[LTP] [RFC] Patch getcpu01 testcase

Suzuki Poulose

author

2009-05-28 17:04 UTC #0 lore raw

RFC 1/1 Patch getcpu01 testcase

Hi,

The getcpu01 always exits with a non-zero exit code (indicating a Failure as per UNIX semantics) on non-NUMA machines.

This is due to the following code :

tst_resm(TWARN,
"/sys not mounted or not a numa system. Assuming one
node: %s",

                        strerror(errno));

Being a non-NUMA machine is not a failure, but an info. So it should not cause the test to have a non zero exit status.

I have attached a patch which could do the same and provide a more friendly o/p.

Please apply.

Thanks
Suzuki

testcases/kernel/syscalls/getcpu/getcpu01.c +6 -2

Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
@@ -47,6 +47,8 @@
  * HISTORY
  *	06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *	05/2009 	Suzuki K P <[email protected]>
+ *			Friendly messages for non-NUMA machines.
  * RESTRICTIONS
  *	none
  */

@@ -234,8 +236,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");
         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming one node: %s",
+                tst_resm(TINFO,
+                        "/sys not mounted or not a numa system. Assuming one node");
+		tst_resm(TINFO,
+			"Reason /sys/devices/system/node :%s",
                         strerror(errno));
                	return 0; //By Default assume it to belong to node Zero 
         } else {

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com

_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Subrata Modak

2009-05-29 12:55 UTC #1 lore raw

Quoted 11 lines · Suzuki Poulose

Hi,
The getcpu01 always exits with a non-zero exit code (indicating a Failure as per UNIX semantics) on non-NUMA machines.
This is due to the following code :
tst_resm(TWARN,
"/sys not mounted or not a numa system. Assuming
one
node: %s",

Quoted 6 lines

                         strerror(errno));
Being a non-NUMA machine is not a failure, but an info. So it should not
cause the test to have a non zero exit status.

Correct,

Quoted 43 lines

I have attached a patch which could do the same and provide a more friendly o/p.
Please apply.
Thanks
Suzuki
plain text
document
attachment
(fix-warning-getcpu01.diff)
Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
---
ltp-20081031.orig/testcases/kernel/syscalls/getcpu/getcpu01.c 2008-10-23 02:23:19.000000000 -0500
+++
ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c 2009-05-28
11:52:54.000000000 -0500
testcases/kernel/syscalls/getcpu/getcpu01.c +2 -0
@@ -47,6 +47,8 @@
  * HISTORY
  *     06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *     05/2009         Suzuki K P <[email protected]>
+ *                     Friendly messages for non-NUMA machines.
  * RESTRICTIONS
  *     none
  */

testcases/kernel/syscalls/getcpu/getcpu01.c line 243

Quoted 10 lines

testcases/kernel/syscalls/getcpu/getcpu01.c +3 -2

@@ -234,8 +236,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");
         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming
 one node: %s",
+                tst_resm(TINFO,
+                        "/sys not mounted or not a numa system.
 Assuming one node");
+               tst_resm(TINFO,

Use TCONF instead of TINFO. That reflects that some system configuration was missing, and, it will not return error value at all.

Regards--
Subrata

Quoted 6 lines

testcases/kernel/syscalls/getcpu/getcpu01.c +1 -0

+                       "Reason /sys/devices/system/node :%s",
                         strerror(errno));

return 0; //By Default assume it to belong to
node Zero

         } else {

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Suzuki Poulose

author

2009-05-29 14:48 UTC #2 lore raw

Subrata,

Good ! Then thats more suitable here :)

Refreshed the patch with TCONF and a better error message.

Thanks

Suzuki

Quoted 73 lines · Subrata Modak

Quoted 6 lines · Suzuki Poulose
Hi,
The getcpu01 always exits with a non-zero exit code (indicating a Failure as per UNIX semantics) on non-NUMA machines.
This is due to the following code :
Quoted 8 lines
tst_resm(TWARN,
"/sys not mounted or not a numa system. Assuming
one node: %s",
                         strerror(errno));
Being a non-NUMA machine is not a failure, but an info. So it should not cause the test to have a non zero exit status.
Correct,
Quoted 41 lines
I have attached a patch which could do the same and provide a more friendly o/p.
Please apply.
Thanks
Suzuki
plain text
document
attachment
(fix-warning-getcpu01.diff)
Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
---
ltp-20081031.orig/testcases/kernel/syscalls/getcpu/getcpu01.c 2008-10-23 02:23:19.000000000 -0500
+++
ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c 2009-05-28
11:52:54.000000000 -0500
(quoted diff) +2 -0
@@ -47,6 +47,8 @@
  * HISTORY
  *     06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *     05/2009         Suzuki K P <[email protected]>
+ *                     Friendly messages for non-NUMA machines.
  * RESTRICTIONS
  *     none
  */
Review line 243
Quoted 10 lines
(quoted diff) +3 -2
@@ -234,8 +236,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");
         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming
 one node: %s",
+                tst_resm(TINFO,
+                        "/sys not mounted or not a numa system.
 Assuming one node");
+               tst_resm(TINFO,
Use TCONF instead of TINFO. That reflects that some system configuration was missing, and, it will not return error value at all.

