RFR: JDK-8306462: Remove NMT gtests that test dead-block-printing
Daniel D. Daugherty
dcubed at openjdk.org
Thu Apr 20 19:41:48 UTC 2023
On Wed, 19 Apr 2023 16:05:51 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:
> [JDK-8304815](https://bugs.openjdk.org/browse/JDK-8304815) introduced using NMT for more precise hs_err file printing. It came with gtests that test correct live- and dead-block printing.
>
> Live/dead-block printing: given a pointer into or very nearby live/dead malloced block, it tests that the printout identifies that block correctly.
>
> Unfortunately, the dead-block-printing tests are prone to intermittent errors, typically on platforms I don't own and cannot test. These tests are not worth the maintenance burden and should just be removed.
Thanks for the explanation. Since you have manually verified the original fix, I'm
okay with accepting your assertion that it is the tests that are buggy.
@dholmes-ora - please chime in on this PR when you get the chance.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/13539#issuecomment-1516852207
More information about the hotspot-runtime-dev
mailing list