Thanks Roger, Naoto for reviewingthe changes
Updated the webrev
- Used ByteBuffer.wrap(inputBytes) instead of ByteBuffer.allocate(size)
- Added Decoder.decode(ByteBuffer)method test. (line 58)
http://cr.openjdk.java.net/~nishjain/8217969_8218265/webrev.01/
Regards,
Nishit Jain
On 05-02-2019 00:03, Naoto Sato wrote:
+1
Naoto
On 2/4/19 10:18 AM, Roger Riggs wrote:
Hi Nishit,
Looks fine.
Thanks for fixing the test and improving the implementation.
The test could save one 2G allocation by using
ByteBuffer.wrap(inputBytes)
instead of ByteBuffer.allocate(size).
Thanks, Roger
On 02/04/2019 10:31 AM, Nishit Jain wrote:
Hi,
Please review the below fix for 8217969 and 8218265
Bug: https://bugs.openjdk.java.net/browse/JDK-8217969
https://bugs.openjdk.java.net/browse/JDK-8218265
Webrev: http://cr.openjdk.java.net/~nishjain/8217969_8218265/webrev.00/
Issue: Fix for 8210583 changed the Base64.Decoder.decode to throw
OOME when an integer value overflows, which is an intermediate
value, not the final value, the final value is always less than
Integer.MAX_VALUE. Also, the test case written for 8210583, fails on
certain platforms and throws "OutOfMemoryError: Requested array size
exceeds VM limit ", due to the array size of length
Integer.MAX_VALUE - 2.
Fix: Handled the integer overflow by storing the intermediate value
as "long" and converting the final value as integer. Also, updating
the test case input array length to Integer.MAX_VALUE - 8, since
MAX_VALUE - 2 was used only to test the overflow condition in
Decoder.decode() which is no longer there.
Regards,
Nishit Jain
