insns.
* value-prof.c (rtl_find_values_to_profile): Do not look for values to
profile in libcalls.
Nathan Sidwell [Mon, 14 Feb 2005 09:36:35 +0000 (09:36 +0000)]
re PR c++/19891 (Covariant returns are broken)
cp:
PR c++/19891
* class.c (build_simple_base_path): Build the component_ref
directly.
(update_vtable_entry_for_fn): Walk the covariant's binfo chain
rather than using lookup_base.
* search.c (dfs_walk_once): Add non-recursive assert check.
* typeck.c (build_class_member_access_expr): It is possible for
the member type to be both const and volatile.
testsuite:
PR c++/19891
* g++.dg/abi/covariant4.C: New.
David Edelsohn [Sun, 13 Feb 2005 21:31:35 +0000 (21:31 +0000)]
re PR target/19019 (GCC ldouble format incompatibility with XLC long double)
PR target/19019
* config/rs6000/aix.h ({TARGET,MASK}_XL_CALL): Rename to
{TARGET,MASK}_XL_COMPAT.
(SUBTARGET_SWITCHES): Rename xl-call to xl-compat. Use
MASK_XL_COMPAT.
* config/rs6000/beos.h ({TARGET,MASK}_XL_CALL): Remove.
* config/rs6000/rs6000.c (function_arg): Change TARGET_XL_CALL to
TARGET_XL_COMPAT.
(rs6000_arg_partial_bytes): Same.
(rs6000_generate_compare): Generate PARALLEL for compare if TFmode
and XL compatibility enabled.
* config/rs6000/rs6000.h (TARGET_XL_CALL): Rename to TARGET_XL_COMPAT.
* config/rs6000/rs6000.md (cmptf_internal1): Add !TARGET_XL_COMPAT
test to final condition.
(cmptf_internal2): New.
* doc/invoke.texi (RS/6000 Subtarget Options): Change xl-call to
xl-compat. Add TFmode information to description.
re PR tree-optimization/14303 ([tree-ssa] gcc.c-torture/execute/20020720-1.c is not fully folded)
2005-02-13 James A. Morrison <phython@gcc.gnu.org>
PR tree-optimization/14303
PR tree-optimization/15784
* fold-const.c (fold): Fold ABS_EXPR<x> >= 0 to true, when possible.
Fold ABS_EXPR<x> < 0 to false. Fold ABS_EXPR<x> == 0 to x == 0 and
ABS_EXPR<x> != 0 to x != 0.
Mark Wielaard [Sat, 12 Feb 2005 13:51:11 +0000 (13:51 +0000)]
Fixes bug libgcj/8170
Fixes bug libgcj/8170
* java/lang/ClassLoader.java (loadClass): Don't rewrap
ClassNotFoundException.
* gnu/java/lang/MainThread.java (run): Chain NoClassDefFoundError.
* gnu/gcj/runtime/NameFinder.java (remove_interpreter): Removed.
(remove_internal): New field superceding remove_interpreter.
(sanitizeStack): Remove all no-package classes starting with "_Jv_".
Remove no-class methods starting with "_Jv_". And Replace null
class or method names with the empty string. Stop at either the
MainThread or a real Thread run() method.
(newElement): Made static.
* java/net/URLClassLoader.java (findClass): Throw
ClassNotFoundExceptions including urls, plus parent using toString().
(thisString): New field.
(toString): New method.
* tree-vectorizer.c (vect_get_base_and_offset): Remove.
(vect_is_simple_iv_evolution): Remove redundant parameter
and step check.
(vect_analyze_scalar_cycles): Call vect_is_simple_iv_evolution
without last parameter.
(vect_analyze_pointer_ref_access): Get access_fn as parameter.
Return pointer step. Call vect_is_simple_iv_evolution without
last parameter. Check only that the step is multiple of size
type. Remove stmt_vinfo updates.
(vect_get_memtag_and_dr): Remove.
(vect_get_memtag): New function.
(vect_address_analysis): New function.
(vect_object_analysis): New function.
(vect_analyze_data_refs): Call vect_object_analysis and
vect_get_memtag. Update stmt_vinfo fields.
Ira Rosen [Sat, 12 Feb 2005 11:47:19 +0000 (11:47 +0000)]
tree-data-ref.c (array_base_name_differ_p): Check that the bases exist and are objects.
* tree-data-ref.c (array_base_name_differ_p): Check that the bases
exist and are objects. Remove checks for pointer.
