[jdk11u-dev] RFR: 8288763: Pack200 extraction failure with invalid size
Volker Simonis
simonis at openjdk.org
Thu Jul 14 09:12:04 UTC 2022
On Wed, 13 Jul 2022 20:51:48 GMT, Tyler Steele <tsteele at openjdk.org> wrote:
> > Do you know if the native zlib on your system contains some extra changes or enhancements compared to the upstream version? And by the way, what Linux distro did you observe this issue?
>
> Yes this was discovered on an IBM zLinux machine with a hardware accelerated zlib. It's special sauce galore! The OS is Ubuntu 20.04, and has been seen on RHEL 8.6 as well.
Thanks for the additional information. You should have said that right from the beginning :)
Interesting to see how the hardware acceleration seems to trade speed for compression ratio (the newly compressed files are all slightly bigger) in the same way the other, purely software based zlib enhancements are doing this (see for example https://github.com/simonis/zlib-bench/blob/master/Results.md).
You should also be aware that this difference in behavior can impact other code, both in the JDK core libraries as well as in user code. [JDK-8253952: Refine ZipOutputStream.putNextEntry() to recalculate ZipEntry's compressed size](https://bugs.openjdk.org/browse/JDK-8253952) has a detailed explanation of why this problem occurs along with a list of some internal classes which have been fixed and some external libraries which are affected.
Ideally, you should downport [JDK-8253952](https://bugs.openjdk.org/browse/JDK-8253952) but unfortunatley that can't be done because it changes the public API specification and has an associated CSR. Maybe it will be acceptable to only downport the code fix part of [JDK-8253952](https://bugs.openjdk.org/browse/JDK-8253952)?
-------------
PR: https://git.openjdk.org/jdk11u-dev/pull/1163
More information about the jdk-updates-dev
mailing list