[LTP] [PATCH] wait403: Reproduce undefined INT_MIN negation

Joerg Vehlow lkml@jv-coder.de
Wed Jul 7 10:52:47 CEST 2021


Hi Richard,

On 7/6/2021 4:23 PM, Richard Palethorpe via ltp wrote:
> Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
> ---
>   runtest/syscalls                           |  1 +
>   testcases/kernel/syscalls/wait4/.gitignore |  1 +
>   testcases/kernel/syscalls/wait4/wait403.c  | 40 ++++++++++++++++++++++
>   3 files changed, 42 insertions(+)
>   create mode 100644 testcases/kernel/syscalls/wait4/wait403.c
>
> diff --git a/runtest/syscalls b/runtest/syscalls
> index 98fe3c02e..128eaadd2 100644
> --- a/runtest/syscalls
> +++ b/runtest/syscalls
> @@ -1663,6 +1663,7 @@ wait02 wait02
>   
>   wait401 wait401
>   wait402 wait402
> +wait403 wait403
>   
>   waitpid01 waitpid01
>   waitpid02 waitpid02
> diff --git a/testcases/kernel/syscalls/wait4/.gitignore b/testcases/kernel/syscalls/wait4/.gitignore
> index 9313eb72c..577f73479 100644
> --- a/testcases/kernel/syscalls/wait4/.gitignore
> +++ b/testcases/kernel/syscalls/wait4/.gitignore
> @@ -1,2 +1,3 @@
>   /wait401
>   /wait402
> +/wait403
> diff --git a/testcases/kernel/syscalls/wait4/wait403.c b/testcases/kernel/syscalls/wait4/wait403.c
> new file mode 100644
> index 000000000..262351b7c
> --- /dev/null
> +++ b/testcases/kernel/syscalls/wait4/wait403.c
> @@ -0,0 +1,40 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2021 SUSE LLC <rpalethorpe@suse.com>
> + */
> +
> +/*
> + * [Description]
> + *
> + * Check wait4(INT_MIN, ...) is not allowed. The pid is negated before
> + * searching for a group with that pid. Negating INT_MIN is not
> + * defined so UBSAN will be triggered if enabled.
> + *
> + * If the bug is present, but UBSAN is not enabled, then it should
> + * result in ECHILD.
> + */
> +
> +#include <stdlib.h>
> +#include <errno.h>
> +#define _USE_BSD
> +#include <sys/types.h>
> +#include <sys/resource.h>
> +#include <sys/wait.h>
> +#include "tst_test.h"
> +
> +static void run(void)
> +{
> +	int status = 1;
> +	struct rusage rusage;
> +
> +	TST_EXP_FAIL2(wait4(-2147483648, &status, 0, &rusage), ESRCH,
> +		      "wait4 fails with ESRCH");
Could this use INT_MIN instead of the literal value? If not a comment 
would be nice, especially because there is already wait4(INT_MIN, ...) 
in the description.
> +}
> +
> +static struct tst_test test = {
> +	.test_all = run,
> +	.taint_check = TST_TAINT_W | TST_TAINT_D,
> +	.tags = (const struct tst_tag[]) {
> +		{"linux-git", "dd83c161fbcc"},
> +	}
> +};

Otherwise:

Acked-by: Joerg Vehlow <joerg.vehlow@aox-tech.de>


Joerg


More information about the ltp mailing list