RFR: 8261702: ClhsdbFindPC can fail due to PointerFinder incorrectly thinking an address is in a .so

[Daniel D.Daugherty]

On Tue, 16 Feb 2021 21:43:53 GMT, Chris Plummer <cjplummer at openjdk.org> wrote:

>> While I was characterizing:
>> 
>>     JDK-8261844 serviceability/sa/ClhsdbFindPC.java#id1 failed with "'In code in NMethod for LingeredAppWithTrivialMain.main' missing from stdout/stderr"
>> 
>> I determined that I could easily reproduce serviceability/sa/ClhsdbFindPC.java failures
>> on my Ubuntu 16.04 test machine. I've applied this version of this PR:
>> 
>> $ git log HEAD^!
>> commit fea1ed851ce6a93030169e7de0daac2c2ac12a31 (HEAD -> pull/2562)
>> Author: Chris Plummer <chris.plummer at oracle.com>
>> Date:   Sat Feb 13 19:02:24 2021 +0000
>> 
>>     Check if addr is in DSO last to avoid finding addr in DSO when it is not due to DSO.size() returning too large of a value.
>> 
>> and retested. I no longer see serviceability/sa/ClhsdbFindPC.java failures.
>> I did a total of 6 runs (X3 configs) and there were no failures.
>
> @dcubed-ojdk Can you also try pull/2563. It's for https://bugs.openjdk.java.net/browse/JDK-8261710, which is an attempt at a more proper fix for this bug (this PR is just avoiding the real issue, which is an incorrect DSO size calculation.)

@plummercj - I tested pull/2563 and added a note over on that PR.
I see failures in 'release' and 'slowdebug' bits with pull/2563.

-------------

PR: https://git.openjdk.java.net/jdk/pull/2562