RFR: 8339711: ZipFile.Source.initCEN needlessly reads END header

[Claes Redestad]

On Sun, 8 Sep 2024 14:39:06 GMT, Eirik Bjørsnøs <eirbjo at openjdk.org> wrote:

> Please review this cleanup PR which makes `ZipFile.Source.initCEN` not include the 22-byte trailing`END` header when reading the `CEN` section of the ZIP file.
> 
> The reading of the END header was probably brought over from native code with the transition to Java in JDK 9.
> 
> In the current JDK, the END header is unused. This needlessly complicates multiple code paths accessing the array since they must account for the trailing END record when calculating the end of CEN position.
> 
> Additionally, the enforcement of the maximum CEN size limit is currently off by one. It allows the construction of a byte array of size `Integer.MAX_VALUE - 1`, but this size is not supported by OpenJDK. Instead, the maximum CEN limit should be such that is does not exceed  `Integer.MAX_VALUE - 2`.
> 
> Testing:
> 
> The `EndOfCenValidation` test is updated to test the rejection of a CEN of size `Integer.MAX_VALUE - 1` as the new minumum rejected CEN size.
> 
> The `ZipFileOpen` benchmark seems neutral to this change.

@eirbjo: an alternative suggested offline by @LanceAndersen is that we merge the two PRs, since they intermingle a bit too much and have the same goal of optimizing initCEN. OK if I merge this into #21133?

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

PR Comment: https://git.openjdk.org/jdk/pull/20905#issuecomment-2374358554