RFR: 8341595: Clean up iteration of CEN headers in ZipFile.Source.initCEN

[Claes Redestad]

On Sun, 6 Oct 2024 16:42:15 GMT, Eirik Bjørsnøs <eirbjo at openjdk.org> wrote:

> Please review this PR which suggests we clean up iteration and validation logic in `ZipFile.Source::initCEN` and some related methods to use a simpler and more consistent style:
> 
> * The main loop in `ZipFile.Source::initCEN` currently updates two iteration variables (`pos` and `entryPos`), where `entryPos` is always just `pos + CENHDR`. One variable should be sufficient. `entryPos` can be moved to an effectively final local variable scoped inside the loop.
> * The local variable `int limit = cen.length` no longer carries its weight and is inlined into its only use site
> * Since `entryPos` is no longer in scope for the loop predicate, this is updated to `pos <= cen.length - CENHDR`, instead of the current `entryPos <= limit`
> * The byte array passed to `countCENHeaders` contains only CEN data now, so the `size` parameter can be removed.
> *  `countCENHeaders` is updated to loop on the same predicate as `initCEN`. Additionally, this method is updated to  throw early if a CEN entry exceeds the CEN size (for consistency with similar logic in `checkAndAddEntry`)
> * The `headerSize` validation in `checkAndAddEntry` is updated to avoid widening conversion to `long` of a variable which can provably never exceed `Integer.MAX_VALUE` and to be consistent with  `countCENHeaders`.
> 
> Testing: 
> 
> A new test `CenSizeMaximum` is added:
> * This produces a ZIP file with a CEN size of exactly `MAX_CEN_SIZE` and verifies that the iteration logic handles ZIP files where the CEN size is on or near the limit.
> * This test also manipulates the END headers  'total number of entries' field, in order to exercise `countCENHeader`. 
> * The test is a bit of a resource hog: It produces a  ~1.5GB ZIP file on disk, requires > 2GB heap and takes > 15 seconds to run on my laptop. Let me know if this should be made a manual test.
> 
> GHA tests are currently pending. I have run the following tests locally:
> 
> 
> % make test TEST="test/jdk/java/util/zip" 
> % make test TEST="test/jdk/java/util/jar"

LGTM

src/java.base/share/classes/java/util/zip/ZipFile.java line 1772:

> 1770:             int pos = 0;
> 1771:             manifestNum = 0;
> 1772:             while (pos <= cen.length - CENHDR) {

The interpreter would probably like something like: 

int limit = cen.length - CENHDR;
while (pos <= limit) {

test/jdk/java/util/zip/ZipFile/CenSizeMaximum.java line 24:

> 22:  */
> 23: 
> 24: /* @test

Yes, this test probably should be either manual or moved to a higher tier.

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

Marked as reviewed by redestad (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/21378#pullrequestreview-2351690639
PR Review Comment: https://git.openjdk.org/jdk/pull/21378#discussion_r1790053062
PR Review Comment: https://git.openjdk.org/jdk/pull/21378#discussion_r1790055277