fix "sigaction" tests

Hi,

POSIX "conformance/interfaces/sigaction/16-1" and other 25 tests failed with following message:
------------
conformance/interfaces/sigaction/16-1: execution: FAILED: Output:
Caught signal 6
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-1.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-10: execution: FAILED: Output:
Caught signal 13
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-10.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-11: execution: FAILED: Output:
Caught signal 3
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-11.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-12: execution: FAILED: Output:
Caught signal 11
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-12.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-13: execution: FAILED: Output:
Caught signal 15
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-13.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-14: execution: FAILED: Output:
Caught signal 20
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-14.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-15: execution: FAILED: Output:
Caught signal 21
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-15.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-16: execution: FAILED: Output:
Caught signal 22
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-16.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-17: execution: FAILED: Output:
Caught signal 10
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-17.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-18: execution: FAILED: Output:
Caught signal 12
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-18.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-19: execution: FAILED: Output:
Caught signal 29
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-19.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-2: execution: FAILED: Output:
Caught signal 14
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-2.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-20: execution: FAILED: Output:
Caught signal 27
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-20.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-21: execution: FAILED: Output:
Caught signal 31
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-21.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-22: execution: FAILED: Output:
Caught signal 5
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-22.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-23: execution: FAILED: Output:
Caught signal 23
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-23.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-24: execution: FAILED: Output:
Caught signal 26
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-24.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-25: execution: FAILED: Output:
Caught signal 24
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-25.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-26: execution: FAILED: Output:
Caught signal 25
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-26.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-3: execution: FAILED: Output:
Caught signal 7
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-3.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-4: execution: FAILED: Output:
Caught signal 17
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-4.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-5: execution: FAILED: Output:
Caught signal 18
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-5.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-6: execution: FAILED: Output:
Caught signal 8
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-6.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-7: execution: FAILED: Output:
Caught signal 1
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-7.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-8: execution: FAILED: Output:
Caught signal 4
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-8.c FAILED: The function returned EINTR while SA_RESTART is set conformance/interfaces/sigaction/16-9: execution: FAILED: Output:
Caught signal 2
[16:42:34]Test ../../../conformance/interfaces/sigaction/16-9.c FAILED: The function returned EINTR while SA_RESTART is set
------------

Environment is as follows:
 - RHEL4.8 (kernel version 2.6.9-89)   --- (x86, x86_64, ia64)
 - RHEL5.5 (kernel version 2.6.18-194) --- (x86, x86_64, ia64)

The cause of above error is that sem_wait() of the testing kernel version always returns with EINTR.

"man sem_wait" says:
------------
[...]

      EINTR  The call was interrupted by a signal handler.

[...]

NOTES
      A  signal handler always interrupts a blocked call to one of these functions, regardless
      of the use of the sigaction(2) SA_RESTART flag.

POSIX.1-2001.
------------

And "man 7 signal" of "RHEL6.0 Beta"(*) says:
------------
      If a blocked call to one of the following interfaces is interrupted by
      a  signal handler, then the call will be automatically restarted after
      the signal handler returns if the SA_RESTART flag was used; otherwise
      the call will fail with the error EINTR:

[...]

          * POSIX  semaphore  interfaces:  sem_wait(3)  and sem_timedwait(3)
            (since Linux 2.6.22; beforehand, always failed with EINTR).

[...]

  Interruption of System Calls and Library Functions by Stop Signals On Linux, even in the absence of  signal  handlers,  certain blocking
      interfaces  can fail with the error EINTR after the process is stopped
      by one of the  stop  signals  and  then  resumed  via  SIGCONT. This
      behavior is not sanctioned by POSIX.1, and doesn’t occur on other sys-
      tems.

[...]

          * Linux 2.6.21 and earlier: futex(2) FUTEX_WAIT, sem_timedwait(3),
            sem_wait(3).

------------
(*) -- Above description didn't exist in "man" of RHEL4.8/5.5. But "man" of RHEL6.0 Beta says that "sem_wait()" failed in the version earlier than Linux 2.6.22.

Therefore, These 26 tests are useful after 2.6.22 version, I think.

