[LTP] [PATCH v2 10/33] keyctl18: Test basic KEYCTL_MOVE

[Cyril Hrubis]

Hi!
> +static key_serial_t ring_a, ring_b;
> +static key_serial_t key;
> +
> +static void setup(void)
> +{
> +	SAFE_KEYCTL(KEYCTL_JOIN_SESSION_KEYRING, 0, 0, 0, 0);
> +
> +	ring_a = new_ring(RING_A_DESC);
> +	ring_b = new_ring(RING_B_DESC);
> +	key = new_user_key(KEY_DESC, PAYLOAD, sizeof(PAYLOAD), ring_a);
> +}
> +
> +static void reset_state(void)
> +{
> +	SAFE_KEYCTL(KEYCTL_LINK, key, ring_a, 0, 0);
> +	TEST(keyctl(KEYCTL_UNLINK, key, ring_b, 0, 0));
> +	if (TST_RET == -1 && TST_ERR != ENOENT)
> +		tst_brk(TBROK | TTERRNO, "failed to unlink key from ring_b");

We are trying to avoid hypotetical TBROK here in the case that the MOVE
didn't work right? I'm not sure it's worth the trouble.

> +}
> +
> +static void run(void)
> +{
> +	reset_state();
> +
> +	TST_EXP_PASS(keyctl(KEYCTL_MOVE, key, ring_a, ring_b, 0));
> +	if (!TST_PASS)
> +		return;
> +
> +	TST_EXP_EQ_LI(search_ring(ring_b, "user", KEY_DESC), key);
> +
> +	TST_EXP_FAIL2(search_ring(ring_a, "user", KEY_DESC), ENOKEY,
> +		      "key no longer found in the source keyring");

Here the search_string() function actually hides the syscall being
used. I would prefer having the keyctl() call in these two.

> +}
> +
> +static struct tst_test test = {
> +	.setup = setup,
> +	.test_all = run,
> +	.min_kver = "5.3",
> +};
> 
> -- 
> 2.51.0
> 
> 
> -- 
> Mailing list info: https://lists.linux.it/listinfo/ltp

-- 
Cyril Hrubis
chrubis@suse.cz