Starting program: /usr/local/src/gcc/head/build/gcc/jc1 bug.java class bugbug.java:4: error: '(' expected. Object ctor = new Object; ^ Program received signal SIGSEGV, Segmentation fault. 0x0804fcc5 in error_if_numeric_overflow (value=0x15d) at lex.c:1727 1727 if (TREE_CODE (value) == INTEGER_CST (gdb) where #0 0x0804fcc5 in error_if_numeric_overflow (value=0x15d) at lex.c:1727 #1 0x08059f27 in java_parse () at parse.y:2328 #2 0x081433c2 in parse_source_file_1 (file=0x404cc86c, finput=0x8830ba8) at ../../gcc/gcc/java/jcf-parse.c:826 #3 0x0814442f in java_parse_file (set_yydebug=0) at ../../gcc/gcc/java/jcf-parse.c:1078 #4 0x085d38a6 in compile_file () at ../../gcc/gcc/toplev.c:1653 #5 0x085d584a in do_compile () at ../../gcc/gcc/toplev.c:2699 #6 0x085d58af in toplev_main (argc=2, argv=0xbffff834) at ../../gcc/gcc/toplev.c:2731 #7 0x40039b45 in __libc_start_main (main=0xbffffac3, argc=-1073743154, ubp_av=0xbffffaeb, init=0x86e38f0 <__libc_csu_init>, fini=0x86e3950 <__libc_csu_fini>, rtld_fini=0x4000baa0 <_dl_fini>, stack_end=0xbffffb51) at ../sysdeps/generic/libc-start.c:209 #8 0x08049c71 in _start () at ../sysdeps/i386/elf/start.S:102 (gdb)
Created attachment 6382 [details] testcase (90 bytes)
Confirmed.
Postponed until GCC 3.4.2.
Postponed all ice-on-invalid bugs to GCC 3.4.3.
FWIW I can't reproduce this with cvs trunk.
I can reproduce on the mainline still (at least updated Wed Sep 8 18:32:57 UTC 2004): Reading specs from /Users/apinski/general_cleanup/lib/gcc/powerpc-apple-darwin8.0.0b2/3.5.0/ specs Reading specs from /Users/apinski/general_cleanup/lib/gcc/powerpc-apple- darwin8.0.0b2/3.5.0/../../../libgcj.spec rename spec lib to liborig Configured with: /Users/apinski/src/devel/general_cleanup/gcc/configure --prefix=/Users/apinski/ general_cleanup Thread model: posix gcc version 3.5.0 20040908 (experimental) /Users/apinski/general_cleanup/libexec/gcc/powerpc-apple-darwin8.0.0b2/3.5.0/jc1 bug.java - fhash-synchronization -fuse-divide-subroutine -fuse-boehm-gc -fnon-call-exceptions -fkeep-inline- functions -fPIC -quiet -dumpbase bug.java -auxbase bug -g1 -version -o /var/tmp//cc5ymwdk.s GNU Java version 3.5.0 20040908 (experimental) (powerpc-apple-darwin8.0.0b2) compiled by GNU C version 3.5.0 20040908 (experimental). GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096 Class path starts here: ./ /Users/apinski/general_cleanup/share/java/libgcj-3.5.0.jar/ (system) (zip) bug.java:4: error: '(' expected. Object ctor = new Object; ^ bug.java:4: internal compiler error: Bus error Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions.
Here is the backtrace: #0 java_parse () at lex.c:1740 #1 0x080c7b19 in java_parse_file (set_yydebug=0) at /home/gates/pinskia/src/gnu/gcc/src/gcc/ java/jcf-parse.c:1115 #2 0x083de193 in toplev_main (argc=4277860656, argv=0xfefb07a4) at /home/gates/pinskia/src/ gnu/gcc/src/gcc/toplev.c:992 #3 0x00125ad4 in __libc_start_main () from /lib/tls/libc.so.6 #4 0x08049d51 in _start ()
I looked in the debugger, and I'm definitely seeing a bad call error_if_numeric_overflow, with an argument of 0x15d (like the reporter). For whatever reason this doesn't cause a crash here, but it is definitely wrong.
I'm testing a fix.
Patch here: <http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02326.html>.
Subject: Bug 15656 CVSROOT: /cvs/gcc Module name: gcc Changes by: tromey@gcc.gnu.org 2004-09-24 15:42:57 Modified files: libjava : ChangeLog Added files: libjava/testsuite/libjava.compile: pr15656.java pr15656.xfail Log message: PR java/15656: * testsuite/libjava.compile/pr15656.xfail: New file. * testsuite/libjava.compile/pr15656.java: new file. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/ChangeLog.diff?cvsroot=gcc&r1=1.3088&r2=1.3089 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/testsuite/libjava.compile/pr15656.java.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/testsuite/libjava.compile/pr15656.xfail.diff?cvsroot=gcc&r1=NONE&r2=1.1
Subject: Bug 15656 CVSROOT: /cvs/gcc Module name: gcc Changes by: tromey@gcc.gnu.org 2004-09-24 15:44:37 Modified files: gcc/java : ChangeLog parse.y Log message: PR java/15656: * parse.y (class_instance_creation_expression): Set `$$' to NULL in error parts of rule. (unary_expression): Don't call error_if_numeric_overflow when $1 is NULL. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/ChangeLog.diff?cvsroot=gcc&r1=1.1462&r2=1.1463 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/parse.y.diff?cvsroot=gcc&r1=1.509&r2=1.510
Fix checked in.