I tried to revised "testcases/open_posix_testsuite/conformance/interfaces/ sigaction/templates/template_16-1.in"

============
---
a/testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/ template_16-1.in        2010-10-14 03:58:31.000000000 +0900
+++
b/testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/ template_16-1.new       2010-10-18 17:47:40.000000000 +0900

testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.new renamed from testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.in +2 -0

@@ -55,6 +55,8 @@
  #include <signal.h>
  #include <errno.h>
 
+#include <sys/utsname.h>
+
 
 /***************************************************************************
 ***/

 /***************************   Test framework
*******************************/

/*************************************************************************** ***/

testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.new renamed from testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.in +23 -0

@@ -124,6 +126,23 @@
        return NULL;
  }
 
+/* Checking for Kernel Version function */
+int kernel_ver_cmp(int k1, int k2, int k3)
+{
+       struct utsname buf;
+       int r1, r2, r3;
+
+       uname(&buf);
+       sscanf(buf.release, "%d.%d.%d", &r1, &r2, &r3);
+       if (r1 > k1)
+               return 0;
+       if (r2 > k2)
+               return 0;
+       if (r3 > k3)
+               return 0;
+       return 1;
+}
+
  /* main function */
  int main()
  {

@@ -133,6 +152,12 @@
 
        struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
        /* Initialize output */
        output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Similar revisions are necessary for other 25 "sigaction" tests.

2010/10/31 Mitani <[email protected]>:

Quoted 25 lines

int main()
{

testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.new renamed from testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.in +6 -0

@@ -133,6 +152,12 @@
 
       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Thanks!
-Garrett

Quoted 1 line · Garrett Cooper

2010/10/31 Mitani <[email protected]>:

...

Quoted 25 lines

Quoted 25 lines

 int main()
 {

testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.new renamed from testcases/open_posix_testsuite/conformance/interfaces/sigaction/templates/template_16-1.in +6 -0

@@ -133,6 +152,12 @@
 
        struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
        /* Initialize output */
        output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 5 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

    Sorry, but Linux-isms can't leak into the open_posix_testsuite. Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Quoted 32 lines · Gui Jianfeng

Quoted 1 line · Garrett Cooper

2010/10/31 Mitani <[email protected]>:

...

Quoted 29 lines

Quoted 18 lines

int main()
{

(quoted diff) +6 -0

@@ -133,6 +152,12 @@
 
       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

============

Quoted 8 lines

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Quoted 1 line

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

Quoted 35 lines · Garrett Cooper

Quoted 33 lines · Gui Jianfeng

Quoted 1 line · Garrett Cooper

2010/10/31 Mitani <[email protected]>:

...

Quoted 25 lines

Quoted 25 lines

 int main()
 {

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +6 -0

@@ -133,6 +152,12 @@
 
        struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
        /* Initialize output */
        output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 4 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

    Sorry, but Linux-isms can't leak into the open_posix_testsuite. Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Quoted 4 lines

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

 This open_posix_suite is the part of LTP(linux test project), must we make sure it can work at all versions of unix?

What about this patch?  If you think so, other similar patchs will be post.

-------------------------------------------------------------------------------

At older linux kernel(2.6.22 before), a signal handler always interrupts a blocked call to one of these functions, regardless of the use of the sigaction SA_RESTART flag, so we should not test these cases at this situation.

Althought uname is not exsit at 4.3 BSD, but this open_posix_suite just the part of LTP(linux test project), so i think it's not a problem.

Signed-off-by Bian Naimeng <[email protected]>

---
 .../conformance/interfaces/sigaction/16-1.c        |   27 ++++++++++++++++++++

 1 files changed, 27 insertions(+), 0 deletions(-)

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +22 -0

index 706dbfc..3d6e1a1 100644
@@ -50,6 +50,7 @@ Anyway, a false negative status cannot be returned.
  #include <stdlib.h>
  #include <string.h>
  #include <unistd.h>
+#include <sys/utsname.h>
 
  #include <semaphore.h>
  #include <signal.h>

@@ -95,6 +96,27 @@ Anyway, a false negative status cannot be returned.
  volatile sig_atomic_t caught = 0;
  sem_t sem;
 
+#define KERNEL_VERSION(a, b, c)        (((a) << 16) + ((b) << 8) + (c))
+
+int check_linux()
+{
+       struct utsname tn;
+       int r1, r2, r3;
+
+       if (0 > uname(&tn)) {
+               UNRESOLVED( errno, "uname failed" );
+       }
+
+       if (strcmp(tn.sysname, "Linux") != 0)
+               return 0;
+
+       sscanf(tn.release, "%d.%d.%d", &r1, &r2, &r3);
+       if (KERNEL_VERSION(r1, r2, r3) >= KERNEL_VERSION(2, 6, 22))
+               return 1;
+
+       return 0;
+}
+
  /* Handler function */
  void handler( int signo )
  {

2010/11/17 Bian Naimeng <[email protected]>:

Quoted 40 lines

Quoted 38 lines · Garrett Cooper

Quoted 28 lines · Gui Jianfeng

Quoted 1 line · Garrett Cooper

2010/10/31 Mitani <[email protected]>:

...

