[kwlug-disc] Identify this exploit?
Mikalai Birukou
mb at 3nsoft.com
Sat Dec 28 09:48:26 EST 2019
## non-specific comment/rumble
Is it called in general a "directory traversal" bug?
I wanted to say that it is a PHP trouble (laughs), and then I thought
for a second:
- 7 months ago confluence (Java) had this style of bug. My confluence
server was hacked.
- A month go, when writing my own thing (NodeJS), I was thinking about
containing path at app level. Like any human I do mistakes.
- Is any listening service runs as root without dropping privileges for
request serving processes?
It is not just PHP. It's Unix problem. In general any process can name
any object in systems whole root. Unis doesn't have true POLA. Ideal
solution is in the future where everything runs as WASI with admin
specified allowed capabilities.
But today, the only action that I can take is to put that confluence
server in LXC container. Container that doesn't have non-system users in
/etc/passwd , cause there is nothing else in confluence LXC, not even jira.
How can I be sure that a human developer asked himself a question about
directory traversal, added code that isn't related to business logic,
and tested it in an adversarial approach? How can I be sure that usual
management will green light and pay for aforementioned steps, if young
developer identifies them? I can't be sure. I have to jail and separate
things from each other.
LXC all linux things!
On 2019-12-28 1:49 a.m., Paul Nijjar via kwlug-disc wrote:
> In my Apache logs I saw something like this, and my search-engine
> skills are weak:
>
> 133.18.209.124 - - [27/Dec/2019:04:09:39 -0500] "GET /download.php?file=../../../../../../../../../../../../etc/passwd HTTP/1.1" 404 209 "-" "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:62.0) Gecko/20100101 Firefox/62.0"
>
> It's pretty obvious what they are trying to do, but I am having
> trouble figuring out what the target is, exactly. Is this an exploit
> in a popular web package I should know about?
>
> - Paul
More information about the kwlug-disc
mailing list