h8300 on testsuite and -wrap

[Jeff Law]

Joel Sherrill wrote:
> Hi,
>
> I am trying to run the gcc test suite on h8300-rtems
> and getting lots of failures which look like this:
>
>
>
> /home/joel/work-gnat/svn/b-gcc1-h8300/gcc/xgcc 
> -B/home/joel/work-gnat/svn/b-gcc1-h8300/gcc/ 
> /home/joel/work-gnat/svn/gcc/gcc/testsuite/gcc.c-torture/execute/20001031-1.c 
> gcc_tg.o  -w  -O1  -DSTACK_SIZE=2048 -mh -mint32 
> -B/home/joel/work-gnat/svn/bsp-install/h8300-rtems4.10/h8sim/lib/ 
> -specs bsp_specs -qrtems 
> /home/joel/work-gnat/svn/b-gcc1-h8300/rtems_gcc_main.o  -Wl,-wrap,exit 
> -Wl,-wrap,_exit -Wl,-wrap,main -Wl,-wrap,abort -lm   -o 
> /home/joel/work-gnat/svn/b-gcc1-h8300/gcc/testsuite/gcc/20001031-1.x1 
> -L /home/joel/work-gnat/svn/install/h8300-rtems4.10/lib/h8300h/ -lm
> /tmp/ccKBizhg.o: In function `.L20':
> 20001031-1.c:(.text+0x99): undefined reference to `__wrap__exit'
> gcc_tg.o: In function `___wrap_exit':
> testglue.c:(.text+0x173): undefined reference to `___real_exit'
> gcc_tg.o: In function `.L9':
> testglue.c:(.text+0x23d): undefined reference to `___real__exit'
> gcc_tg.o: In function `___wrap_abort':
> testglue.c:(.text+0x271): undefined reference to `___real_abort'
> gcc_tg.o: In function `___wrap_main':
> testglue.c:(.text+0x2b9): undefined reference to `___real_main'
> testglue.c:(.text+0x2c1): undefined reference to `__wrap__exit'
> /home/joel/work-gnat/svn/bsp-install/h8300-rtems4.10/h8sim/lib/librtemscpu.a(libposix_a-psignal.o): 
> In function `_POSIX_signals_Abnormal_termination_handler':
> /home/joel/work-gnat/svn//rtems/c/src/../../cpukit/posix/src/psignal.c:41: 
> undefined reference to `__wrap__exit'
>
> The arguments passed to ld look correct but looking
> into this with a simple h8300-elf-gcc build, I noticed
> that since gcc puts a leading _ on the symbols, you
> have to pass "-Wl,-wrap,_malloc" NOT -Wl,-wrap,malloc".
>
> Where is the leading underscore not getting dealt with
> so the wrap functions get invoked.
>
> FWIW I didn't see any h8300-elf results since last year
> and wonder if this is a known issue.
I ran into it as well when testing IRA.  Hell if I can remember what I 
did to fix it though....

jeff