Quoted 25 lines

Quoted 25 lines

int main()
{

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +6 -0

@@ -133,6 +152,12 @@
 
       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 5 lines

Quoted 4 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

Quoted 2 lines

This open_posix_suite is the part of LTP(linux test project), must we
make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

Quoted 68 lines

What about this patch? If you think so, other similar patchs will be post.

-------------------------------------------------------------------------------

At older linux kernel(2.6.22 before), a signal handler always interrupts a blocked call to one of these functions, regardless of the use of the sigaction SA_RESTART flag, so we should not test these cases at this situation.

Althought uname is not exsit at 4.3 BSD, but this open_posix_suite just the part of LTP(linux test project), so i think it's not a problem.

Signed-off-by Bian Naimeng <[email protected]>

---
 .../conformance/interfaces/sigaction/16-1.c        |   27 ++++++++++++++++++++
 1 files changed, 27 insertions(+), 0 deletions(-)

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +27 -0

index 706dbfc..3d6e1a1 100644
@@ -50,6 +50,7 @@ Anyway, a false negative status cannot be returned.
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <sys/utsname.h>
 
 #include <semaphore.h>
 #include <signal.h>

@@ -95,6 +96,27 @@ Anyway, a false negative status cannot be returned.
 volatile sig_atomic_t caught = 0;
 sem_t sem;
 
+#define KERNEL_VERSION(a, b, c)        (((a) << 16) + ((b) << 8) + (c))
+
+int check_linux()
+{
+       struct utsname tn;
+       int r1, r2, r3;
+
+       if (0 > uname(&tn)) {
+               UNRESOLVED( errno, "uname failed" );
+       }
+
+       if (strcmp(tn.sysname, "Linux") != 0)
+               return 0;
+
+       sscanf(tn.release, "%d.%d.%d", &r1, &r2, &r3);
+       if (KERNEL_VERSION(r1, r2, r3) >= KERNEL_VERSION(2, 6, 22))
+               return 1;
+
+       return 0;
+}
+
 /* Handler function */
 void handler( int signo )
 {

@@ -133,6 +155,11 @@ int main()
 
       struct sigaction sa;
 
+       if (!check_linux())
+               UNTESTED( "A signal handler always interrupts sem_wait "
+                         "at linux release, regardless of the use of "
+                         "the sigaction SA_RESTART flag" );
+
       /* Initialize output */
       output_init();

-----Original Message-----
From: Bian Naimeng [mailto:[email protected]]
Sent: Thursday, November 18, 2010 2:00 PM
To: Garrett Cooper
Cc: Gui Jianfeng; [email protected]; 當座 健市; Mitani
Subject: Re: [LTP] [PATCH] fix "sigaction" tests

Quoted 4 lines · Garrett Cooper

2010/11/17 Bian Naimeng <[email protected]>:

Quoted 3 lines

Quoted 1 line · Garrett Cooper

On Mon, Nov 1, 2010 at 1:00 AM, Gui Jianfeng

Quoted 7 lines · <[email protected]> wrote:

Quoted 7 lines

Quoted 7 lines

Quoted 7 lines

Quoted 1 line · Garrett Cooper

2010/10/31 Mitani <[email protected]>:

...

Quoted 4 lines

Quoted 4 lines

int main()
{

(quoted diff) +0 -0

@@ -133,6 +152,12 @@
 

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Review line 158

Quoted 6 lines

(quoted diff) +4 -0

       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel

versions

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +2 -0

 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd

testcases/open_posix_testsuite/conformance/interfaces/sigaction/

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 4 lines

#./gentests.pl
------------

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap

me

Quoted 7 lines

Quoted 7 lines

Quoted 6 lines

Quoted 2 lines

Quoted 1 line

or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

This open_posix_suite is the part of LTP(linux test project), must

we

Quoted 8 lines

Quoted 1 line

make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

OK, i get it. Thanks. ^_^

Then, whether should we remove the following case, because it invokes the uname that it's not exsit at 4.3 BSD. conformance/interfaces/pthread_mutex_init/speculative/5-2.c

Thanks
Bian

Quoted 1 line

Quoted 1 line

What about this patch? If you think so, other similar patchs will

be post.

