dtrace discuss - Dec 2006 - compile error in jdk1.6 sample scripts

I am running the sample scripts included in java 1.6, and I am seeing errors.
For example:

#  dtrace -s class_loading_stat.d -p 7341
dtrace: failed to compile script class_loading_stat.d: line 156: probe
description syscall::exit:entry does not match any probes
# dtrace -s gc_time_stat.d -p 7341
dtrace: failed to compile script gc_time_stat.d: line 202: probe description
syscall::exit:entry does not match any probes

Other sample scripts are showing errors too.

Lines 155 to 160 from class_loading_stat.d look like this:

syscall::rexit:entry,
syscall::exit:entry
/pid == $target/
{
   exit(0);
}

If I remove line 156, the script seems to run OK.  Is this expected behavior? In
other words, is "syscall::exit:entry" valid for some solaris versions?

The details for my system are:
# uname -a
SunOS unknown 5.11 snv_52 i86pc i386 i86pc
# java -version
java version "1.6.0-rc"
Java(TM) SE Runtime Environment (build 1.6.0-rc-b100)
Java HotSpot(TM) Client VM (build 1.6.0-rc-b100, mixed mode, sharing)

The sample scripts are located in 

${JAVA_HOME}/sample/dtrace/hotspot

thanks
Steven
 
 
This message posted from opensolaris.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: class_loading_stat.d
Type: application/octet-stream
Size: 4811 bytes
Desc: not available
URL:
<http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20061203/d3445571/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gc_time_stat.d
Type: application/octet-stream
Size: 6717 bytes
Desc: not available
URL:
<http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20061203/d3445571/attachment-0001.obj>

Steven,

There is no actual exit system call.  "exit" is a library subroutine 
that eventually calls "rexit" which is actually the name of the system
call, according to the switch table.

As far as why the canned scripts are not working, I noticed the D script 
preambles specify the -Z option which allows for probe descriptions that 
don''t match any probes.  However, in your E-mail, you invoke the
scripts
with the dtrace command, not using -Z.  If you had made the D scripts 
executables, and run them directly from the command line, the -Z option 
in the script preamble would have prevented the lack of a match for 
probe description "syscall::exit:entry" from causing the script to
fail.

Chip

Steven Reynolds wrote:
> I am running the sample scripts included in java 1.6, and I am seeing
errors. For example:
>
> #  dtrace -s class_loading_stat.d -p 7341
> dtrace: failed to compile script class_loading_stat.d: line 156: probe
description syscall::exit:entry does not match any probes
> # dtrace -s gc_time_stat.d -p 7341
> dtrace: failed to compile script gc_time_stat.d: line 202: probe
description syscall::exit:entry does not match any probes
>
> Other sample scripts are showing errors too.
>
> Lines 155 to 160 from class_loading_stat.d look like this:
>
> syscall::rexit:entry,
> syscall::exit:entry
> /pid == $target/
> {
>    exit(0);
> }
>
> If I remove line 156, the script seems to run OK.  Is this expected
behavior? In other words, is "syscall::exit:entry" valid for some
solaris versions?
>
> The details for my system are:
> # uname -a
> SunOS unknown 5.11 snv_52 i86pc i386 i86pc
> # java -version
> java version "1.6.0-rc"
> Java(TM) SE Runtime Environment (build 1.6.0-rc-b100)
> Java HotSpot(TM) Client VM (build 1.6.0-rc-b100, mixed mode, sharing)
>
> The sample scripts are located in 
>
> ${JAVA_HOME}/sample/dtrace/hotspot
>
> thanks
> Steven
>  
>  
> This message posted from opensolaris.org
> ------------------------------------------------------------------------
>
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org

Steven Reynolds wrote:
>I am running the sample scripts included in java 1.6, and I am seeing
errors. For example:
>
>#  dtrace -s class_loading_stat.d -p 7341
>dtrace: failed to compile script class_loading_stat.d: line 156: probe
description syscall::exit:entry does not match any probes
># dtrace -s gc_time_stat.d -p 7341
>dtrace: failed to compile script gc_time_stat.d: line 202: probe description
syscall::exit:entry does not match any probes
>
>Other sample scripts are showing errors too.
>
>Lines 155 to 160 from class_loading_stat.d look like this:
>
>syscall::rexit:entry,
>syscall::exit:entry
>/pid == $target/
>{
>   exit(0);
>}
>
>If I remove line 156, the script seems to run OK.  
>Is this expected behavior? In other words, is
"syscall::exit:entry" valid for some solaris versions?
>
>
I do not think so.
I believe syscall::exit:entry is artifact left in these samples.
I''ll file bug to take care of it.

As workaround you can either use ''dtrace -Zs ...'' or just
delete the
line "syscall::exit:entry".

thanks,
-katya
>The details for my system are:
># uname -a
>SunOS unknown 5.11 snv_52 i86pc i386 i86pc
># java -version
>java version "1.6.0-rc"
>Java(TM) SE Runtime Environment (build 1.6.0-rc-b100)
>Java HotSpot(TM) Client VM (build 1.6.0-rc-b100, mixed mode, sharing)
>
>The sample scripts are located in 
>
>${JAVA_HOME}/sample/dtrace/hotspot
>
>thanks
>Steven
> 
> 
>This message posted from opensolaris.org
>  
>
>------------------------------------------------------------------------
>
>_______________________________________________
>dtrace-discuss mailing list
>dtrace-discuss at opensolaris.org
>  
>
-- 
Ekaterina Pavlova,
VMSQE Team, St. Petersburg.
http://blogs.sun.com/roller/page/vmrobot
http://blogs.sun.com/theaquarium_ru/

Thanks for the tip about -Z, I missed that one.
> I''ll file bug to take care of it
Thanks.

I noticed this description from the DTrace Guide itself:

System Call Anachronisms
In some cases, the name of the system call as provided by the syscall provider
is
actually a re?ection of an ancient implementation detail. For example, for
reasons
dating back to UNIX? antiquity, the name of exit(2) in /etc/name_to_sysnum is
rexit.
 
 
This message posted from opensolaris.org