Skip to contentSkip to conversation

POSIX "sched_yield/1-1.c" build failed

LTP Mitani 4 participants 9 messages Sep 14, 2010 – Sep 21, 2010

lore
Share & tools
lore

Mitani

author
#0 lore raw

Hi,

POSIX "open_posix_testsuite/conformance/interfaces/sched_yield/1-1.c" build failed in RHEL4.8 (x86, x86_64, ia64):
------------
conformance/interfaces/sched_yield/1-1 compile FAILED; SKIPPING In file included from ../../../conformance/interfaces/sched_yield/1-1.c:35:
../../../include/posixtest.h:14:2: #error "Contains GNU-isms that need fixing."
../../../conformance/interfaces/sched_yield/1-1.c: In function `set_thread_affinity':
../../../conformance/interfaces/sched_yield/1-1.c:91: warning: implicit declaration of function `pthread_setaffinity_np'
------------

This testset seems to test "sched_yield()", and uses "sched_setafinity()" for CPU fixing.
I tried to remove "_GNU_SOURCE" definition from "sched_yield/1-1.c". But "sched_setafinity()" needs "_GNU_SOURCE". "sched_setafinity()" is necessary for this testset, and I considered the substitution for "sched_setafinity()".
But I don't have any ideas. There is no function to be substitute for "sched_setafinity()" by POSIX conformity, I think.

Regards--

-Tomonori Mitani

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#1 lore raw
Quoted 25 lines · ext Mitani

Hi,

POSIX "open_posix_testsuite/conformance/interfaces/sched_yield/1-1.c" build failed in RHEL4.8 (x86, x86_64, ia64):
------------
conformance/interfaces/sched_yield/1-1 compile FAILED; SKIPPING In file included from ../../../conformance/interfaces/sched_yield/1-1.c:35:
../../../include/posixtest.h:14:2: #error "Contains GNU-isms that need fixing."
../../../conformance/interfaces/sched_yield/1-1.c: In function `set_thread_affinity':
../../../conformance/interfaces/sched_yield/1-1.c:91: warning: implicit declaration of function `pthread_setaffinity_np'
------------

This testset seems to test "sched_yield()", and uses "sched_setafinity()" for CPU fixing.
I tried to remove "_GNU_SOURCE" definition from "sched_yield/1-1.c". But "sched_setafinity()" needs "_GNU_SOURCE". "sched_setafinity()" is necessary for this testset, and I considered the substitution for "sched_setafinity()".
But I don't have any ideas. There is no function to be substitute for "sched_setafinity()" by POSIX conformity, I think.

Quoted 4 lines

Regards--

-Tomonori Mitani

Hi,

could try this out?

posixtest.h +8 -0
@@ -11,8 +11,16 @@
  */
 
 #if defined(_GNU_SOURCE)
+#if defined (PTS_GNU_SOURCE_WARNING)
+#warning "Contains GNU-isms that need fixing."
+#else
 #error "Contains GNU-isms that need fixing."
+#endif
+
 #elif defined(_BSD_SOURCE)
+#if defined (PTS_BSD_SOURCE_WARNING)
+#warning "Contains BSD-isms that need fixing."
+#else
 #error "Contains BSD-isms that need fixing."
 #endif
 

Add to yr source #define PTS_GNU_SOURCE_WARNING 1 and compile, or define in posixtest.h and include that hdr.

br,
Hannu

Signature · 12 lines
Hannu Heikkinen
http://www.nixu.com

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
#2 lore raw
Quoted 55 lines · Hannu Heikkinen
Quoted 25 lines · ext Mitani

Hi,

POSIX "open_posix_testsuite/conformance/interfaces/sched_yield/1-1.c" build failed in RHEL4.8 (x86, x86_64, ia64):
------------
conformance/interfaces/sched_yield/1-1 compile FAILED; SKIPPING In file included from ../../../conformance/interfaces/sched_yield/1-1.c:35:
../../../include/posixtest.h:14:2: #error "Contains GNU-isms that need fixing."
../../../conformance/interfaces/sched_yield/1-1.c: In function `set_thread_affinity':
../../../conformance/interfaces/sched_yield/1-1.c:91: warning: implicit declaration of function `pthread_setaffinity_np'
------------

This testset seems to test "sched_yield()", and uses "sched_setafinity()" for CPU fixing.
I tried to remove "_GNU_SOURCE" definition from "sched_yield/1-1.c". But "sched_setafinity()" needs "_GNU_SOURCE". "sched_setafinity()" is necessary for this testset, and I considered the substitution for "sched_setafinity()".
But I don't have any ideas. There is no function to be substitute for "sched_setafinity()" by POSIX conformity, I think.

Quoted 4 lines

Regards--

-Tomonori Mitani

Hi,

could try this out?

posixtest.h  2010-09-14 19:01:02.000000000 +0000 renamed from posixtest.h 2010-09-14 14:05:21.000000000 +0000 +8 -0
@@ -11,8 +11,16 @@
  */
 
  #if defined(_GNU_SOURCE)
+#if defined (PTS_GNU_SOURCE_WARNING)
+#warning "Contains GNU-isms that need fixing."
+#else
  #error "Contains GNU-isms that need fixing."
+#endif
+
  #elif defined(_BSD_SOURCE)
+#if defined (PTS_BSD_SOURCE_WARNING)
+#warning "Contains BSD-isms that need fixing."
+#else
  #error "Contains BSD-isms that need fixing."
  #endif
 
Quoted 2 lines

Add to yr source #define PTS_GNU_SOURCE_WARNING 1 and compile, or define in posixtest.h and include that hdr.

That seems like a reasonable thing to do, but here's an alternative way to do this (less variables):

testcases/open_posix_testsuite/include/posixtest.h +14 -0
@@ -10,10 +10,24 @@
  * return codes
  */
 
+/*
+ * Define PTS_DEVELOPER_MODE if you want to compile for developer scenarios,
+ * including reporting errors (as opposed to warnings), when compiling some
+ * test programs.
+ */
+
 #if defined(_GNU_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
 #error "Contains GNU-isms that need fixing."
+#else
+#warning "Contains GNU-isms that need fixing."
+#endif
 #elif defined(_BSD_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
 #error "Contains BSD-isms that need fixing."
+#else
+#warning "Contains BSD-isms that need fixing."
+#endif
 #endif
 
 #define PTS_PASS        0

Thanks for the idea Hannu :),
-Garrett

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#3 lore raw
Quoted 31 lines · Garrett Cooper

That seems like a reasonable thing to do, but here's an alternative way to do this (less variables):

testcases/open_posix_testsuite/include/posixtest.h +14 -0
@@ -10,10 +10,24 @@
  * return codes
  */
 
+/*
+ * Define PTS_DEVELOPER_MODE if you want to compile for developer scenarios,
+ * including reporting errors (as opposed to warnings), when compiling some
+ * test programs.
+ */
+
 #if defined(_GNU_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
 #error "Contains GNU-isms that need fixing."
+#else
+#warning "Contains GNU-isms that need fixing."
+#endif
 #elif defined(_BSD_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
 #error "Contains BSD-isms that need fixing."
+#else
+#warning "Contains BSD-isms that need fixing."
+#endif
 #endif
 
 #define PTS_PASS        0
Quoted 3 lines

Thanks for the idea Hannu :),
-Garrett

Ok, even better, less is always good! Developer mode sounds good.

Should this posixtest.h be kind of a must include for all test cases? Currently not all cases include this.

br,
Hannu

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#4 lore raw
Quoted 32 lines · Hannu Heikkinen
Quoted 31 lines · Garrett Cooper

That seems like a reasonable thing to do, but here's an alternative way to do this (less variables):

testcases/open_posix_testsuite/include/posixtest.h +14 -0
@@ -10,10 +10,24 @@
   * return codes
   */
 
+/*
+ * Define PTS_DEVELOPER_MODE if you want to compile for developer scenarios,
+ * including reporting errors (as opposed to warnings), when compiling some
+ * test programs.
+ */
+
  #if defined(_GNU_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
  #error "Contains GNU-isms that need fixing."
+#else
+#warning "Contains GNU-isms that need fixing."
+#endif
  #elif defined(_BSD_SOURCE)
+#if defined(PTS_DEVELOPER_MODE)
  #error "Contains BSD-isms that need fixing."
+#else
+#warning "Contains BSD-isms that need fixing."
+#endif
  #endif
 
  #define PTS_PASS        0
Quoted 8 lines
Quoted 3 lines

Thanks for the idea Hannu :),
-Garrett

Ok, even better, less is always good! Developer mode sounds good.

