We use `Math.ceil()` at 2 places in BMPImageWriter and both of them are under 
encodeRLE4() function. Also they are used only under absolute mode of encoding.

First usage `(int)Math.ceil((absVal-1)/2)` is while encoding data in each 
scanline. Here absVal always starts with 3 and increments by 2, so 
`Math.ceil()` is basically a no-op here and its usage can be removed.

Second usage `(int)Math.ceil((absVal+1)/2)` is while encoding end of scanline 
data. Here absVal can be even and we need to use `Math.ceil()` properly 
otherwise padding will not be added properly. I have added regression test 
which actually identifies this issue and it creates a corrupt encoded BMP 
image. In this regression test we have odd number of distinct pixels at the end 
of scanline and encoding without proper padding corrupts next scanline of data.

---------
- [x] I confirm that I make this contribution in accordance with the [OpenJDK 
Interim AI Policy](https://openjdk.org/legal/ai).

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

Commit messages:
 - Update test message
 - Add test, remove ceil() and floating point usage
 - 8384512: BMPImageWriter uses integer division before Math.ceil causing 
incorrect calculation

Changes: https://git.openjdk.org/jdk/pull/31447/files
  Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=31447&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8384512
  Stats: 114 lines in 2 files changed: 112 ins; 0 del; 2 mod
  Patch: https://git.openjdk.org/jdk/pull/31447.diff
  Fetch: git fetch https://git.openjdk.org/jdk.git pull/31447/head:pull/31447

PR: https://git.openjdk.org/jdk/pull/31447

Reply via email to