Re: [PATCH v3 1/2] ptrace: PTRACE_SET_SYSCALL_INFO syscall skipping support

[Oleg Nesterov]

On 07/07, Renzo Davoli wrote:
>
> +	child_op = ptrace_get_syscall_info_op(child);
> +	if (child_op != info.op) {
> +		if (info.op == PTRACE_SYSCALL_INFO_EXIT &&
> +				 child_op == PTRACE_SYSCALL_INFO_SECCOMP)
> +			skip_syscall = true;
> +		else
> +			return -EINVAL;

OK, I have to agree, we can't allow _ENTRY -> _EXIT transition right now.

Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>

---------------------------------------------------------------------------
Hopefully we can allow _ENTRY -> _EXIT later, when/if we have some help from
arch/ maintainers.

Although I think that the much more likely case is when Dmitry who understands
_every_ arch/ will have time/motivation to make the neccessary changes ;) Hmm...
but you removed him from CC list, please don't do this. Add him back.

And... As I said, I agree with any naming. But somehow I don't like "child_op".
Again, I won't argue, but if you will have any reason to send V4, please think
about something else. "reported_op" or just "op". To me "child_op" looks a bit
confusing, but this is minor/subjective, feel free to ignore.

Oleg.