On Thu, Sep 28, 2006 at 12:46:20PM +0000, Christos Zoulas wrote:
> In article <20060928114656.GA9756@panix.com>,
> Thor Lancelot Simon  <tls@rek.tjls.com> wrote:
> >How, for example, do you intend to solve the problem that many functions
> >in the C library in Linux have the same name, but different signatures
> >than their NetBSD counterparts?
> 
> It is also sizes of structs. You'll need to wrap most of them with small
> translation shims and then call the real linux code. Same for syscalls that
> do not have the same numbers. Finally you need to fake other ELF library
> dependencies.

	Right, but you only have to do that for the linux functions that you use
from NetBSD code.  Then the shim does:
foo_shim()
{
	convert args
	syscall to switch emul mode to linux
	call linux func
	syscall to switch emul mode back to netbsd
	convert return values
}
but anything the linux func calls would be just other linux funcs.

i.e. basically the exact same thing the in-kernel emul code does, but the
other way around.

eric