On 12/08/17 01:15, Segher Boessenkool wrote:
On Fri, Aug 11, 2017 at 11:33:08AM +0200, Sebastian Huber wrote:
crtstuff.s: Assembler messages:
crtstuff.s:44: Error: can't resolve `.tm_clone_table' {.tm_clone_table
section} - `.LCTOC1' {.toc1 section}
crtstuff.s:45: Error: can't resolve `__TMC_END__' {*UND* section} -
`.LCTOC1' {.toc1 section}
Maybe you want to
#define USE_TM_CLONE_REGISTRY 0
for your subtarget?
I don't see what is the real problem causing your issues, but that
should allow you to get further at least.
I had to do this:
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=251275
The main issue was that I didn't define RS6000_ABI_NAME, so
SUBTARGET_OVERRIDE_OPTIONS() from sysv4.h selected ABI_V4. This triggered:
if (flag_pic > 1 && DEFAULT_ABI == ABI_V4) \
{ \
/* Note: flag_pic should not change any option flags that would \
be invalid with or pessimise -fno-PIC code. LTO turns off \
flag_pic when linking/recompiling a fixed position executable. \
However, if the objects were originally compiled with -fPIC, \
then other target options forced on here by -fPIC are restored \
when recompiling those objects without -fPIC. In particular \
TARGET_RELOCATABLE must not be enabled here by flag_pic. */ \
rs6000_isa_flags |= OPTION_MASK_MINIMAL_TOC; \
TARGET_NO_FP_IN_TOC = 1; \
}
The minimal TOC option apparently didn't work well for the crtbeginS.o.
With this patch I am now able to build RTEMS with the FreeBSD network
stack utilizing > 4GiB of RAM on a T4240. Segher, thanks for your kind
help. Without your help I would still struggle with the GCC configuration.
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.hu...@embedded-brains.de
PGP : Public key available on request.
Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
