Re: [LTP] [PATCH] syscalls/semctl08: Skip semctl08 when __USE_TIME64_REDIRECTS is defined

[Petr Vorel <pvorel@suse.cz>]

Hi Jiaying,

[ NOTE: Ricardo was wrongly set the sender by patchwork ]

> From: Jiaying Song <jiaying.song.cn@windriver.com>

> When __USE_TIME64_REDIRECTS is defined, glibc redirects struct semid_ds to a
> 64-bit time-safe version that omits the sem_otime_high and sem_ctime_high
> fields. As a result, the case becomes invalid and leads to incorrect behavior.

> This patch adds a check to skip the test when __USE_TIME64_REDIRECTS is
> defined, ensuring the test only runs when semid_ds structurally matches
> semid64_ds and the *_high fields are accessible.

Looking into this old patch. Depending on glibc specific internal definition is
not a preferred approach (definition can be renamed). I wonder if this could be
detected with m4 macro or otherwise (some kind of sizeof check).

IMHO glibc has not switched to 64bit time-safe version by default yet, right?
And it depends on glibc being compiled with -D_TIME_BITS=64 and
-D_FILE_OFFSET_BITS=64, right? But these aren't detectable, right?

Looking at glibc sources [1], _TIME_BITS and _FILE_OFFSET_BITS really determine
__USE_TIME64_REDIRECTS:

#if defined _TIME_BITS
# if _TIME_BITS == 64
#  if ! defined (_FILE_OFFSET_BITS) || _FILE_OFFSET_BITS != 64
#   error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64"
#  endif
#  define __USE_TIME_BITS64     1
# elif _TIME_BITS == 32
#  if __TIMESIZE > 32
#   error "_TIME_BITS=32 is not compatible with __TIMESIZE > 32"
#  endif
# else
#  error Invalid _TIME_BITS value (can only be 32 or 64-bit)
# endif
#elif __TIMESIZE == 64
# define __USE_TIME_BITS64      1
#endif

#if defined __USE_TIME_BITS64 && __TIMESIZE == 32
# define __USE_TIME64_REDIRECTS 1
#endif

_TIME_BITS toolchain support is detected by configure.ac [2]

and nowadays is probably 64 bit by default, see  timezone/private.h [3]:

#if !defined _TIME_BITS && _FILE_OFFSET_BITS == 64
# define _TIME_BITS 64
#endif

And the same question applies to your other msgctl05 related patch [4].

Kind regards,
Petr

[1] https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/features-time64.h;h=8d573cd23e48c0f771c72350168b04c6a38c7864;hb=a4ed0471d71739928a0d0fa3258b3ff3b158e9b9
[2] https://sourceware.org/git/?p=glibc.git;a=blob;f=configure.ac;h=af57b0cbae8c1b22bfdc7a4e47d46c4e5064873a;hb=HEAD#l1826
[3] https://sourceware.org/git/?p=glibc.git;a=blob;f=timezone/private.h;h=c33041049f4c8f40b874c23b6fe0556967069b2c;hb=HEAD#l167
[4] https://patchwork.ozlabs.org/project/ltp/patch/20250522093003.1958528-1-jiaying.song.cn@windriver.com/

> Signed-off-by: Jiaying Song <jiaying.song.cn@windriver.com>
> ---
>  testcases/kernel/syscalls/ipc/semctl/semctl08.c | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)

> diff --git a/testcases/kernel/syscalls/ipc/semctl/semctl08.c b/testcases/kernel/syscalls/ipc/semctl/semctl08.c
> index 1878bd4..3b799fa 100644
> --- a/testcases/kernel/syscalls/ipc/semctl/semctl08.c
> +++ b/testcases/kernel/syscalls/ipc/semctl/semctl08.c
> @@ -10,7 +10,11 @@
>  #include "tst_test.h"
>  #include "libnewipc.h"

> -#ifdef HAVE_SEMID64_DS_TIME_HIGH
> +#if !defined(HAVE_SEMID64_DS_TIME_HIGH)
> +TST_TEST_TCONF("test requires struct semid64_ds to have the time_high fields");
> +#elif defined(__USE_TIME64_REDIRECTS)
> +TST_TEST_TCONF("test requires __USE_TIME64_REDIRECTS to be undefined");
> +#else

>  static void run(void)
>  {
> @@ -47,6 +51,4 @@ static struct tst_test test = {
>  	.test_all = run,
>  	.needs_tmpdir = 1,
>  };
> -#else
> -TST_TEST_TCONF("test requires struct semid64_ds to have the time_high fields");
>  #endif

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