by Geert Uytterhoeven

[permalink] [raw]

Subject: Re: [PATCH 2/4] m68k: Replace NR_syscalls macro from asm/unistd.h

Hi Firoz,

On Thu, Sep 20, 2018 at 10:12 AM Firoz Khan <[email protected]> wrote:
> On 18 September 2018 at 15:34, Geert Uytterhoeven <[email protected]> wrote:
> > On Tue, Sep 18, 2018 at 9:16 AM Firoz Khan <[email protected]> wrote:
> >> On 9 August 2018 at 13:00, Geert Uytterhoeven <[email protected]> wrote:
> >> > One first comment below...
> >> >
> >> > On Thu, Aug 9, 2018 at 7:16 AM Firoz Khan <[email protected]> wrote:
> >> >> NR_syscalls macro holds the number of system call exist in m68k
> >> >> architecture. This macro is currently the part of asm/unistd.h file.
> >> >> We have to change the value of NR_syscalls, if we add or delete a
> >> >> system call.
> >> >>
> >> >> One of patch in this patch series has a script which will generate
> >> >> a uapi header based on syscall.tbl file. The syscall.tbl file
> >> >> contains the number of system call information. So we have two
> >> >> option to update NR_syscalls value.
> >> >>
> >> >> 1. Update NR_syscalls in asm/unistd.h manually by counting the
> >> >> no.of system calls. No need to update NR_syscalls untill
> >> >> we either add a new system call or delete an existing system
> >> >> call.
> >> >>
> >> >> 2. We can keep this feature it above mentioned script, that'll
> >> >> count the number of syscalls and keep it in a generated file.
> >> >> In this case we don't need to explicitly update NR_syscalls
> >> >> in asm/unistd.h file.
> >> >>
> >> >> The 2nd option will be the recommended one. For that, I moved the
> >> >> NR_syscalls macro from asm/unistd.h to uapi/asm/unistd.h. The macro
> >> >> name also changed form NR_syscalls to __NR_syscalls 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.
> >> >
> >> > It can indeed not be part of the UAPI, as UAPI definitions can never change,
> >> > while new syscalls will be added in the future, increasing the number ;-)
> >>
> >> Thanks for your reply :)
> >> Sorry for the delayed response :(
> >>
> >> I would like to keep __NR_syscalls macro to uapi header in order to simplify
> >> the system call table generation script. Otherwise the __NR_syscalls
> >> macro need to update manually. That become a problem.
> >>
> >> Please check the below implementation in uapi file make sense?
> >> It is an easy workaround to leave __NR_syscalls macro in uapi/asm/unistd.h
> >> and enclose it in #ifdef __KERNEL__
> >>
> >> ...
> >> ...
> >> #define __NR_pwritev2 378
> >> #define __NR_statx 379
> >>
> >> #ifdef __KERNEL__
> >> #define __NR_syscalls 380
> >> #endif
> >> ...
> >> ...
> >
> > #ifdef __KERNEL__ sounds fine to me.
>
> I posted similar script for 10 different architectures. I got few good review
> from the maintainers and it will be applicable for all the
> architectures including
> m68k. There are few area which I identified need to improve. This will take
> couple of days.
>
> But it will be very helpful if you can perform the boot test on the
> actual platform
> and share the result.

Builds and boots fine on ARAnyM (virtual Atari).

So for the full series:
Tested-by: Geert Uytterhoeven <[email protected]>

However, I noticed the following effective difference between the old
arch/m68k/include/uapi/asm/unistd.h and the new generated
arch/m68k/include/generated/uapi/asm/unistd_32.h:

-/*#define __NR_break 17*/
-/*#define __NR_stty 31*/
-/*#define __NR_gtty 32*/
-/*#define __NR_ftime 35*/
-/*#define __NR_prof 44*/
-/*#define __NR_lock 53*/
-/*#define __NR_mpx 56*/
-/*#define __NR_ulimit 58*/
-/*#define __NR_oldolduname 59*/
-/*#define __NR_profil 98*/
-/*#define __NR_ioperm 101*/
-/*#define __NR_olduname 109*/
-/*#define __NR_iopl 110*/ /* not supported */
-/*#define __NR_idle 112*/ /* Obsolete */
-/*#define __NR_vm86 113*/ /* not supported */
-/*#define __NR_afs_syscall 137*/ /* Syscall for Andrew File System */
-/*#define __NR_vserver 278*/
+#define __NR_break 17
+#define __NR_stty 31
+#define __NR_gtty 32
+#define __NR_ftime 35
+#define __NR_prof 44
+#define __NR_lock 53
+#define __NR_mpx 56
+#define __NR_ulimit 58
+#define __NR_oldolduname 59
+#define __NR_profil 98
+#define __NR_ioperm 101
+#define __NR_olduname 109
+#define __NR_iopl 110
+#define __NR_idle 112
+#define __NR_vm86 113
+#define __NR_afs_syscall 137
+#define __NR_vserver 278

Given userspace code may contain checks for the presence of these
defines, I think they should not be present.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

2018-09-22 11:34:16

by Firoz Khan

[permalink] [raw]

