<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Hi<br>
<br>
I've updated fix <br>
<br>
The vm.simpleArch variable has been added which corresponds to
os.simpleArch of tested platform. All hotspot tests have been
updated to use vm.simpleArch instead of os.simpleArch. Once jtreg
is fixed to read 'os.arch' from tested JDK then it would be possible
just revert this fix and preserver same behavior (See <a
href="https://bugs.openjdk.java.net/browse/CODETOOLS-7901695">CODETOOLS-7901695</a>)
. <br>
<br>
Updated webrev:<br>
<a
href="http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.01/root/"><a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~lmesnik/8157831/webrev.01/root/">http://cr.openjdk.java.net/~lmesnik/8157831/webrev.01/root/</a></a><br>
<a
href="http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.01/hotspot/"><a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~lmesnik/8157831/webrev.01/hotspot/">http://cr.openjdk.java.net/~lmesnik/8157831/webrev.01/hotspot/</a></a><br>
<br>
Testing is still in progress.<br>
<br>
Leonid<br>
<br>
<div class="moz-cite-prefix">On 16.06.2016 13:55, David Holmes
wrote:<br>
</div>
<blockquote
cite="mid:aa768e63-b7ec-18b5-3ee7-30dda3038dd7@oracle.com"
type="cite">On 16/06/2016 8:47 PM, Leonid Mesnik wrote:
<br>
<blockquote type="cite">Hi
<br>
<br>
On 09.06.2016 03:43, David Holmes wrote:
<br>
<blockquote type="cite">Hi Leonid,
<br>
<br>
Sorry for the delay.
<br>
<br>
On 7/06/2016 7:34 PM, Leonid Mesnik wrote:
<br>
<blockquote type="cite">Hi
<br>
<br>
Added <a class="moz-txt-link-abbreviated" href="mailto:jtreg-use@openjdk.java.net">jtreg-use@openjdk.java.net</a>
<br>
<br>
I think that you are right. Here is the documentation:
<br>
<a class="moz-txt-link-freetext" href="http://openjdk.java.net/jtreg/tag-spec.html">http://openjdk.java.net/jtreg/tag-spec.html</a>
<br>
<br>
<br>
"@requires <expression>
<br>
<br>
Express a dependence on characteristics of the system
being tested.
<br>
Some harnesses allow tests to be selected according to
the
<br>
characteristics of the system being tested. The
expression may be
<br>
composed of the following elements:"
<br>
<br>
"os.arch The operating system architecture, as given by the
<br>
corresponding system property."
<br>
<br>
So user could expect to have "os.arch" of tested VM.
<br>
<br>
If filed jtreg issues for this:
<br>
<br>
1. CODETOOLS-7901695
<br>
<a class="moz-txt-link-rfc2396E" href="https://bugs.openjdk.java.net/browse/CODETOOLS-7901695"><https://bugs.openjdk.java.net/browse/CODETOOLS-7901695></a>jtreg
uses
<br>
value 'os.arch' property of JDK which executed JDK and
not of
<br>
tested JDK
<br>
<br>
<br>
Following fix could be used as a temporary workaround while
jtreg fix is
<br>
not ready. Does it make a sense? I this case it is needed
to change
<br>
linux-arm64 back to linux-aarch64 to minimize changes.
<br>
</blockquote>
<br>
I still think we have a fundamental problem concerning what
os.arch
<br>
means. This workaround seems to work but I find it all very
confusing.
<br>
We really need a vm.arch property, distinct from os.arch.
<br>
</blockquote>
I see that CODETOOLS-7901695 is going to be fixed. After it is
<br>
implemented the 'os.arch' property will point to 'os.arch' of
tested JDK
<br>
as it described in jtreg documentation.
<br>
However there are 64 failures in nightly are caused by failures
of JVMCI
<br>
tests. Does it make a sense to implement this fix as a
workaround to
<br>
remove noise until jtreg is fixed?
<br>
</blockquote>
<br>
There may be some delay between jtreg being fixed and the updated
version being put in to use.
<br>
<br>
Implementing the workaround seems reasonable.
<br>
<br>
Thanks,
<br>
David
<br>
<br>
<blockquote type="cite">Leonid
<br>
<blockquote type="cite">
<br>
David
<br>
-----
<br>
<br>
<br>
<br>
<blockquote type="cite">Leonid
<br>
<br>
On 31.05.2016 04:26, David Holmes wrote:
<br>
<blockquote type="cite">Hi Leonid,
<br>
<br>
This really strikes me as as a jtreg problem that should
be fixed in
<br>
jtreg. When writing an @requires clause in a test the test
writer
<br>
should not have to be thinking "oh wait! Is this going to
query the VM
<br>
running jtreg or the VM running the test?". It should
obviously be the
<br>
VM running the test.
<br>
<br>
That said we also seem to have a problem with the
definition of
<br>
os.arch:
<br>
<br>
os.arch Operating system architecture
<br>
<br>
if it is returning the build architecture of the VM and
not the OS it
<br>
is running on. That in itself argues for two distinct
properties.
<br>
<br>
David
<br>
<br>
On 26/05/2016 11:45 PM, Leonid Mesnik wrote:
<br>
<blockquote type="cite">Hi
<br>
<br>
Could you please review following fix:
<br>
root
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~lmesnik/8157831/webrev.00/root/">http://cr.openjdk.java.net/~lmesnik/8157831/webrev.00/root/</a>
<br>
<a class="moz-txt-link-rfc2396E" href="http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.00/root/"><http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.00/root/></a>
<br>
hotspot
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~lmesnik/8157831/webrev.00/hotspot/">http://cr.openjdk.java.net/~lmesnik/8157831/webrev.00/hotspot/</a>
<br>
<a class="moz-txt-link-rfc2396E" href="http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.00/hotspot/"><http://cr.openjdk.java.net/%7Elmesnik/8157831/webrev.00/hotspot/></a>
<br>
for bug
<br>
<a class="moz-txt-link-freetext" href="https://bugs.openjdk.java.net/browse/JDK-8157831">https://bugs.openjdk.java.net/browse/JDK-8157831</a>
<br>
<a class="moz-txt-link-rfc2396E" href="https://bugs.openjdk.java.net/browse/JDK-8157831"><https://bugs.openjdk.java.net/browse/JDK-8157831></a>
<br>
<br>
The property "os.name" which was used to filter tests
depends on the
<br>
arch of jdk which is used to run jtreg. It might differ
from arch of
<br>
tested jdk.
<br>
This fix introduce new property "vm.arch" which depends
on the arch of
<br>
tested jdk and could be used to filter tests with
@requires.
<br>
<br>
I verified that tests are filtered where it is expected.
<br>
Note: I am going to push this fix in jdk9/hs to fix
regular hotspot
<br>
testing.
<br>
<br>
Leonid
<br>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>