Java.io.IOException: Corrupt GZIP trailer

Question

I have some code to gunzip a byte array of data. I've used this code frequently and it works well. However, I just tried unzipping a byte[] array that is 4,529,218 bytes which expands to a file that is 2,157,763,194 bytes (on an SGI).
I get a:
java.io.IOException: Corrupt GZIP trailer
thrown from the while ( ( length = zipin.read(buffer, 0, sChunk) ) != -1) line below.
I did a quick Google search and found these links:
http://www.aoindustries.com/docs/aocode-public/com/aoindustries/util/zip/CorrectedGZIPInputStream.html
http://www.sanger.ac.uk/Software/Artemis/artemis_docs/uk.ac.sanger.pathogens.WorkingGZIPInputStream.html
with comments like:
Works around the "Corrupt GZIP trailer" problem in GZIPInputStream by catching and ignoring this exception.
A wrapper class to work around a bug in GZIPInputStream. The read() method sometimes throws a IOException complaining about a Corrupt GZIP trailer on the jdk 1.1.8-13 on the Alphas. This wrapper catches and ignores that exception.
The code below writes out the correct number of bytes to the filesystem, but I get the IOException. From my Google search, it appears that some people have run into this problem before. Does anyone here know if this is a bogus error, or if it is valid? Is this a known "bug" in Java?
Note: the file size is larger than the Integer.MAX_SIZE. Is there an integer somewhere in the code that checks the trailer that was exceeded because of the file size?
(I ommited some catch code to make reading it easier)
private void gunzip(byte[] data, File file)
    int sChunk = 8192;
    // create input stream
    GZIPInputStream zipin;
    try
        ByteArrayInputStream in = new ByteArrayInputStream(data);
        zipin = new GZIPInputStream(in);
    catch (IOException e)
        // omitted for clarity
    byte[] buffer = new byte[sChunk];
    FileOutputStream out = null;
    // decompress the data
    try
        out = new FileOutputStream(file);
        while ( ( length = zipin.read(buffer, 0, sChunk) ) != -1) // error here
            out.write(buffer, 0, length);
    catch (IOException e)
        // omitted for clarity
    finally
        try
            out.close();
            zipin.close();
        catch(IOException e)
            // omitted for clarity
}

Answer 1

In GZIPInputStream.java:
     * Reads GZIP member trailer.
    private void readTrailer() throws IOException {
     InputStream in = this.in;
     int n = inf.getRemaining();
     if (n > 0) {
         in = new SequenceInputStream(
               new ByteArrayInputStream(buf, len - n, n), in);
     long v = crc.getValue();
     if (readUInt(in) != v || readUInt(in) != inf.getTotalOut()) {
         throw new IOException("Corrupt GZIP trailer");
    }This line
if (readUInt(in) != v || readUInt(in) != inf.getTotalOut())appears to be verifying the CRC value and data size from the trailer with data actually read from the stream.
The inf.getTotalOut() returns an integer. My file size is greater than a 32 bit integer value.
I looked at the GZip file specification at:
http://www.faqs.org/rfcs/rfc1952.html
and see that the trailer input size is defined as:
ISIZE (Input SIZE)
            This contains the size of the original (uncompressed) input
            data modulo 2^32.I assume data modulo 2^32 means that this is supposed to be a 32 bit integer.