Re: [LTP] [PATCH v2 07/33] keyctl15: Test KEYCTL_GET_SECURITY label retrieval

[Cyril Hrubis <chrubis@suse.cz>]

Hi!
> diff --git a/testcases/kernel/syscalls/keyctl/keyctl15.c b/testcases/kernel/syscalls/keyctl/keyctl15.c
> new file mode 100644
> index 000000000..c281e9c69
> --- /dev/null
> +++ b/testcases/kernel/syscalls/keyctl/keyctl15.c
> @@ -0,0 +1,66 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2026 Andrea Cervesato <andrea.cervesato@suse.com>
> + */
> +
> +/*\
> + * Test ``KEYCTL_GET_SECURITY`` label retrieval of :manpage:`keyctl(2)`.
> + *
> + * ``KEYCTL_GET_SECURITY`` reads the LSM security label of a key.
> + * When no label is set (no LSM enabled or the LSM does not label keys)
> + * the operation returns 1 and an empty string.
> + *
> + * [Algorithm]
> + *
> + * - read the label of a valid key into a large buffer, verify the return
> + *   value is at least 1 and an empty string is returned when no label is set
> + */
> +
> +#include "keyctl_common.h"
> +
> +#define KEY_DESC	"ltpkeyctl15"
> +#define PAYLOAD		"payload"
> +#define BUF_SIZE	128
> +
> +static key_serial_t key;
> +static char buf[BUF_SIZE];
> +
> +static void setup(void)
> +{
> +	SAFE_KEYCTL(KEYCTL_JOIN_SESSION_KEYRING, 0, 0, 0, 0);
> +
> +	key = new_user_key(KEY_DESC, PAYLOAD, sizeof(PAYLOAD),
> +			   KEY_SPEC_PROCESS_KEYRING);
> +	SAFE_KEYCTL(KEYCTL_SETPERM, key, KEY_PERM_SET, 0, 0);
> +}
> +
> +static void run(void)
> +{
> +	memset(buf, 0, sizeof(buf));
> +
> +	TEST(keyctl(KEYCTL_GET_SECURITY, key, (unsigned long)buf, sizeof(buf), 0));
> +	if (TST_RET < 0)
> +		tst_brk(TBROK | TTERRNO, "KEYCTL_GET_SECURITY failed");
> +
TST_EXP_POSITIVE()?

> +	if (TST_RET < 1) {
> +		tst_res(TFAIL, "returned %ld, expected at least 1", TST_RET);
> +		return;
> +	}
> +	if (TST_RET == 1) {
> +		if (buf[0] != '\0') {
> +			tst_res(TFAIL, "empty label is not NUL terminated");
> +			return;
> +		}
> +
> +		tst_res(TPASS, "no label set, empty string returned");
> +		return;
> +	}

Is the string supposed to be null terminated? Shouldn't we just check
that the is null at the end regardless the length (we fail the test if <
1 either way)?

> +	tst_res(TPASS, "security label returned, full length %ld", TST_RET);
> +}
> +
> +static struct tst_test test = {
> +	.setup = setup,
> +	.test_all = run,
> +};
> 
> -- 
> 2.51.0
> 
> 
> -- 
> Mailing list info: https://lists.linux.it/listinfo/ltp

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp