Hi Mark,
On Fri, May 03, 2019 at 11:21:28AM +0100, Mark Rutland wrote:
> Generally, this makes sense, but I'm not sure that this is complete.
>
> IIUC this introduces a new type mismatch with sys_ni_syscall() in some
> cases.
Thanks for the review. You're correct, sys_ni_syscall needs to be fixed
too. I'll include this in v2.
> We probably need that to use SYSCALL_DEFINE0(), and maybe have a
> ksys_ni_syscall() for in-kernel wrappers.
Why would we need ksys_ni_syscall? It seems something like this should
be sufficient:
asmlinkage long sys_ni_syscall(void);
SYSCALL_DEFINE0(ni_syscall)
{
return sys_ni_syscall();
}
Sami
