> Regarding mainline: > - I decided not to 'unroll' the top while loop (i.e. `engineUpdate(byte[] > input, int offset, int len)` is unrolled) > - It is debatable which version is easier to understand. If this version > is 'too complex', I can unroll the top while loop. > - I do think this version is incremental (i.e. easier to review?): > - Move `remaining -= bytesToWrite;` into each `if` branch > - Change first `if` case to process multiple blocks instead of one > > This `while` loop has a lot of cases to remember; Very roughly: > > 1. process from previous call > 2. process current data > 3. store overflow > > Interesting situations: > - `blockOffset` might be non-`0` > - `remaining+blockOffset` might not be enough to fill a single block. Or just > enough for one block and to leave an overflow again.. > - etc. > > Regarding testing > - Correctness of intrinsic was already tested in > https://github.com/openjdk/jdk/pull/10582 so not adding any tests there (i.e. > no KAT) > - In principle, fuzz test should also be sufficient to test bytebuffer (did > increase repetitions)
Volodymyr Paprotski has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains four additional commits since the last revision: - bench and handle buf.position() - Merge remote-tracking branch 'origin/master' into avx512-poly-buf - whitespace - ByteBuffer support and tests ------------- Changes: - all: https://git.openjdk.org/jdk/pull/11338/files - new: https://git.openjdk.org/jdk/pull/11338/files/b31d7902..e307e344 Webrevs: - full: https://webrevs.openjdk.org/?repo=jdk&pr=11338&range=01 - incr: https://webrevs.openjdk.org/?repo=jdk&pr=11338&range=00-01 Stats: 13608 lines in 487 files changed: 8557 ins; 2924 del; 2127 mod Patch: https://git.openjdk.org/jdk/pull/11338.diff Fetch: git fetch https://git.openjdk.org/jdk pull/11338/head:pull/11338 PR: https://git.openjdk.org/jdk/pull/11338
