Re: Cobalt Qube 2 kernel?
On Fri, May 02, 2003 at 04:11:20PM +0200, Guido Guenther wrote:
> In what ways are binutils broken, are there other things besides the
> "branch to globally visible symbol" issue (which should be fixed in the
> kernel anyway)?
> -- Guido
The problem I had was that binutils was generating a
BFD_RELOC_16_PCREL_S2 relocation, which is a Cygnusism. I suppose that
this is what you say... the first file in the CVS kernel that triggers
it is scall_o32.S -- so isn't there a simple rewrite to
dodge the problem?
LEAF(sys_sysmips)
beq a0, MIPS_ATOMIC_SET, mips_atomic_set
j _sys_sysmips
END(sys_sysmips)
_sys_sysmips is a global symbol: la, jr...
I tried to back-port code related to BFD_RELOC_16_PCREL into the
unstable binutils, and so far I've managed to break gas in various
weird ways without fixing the problem.
-- greg
Reply to: