This is the mail archive of the
newlib@sourceware.org
mailing list for the newlib project.
Re: [uClinux-dev] _vfork libc implementation on ARM
- From: "Shaun Jackman" <sjackman at gmail dot com>
- To: "David McCullough" <david_mccullough at au dot securecomputing dot com>
- Cc: "uClinux development list" <uclinux-dev at uclinux dot org>, "ARM Linux" <linux at arm dot linux dot org dot uk>, newlib at sources dot redhat dot com
- Date: Wed, 3 May 2006 14:37:53 -0600
- Subject: Re: [uClinux-dev] _vfork libc implementation on ARM
- References: <7f45d9390604281251x48db24cqc5007c46e3da4db7@mail.gmail.com> <20060429110626.GE17133@beast>
- Reply-to: "Shaun Jackman" <sjackman at gmail dot com>
On 4/29/06, David McCullough <david_mccullough@au.securecomputing.com> wrote:
Have a look at uClibc/glibc is the best bet. Looks something like this:
vfork:
swi __NR_vfork
...
I'm in doing something a little different than usual, though. I'm
running a Thumb user-space, and the system call number is passed in
$r7. The ABI says $r7 is saved by the callee. The callee (vfork) can't
use the stack to save the register. I was thinking of saving $r7 in
$r2 or $r3, which should be preserved through the swi by Linux.
Cheers,
Shaun