[LTP] [PATCH] realtime/m4: Re-add define for robust mutexattr APIs

[LTP] [PATCH] realtime/m4: Re-add define for robust mutexattr APIs

[Joerg Vehlow]

From: Mathias Fiedler <mathias.fiedler@aox-tech.de>

This define is required by sbrk_mutex.c and testpi-6.c.
This define was dropped in
commit 3f6e7fe1e7d6afffb1db593e0dc14ca88a01cf66:
realtime: Fix pthread_mutexattr_{g,s}etrobust_np detection

Signed-off-by: Mathias Fiedler <mathias.fiedler@aox-tech.de>
Fixes: 3f6e7fe1e
---
 testcases/realtime/m4/check.m4 | 1 +
 1 file changed, 1 insertion(+)

diff --git a/testcases/realtime/m4/check.m4 b/testcases/realtime/m4/check.m4
index 5aa53bd42..e4d4fac75 100644
--- a/testcases/realtime/m4/check.m4
+++ b/testcases/realtime/m4/check.m4
@@ -17,6 +17,7 @@ AC_DEFUN([REALTIME_CHECK_ROBUST_APIS],[
 	AC_CHECK_DECLS([pthread_mutexattr_getrobust, pthread_mutexattr_setrobust],[],[has_robust="no"],[[#include <pthread.h>]])
 	AC_MSG_CHECKING([for pthread_mutexattr_*robust* APIs])
 if test "x$has_robust" != "xno"; then
+	AC_DEFINE(HAS_PTHREAD_MUTEXTATTR_ROBUST_APIS,1,[Define to 1 if you have pthread_mutexattr_*robust* APIs])
 	AC_MSG_RESULT(yes)
 else
 	AC_MSG_RESULT(no)
-- 
2.25.1

[LTP] [PATCH] realtime/m4: Re-add define for robust mutexattr APIs

[Cyril Hrubis]

Hi!
The code switched to the POSIX variants later on so I wonder if we even
need this configure check or if we can remove the ifdefs from the
testcases completely.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [PATCH] realtime/m4: Re-add define for robust mutexattr APIs

[Joerg Vehlow]

Hi,

On 11/19/2020 12:38 PM, Cyril Hrubis wrote:
> Hi!
> The code switched to the POSIX variants later on so I wonder if we even
> need this configure check or if we can remove the ifdefs from the
> testcases completely.
either that or this patch. At least at the moment these tests are 
completely broken.
I am not sure how long pthread_mutexattr_*roboust* api is part of posix 
and if ltp is supposed to be supported on platforms, that may not 
implement it.

But should accepting this patch wouldn't do any harm for now, but would 
fix the tests. In the future it can still be removed completely.

J?rg

PS: @Cyril Sorry for the resending this, forgot to include the mail list

[LTP] [PATCH] realtime/m4: Re-add define for robust mutexattr APIs

[Cyril Hrubis]

Hi!
> > The code switched to the POSIX variants later on so I wonder if we even
> > need this configure check or if we can remove the ifdefs from the
> > testcases completely.
> either that or this patch. At least at the moment these tests are 
> completely broken.
> I am not sure how long pthread_mutexattr_*roboust* api is part of posix 
> and if ltp is supposed to be supported on platforms, that may not 
> implement it.

As for LTP the only part that is supposed to be used on anything else
than Linux is the open posix testsuite. The rest, including the relatime
tests, is strictly Linux only.

Looking at the glibc the pthread_mutexattr_*robust* calls were
implemented in 2010 which is 10 years from now, which means that it's
old enough to just remove the checks.

Will you send a patch or should I do so?

-- 
Cyril Hrubis
chrubis@suse.cz