Quoted 2 lines

Quoted 2 lines

--------------------------------------------------------------------
-----------

Quoted 2 lines

Quoted 2 lines

At older linux kernel(2.6.22 before), a signal handler always

interrupts a

Quoted 1 line

Quoted 1 line

blocked call to one of these functions, regardless of the use of the

sigaction

Quoted 3 lines

Quoted 3 lines

SA_RESTART flag, so we should not test these cases at this situation.

Althought uname is not exsit at 4.3 BSD, but this open_posix_suite

just

Quoted 6 lines

Quoted 6 lines

the part of LTP(linux test project), so i think it's not a problem.

Signed-off-by Bian Naimeng <[email protected]>

---
 .../conformance/interfaces/sigaction/16-1.c        |   27

++++++++++++++++++++

Quoted 3 lines

Quoted 3 lines

 1 files changed, 27 insertions(+), 0 deletions(-)

diff --git

a/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c b/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 2 lines

Quoted 2 lines

index 706dbfc..3d6e1a1 100644
---

a/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 1 line

Quoted 1 line

+++

b/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 6 lines

Quoted 6 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +1 -0

@@ -50,6 +50,7 @@ Anyway, a false negative status cannot be returned.
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <sys/utsname.h>
 

Quoted 7 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c Anyway, a false negative status cannot be returned. · line 99

Quoted 7 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +1 -0

 #include <semaphore.h>
 #include <signal.h>

@@ -95,6 +96,27 @@ Anyway, a false negative status cannot be returned.
 volatile sig_atomic_t caught = 0;
 sem_t sem;
 
+#define KERNEL_VERSION(a, b, c)        (((a) << 16) + ((b) << 8) +

(c))

Quoted 25 lines

Quoted 25 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +20 -0

+
+int check_linux()
+{
+       struct utsname tn;
+       int r1, r2, r3;
+
+       if (0 > uname(&tn)) {
+               UNRESOLVED( errno, "uname failed" );
+       }
+
+       if (strcmp(tn.sysname, "Linux") != 0)
+               return 0;
+
+       sscanf(tn.release, "%d.%d.%d", &r1, &r2, &r3);
+       if (KERNEL_VERSION(r1, r2, r3) >= KERNEL_VERSION(2, 6, 22))
+               return 1;
+
+       return 0;
+}
+
 /* Handler function */
 void handler( int signo )
 {

@@ -133,6 +155,11 @@ int main()
 

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c int main()

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +2 -0

       struct sigaction sa;
 
+       if (!check_linux())
+               UNTESTED( "A signal handler always interrupts sem_wait

"

Quoted 1 line

Review int main() · line 160

Quoted 1 line

(quoted diff) +1 -0

+                         "at linux release, regardless of the use

of "

Quoted 6 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c int main() · line 164

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +2 -0

+                         "the sigaction SA_RESTART flag" );
+
       /* Initialize output */
       output_init();

--
Regards
Bian Naimeng

Quoted 81 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 6 lines

Quoted 4 lines

#./gentests.pl
------------

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap

me

Quoted 7 lines

Quoted 7 lines

Quoted 6 lines

Quoted 2 lines

Quoted 1 line

or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

This open_posix_suite is the part of LTP(linux test project), must

we

Quoted 7 lines

Quoted 1 line

make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

OK, i get it. Thanks. ^_^

Then, whether should we remove the following case, because it invokes the uname that it's not exsit at 4.3 BSD. conformance/interfaces/pthread_mutex_init/speculative/5-2.c

Thanks
Bian

Quoted 1 line

Quoted 1 line

What about this patch? If you think so, other similar patchs will

be post.

Quoted 1 line

Quoted 1 line

--------------------------------------------------------------------
-----------

Quoted 1 line

Quoted 1 line

At older linux kernel(2.6.22 before), a signal handler always

interrupts a

Quoted 1 line

Quoted 1 line

blocked call to one of these functions, regardless of the use of the

sigaction

Quoted 3 lines

Quoted 3 lines

SA_RESTART flag, so we should not test these cases at this situation.

