[patch 16/18] x86/entry: Get rid of the sys_ni_syscall() indirection

[H. Peter Anvin]

On 2026-10-22 08:02, Sebastian Andrzej Siewior wrote:
> On 2026-10-22 16:24:30 [+0200], To Thomas Gleixner wrote:
>> The test could be updated via the hunk below in case the new behaviour
>> is what is expected. hpa made the extension to the test in commit
>> 795e2a023b808 ("selftests/x86/syscall: Add tests under ptrace to
>> syscall_numbering_64").
>>
>> diff --git a/tools/testing/selftests/x86/syscall_numbering.c b/tools/testing/selftests/x86/syscall_numbering.c
>> index ca0eca7b9dced..1435394fba022 100644
>> --- a/tools/testing/selftests/x86/syscall_numbering.c
>> +++ b/tools/testing/selftests/x86/syscall_numbering.c
> 
> Now I notice the fix depends on CONFIG_X86_X32_ABI=y (fails for =n).
> But we want to get rid of it anyway so…
> CONFIG_IA32_EMULATION is not affected (with the fix).
> 
It would be very simple to probe for by issuing a simple system call like
getppid() with the x32 bit set.

	-hpa