Should this posixtest.h be kind of a must include for all test cases? Currently not all cases include this.

Well, every affected testcase should include this :). Not all
files need to include this (there are a number of applications that don't need what's defined here) -- so there's no sense in generating more work than necessary :).
Thanks,
-Garrett

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#5 lore raw
Quoted 10 lines · Garrett Cooper
Quoted 3 lines

Should this posixtest.h be kind of a must include for all test cases? Currently not all cases include this.

Well, every affected testcase should include this :). Not all
files need to include this (there are a number of applications that don't need what's defined here) -- so there's no sense in generating more work than necessary :).
Thanks,
-Garrett

I must have been hald asleep writing that prev email. I meant that posixtest cases should include that, and of course not all LTP cases.

Might be a consideration for having global ltp header file for all C test cases
so that generic functionality could be better tweaked by changing certain defines. Of course that needs more discussion and analysis, thoug.

br,
Hannu

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#6 lore raw
Quoted 12 lines · Hannu Heikkinen
Quoted 10 lines · Garrett Cooper
Quoted 3 lines

Should this posixtest.h be kind of a must include for all test cases? Currently not all cases include this.

    Well, every affected testcase should include this :). Not all files need to include this (there are a number of applications that don't need what's defined here) -- so there's no sense in generating more work than necessary :).
Thanks,
-Garrett

Quoted 2 lines

I must have been hald asleep writing that prev email. I meant that posixtest cases should include that, and of course not all LTP cases.

I must have doubly been so because I didn't catch that either :).

Quoted 4 lines

Might be a consideration for having global ltp header file for all C test cases
so that generic functionality could be better tweaked by changing certain defines. Of course that needs more discussion and analysis, thoug.

Already done:

include/: http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=tree;f=include;h=93d0ff91530c366d3af0cfc0c6b4f27c779cb866;hb=HEAD
lib/: http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=tree;f=lib;h=85556e60de93da651c3ffbab56eeef2b3a68c8bf;hb=HEAD

The above changes mentioned only affect the open posix testsuite
(http://sf.net/projects/posixtest/), which is a component completely orthogonal to the rest of the code in the LTP sourcebase, so these changes are only required for that section of the codebase. However, certain tunable features can be added in the LTP buildsystem and the open posix testsuite to support automatic `developer mode' if you will. This will take a while to do properly though.

Thanks,
-Garrett

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#7 lore raw
Quoted 36 lines · ext Garrett Cooper
Quoted 12 lines · Hannu Heikkinen
Quoted 10 lines · Garrett Cooper
Quoted 3 lines

Should this posixtest.h be kind of a must include for all test cases? Currently not all cases include this.

    Well, every affected testcase should include this :). Not all files need to include this (there are a number of applications that don't need what's defined here) -- so there's no sense in generating more work than necessary :).
Thanks,
-Garrett

Quoted 2 lines

I must have been hald asleep writing that prev email. I meant that posixtest cases should include that, and of course not all LTP cases.

I must have doubly been so because I didn't catch that either :).

Quoted 4 lines

Might be a consideration for having global ltp header file for all C test cases
so that generic functionality could be better tweaked by changing certain defines. Of course that needs more discussion and analysis, thoug.

Already done:

include/: http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=tree;f=include;h=93d0ff91530c366d3af0cfc0c6b4f27c779cb866;hb=HEAD
lib/: http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=tree;f=lib;h=85556e60de93da651c3ffbab56eeef2b3a68c8bf;hb=HEAD

The above changes mentioned only affect the open posix testsuite
(http://sf.net/projects/posixtest/), which is a component completely orthogonal to the rest of the code in the LTP sourcebase, so these changes are only required for that section of the codebase. However, certain tunable features can be added in the LTP buildsystem and the open posix testsuite to support automatic `developer mode' if you will. This will take a while to do properly though.

