Hi guys,
I have some strange, very reproducible segfaults in clamd on a
Debian-based system. I tried 0.83 and the cvs version. I built it from
source with debug symbols and static libs (to simplify debugging).
Anyway, my segfaults occur only when I use clamd in a setup where I do a
lot of
Hi guys,
This is my fourth submission for an environment variable patch for
VirusEvents. The other 3 I submitted all had various bugs.
This patch makes the virusaction function reentrant (my previous patches
broke reentrancy).
It didn't occur to me initially that modifying the environment is
On Mon, 14 Mar 2005, John Giammarche wrote:
Hello everyone and thanks for reading my message.
I want to use clamd to scan files that are uploaded to
a Java Servlet. So far, I've connected to clamd and
clamd answered the PORT that I should connect to send
the file. So far so good.
When I connect to
Hello? Is my patch going to make it into clamav? I really would like
VirusEvent scripts to get the infected filename in the environment.
Please accept my patch!
___
http://lurker.clamav.net/list/clamav-devel.html
Would anyone be opposed if I were to submit a patch so that clamuko and
clamd's scanner could support more than just %v in VirusEvent strings?
The function in question is clamd/others.c::virusaction().
In particular, I am thinking of supporting at least:
%f - infected filename
For clamuko,
On Wed, 9 Mar 2005, Tomasz Kojm wrote:
In particular, I am thinking of supporting at least:
%f - infected filename
It was already supported but has been removed due to security
issues.
Why is this a security issue?
Because it's hard to implement in a secure way. Read the change log,
google, ...
On Wed, 9 Mar 2005, Jeremy Kitchen wrote:
[...]
well, I know they'll probably tell you that the %f flag won't be resurrected..
but it might be kinda nice to add the filename to an environment variable
that the script can use...
Ooooh that would be nice! Can I do that? Please?! Pretty please?!
On Wed, 9 Mar 2005, Jeremy Kitchen wrote:
Shouldn't be overly difficult to set an environment variable with the filename
before calling the event program.
True. Now, what to call this variable? I was thinking
CLAM_VIRUSEVENT_*. Where * is one of FILENAME, VIRUSNAME, etc...
How does that suit
Also please create an example script that can use all the variables
which we can put into /contrib
The example script is attached...
-Calin
#!/bin/sh
# clam_email_alert.sh
# Calin A. Culianu calin at ajvar.org
#
# A small sample script to email alerts. To use it, put the following in your
NOTE: This replaces my previous patch. I fixed it up to use the more
compatible putenv() function which is POSIX. The setenv() function I was
using before was not POSIX.
Summary: This adds two environment variables for the purpose of VirusEvent
when executing an external command.
Argh, my last two patches are bad. This one should work fine.
Tomasz: Please use this one and disregard the last two I sent. This one
is 100% correct -- the other one had a problem with its use of putenv().
Sheesh putenv() is a retarded function! :)
diff -ur clamav-devel/clamd/clamuko.c
On Wed, 9 Mar 2005, Andy Fiddaman wrote:
This is looking good, just one problem and one comment.
mcalloc() is unfortunately not portable, it's probably best that you use the
ClamAV library function cli_malloc() for consistency.
Hmm. Well, why do other parts of the clamd program then call
How would people feel about adding some sort of caching feature to the
clamuko thread in order to cache the results of previous scans? This
could dramatically speed things up for on-access scanning of often-used
files..
Of course, the cache entry would have to be invalidated if the file is
13 matches
Mail list logo