[LTP] [PATCH v2 1/1] [STAGING] ioctl_pidfd06: accept EREMOTE/ESRCH for PIDFD_GET_INFO across pidns

[Cyril Hrubis]

Hi!
> PIDFD_GET_INFO semantics changed for pidfds referring to tasks without a PID
> in the caller's current pid namespace (commit ab89060fbc92e ("pidfs: return
> -EREMOTE when PIDFD_GET_INFO is called on another ns")).
> 
> Accept both errnos to keep the test working on kernels with and without the
> change (including backports).
> 
> Signed-off-by: Jan Polensky <japo@linux.ibm.com>
> ---
>  testcases/kernel/syscalls/ioctl/ioctl_pidfd06.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/testcases/kernel/syscalls/ioctl/ioctl_pidfd06.c b/testcases/kernel/syscalls/ioctl/ioctl_pidfd06.c
> index 751281bacb74..6b4bc3191e37 100644
> --- a/testcases/kernel/syscalls/ioctl/ioctl_pidfd06.c
> +++ b/testcases/kernel/syscalls/ioctl/ioctl_pidfd06.c
> @@ -14,6 +14,7 @@
>  
>  static struct tst_clone_args *args;
>  static struct pidfd_info *info;
> +static const int exp_errnos[] = {EREMOTE, ESRCH};
>  
>  static void run(void)
>  {
> @@ -41,7 +42,8 @@ static void run(void)
>  	args->exit_signal = SIGCHLD;
>  

I would be stricter for new kernels, we do not want the new value
reverted back by an accident. So what about:

static int err_nr = ARRAY_SIZE(exp_errnos);

static void setup(void)
{
	...
	if (tst_kvercmp(9, 19, 0) >= 0)
		err_nr = 1;
	...
}

>  	if (!SAFE_CLONE(args)) {
> -		TST_EXP_FAIL(ioctl(pidfd, PIDFD_GET_INFO, info), ESRCH);
> +		TST_EXP_FAIL_ARR(ioctl(pidfd, PIDFD_GET_INFO, info), exp_errnos,
> +				ARRAY_SIZE(exp_errnos));
				^
				err_nr

>  		exit(0);
>  	}
>  
> -- 
> 2.53.0
> 

-- 
Cyril Hrubis
chrubis@suse.cz