Subject: Re: [PATCH 2/4] m68k: Replace NR_syscalls macro from asm/unistd.h

Hi Geert,

On Thu, 20 Sep 2018 at 14:54, Geert Uytterhoeven <[email protected]> wrote:
>
> Hi Firoz,
>
> On Thu, Sep 20, 2018 at 10:12 AM Firoz Khan <[email protected]> wrote:
> > On 18 September 2018 at 15:34, Geert Uytterhoeven <[email protected]> wrote:
> > > On Tue, Sep 18, 2018 at 9:16 AM Firoz Khan <[email protected]> wrote:
> > >> On 9 August 2018 at 13:00, Geert Uytterhoeven <[email protected]> wrote:
> > >> > One first comment below...
> > >> >
> > >> > On Thu, Aug 9, 2018 at 7:16 AM Firoz Khan <[email protected]> wrote:
> > >> >> NR_syscalls macro holds the number of system call exist in m68k
> > >> >> architecture. This macro is currently the part of asm/unistd.h file.
> > >> >> We have to change the value of NR_syscalls, if we add or delete a
> > >> >> system call.
> > >> >>
> > >> >> One of patch in this patch series has a script which will generate
> > >> >> a uapi header based on syscall.tbl file. The syscall.tbl file
> > >> >> contains the number of system call information. So we have two
> > >> >> option to update NR_syscalls value.
> > >> >>
> > >> >> 1. Update NR_syscalls in asm/unistd.h manually by counting the
> > >> >> no.of system calls. No need to update NR_syscalls untill
> > >> >> we either add a new system call or delete an existing system
> > >> >> call.
> > >> >>
> > >> >> 2. We can keep this feature it above mentioned script, that'll
> > >> >> count the number of syscalls and keep it in a generated file.
> > >> >> In this case we don't need to explicitly update NR_syscalls
> > >> >> in asm/unistd.h file.
> > >> >>
> > >> >> The 2nd option will be the recommended one. For that, I moved the
> > >> >> NR_syscalls macro from asm/unistd.h to uapi/asm/unistd.h. The macro
> > >> >> name also changed form NR_syscalls to __NR_syscalls 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.
> > >> >
> > >> > It can indeed not be part of the UAPI, as UAPI definitions can never change,
> > >> > while new syscalls will be added in the future, increasing the number ;-)
> > >>
> > >> Thanks for your reply :)
> > >> Sorry for the delayed response :(
> > >>
> > >> I would like to keep __NR_syscalls macro to uapi header in order to simplify
> > >> the system call table generation script. Otherwise the __NR_syscalls
> > >> macro need to update manually. That become a problem.
> > >>
> > >> Please check the below implementation in uapi file make sense?
> > >> It is an easy workaround to leave __NR_syscalls macro in uapi/asm/unistd.h
> > >> and enclose it in #ifdef __KERNEL__
> > >>
> > >> ...
> > >> ...
> > >> #define __NR_pwritev2 378
> > >> #define __NR_statx 379
> > >>
> > >> #ifdef __KERNEL__
> > >> #define __NR_syscalls 380
> > >> #endif
> > >> ...
> > >> ...
> > >
> > > #ifdef __KERNEL__ sounds fine to me.
> >
> > I posted similar script for 10 different architectures. I got few good review
> > from the maintainers and it will be applicable for all the
> > architectures including
> > m68k. There are few area which I identified need to improve. This will take
> > couple of days.
> >
> > But it will be very helpful if you can perform the boot test on the
> > actual platform
> > and share the result.
>
> Builds and boots fine on ARAnyM (virtual Atari).

Thanks for the support :)

>
> So for the full series:
> Tested-by: Geert Uytterhoeven <[email protected]>
>
> However, I noticed the following effective difference between the old
> arch/m68k/include/uapi/asm/unistd.h and the new generated
> arch/m68k/include/generated/uapi/asm/unistd_32.h:
>
> -/*#define __NR_break 17*/
> -/*#define __NR_stty 31*/
> -/*#define __NR_gtty 32*/
> -/*#define __NR_ftime 35*/
> -/*#define __NR_prof 44*/
> -/*#define __NR_lock 53*/
> -/*#define __NR_mpx 56*/
> -/*#define __NR_ulimit 58*/
> -/*#define __NR_oldolduname 59*/
> -/*#define __NR_profil 98*/
> -/*#define __NR_ioperm 101*/
> -/*#define __NR_olduname 109*/
> -/*#define __NR_iopl 110*/ /* not supported */
> -/*#define __NR_idle 112*/ /* Obsolete */
> -/*#define __NR_vm86 113*/ /* not supported */
> -/*#define __NR_afs_syscall 137*/ /* Syscall for Andrew File System */
> -/*#define __NR_vserver 278*/
> +#define __NR_break 17
> +#define __NR_stty 31
> +#define __NR_gtty 32
> +#define __NR_ftime 35
> +#define __NR_prof 44
> +#define __NR_lock 53
> +#define __NR_mpx 56
> +#define __NR_ulimit 58
> +#define __NR_oldolduname 59
> +#define __NR_profil 98
> +#define __NR_ioperm 101
> +#define __NR_olduname 109
> +#define __NR_iopl 110
> +#define __NR_idle 112
> +#define __NR_vm86 113
> +#define __NR_afs_syscall 137
> +#define __NR_vserver 278
>
> Given userspace code may contain checks for the presence of these
> defines, I think they should not be present.

My patch series had some different way to handle the above one. Some
how the plan got changed and missed it :(

I posted v2 patch series Thursday evening. This contain some modi-
fications suggested by different architecture maintainers including
you.

Please help us to review the patch series and please perform boot
test on the actual platform.

Thanks
Firoz

>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds