Re: [PATCH v2 1/3] xtensa: add __NR_syscalls along with __NR_syscall_count

From: Firoz Khan
Date: Thu Nov 15 2018 - 05:05:38 EST


Hi Max,

On Wed, 14 Nov 2018 at 05:19, Max Filippov <jcmvbkbc@xxxxxxxxx> wrote:
>
> Hi Firoz,
>
> I have one more question:
>
> On Tue, Nov 13, 2018 at 2:20 AM Firoz Khan <firoz.khan@xxxxxxxxxx> wrote:
> > The 2nd option will be the recommended one. For that, I
> > added the __NR_syscalls macro in uapi/asm/unistd.h along
> > with __NR_syscall_count asm/unistd.h. The macro __NR_sys-
> > calls also added for making the name convention same across
> > all architecture. While __NR_syscalls isn't strictly part
> > of the uapi, having it as part of the generated header to
> > simplifies the implementation. We also need to enclose
> > this macro with #ifdef __KERNEL__ to avoid side effects.
>
> Looking at the include/uapi/asm-generic/unistd.h I see that
> __NR_syscalls is not guarded by the #ifdef __KERNEL__,
> why should it be guarded for xtensa?

__NR_syscalls for kernel. So this macro is present in asm/unistd.h
rather than uapi/asm/unistd.h. As I mentioned in the patch, it would
be better to keep in uapi/asm/unistd.h to count the number of syscalls.
But this will create some side effect. So I was guarded with __KERNEL__.
In order to come up with common implementation, I kept this for all
architecture.

>
> If we remove __NR_syscall_count from the uapi header I'd
> suggest dropping it completely and switching its two current
> users to __NR_syscalls.

I'm not removing the __NR_syscall_count macro; just place it
in asm/unistd.h file for the above reason.

FYI, I made sure that the kernel will build with my patch. I would
appreciate if you can perform the boot test on the actual platform.

Thanks
Firoz
>
> --
> Thanks.
> -- Max