This is the mail archive of the gcc@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]

Re: undef'ed C++/dwarf labels, was: Failed compile


Jean-Pierre Radley wrote:

[ cvs update as of monday afternoon results in failure ] 

> /bin/sh ../../../libio/gen-params LIB_VERSION=2.8.0 MATH_H_INLINES=1 >tmp-params.h
> (EOF):undefined name: .L_T87
> gen-params: could not compile dummy.C with /s/tools/egcs/obj/gcc/xgcc -B/s/tools/egcs/obj/gcc/ -I. -I../../../libio -nostdinc++ -nostdinc++ -g -O2

Mine keels in a slightly different place, but it still keels.  I'm pretty
sure that it did work on Saturday.

My build dies here:

/home/play/negcs/gcc/xgcc -B/home/play/negcs/gcc/ -c -g -O2 -Wall  -fno-implicit-templates -I. -I../../../egcs/libio -nostdinc++  ../../../egcs/libio/builtinbuf.cc
(EOF):undefined name: .L_T87
make[1]: *** [builtinbuf.o] Error 1

The exact same source hosting and targeting Linux does finish the build,
making me wonder if it's a dwarf thing, though I see no changelogs for
dwarf*.c.


The occurrences in the source are mostly of the form:

        .section        .debug
.L_D400:
        .4byte  .L_D400_e-.L_D400
        .2byte  0x16    / TAG_typedef
        .2byte  0x12    / AT_sibling
        .4byte  .L_D401
        .2byte  0x38    / AT_name
        .string "global type"
        .2byte  0x72    / AT_user_def_type
        .4byte  .L_T87
.L_D400_e:
        .previous


I don't see this temporary label referenced anywhere outside of the 
dwarf debugging records.    If we remove '-g', the error goes away, 
of course.

Copies of generated assembly source or other help available to those
that understand such things and that would find them useful.


RJL


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