The changes to bsd.cpu.mk that append to LDFLAGS will most likely
break building kernel modules. Can you build a kernel module in a tree
with your patch?
The problem is that LDFLAGS are directly passed as arguments to the
linker in bsd.kmod.mk and the linker does not like the
-Wl,-linker_option style of setting options.
Do you really need to set LDFLAGS in bsd.cpu.mk? It would seem that
simply setting LD would work.
On Mon, Mar 1, 2010 at 6:28 AM, C. Jayachandran
> I'm in the process of getting n32 ABI support for the RMI processors.
> The plan is have n32 compiled kernel with 64-bit register_t and
> physaddr_t and 32 bit long and pointer types.
> I've attached two patches, one for support for n32 in toolchain and
> one for n32 support in kernel compilation. With this I am able to
> compile the kernel and user space with n32, and the boot-up reaches
> until init. There is a lot more work on user-space and kernel (esp in
> mips/mips/*.S) before it can complete boot-up.
> Please review and let me know if you have any comments or objections
> on this approach. The patches are:
> Toolchain support for N32
> - Adds the linker emulations needed for n32
> - Common preprocessor defines for ABI (_ABI_MIPS_SIM and _ABI???).
> - Sets the long double type as 64 bit (this should be 128 bit in n32,
> but there is some work needed to get the 128 bit soft-float working).
> N32 compilation - makefiles and conf
> - Adds ldscript.mips.n32.
> - Some cleanup in Makefile.mips, add ABI flags
> - bsd.cpu.mk CFLAGS for n32 compilation and linking
> I have introduced a TARGET_N32 similart to TARGET_64 for n32
> compilation. But I think on the long term, we need clean up the
> different flags that affect architecture and ABI. Currently there is
> an overlap between the TARGET_CPUTYPE flag, the ISA_<arch> flags and
> the TARGET_<type> flags.
> email@example.com mailing list
> To unsubscribe, send any mail to "firstname.lastname@example.org"