8219196: DataOutputStream.writeUTF may throw unexpected exceptions

[Martin Buchholz]

On Fri, Mar 15, 2019 at 12:46 PM Brian Burkhalter <
brian.burkhalter at oracle.com> wrote:

> May I ask “why” if it cannot exceed 65535?
>

You're right, that was a bad idea ...

The repetition of the overflow check bothers me.

Another idea: simply move the existing utflen > 65535 into the loop.
With modern processors, that should be close to free.


> On Mar 15, 2019, at 12:40 PM, Martin Buchholz <martinrb at google.com> wrote:
>
> Consider changing utflen to a long.
>
> On Fri, Mar 15, 2019 at 12:25 PM Brian Burkhalter <
> brian.burkhalter at oracle.com> wrote:
>
>> For [1] please review the proposed fix [2]. It is possible to
>> preemptively detect a sufficient condition for when the length of the
>> modified UTF-8 encoding of the string parameter will exceed the maximum
>> allowed value and thereby avert any numerical overflow in incrementing the
>> accumulator which counts the number of encoded characters.
>>
>> Thanks,
>>
>> Brian
>>
>> [1] https://bugs.openjdk.java.net/browse/JDK-8219196
>> [2] http://cr.openjdk.java.net/~bpb/8219196/webrev.00/
>
>
>