[PATCH v4] tests/fcntl.c: add test for struct f_owner_ex

[Dmitry V. Levin]

On Fri, Mar 30, 2018 at 05:02:27PM +0800, Zhibin Li wrote:
> *tests/fcntl.c (test_f_owner_ex_type_pid)
> (is_esrch_test_f_owner_ex, test_f_owner_ex): New functions.
> (main): Use test_f_owner_ex.
> ---
>  tests/fcntl.c | 36 ++++++++++++++++++++++++++++++++++++
>  1 file changed, 36 insertions(+)
> 
> diff --git a/tests/fcntl.c b/tests/fcntl.c
> index 4f62ca2a..7bbf2024 100644
> --- a/tests/fcntl.c
> +++ b/tests/fcntl.c
> @@ -69,12 +69,48 @@ test_flock64(void)
>  #endif
>  }
>  
> +static long
> +test_f_owner_ex_type_pid(const int cmd, const char *const cmd_name,
> +			 const int type, const char *const type_name,
> +			 const pid_t pid)
> +{
> +	struct f_owner_ex fo = { .type = type, .pid = pid };
> +
> +	long rc = invoke_test_syscall(cmd, &fo);
> +	printf("%s(0, %s, {type=%s, pid=%d}) = %s\n",
> +	       TEST_SYSCALL_STR, cmd_name, type_name, pid, sprintrc(rc));
> +	return rc;
> +}
> +
> +static void
> +is_esrch_test_f_owner_ex(const int type, const char *const type_name,
> +			 pid_t pid)

Now I wonder what does "is_esrch_" prefix mean here.

> +{
> +	long rc = test_f_owner_ex_type_pid(ARG_STR(F_SETOWN_EX),
> +				 type, type_name, pid);
> +	if (!rc)
> +		test_f_owner_ex_type_pid(ARG_STR(F_GETOWN_EX),
> +					 type, type_name, pid);

The indentation of test_f_owner_ex_type_pid arguments
is good only in the second invocation.

> +}
> +
> +static void
> +test_f_owner_ex(void)
> +{
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_TID), 1234567890);
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_PID), 1298126790);
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_PGRP), 1294567890);
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_TID), 20);
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_PID), 30);
> +	is_esrch_test_f_owner_ex(ARG_STR(F_OWNER_PGRP), 40);
> +}
> +
>  int
>  main(void)
>  {
>  	create_sample();
>  	test_flock();
>  	test_flock64();
> +	test_f_owner_ex();
>  
>  	puts("+++ exited with 0 +++");
>  	return 0;

This looks good, but the case when umove_or_printaddr(tcp, addr, &owner)
in print_f_owner_ex decoder return nonzero remains not covered by this
test.  Could you extend the test to check that case, too?


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20180404/76b948d1/attachment.bin>