Re: [PATCH v2] arch: Reserve map_shadow_stack() syscall number for all architectures

[Catalin Marinas]

On Thu, Sep 14, 2023 at 06:58:03PM +0000, Sohil Mehta wrote:
> commit c35559f94ebc ("x86/shstk: Introduce map_shadow_stack syscall")
> recently added support for map_shadow_stack() but it is limited to x86
> only for now. There is a possibility that other architectures (namely,
> arm64 and RISC-V), that are implementing equivalent support for shadow
> stacks, might need to add support for it.
> 
> Independent of that, reserving arch-specific syscall numbers in the
> syscall tables of all architectures is good practice and would help
> avoid future conflicts. map_shadow_stack() is marked as a conditional
> syscall in sys_ni.c. Adding it to the syscall tables of other
> architectures is harmless and would return ENOSYS when exercised.
> 
> Note, map_shadow_stack() was assigned #453 during the merge process
> since #452 was taken by fchmodat2().
> 
> For Powerpc, map it to sys_ni_syscall() as is the norm for Powerpc
> syscall tables.
> 
> For Alpha, map_shadow_stack() takes up #563 as Alpha still diverges from
> the common syscall numbering system in the other architectures.
> 
> Link: https://lore.kernel.org/lkml/20230515212255.GA562920@xxxxxxxxxxxxxxxxxxxxx/
> Link: https://lore.kernel.org/lkml/b402b80b-a7c6-4ef0-b977-c0f5f582b78a@xxxxxxxxxxxxx/
> 
> Signed-off-by: Sohil Mehta <sohil.mehta@xxxxxxxxx>
> ---
> v2:
> - Skip syscall table changes to tools/. They will be handled separetely by the
>   perf folks.
> - Map Powerpc to sys_ni_syscall (Rick Edgecombe)
> ---
>  arch/alpha/kernel/syscalls/syscall.tbl      | 1 +
>  arch/arm/tools/syscall.tbl                  | 1 +
>  arch/arm64/include/asm/unistd.h             | 2 +-
>  arch/arm64/include/asm/unistd32.h           | 2 ++

For arm64 (compat):

Acked-by: Catalin Marinas <catalin.marinas@xxxxxxx>