full-disclosure - Strange netcat behavior

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

From: thalm at netcabo.pt (thalm)
Subject: Strange netcat behavior

Ok, the difference between the tools is one thing. I only saw it through one perspective and I was wrong.
 
Now concerning the difference of behaviour of WebServers in terms of HTTP RFC and the need to have CRLF separating the HTTP "lines" (sorry for the term "lines" but I am sure you all understand it).
 
a) Why do Web Servers behave this way?
 
b) Which Web Browsers work by sending only LF?
I think a version of Internet Explorer 6 in Win98 does work this way (only LF)...but all others work the correct way, i.e, they send CRLF
 
Tiago Halm
http://www.kodeit.org

	-----Original Message----- 
	From: full-disclosure-admin@...ts.netsys.com on behalf of Giovanni Giacobbi 
	Sent: Fri 10/17/2003 3:43 PM 
	To: full-disclosure@...ts.netsys.com 
	Cc: 
	Subject: Re: [Full-Disclosure] Strange netcat behavior
	
	

	On Fri, Oct 17, 2003 at 01:56:02PM +0100, thalm wrote:
	> Since netcat is a widely used network tool, this may have been discussed already, but since I wasn't able to found such discussion, here goes:
	> 
	> When using netcat (Windows and Linux versions) to connect to a web server, and everytime ENTER is pressed in the command line, netcat only sends LF (0x0A) instead of CRLF (0x0D 0x0A).
	> ex: GET / HTTP/1.0[LF][LF]
	> 
	> when using telnet, the behavior is different:
	> ex: GET / HTTP/1.0[CRLF][CRLF]
	> 
	> Although webservers (IIS and probably Apache) don't mind such behavior and accept it (LF) as if it was CRLF, RFC 2616 clearly states that the HTTP Request/Response Line and HTTP Headers *MUST* be separated by a CRLF and not only by a LF.
	> 
	> Why is there such a difference between netcat and telnet behavior?
	> 
	
	Because they are different programs, with different purposes.
	Netcat is a stream handler, just like "cat". What it actually does is to take the source input stream
	and forward it to the output stream (a socket, in this case). Everything else is outside its scope.
	
	In your example, the difference is made by your TTY. Your TTY is actually feeding Netcat with a single
	linefeed, while telnet parses the input and converts manually those linefeeds in cr/lf combinations.
	Also, Netcat has no clue about what a "line" is, it doesn't make any difference between sent bytes,
	as it is binary-safe.
	
	I yet have to find a way to tell linux terminal to convert carriage returns to cr/lf rather than a
	single lf or cr (see stty(1) manpage).
	
	A quick workaround for your problem would be:
	 $ echo -en 'GET / HTTP/1.0\r\n\r\n' | nc localhost 80
	
	Regards
	
	--
	Giovanni Giacobbi
	
	_______________________________________________
	Full-Disclosure - We believe in it.
	Charter: http://lists.netsys.com/full-disclosure-charter.html