linux-kernel - Re: [PATCH] ioctl_pidfd05: accept both EINVAL and ENOTTY as valid errors

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20251022121203.GA481852@pevik>
Date: Wed, 22 Oct 2025 14:12:03 +0200
From: Petr Vorel <pvorel@...e.cz>
To: Naresh Kamboju <naresh.kamboju@...aro.org>
Cc: ltp@...ts.linux.it, lkft@...aro.org, lkft-triage@...aro.org,
	arnd@...nel.org, dan.carpenter@...aro.org, jack@...e.cz,
	brauner@...nel.org, chrubis@...e.cz, linux-kernel@...r.kernel.org,
	linux-fsdevel@...r.kernel.org, lkft-triage@...ts.linaro.org,
	regressions@...ts.linux.dev, aalbersh@...nel.org, arnd@...db.de,
	viro@...iv.linux.org.uk, anders.roxell@...aro.org,
	benjamin.copeland@...aro.org, andrea.cervesato@...e.com
Subject: Re: [PATCH] ioctl_pidfd05: accept both EINVAL and ENOTTY as valid
 errors

Hi Naresh,

> Latest kernels return ENOTTY instead of EINVAL when invoking
> ioctl(pidfd, PIDFD_GET_INFO_SHORT, info_invalid).  Update the test to
> accept both EINVAL and ENOTTY as valid errors to ensure compatibility
> across different kernel versions.

Thanks a lot for contributing to LTP, we really appreciate it.

> Link: https://lore.kernel.org/all/CA+G9fYtUp3Bk-5biynickO5U98CKKN1nkE7ooxJHp7dT1g3rxw@mail.gmail.com
very nit: +1 for this. I prefer to reference it differently (e.g. [1]) as I add
Link: for referencing your actual patch the same way how it's used in kernel.
(e.g. Link: https://lore.kernel.org/ltp/20251022115704.46936-1-naresh.kamboju@linaro.org/)

> +++ b/testcases/kernel/syscalls/ioctl/ioctl_pidfd05.c
> @@ -4,7 +4,7 @@
>   */

>  /*\
> - * Verify that ioctl() raises an EINVAL error when PIDFD_GET_INFO is used. This
> + * Verify that ioctl() raises an EINVAL or ENOTTY error when PIDFD_GET_INFO is used. This
nit: maybe note for ENOTTY: (from v6.18)?
>   * happens when:
>   *
>   * - info parameter is NULL
> @@ -14,6 +14,7 @@
>  #include "tst_test.h"
>  #include "lapi/pidfd.h"
>  #include "lapi/sched.h"
> +#include <errno.h>
>  #include "ioctl_pidfd.h"

>  struct pidfd_info_invalid {
> @@ -43,7 +44,22 @@ static void run(void)
>  		exit(0);

>  	TST_EXP_FAIL(ioctl(pidfd, PIDFD_GET_INFO, NULL), EINVAL);
> -	TST_EXP_FAIL(ioctl(pidfd, PIDFD_GET_INFO_SHORT, info_invalid), EINVAL);
> +	/* Expect ioctl to fail; accept either EINVAL or ENOTTY */
> +	TEST(ioctl(pidfd, PIDFD_GET_INFO_SHORT, info_invalid));

I'm sorry, we prefer these macros in include/tst_test_macros.h which shorten the
code. Could you please use TST_EXP_FAIL_ARR() [1]?

Kind regards,
Petr

[1] https://linux-test-project.readthedocs.io/en/latest/developers/api_c_tests.html#macro-tst-exp-fail-arr

> +	if (TEST_RETURN == -1) {
> +		if (TEST_ERRNO == EINVAL || TEST_ERRNO == ENOTTY) {
> +			tst_res(TPASS,
> +				"ioctl(PIDFD_GET_INFO_SHORT) failed as expected with %s",
> +				tst_strerrno(TEST_ERRNO));
> +		} else {
> +			tst_res(TFAIL,
> +				"Unexpected errno: %s (expected EINVAL or ENOTTY)",
> +				tst_strerrno(TEST_ERRNO));
> +		}
> +	} else {
> +		tst_res(TFAIL, "ioctl(PIDFD_GET_INFO_SHORT) unexpectedly succeeded");
> +	}
> +

>  	SAFE_CLOSE(pidfd);
>  }