cyclictest - fix comparison about flexible word counts of available_tracers file.

cyclictest - fix comparison about flexible word counts of available_tracers file.

[GeunSik Lim]

[-- Attachment #1: Type: text/plain, Size: 2955 bytes --]

Dear Thomas,

I found bug at settracer function of cyclictest.c source (rt-test ver 0.33).
settracer func use 10 times in the for() statement to searching tracer
like preemptoff , irqsoff , preemptirqsoff.
But, I think that we have to consider 10+ words in the available_tracers file.

[root@fedora9 tracing]# cd /debug/tracing/
[root@fedora9 tracing]# cat available_tracers
syscall blk kmemtrace power branch function_graph mmiotrace wakeup_rt
wakeup preemptirqsoff preemptoff irqsoff function sched_switch
initcall nop
[root@fedora9 tracing]# cat available_tracers  | wc -w
16
[root@fedora9 tracing]#
[root@fedora9 tracing]# git clone
git://git.kernel.org/pub/scm/linux/kernel/git/clrkwllms/rt-tests.git
[root@fedora9 tracing]# cd rt-tests
[root@fedora9 tracing]# make clean ; make all
[root@fedora9 tracing]# ./cyclictest  -n -t 3 -b 1000000 -P-D 2 --> Success
[root@fedora9 tracing]#./cyclictest  -n -t 3 -b 1000000 -I -D 2  --> Failed
[root@fedora9 tracing]#./cyclictest  -n -t 3 -b 1000000 -B-D 2 --> Failed


So I made  patch file to consider above case.
If I mistake, correct me.

--------------------------------------------------
[root@fedora9 cyclictest]# diff -urN ./cyclictest.c.ori  ./cyclictest.c
--- ./cyclictest.c.ori	2009-04-05 23:48:03.000000000 +0900
+++ ./cyclictest.c	2009-04-06 00:15:51.000000000 +0900
@@ -320,7 +320,7 @@
 	char name[100];
 	FILE *fp;
 	int ret;
-	int i;
+	//int i;

 	/* Make sure tracer is available */
 	strncpy(filename, debugfileprefix, sizeof(filename));
@@ -330,7 +330,7 @@
 	if (!fp)
 		return -1;

-	for (i = 0; i < 10; i++) {
+	while(1) {
 		ret = fscanf(fp, "%99s", name);
 		if (!ret) {
 			ret = -1;


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

After patching .....

[root@fedora9 tracing]#./cyclictest  -n -t 3 -b 1000000 -I -D 2
[root@fedora9 tracing]# cat trace
# tracer: irqsoff
#
# irqsoff latency trace v1.1.5 on 2.6.29.1-rt4.20090403-ftrace-lgs
# --------------------------------------------------------------------
# latency: 108 us, #3/3, CPU#0 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:2)
#    -----------------
#    | task: swapper-0 (uid:0 nice:0 policy:0 rt_prio:0)
#    -----------------
#
#                  _------=> CPU#
#                 / _-----=> irqs-off
#                | / _----=> need-resched
#                || / _---=> hardirq/softirq
#                ||| / _--=> preempt-depth
#                |||| /
#                |||||     delay
#  cmd     pid   ||||| time  |   caller
#     \   /      |||||   \   |   /
  <idle>-0       0d..1    3us!: acpi_idle_do_entry <-acpi_idle_enter_bm
  <idle>-0       0.N.1  107us+: cpu_idle <-rest_init
  <idle>-0       0.N.1  110us : trace_preempt_on <-rest_init


-- 
Regards,
GeunSik Lim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[-- Attachment #2: fix-comparison-wordcount-available-tracers.patch --]
[-- Type: application/octet-stream, Size: 444 bytes --]

--- ./cyclictest.c.ori	2009-04-05 23:48:03.000000000 +0900
+++ ./cyclictest.c	2009-04-06 00:15:51.000000000 +0900
@@ -320,7 +320,7 @@
 	char name[100];
 	FILE *fp;
 	int ret;
-	int i;
+	//int i;
 
 	/* Make sure tracer is available */
 	strncpy(filename, debugfileprefix, sizeof(filename));
@@ -330,7 +330,7 @@
 	if (!fp)
 		return -1;
 
-	for (i = 0; i < 10; i++) {
+	while(1) {
 		ret = fscanf(fp, "%99s", name);
 		if (!ret) {
 			ret = -1;

Re: cyclictest - fix comparison about flexible word counts of available_tracers file.

[Thomas Gleixner]

On Sun, 5 Apr 2009, GeunSik Lim wrote:
> [root@fedora9 tracing]# ./cyclictest  -n -t 3 -b 1000000 -P-D 2 --> Success
> [root@fedora9 tracing]#./cyclictest  -n -t 3 -b 1000000 -I -D 2  --> Failed
> [root@fedora9 tracing]#./cyclictest  -n -t 3 -b 1000000 -B-D 2 --> Failed
> 
> So I made  patch file to consider above case.
> If I mistake, correct me.