Althought uname is not exsit at 4.3 BSD, but this open_posix_suite

just

Quoted 6 lines

Quoted 6 lines

the part of LTP(linux test project), so i think it's not a problem.

Signed-off-by Bian Naimeng <[email protected]>

---
 .../conformance/interfaces/sigaction/16-1.c        |   27

++++++++++++++++++++

Quoted 3 lines

Quoted 3 lines

 1 files changed, 27 insertions(+), 0 deletions(-)

diff --git

a/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c b/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 2 lines

Quoted 2 lines

index 706dbfc..3d6e1a1 100644
---

a/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 1 line

Quoted 1 line

+++

b/testcases/open_posix_testsuite/conformance/interfaces/sigaction/16 -1.c

Quoted 6 lines

Quoted 6 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +1 -0

@@ -50,6 +50,7 @@ Anyway, a false negative status cannot be returned.
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <sys/utsname.h>
 

Quoted 7 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c Anyway, a false negative status cannot be returned. · line 99

Quoted 7 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +1 -0

 #include <semaphore.h>
 #include <signal.h>

@@ -95,6 +96,27 @@ Anyway, a false negative status cannot be returned.
 volatile sig_atomic_t caught = 0;
 sem_t sem;
 
+#define KERNEL_VERSION(a, b, c)        (((a) << 16) + ((b) << 8) +

(c))

Quoted 5 lines

Quoted 5 lines

Quoted 5 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +5 -0

+
+int check_linux()
+{
+       struct utsname tn;
+       int r1, r2, r3;

Quoted 8 lines

Quoted 8 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c Anyway, a false negative status cannot be returned. · line 112

Quoted 8 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +8 -0

+
+       if (0 > uname(&tn)) {
+               UNRESOLVED( errno, "uname failed" );
+       }
+
+       if (strcmp(tn.sysname, "Linux") != 0)
+               return 0;
+

Quoted 24 lines

Quoted 12 lines

Quoted 12 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +7 -0

+       sscanf(tn.release, "%d.%d.%d", &r1, &r2, &r3);
+       if (KERNEL_VERSION(r1, r2, r3) >= KERNEL_VERSION(2, 6, 22))
+               return 1;
+
+       return 0;
+}
+
 /* Handler function */
 void handler( int signo )
 {

@@ -133,6 +155,11 @@ int main()
 

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c int main()

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +2 -0

       struct sigaction sa;
 
+       if (!check_linux())
+               UNTESTED( "A signal handler always interrupts sem_wait

"

Quoted 1 line

Review int main() · line 160

Quoted 1 line

(quoted diff) +1 -0

+                         "at linux release, regardless of the use

of "

Quoted 5 lines

Quoted 4 lines

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +2 -0

+                         "the sigaction SA_RESTART flag" );
+
       /* Initialize output */
       output_init();

Quoted 3 lines

--
Regards
Bian Naimeng

I applied Bian's patch in my system, and build succeeded.

I'm OK with the Bian's patch.

Quoted 35 lines

Quoted 35 lines

Quoted 35 lines

Quoted 26 lines

...

Quoted 25 lines

Quoted 25 lines

int main()
{

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +6 -0

@@ -133,6 +152,12 @@
 
       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 5 lines

Quoted 4 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

Quoted 9 lines

Quoted 2 lines

This open_posix_suite is the part of LTP(linux test project), must we
make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

OK, i get it. Thanks. ^_^

Then, whether should we remove the following case, because it invokes the uname that it's not exsit at 4.3 BSD. conformance/interfaces/pthread_mutex_init/speculative/5-2.c

Quoted 50 lines

Quoted 49 lines

Quoted 35 lines

Quoted 35 lines

Quoted 35 lines

Quoted 26 lines

...

Quoted 25 lines

Quoted 25 lines

int main()
{

testcases/open_posix_testsuite/conformance/interfaces/sigaction/16-1.c +6 -0

@@ -133,6 +152,12 @@
 
       struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
       /* Initialize output */
       output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 5 lines

Quoted 4 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

Sorry, but Linux-isms can't leak into the open_posix_testsuite.
Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

Quoted 8 lines

Quoted 2 lines

This open_posix_suite is the part of LTP(linux test project), must we
make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

OK, i get it. Thanks. ^_^

Then, whether should we remove the following case, because it invokes the uname that it's not exsit at 4.3 BSD. conformance/interfaces/pthread_mutex_init/speculative/5-2.c

Quoted 2 lines

The uname() is part of the POSIX, for sure it's available on BSD.

