tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vanlerberghe, Luc" <luc.vanlerber...@bvdep.com>
Subject [PATCH] Tomcat4.0: org.apache.catalina.connector.http.SocketInput Stream
Date Mon, 08 Jan 2001 11:02:27 GMT
I noticed the possibility for an infinite loop in
org.apache.catalina.connector.http.SocketInputStream
after the commit on thursday...

I've attached a patch, but the diff is pretty unreadable...

I changed the following piece of code:

        do { // Skipping CR or LF
            try {
                chr = read();
            } catch (IOException e) {
                chr = -1;
            }
        } while ((chr == CR) || (chr == LF));
        if (chr == -1)
            throw new EOFException
                (sm.getString("requestStream.readline.error"));
        if ((chr != CR) || (chr != LF)) {
            pos--;
        }

1. When there's a IOException, there's no way out of the loop...
2. After the loop chr is neither CR nor LF, so the test looks obsolete. pos
should always be decremented.  (I didn't check why.  I only looked at this
piece of code, not the reasons behind the logic)
3. I moved the try {} catch outside the loop to make it a bit clearer (a
matter of taste, I guess but also I'm not so sure entering/leaving a try {}
catch doesn't involve a slight overhead)

This is the result:

        try {
            do { // Skipping CR or LF
                chr = read();
            } while ((chr == CR) || (chr == LF));

            pos--;
        } catch (IOException e) {
            throw new EOFException
                (sm.getString("requestStream.readline.error"));
        }

I compiled the code and tested some of the example pages, so I don't think I
did any damage...

Luc Vanlerberghe
 <<SocketInputStream.java.diff>> 

Mime
View raw message