[PATCH v2] selftests/seccomp: fix ptrace tests on powerpc

[Michael Ellerman]

Kees Cook <keescook at chromium.org> writes:
> On Fri, Sep 11, 2020 at 03:10:12PM -0300, Thadeu Lima de Souza Cascardo wrote:
...
>> diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c
>> index 7a6d40286a42..0ddc0846e9c0 100644
>> --- a/tools/testing/selftests/seccomp/seccomp_bpf.c
>> +++ b/tools/testing/selftests/seccomp/seccomp_bpf.c
>> @@ -1916,10 +1957,15 @@ void tracer_ptrace(struct __test_metadata *_metadata, pid_t tracee,
>>  	EXPECT_EQ(entry ? PTRACE_EVENTMSG_SYSCALL_ENTRY
>>  			: PTRACE_EVENTMSG_SYSCALL_EXIT, msg);
>>  
>> -	if (!entry)
>> +	if (!entry && !variant)
>>  		return;
>>  
>> -	nr = get_syscall(_metadata, tracee);
>> +	if (entry)
>> +		nr = get_syscall(_metadata, tracee);
>> +	else if (variant)
>> +		nr = variant->syscall_nr;
>> +	if (variant)
>> +		variant->syscall_nr = nr;
>
> So, to be clear this is _only_ an issue for the ptrace side of things,
> yes? i.e. seccomp's setting of the return value will correct stick?

Yes. There's a comment which (hopefully) explains the difference here:

  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/powerpc/kernel/ptrace/ptrace.c?commit=ab29a807a7ddaa7c84d2f4cb8d29e74e33759072#n239

Which says:

static int do_seccomp(struct pt_regs *regs)
{
	if (!test_thread_flag(TIF_SECCOMP))
		return 0;

	/*
	 * The ABI we present to seccomp tracers is that r3 contains
	 * the syscall return value and orig_gpr3 contains the first
	 * syscall parameter. This is different to the ptrace ABI where
	 * both r3 and orig_gpr3 contain the first syscall parameter.
	 */
	regs->gpr[3] = -ENOSYS;


cheers