RFR: 8242188: [TESTBUG] error in jtreg test jdk/jfr/api/consumer/TestRecordedFrame.java on linux-aarch64

[Erik Gahlin]

Looks reasonable (but perhaps still fragile??)

Erik

> On 12 May 2020, at 11:20, Nick Gasson <nick.gasson at arm.com> wrote:
> 
> Hi,
> 
> Any feedback on this patch? It's just tweaking the test to work around
> an oddity in -Xcomp on AArch64.
> 
> --
> Thanks,
> Nick
> 
> 
> On 04/29/20 16:11 pm, Nick Gasson wrote:
>> Hi,
>> 
>> A small fix for a jtreg test failure on AArch64:
>> 
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8242188
>> Webrev: http://cr.openjdk.java.net/~ngasson/8242188/webrev.0/
>> 
>> This test is run with -Xcomp and checks the main() method in the stack
>> trace is not interpreted. This fails on AArch64 because this port does
>> not support patching the instructions of a method after it is compiled,
>> so -Xcomp is very sensitive to the order classes are loaded in. In this
>> case the call to System.out.println() causes main() to be deoptimized
>> almost immediately and run in the interpreter.
>> 
>> Fix by calling an empty method instead, which has the same effect of
>> non-zero BCI for the test() call.
>> 
>> 
>> Thanks,
>> Nick
>