[users@httpd] possible memory leak in line ending conversion?

[Charlie Cook <cook@xxxxxxxxxxxxxxxx>]

        Before submitting this to bugzilla, I wanted to see if 
anyone else has run across this and found a solution.  I didn't run 
across anything in the archives, but it could have been my search 
word choices.  We are seeing huge amounts of memory being used when 
text is converted from UNIX line endings to Windows line endings on 
very large text returns.  We have a cgi that gives back lots and lots 
of formatted text as text/html.  The columns are no more than 70 and 
have hard returns.  As Apache converts the line endings from UNIX to 
Windows, you can watch the memory size of httpd grow.  One of our 
developers created a little test script to help show off the 
problem.  The script is:

#!/bin/bash
echo "Content-Type: text/plain;"
echo ""
echo "<pre>"
for i in `seq 1 $1` ; do
        echo "A"
done
echo "</pre>"
exit 0

        You can then call it with an argument of how many lines you 
want returned.  A good example would be if you call the script 
test.cgi and call it with 
"http://my_test_server/test.cgi?333333333";.  Going from UNIX to UNIX, 
you do not see a problem.  However, if you use a Windows client or 
just have wget say it is a Windows client with 
'--user-agent='Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; 
.NET)' you can watch the httpd grow fairly quickly.
        This has been tested on 2.0.55 and 2.2.0.  They both have 
the apparent leak, but 2.2.0 was about half as large as 2.0.55 when 
the same amount of data (~200M) had been sent.

Charlie


Contractor  


---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx
  "   from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx
For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx