<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
<div class="moz-cite-prefix">On 11/12/2013 17:55, Jonathan Gibbons
wrote:<br>
</div>
<blockquote cite="mid:52A8A72C.7050300@oracle.com" type="cite">Balchandra,
<br>
<br>
The important part of that of interest here is the part about
"calls jtreg commands
<br>
(one each for jdk, langtools and hotspot) with those recommended
<br>
options".
<br>
<br>
Is there a way you could post here the segment of the script that
does that?
<br>
<br>
</blockquote>
Here it is<br>
<br>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
<table border="0" height="300" width="749">
<tbody>
<tr>
<td>2.1 Running tests in jdk/test
</td>
</tr>
<tr>
<td>
<div style="margin-left: 40px;"> $ jtreg -dir:{openjdk
source top directory}/jdk/test -verbose:summary
-exclude:{openjdk source top
directory}/jdk/test/ProblemList.txt -conc:auto -a
-ignore:quiet -timeoutFactor:5 -othervm -testjdk:{location
of the test jdk} `cat <a
href="http://download.java.net/jdk8/testresults/docs/dir.list">dir.list</a>`
<br>
<br>
</div>
</td>
</tr>
<tr>
<td>
2.2 Running tests in langtools/test
</td>
</tr>
<tr>
<td>
<div style="margin-left: 40px;"> $ jtreg -dir:{openjdk
source top directory}/langtools/test -verbose:summary
-conc:auto -a -ignore:quiet -timeoutFactor:5 -agentvm
-testjdk:{location of the test jdk} com tools <br>
<br>
</div>
</td>
</tr>
<tr>
<td>
2.3 Running tests in hotspot/test
</td>
</tr>
<tr>
<td>
<div style="margin-left: 40px;"> $ jtreg -dir:{openjdk
source top directory}/hotspot/test -verbose:summary
-conc:auto -a -ignore:quiet -timeoutFactor:5 -agentvm
-testjdk:{location of the test jdk} compiler gc runtime
sanity serviceability
</div>
</td>
</tr>
</tbody>
</table>
<br>
The instructions is at
<a class="moz-txt-link-freetext" href="http://download.java.net/jdk8/testresults/docs/howtoruntests.html">http://download.java.net/jdk8/testresults/docs/howtoruntests.html</a><br>
and is linked from
<a class="moz-txt-link-freetext" href="http://www.java.net/download/jdk8/testresults/testresults.html">http://www.java.net/download/jdk8/testresults/testresults.html</a><br>
<br>
The caveat in this approach is a human error where one (me) forget
to update the instruction<br>
or forget to remove any additional flags used in the script
temporarily - a mismatch<br>
of the results could occur. <br>
<br>
From the next build (b120), I am going to update "2.1 Running tests
in jdk/test" to<br>
<br>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
$ jtreg -dir:{openjdk source top directory}/jdk/test
-verbose:summary -exclude:{openjdk source top
directory}/jdk/test/ProblemList.txt -conc:auto -a -ignore:quiet
-timeoutFactor:5 -agentvm -testjdk:{location of the test jdk}
:jdk_core :jdk_svc :jdk_beans :jdk_imageio :jdk_sound :jdk_sctp
javax/accessibility com/sun/java/swing javax/print sun/pisces
com/sun/awt <br>
<br>
<br>
<blockquote cite="mid:52A8A72C.7050300@oracle.com" type="cite">Alternatively,
that segment of the script could be a candidate for a target in
<br>
one or more test/Makefile files.
<br>
</blockquote>
<br>
This is good idea, but my experience with the 'make' is that if one
target critically fail, all<br>
subsequent targets will not run. I thought it is a restriction of
'make'.<br>
<br>
<br>
Thanks<br>
Balchandra<br>
<br>
<blockquote cite="mid:52A8A72C.7050300@oracle.com" type="cite">
<br>
-- Jon
<br>
<br>
<br>
On 12/11/2013 08:58 AM, Balchandra Vaidya wrote:
<br>
<blockquote type="cite">
<br>
Hi Volker,
<br>
<br>
The custom script I use just wrap the instruction posted (except
<br>
building jtreg). It just clones the openjdk repository, calls
jtreg commands
<br>
(one each for jdk, langtools and hotspot) with those recommended
<br>
options, runs diffs, runs nashorn tests, and archives results.
<br>
<br>
Such convenient script may not fit under openjdk repo, but think
of
<br>
adding it in code-tools project in the future.
<br>
<br>
<br>
Thanks
<br>
Balchandra
<br>
<br>
On 11/12/2013 13:45, Volker Simonis wrote:
<br>
<blockquote type="cite">Hi Balchandra,
<br>
<br>
what do you think about checking in a little script or
Makefile target
<br>
which produces the test results you report for the ea builds?
<br>
<br>
This would be of great help to produce comparable test
results.
<br>
<br>
Regards,
<br>
Volker
<br>
<br>
On Tue, Dec 10, 2013 at 5:39 PM, Balchandra Vaidya
<br>
<a class="moz-txt-link-rfc2396E" href="mailto:balchandra.vaidya@oracle.com"><balchandra.vaidya@oracle.com></a> wrote:
<br>
<blockquote type="cite">On 12/10/13 03:25 PM, Alan Bateman
wrote:
<br>
<blockquote type="cite">On 10/12/2013 11:20, Balchandra
Vaidya wrote:
<br>
<blockquote type="cite">:
<br>
<br>
2) Testcase do not run with -agentvm option. It might be
an
<br>
issue in the testcase (or how the testcase was
run). I will look into
<br>
it.
<br>
<br>
I missed this issue because I had changed my
scripts to use -othervm
<br>
sometime back for debugging/analyzing some testcase
<br>
failure but forgot to change it back. Thank you for
pointing it out.
<br>
</blockquote>
For Volker's benefit, othervm is where jtreg spins up a
new VM for each
<br>
test, agentvm is where jtreg re-uses the VM if possible.
There are tests
<br>
that don't or can't clean up and those tests need to
ensure that their @run
<br>
tests have the /othervm option so that the tests runs in
its own VM.
<br>
Alternatively, the TEST.ROOT file has a key that lists
directories where all
<br>
the tests in those directory trees must run in their own
VM.
<br>
<br>
Balchandra - one thing that be useful is to look at the
client tests to
<br>
see what can run in agentvm and what can't. For the
jdk_core and jdk_svc
<br>
tests then they can be run in either (although we're still
picking off
<br>
issues as they found in agentvm mode). I'm not aware of
any work done yet on
<br>
getting the tests in the jdk_desktop test group to run in
agentvm mode. It
<br>
may be that TEST.ROOT needs to be updated to see a few
additional top-level
<br>
directories so that all the tests in these areas
temporarily run in othervm
<br>
mode.
<br>
</blockquote>
<br>
In the subset of the tests I have been running, it appears
only one test [1]
<br>
is failing with agentvm option. I will try to run other
tests in client
<br>
groups [Note: I will update the instruction to use jtreg
groups with b119
<br>
results posting] and see what tests can be run in agentvm
mode, but main
<br>
issue
<br>
in those remaining client tests are some tests may not
consistently produce
<br>
same results - could stabilizing those tests be a goal for
jdk9 repo ?
<br>
<br>
<br>
Thanks
<br>
Balchandra
<br>
<br>
[1] com/sun/java/swing/plaf/windows/8016551/bug8016551.java
<br>
<br>
<blockquote type="cite">-Alan
<br>
</blockquote>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>