Maybe i miss somethings, but the IEEE Std 1003.1-2008 said:
"The uname() function originated in System III, System V, and related implementations, and it does not exist in Version 7 or 4.3 BSD. The values it returns are set at system compile time in those historical implementations."

Quoted 6 lines

Thinking of it, the patch that checks for kernel version only when sysname == Linux should compile and work. But the question here is if we want to bloat the tests with workarounds for broken systems. I would rather see this covered by a database of known issues, so you can easily copare results and read why your system is broken.

OK, if we have this database, it looks good to me.

Quoted 53 lines

Quoted 53 lines

Quoted 52 lines

Quoted 43 lines

Quoted 35 lines

Quoted 35 lines

Quoted 35 lines

Quoted 26 lines

...

Quoted 25 lines

Quoted 25 lines

 int main()
 {

(quoted diff) +6 -0

@@ -133,6 +152,12 @@
 
        struct sigaction sa;
 
+       /* Checking for Kernel Version */
+       if ( kernel_ver_cmp(2, 6, 21) != 0 )
+       {
+               UNTESTED( "sem_wait() returned EINTR on kernel versions
 lower than 2.6.22" );
+       }
+
        /* Initialize output */
        output_init();

============

Above patch is revision only for templates.
The following procedure is necessary after applying patch:
------------
#cd testcases/open_posix_testsuite/conformance/interfaces/sigaction/ #./gentests.pl
------------

Quoted 5 lines

Quoted 4 lines

Quoted 1 line

Similar revisions are necessary for other 25 "sigaction" tests.

    Sorry, but Linux-isms can't leak into the open_posix_testsuite. Same with BSDisms (at least they shouldn't... if they do, then slap me or whoever did the commit).

Do you mean the culprit is Linux-isms?

Mitani-san is proposing that I `fix' the testcases to work on RHEL 4.8. These testcases should function with minimal change for all versions of Unix.

Quoted 8 lines

Quoted 2 lines

 This open_posix_suite is the part of LTP(linux test project), must we make sure it can work at all versions of unix?

I use it in FreeBSD and others have used it in other versions of Unix (in the past), based on the README. There's no reason why we need to maintain something that philosophically diverges from the upstream project's goal because it will make merging more painful and unnecessary.

Quoted 6 lines

  OK, i get it. Thanks. ^_^

  Then, whether should we remove the following case, because it invokes the uname that it's not exsit at 4.3 BSD. conformance/interfaces/pthread_mutex_init/speculative/5-2.c

The uname() is part of the POSIX, for sure it's available on BSD.

Quoted 7 lines

Quoted 4 lines

  Maybe i miss somethings, but the IEEE Std 1003.1-2008 said:
  "The uname() function originated in System III, System V, and related implementations, and it does not exist in Version 7 or 4.3 BSD. The values it returns are set at system compile time in those historical implementations."

Well BSD 4.3 has been released in 1986. The recent BSD systems (tested on FreeBSD 8.0) should be fine.

The uname system call is fairly standard on all versions of Unix
(and guess what folks -- it's POSIX standard too!).
The point wasn't to discount the value of adding uname, but where
do you stop in trying to ?
If you guys want functionality to _remove_ testcases for local
execution, then I'm willing to help there. Otherwise, there will be no end to this hack mess.

Quoted 7 lines

Quoted 7 lines

Quoted 6 lines

Thinking of it, the patch that checks for kernel version only when sysname == Linux should compile and work. But the question here is if we want to bloat the tests with workarounds for broken systems. I would rather see this covered by a database of known issues, so you can easily copare results and read why your system is broken.

Quoted 5 lines

Quoted 2 lines

  OK, if we have this database, it looks good to me.

Garrett pointed out the results mailing list. I'm however not reading it either.

Yes, because it's mostly spam and a support email alias nowadays (instead of the emails coming here they go there >:(...). I need to come up with a script anyhow to aggregate the results and ship them off via LTP (not the open_posix_testsuite because I'm not going to commit those bits there). IIRC there's a script in tools that can do that.