Quoted 10 lines

Regards--
Subrata
Quoted 6 lines
(quoted diff) +1 -0
+                       "Reason /sys/devices/system/node :%s",
                         strerror(errno));
return 0; //By Default assume it to belong to
node Zero
         } else {

testcases/kernel/syscalls/getcpu/getcpu01.c +7 -2

Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
@@ -47,6 +47,9 @@
  * HISTORY
  *	06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *	05/2009 	Suzuki K P <[email protected]>
+ *			Use TCONF instead of TWARN for non-NUMA machines
+ *	
  * RESTRICTIONS
  *	none
  */

@@ -234,8 +237,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");
         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming one node: %s",
+                tst_resm(TCONF,
+                   "/sys not mounted or not a numa system. Assuming one node");
+		tst_resm(TCONF,
+			"Error opening: /sys/devices/system/node :%s",
                         strerror(errno));
                	return 0; //By Default assume it to belong to node Zero 
         } else {

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com

_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Subrata Modak

2009-05-30 18:41 UTC #3 lore raw

Quoted 5 lines · Suzuki Poulose

Subrata,
Good ! Then thats more suitable here :)
Refreshed the patch with TCONF and a better error message.

Thanks.

Regards--
Subrata

Quoted 111 lines

Thanks
Suzuki
Quoted 77 lines · Subrata Modak
Quoted 11 lines · Suzuki Poulose
Hi,
The getcpu01 always exits with a non-zero exit code (indicating a Failure as per UNIX semantics) on non-NUMA machines.
This is due to the following code :
tst_resm(TWARN,
"/sys not mounted or not a numa system. Assuming
one
node: %s",
Quoted 6 lines
                         strerror(errno));
Being a non-NUMA machine is not a failure, but an info. So it should not
cause the test to have a non zero exit status.
Correct,
Quoted 42 lines
I have attached a patch which could do the same and provide a more friendly o/p.
Please apply.
Thanks
Suzuki
plain text
document
attachment
(fix-warning-getcpu01.diff)
Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
---
ltp-20081031.orig/testcases/kernel/syscalls/getcpu/getcpu01.c 2008-10-23 02:23:19.000000000 -0500
+++
ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c 2009-05-28
11:52:54.000000000 -0500
testcases/kernel/syscalls/getcpu/getcpu01.c +2 -0
@@ -47,6 +47,8 @@
  * HISTORY
  *     06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *     05/2009         Suzuki K P <[email protected]>
+ *                     Friendly messages for non-NUMA machines.
  * RESTRICTIONS
  *     none
  */
testcases/kernel/syscalls/getcpu/getcpu01.c line 243
Quoted 10 lines
testcases/kernel/syscalls/getcpu/getcpu01.c +3 -2
@@ -234,8 +236,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");
         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming
 one node: %s",
+                tst_resm(TINFO,
+                        "/sys not mounted or not a numa system.
 Assuming one node");
+               tst_resm(TINFO,
Use TCONF instead of TINFO. That reflects that some system configuration was missing, and, it will not return error value at all.
Quoted 10 lines
Regards--
Subrata
Quoted 6 lines
(quoted diff) +1 -0
+                       "Reason /sys/devices/system/node :%s",
                         strerror(errno));
return 0; //By Default assume it to belong to
node Zero
         } else {
plain text document attachment (fix-warning-getcpu01.diff)
Index: ltp-20081031/testcases/kernel/syscalls/getcpu/getcpu01.c
===================================================================
testcases/kernel/syscalls/getcpu/getcpu01.c +3 -0
@@ -47,6 +47,9 @@
  * HISTORY
  *	06/2008 written by Sharyathi Nagesh <[email protected]>
  *
+ *	05/2009 	Suzuki K P <[email protected]>
+ *			Use TCONF instead of TWARN for non-NUMA machines
+ *	
  * RESTRICTIONS
  *	none
  */
@@ -234,8 +237,10 @@
 
         directory_parent = opendir("/sys/devices/system/node");

testcases/kernel/syscalls/getcpu/getcpu01.c line 246

Quoted 10 lines

testcases/kernel/syscalls/getcpu/getcpu01.c +4 -2

         if (!directory_parent)  {
-                tst_resm(TWARN,
-                   "/sys not mounted or not a numa system. Assuming one node: %s",
+                tst_resm(TCONF,
+                   "/sys not mounted or not a numa system. Assuming one node");
+		tst_resm(TCONF,
+			"Error opening: /sys/devices/system/node :%s",
                         strerror(errno));
                	return 0; //By Default assume it to belong to node Zero 
         } else {

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list