Fix is good.
I think next are reasons you don't get MDO in this scenario.
Tier1 (C1 compilation) does not generate profiling code and does not created MDO. C1 request MDO only with tiers 2 and 3
[1][2].
With -Xcomp flag a Java method is not executed in Interpreter but requests its compilation and waits when it is
finished. As result MDO is not created in Interpreter too. May be late if a method is deoptimized it will be executed in
Interpreter and MDO will be created.
Thanks,
Vladimir
[1]
http://hg.openjdk.java.net/jdk/jdk/file/796c9fa50850/src/hotspot/share/c1/c1_Compilation.hpp#l226
[2]
http://hg.openjdk.java.net/jdk/jdk/file/796c9fa50850/src/hotspot/share/c1/c1_Compilation.cpp#l381
On 7/7/20 8:47 PM, Fairoz Matte wrote:
Thanks Chris, for the review comments.
I have updated the suggested change.
Thanks,
Fairoz
-----Original Message-----
From: Chris Plummer
Sent: Wednesday, July 8, 2020 3:38 AM
To: Fairoz Matte <fairoz.ma...@oracle.com>; hotspot-compiler-
d...@openjdk.java.net; serviceability-dev@openjdk.java.net
Subject: Re: RFR(s): 8236042: [TESTBUG] serviceability/sa/ClhsdbCDSCore.java
fails with -Xcomp -XX:TieredStopAtLevel=1
Hi Fairoz,
Looks good, except for the missing space in "if(testJavaOpts...".
thanks,
Chris
On 7/7/20 7:49 AM, Fairoz Matte wrote:
Hi,
Please review this small test change to consider the scenario when there is no
"printmdo" output
JBS - https://bugs.openjdk.java.net/browse/JDK-8236042
Webrev - http://cr.openjdk.java.net/~fmatte/8236042/webrev.00/
Thanks,
Fairoz