* tree-vectorizer.c (vect_create_addr_base_for_vector_ref): Use
STMT_VINFO_VECT_DR_BASE_ADDRESS instead of DR_BASE_NAME.
(vect_create_data_ref_ptr): Likewise.
(vect_base_addr_differ_p): New function.
(vect_analyze_data_ref_dependence): Call vect_base_addr_differ_p.
(vect_analyze_pointer_ref_access): Add output parameter - ptr_init.
Don't set the DR_BASE_NAME field of data-ref.
(vect_get_memtag_and_dr): Use ptr_init instead of DR_BASE_NAME.
Ira Rosen [Sat, 12 Feb 2005 10:24:40 +0000 (10:24 +0000)]
tree-vectorizer.h (struct _stmt_vec_info): Rename a field: base to base_address.
2005-02-13 Ira Rosen <irar@il.ibm.com>
* tree-vectorizer.h (struct _stmt_vec_info): Rename a field: base
to base_address.
* tree-vectorizer.c (new_stmt_vec_info): Rename the above field of
stmt_vec_info.
(vect_get_base_and_offset): Always return an address.
(vect_create_addr_base_for_vector_ref): Remove treatment for
different data reference types.
(vect_compute_data_ref_alignment): Rename base to base_address in
stmt_vec_info. Get the object in order to force its alignment.
(vect_get_memtag_and_dr): Rename base to base_address in
stmt_vec_info. Extract the object for memtag analysis.
Per Bothner [Sat, 12 Feb 2005 06:12:40 +0000 (22:12 -0800)]
re PR java/15543 ("jv-scan --complexity" segfaults)
PR java/15543
* parse-scan.y (input_location): Remove variable.
(main_input_filename): New - replaces input_filename, which isn't
settable if USE_MAPPED_LOCATION.
* lex.c (java_init_lex): Wrap some more places in #ifndef JC1-LITE,
so we don't reference input_location or wfl_operator in that case.
* jv-scan.c (expand_location): Remove - no longer used.
(main): Set main_input_filename rather than input_filename.
re PR regression/19898 (cris-elf testsuite failure: gcc.c-torture/execute/strlen-1.c execution, -O2 and higher)
PR regression/19898.
* config/cris/cris.c (cris_notice_update_cc): When testing if insn
changes cc_status, use apply modified_in_p to part of cc_status
and insn, not cris_reg_overlap_mentioned_p on SET_DEST of insn
body.
Steven Bosscher [Fri, 11 Feb 2005 21:52:34 +0000 (21:52 +0000)]
re PR tree-optimization/19876 (g++ starts eating all the memory and the CPU)
PR tree-optimization/19876
Partially revert my change from 2005-01-14
* tree-ssa-pre.c (compute_antic_aux): Make recursive once again...
(compute_antic): ...and remove the loop here.
Ulrich Weigand [Fri, 11 Feb 2005 14:40:55 +0000 (14:40 +0000)]
s390.md ("*cmpdi_cct", [...]): Use %h instead of %c to print immediate operand.
* config/s390/s390.md ("*cmpdi_cct", "*cmpsi_cct", "*cmpdi_ccs",
"*cmpsi_ccs"): Use %h instead of %c to print immediate operand.
("*zero_extendhisi2_64", "*zero_extendhisi2_31"): Choose based on
TARGET_ZARCH instead of TARGET_64BIT.
Steven Bosscher [Thu, 10 Feb 2005 22:57:30 +0000 (22:57 +0000)]
re PR tree-optimization/17549 (10% increase in codesize with C code compared to GCC 3.3)
PR tree-optimization/17549
* tree-outof-ssa.c (find_replaceable_in_bb): Do not allow
TER to replace a DEF with its expression if the DEF and the
rhs of the expression we replace into have the same root
variable.
Tom Tromey [Thu, 10 Feb 2005 18:33:54 +0000 (18:33 +0000)]
Makefile.in: Rebuilt.
* external/sax/Makefile.in: Rebuilt.
* external/sax/Makefile.am (all-local): Removed.
(BUILT_SOURCES): New variable.
(mostlyclean-local): No error if no classes to delete.
* external/w3c_dom/Makefile.in: Rebuilt.
* external/w3c_dom/Makefile.am (all-local): Removed.
(BUILT_SOURCES): New variable.
(mostlyclean-local): No error if no classes to delete.