[OpenJDK 2D-Dev] [9] request for review: 8074954: ImageInputStreamImpl.readShort/readInt do not behave correctly at EOF

[Sergey Bylokhov]

Right. It will be thrown on the next iteration.
The fix looks fine then.

20.03.15 20:56, Phil Race wrote:
> I see. You were referring to the implementation. But your snippet
> doesn't show the loop which should handle this case properly :
>   350          while (len > 0) {
>   351             int nbytes = read(b, off, len);
>   352             if (nbytes == -1) {
>   353                 throw new EOFException();
>   354             }
>   355             off += nbytes;
>   356             len -= nbytes;
>   357         }
>
> -phil.
>
> On 03/20/2015 10:54 AM, Sergey Bylokhov wrote:
>>> Andrew, The fix looks good to me.
>>>
>>> Sergey, are you asking for that text to be added ? Looks to me like 
>>> its already there,
>>> so probably I am not following what you mean.
>> No, readFully() has this code:
>>             int nbytes = read(b, off, len);
>>             if (nbytes == -1) {
>>                 throw new EOFException();
>>             }
>> I assume this method should throw an exception if nbytes!=len as 
>> described in the specification. same as readShort/readInt.
>>
>>>
>>> -phil.
>>>
>>> On 03/20/2015 10:32 AM, Sergey Bylokhov wrote:
>>>> Hi, Andrew?
>>>> Hi, Andrew.
>>>> Should we update a readFully() also?
>>>>      * @exception java.io.EOFException if the stream reaches the 
>>>> end before
>>>>      * reading all the bytes.
>>>>      * @exception IOException if an I/O error occurs.
>>>>      */
>>>>     void readFully(byte[] b, int off, int len) throws IOException;
>>>>
>>>> 20.03.15 14:30, Andrew Brygin wrote:
>>>>> Hello,
>>>>>
>>>>>  could you please review a fix for CR 8074954?
>>>>>
>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8074954
>>>>> Webrev: http://cr.openjdk.java.net/~bae/8074954/9/webrev.00/
>>>>>
>>>>>  The problem happens if an input stream does not contain enough data
>>>>>  to read a multi-byte type (as 4-bytes integer or 2-bytes short)
>>>>>  completely. In this case the actual number of obtained bytes is
>>>>>  returned, and if we get at least one byte, the EOF exception in
>>>>>  not triggered.
>>>>>  As a result, an incorrect value is returned.
>>>>>
>>>>>  Suggested fix is to check explicitly whether required number of 
>>>>> bytes
>>>>>  has been read.
>>>>>
>>>>>  Supplied regression test demonstrates the problem.
>>>>>
>>>>>  Please take a look.
>>>>>
>>>>> Thanks,
>>>>> Andrew.
>>>>
>>>>
>>>> -- 
>>>> Best regards, Sergey.
>>>
>>
>>
>> -- 
>> Best regards, Sergey.
>


-- 
Best regards, Sergey.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/2d-dev/attachments/20150320/e078c7e5/attachment.html>