New Miscompilation Into Bytecode Error

Thu Jul 1 09:29:00 GMT 2004

There have been only three substantive changes in the last few days:
revision 1.1396
date: 2004年07月01日 07:51:11; author: rth; state: Exp; lines: +16 -0
 * cgraph.h (cgraph_build_static_cdtor): Declare.
 * cgraphunit.c (cgraph_build_static_cdtor): New.
 * c-objc-common.c (build_cdtor): Use it.
 * coverage.c (create_coverage): Likewise.
 * libfuncs.h (LTI_gcov_init, gcov_init_libfunc): Remove.
 * optabs.c (init_optabs): Don't set gcov_init_libfunc.
java/
 * class.c (registerClass_libfunc): Remove.
 (init_class_processing): Don't set it.
 (emit_register_classes): Take list_p parameter. Fill it in
 with _Jv_RegisterClass calls.
 * decl.c (java_init_decl_processing): Don't call
 init_resource_processing.
 * jcf-parse.c (java_emit_static_constructor): New.
 (java_parse_file): Call it.
 * resource.c (registerResource_libfunc): Remove.
 (init_resource_processing): Remove.
 (write_resource_constructor): Take list_p parameter. Fill it in
 with _Jv_RegisterResource calls.
 * java-tree.h: Update prototypes.
----------------------------
revision 1.1395
date: 2004年06月29日 17:58:44; author: bryce; state: Exp; lines: +13 -0
 PR java/1262
 * class.c (layout_class_method): Do not override package-private
 method if its in a different package.
 (split_qualified_name): Move here from parse.y. Rename from
 breakdown_qualified. Add comment.
 (in_same_package): Move here from parse.y. Add comment.
 * java-tree.h (break_down_qualified, in_same_package): Declare.
 (in_same_package): Likewise.
 * parse.y (breakdown_qualified, in_same_package): Moved to class.c.
 Callers updated.
----------------------------
revision 1.1394
date: 2004年06月29日 16:18:45; author: aph; state: Exp; lines: +18 -0
2004年06月29日 Andrew Haley <aph@redhat.com>
 * except.c (expand_start_java_handler): Push a new binding level.
 Don't build a TRY_CATCH_EXPR now, we'll do it later. Call
 register_exception_range() to register where we'll do it.
 (expand_end_java_handler): Remove old bogus code. Replace with
 new logic that simply builds TRY_CATCH_EXPRs and inserts them at
 the top of the expression we're curently building.
 (maybe_end_try): Delete.
 * decl.c (binding_level.exception_range): New field.
 (clear_binding_level): Add field exception_range. Reformat.
 (poplevel): Call expand_end_java_handler().
 (poplevel): Call java_add_stmt only if functionbody is false.
 (maybe_poplevels): Don't call maybe_end_try() from here.
 (end_java_method): Clear no longer used trees in function decl.
 (register_exception_range): New function.
 * java-tree.h (register_exception_range, struct eh_range): Declare.
----------------------------
Can you please find out which of these is causing the problem?
Thanks,
Andrew.