This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: bootstrapping with objc enabled busted


On Mon, Jul 29, 2002 at 10:15:26PM +0930, Alan Modra wrote:
> When bootstrapping mainline on powerpc-linux
> 
> /usr/tmp/ppc/gcc-curr/gcc/xgcc -B/usr/tmp/ppc/gcc-curr/gcc/ -B/usr/local/powerpc-linux/bin/ -B/usr/local/powerpc-linux/lib/ -isystem /usr/local/powerpc-linux/include -fgnu-runtime -c -I. -I/src/gcc-current/libobjc -O2 -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes -DIN_GCC -DIN_TARGET_LIBS -I/src/gcc-current/libobjc/objc -I/src/gcc-current/libobjc/../gcc -I/src/gcc-current/libobjc/../gcc/config -I../../gcc -I/src/gcc-current/libobjc/../include /src/gcc-current/libobjc/NXConstStr.m -o NXConstStr.o
> /src/gcc-current/libobjc/objc/hash.h:204: internal compiler error: internal error: Segmentation fault

Sorry for the confused bug-report.  I think a problem with my
bootstrap compiler caused the above, while adding -v -save-temps to
the above command line caused the problem below.

So a recipe for reproducing the problem is to take the NXConstStr.m
command line from your latest bootstrap, and add -save-temps.

> Seems to be fallout from 2002-07-07 Neil Booth.  cc1obj -E causes some
> initializations in lang_independent_init to be skipped, and later
> objc-act.c:synth_module_prologue does
> 	  create_builtin_decl (VAR_DECL, temp_type,
> 			       "_OBJC_SELECTOR_TABLE");
> 
> Starting program: /usr/tmp/ppc/gcc-curr/gcc/cc1obj -E ...
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x10346344 in splay_tree_splay (sp=0x0, key=3) at /src/gcc-current/libiberty/splay-tree.c:195
> 
> (gdb) bt
> #0  0x10346344 in splay_tree_splay (sp=0x0, key=3) at /src/gcc-current/libiberty/splay-tree.c:195
> #1  0x10346788 in splay_tree_lookup (sp=0x0, key=3) at /src/gcc-current/libiberty/splay-tree.c:409
> #2  0x10077494 in get_alias_set_entry (alias_set=3) at /src/gcc-current/gcc/alias.c:215
> #3  0x10078bb0 in record_alias_subset (superset=2594073677423181827, subset=4) at /src/gcc-current/gcc/alias.c:638
> #4  0x10078e94 in record_component_aliases (type=0x40045d00) at /src/gcc-current/gcc/alias.c:714
> #5  0x10078808 in get_alias_set (t=0x40045c00) at /src/gcc-current/gcc/alias.c:595
> #6  0x10078d50 in record_component_aliases (type=0x40045e80) at /src/gcc-current/gcc/alias.c:718
> #7  0x10078808 in get_alias_set (t=0x40045e80) at /src/gcc-current/gcc/alias.c:595
> #8  0x100cdaf4 in set_mem_attributes (ref=0x2, t=0x40045f00, objectp=271908864) at /src/gcc-current/gcc/emit-rtl.c:1714
> #9  0x102a3b58 in make_decl_rtl (decl=0x40045f00, asmspec=0x0) at /src/gcc-current/gcc/varasm.c:955
> #10 0x10009054 in create_builtin_decl (code=VAR_DECL, type=0x40045f00, name=0x0) at /src/gcc-current/gcc/objc/objc-act.c:1002
> #11 0x10009420 in synth_module_prologue () at /src/gcc-current/gcc/objc/objc-act.c:1145
> #12 0x1001b2f0 in init_objc () at /src/gcc-current/gcc/objc/objc-act.c:8022
> #13 0x1000790c in objc_init (filename=0x0) at /src/gcc-current/gcc/objc/objc-act.c:505
> #14 0x10285094 in lang_dependent_init (name=0x103e0000 'c' <repeats 64 times>, "%d") at /src/gcc-current/gcc/toplev.c:5161
> #15 0x10285320 in do_compile (no_backend=1) at /src/gcc-current/gcc/toplev.c:5277
> #16 0x102853f8 in toplev_main (argc=272498688, argv=0xffffe964) at /src/gcc-current/gcc/toplev.c:5310
> #17 0x10077468 in main (argc=0, argv=0x3) at /src/gcc-current/gcc/main.c:35
> #18 0x0feca9a0 in __libc_start_main () from /lib/libc.so.6
> 
> Why #10 has name=0, I don't know.  gdb bug probably, it really is
> "_OBJC_SELECTOR_TABLE"
> 
> -- 
> Alan Modra
> IBM OzLabs - Linux Technology Centre

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]