[rrd-users] Re: rrdcgi segfault

Wed Apr 19 23:36:24 MEST 2000

Hi Tobi,

On Wed, Apr 19, 2000 at 09:46:16PM +0200, Tobias Oetiker wrote:

> first there is a braket missing
> 
>  <RRD::GRAPH <RRD::CV::PATH DS1>>

OK, that may happend during the stripping of the original cgi, but appen-
ding the braket didn't help.

> second you might want to run this from within ddd which is great
> for finding the place in the code where the segfault occurs

I ran it unter ddd and found the position where the segfault appears.
It's in free() in /lib/libc.so.6 called from line 339 of rrd_cgi.c.
I have no idea what happens. I single stepped cgigetqp, but don't see
any reason for the sigsev.
The buffer was not changed after allocating and i saw no buffer overflow
happen.
For obvious reasons i don't want to debug my libc, so i think i have to
give up.
The interesting point is, that the problem seams to depend on the content
of the RRD::CV::PATH parameter and the size:
DS1=
blablubbl			OK
blablubbla			segfault
DS1=blablubblablubblablub	OK

Do you have any idea?
Can i do some more digging?
Any other thing to do?

> if I trie it here (with the brace added)
> 
> I get:
> 
> oetiker.ee> ./cgitest.cgi
> (offline mode: enter name=value pairs on standard input)
> DS1=c_h_errors
> Content-Type: text/html
> Content-Length: 136
> 
> <HTML>
> <HEAD><TITLE>Tux statistics</TITLE></HEAD>
> <BODY BGCOLOR="#FFFFFF">
> [ERROR: can't make a graph without contents]
> </BODY>
> </HTML>

That's the thing i feared :-)

Ciao
-- 
Rainer Nagel
Rainer.Nagel at tashrah.com
Duesseldorfer Linux User Group - http://www.dlug.de

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users