Good catch. Can you please clean up the patch and resend it with a
patch description and your Signed-off-by line ? 

> --------------------------------------------------
> [root@fedora9 cyclictest]# diff -urN ./cyclictest.c.ori  ./cyclictest.c
> --- ./cyclictest.c.ori	2009-04-05 23:48:03.000000000 +0900
> +++ ./cyclictest.c	2009-04-06 00:15:51.000000000 +0900
> @@ -320,7 +320,7 @@
>  	char name[100];
>  	FILE *fp;
>  	int ret;
> -	int i;
> +	//int i;

  Please remove

>  	/* Make sure tracer is available */
>  	strncpy(filename, debugfileprefix, sizeof(filename));
> @@ -330,7 +330,7 @@
>  	if (!fp)
>  		return -1;
> 
> -	for (i = 0; i < 10; i++) {
> +	while(1) {

  	while (1) {

  please

>  		ret = fscanf(fp, "%99s", name);
>  		if (!ret) {
>  			ret = -1;

Thanks,

	tglx

Re: cyclictest - fix comparison about flexible word counts of available_tracers file.

[Clark Williams]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, 5 Apr 2009 15:45:33 +0900
GeunSik Lim <leemgs1@gmail.com> wrote:

> Dear Thomas,
> 
> I found bug at settracer function of cyclictest.c source (rt-test ver 0.33).
> settracer func use 10 times in the for() statement to searching tracer
> like preemptoff , irqsoff , preemptirqsoff.
> But, I think that we have to consider 10+ words in the available_tracers file.
> 

GeunSik,

Good catch!  I applied your patch, but then found a bug in either
fscanf or debugfs (probably debugfs). Since I didn't have IRQSOFF or
PREEMPTOFF set in my kernel, those tracers weren't available and fscanf
never returned EOF (it kept returning the "nop" tracer). So I reworked
your patch to read everything in and parse it using strtok(3). 

The git tree has been updated and an rt-tests-0.34.tar.gz tarball is
available at:

http://www.kernel.org/pub/linux//kernel/people/clrkwllms/rt-tests/

Clark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.10 (GNU/Linux)

iEYEARECAAYFAknYxR4ACgkQHyuj/+TTEp3TNwCbBNrEHMLjPmAS6JciwjyXWqME
wpcAn3+nUGdOwCJR6sFGfwundPtaITrk
=dT+3
-----END PGP SIGNATURE-----

Re: cyclictest - fix comparison about flexible word counts of available_tracers file.

[GeunSik Lim]

Dear Clark

Thanks your patching.

Since I didn't have IRQSOFF or
> PREEMPTOFF set in my kernel, those tracers weren't available and fscanf
> never returned EOF (it kept returning the "nop" tracer).

Clark. I think that Your approach is right.

In fact , I also thought  this case that  a developer disable
PREEMPTOFF or IRQSOFF on "#>make menuconfig" setting menu.
As you know, Most peoples will use -P or -I option when they enabled
PREEMPTOFF or IRQSOFF in Linux Kernel just.

But, Exception Handling is important as your explanation.
Thanks again.


2009/4/5 Clark Williams <williams@redhat.com>:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Sun, 5 Apr 2009 15:45:33 +0900
> GeunSik Lim <leemgs1@gmail.com> wrote:
>
>> Dear Thomas,
>>
>> I found bug at settracer function of cyclictest.c source (rt-test ver 0.33).
>> settracer func use 10 times in the for() statement to searching tracer
>> like preemptoff , irqsoff , preemptirqsoff.
>> But, I think that we have to consider 10+ words in the available_tracers file.
>>
>
> GeunSik,
>
> Good catch!  I applied your patch, but then found a bug in either
> fscanf or debugfs (probably debugfs). Since I didn't have IRQSOFF or
> PREEMPTOFF set in my kernel, those tracers weren't available and fscanf
> never returned EOF (it kept returning the "nop" tracer). So I reworked
> your patch to read everything in and parse it using strtok(3).
>
> The git tree has been updated and an rt-tests-0.34.tar.gz tarball is
> available at:
>
> http://www.kernel.org/pub/linux//kernel/people/clrkwllms/rt-tests/
>
> Clark
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2.0.10 (GNU/Linux)
>
> iEYEARECAAYFAknYxR4ACgkQHyuj/+TTEp3TNwCbBNrEHMLjPmAS6JciwjyXWqME
> wpcAn3+nUGdOwCJR6sFGfwundPtaITrk
> =dT+3
> -----END PGP SIGNATURE-----
>



-- 
Regards,
GeunSik Lim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html