Quoted 2 lines

Thanks,
-Garrett

Thanks for the info! That's good news!

br,
Hannu

Signature · 13 lines
Hannu Heikkinen
+358-40-1586858
http://www.nixu.com

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
author
#8 lore raw
Quoted 63 lines

-----Original Message-----
From: Hannu Heikkinen [mailto:[email protected]]
Sent: Saturday, September 18, 2010 1:48 AM
To: ext Garrett Cooper
Cc: Hannu Heikkinen; [email protected]; ext Mitani
Subject: Re: [LTP] POSIX "sched_yield/1-1.c" build failed

Quoted 1 line · ext Garrett Cooper

On Thu, Sep 16, 2010 at 12:00 PM, Hannu Heikkinen <[email protected]>

Quoted 3 lines · wrote:
Quoted 3 lines · Garrett Cooper
Quoted 2 lines
Quoted 2 lines

Should this posixtest.h be kind of a must include for all test

cases?

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

Currently not all cases include this.

    Well, every affected testcase should include this :). Not

all

Quoted 1 line
Quoted 1 line
Quoted 1 line

files need to include this (there are a number of applications

that

Quoted 1 line
Quoted 1 line
Quoted 1 line

don't need what's defined here) -- so there's no sense in

generating

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

more work than necessary :).
Thanks,
-Garrett

I must have been hald asleep writing that prev email. I meant that

posixtest

Quoted 5 lines
Quoted 1 line

cases should include that, and of course not all LTP cases.

I must have doubly been so because I didn't catch that either :).

Quoted 1 line

Might be a consideration for having global ltp header file for all

C test

Quoted 2 lines
Quoted 2 lines

cases
so that generic functionality could be better tweaked by changing

certain

Quoted 5 lines
Quoted 1 line

defines. Of course that needs more discussion and analysis, thoug.

Already done:

include/:

http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=
tree;f=include;h=93d0ff91530c366d3af0cfc0c6b4f27c779cb866;hb=HEAD

Quoted 1 line

lib/:

http://ltp.git.sourceforge.net/git/gitweb.cgi?p=ltp/ltp-dev.git;a=
tree;f=lib;h=85556e60de93da651c3ffbab56eeef2b3a68c8bf;hb=HEAD

Quoted 3 lines

The above changes mentioned only affect the open posix testsuite
(http://sf.net/projects/posixtest/), which is a component

completely

Quoted 8 lines

orthogonal to the rest of the code in the LTP sourcebase, so these changes are only required for that section of the codebase. However, certain tunable features can be added in the LTP buildsystem and the open posix testsuite to support automatic `developer mode' if you will. This will take a while to do properly though.

Thanks,
-Garrett

Thanks for the info! That's good news!

Quoted 7 lines

br,
Hannu

--
Hannu Heikkinen
+358-40-1586858
http://www.nixu.com

The patch was already included in following gits:
"ltp-dev-121ad2a296f4ade04e2c2e351820df32a6c441d5.tar.gz" "ltp-dev-8b0ca7e9a5e7dcab8d7cbad0b6f0c808798bdc53.tar.gz" and I tried to build with these gits instead of the patch.

POSIX "sched_yield/1-1.c" build successed in RHEL5.5 (x86, x86_64, ia64), but failed in RHEL4.8 (x86, x86_64, ia64):
------------(RHEL5.5 x86) conformance/interfaces/sched_yield/1-1 compile PASSED
------------

------------(RHEL4.8 x86)
conformance/interfaces/sched_yield/1-1 compile FAILED; SKIPPING In file included from ../../../conformance/interfaces/sched_yield/1-1.c:35:
../../../include/posixtest.h:23:2: warning: #warning "Contains GNU-isms that need fixing."
../../../conformance/interfaces/sched_yield/1-1.c: In function `set_thread_affinity':
../../../conformance/interfaces/sched_yield/1-1.c:91: warning: implicit declaration of function pthread_setaffinity_np' /tmp/ccuhCnVY.o(.text+0x1c1): In function set_thread_affinity':
: undefined reference to `pthread_setaffinity_np'
collect2: ld returned 1 exit status
------------

But, in all my RHEL5.5/4.8 systems, this revision brought the build successes
of the following testsets that were concern from the past:
- "pthread_attr_setstack/2-1.c"
- "pthread_attr_setstacksize/2-1.c"

Now in all my RHEL5.5 systems, all build failes were settled by this revision.

Thank you--

-Tomonori Mitani

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list