Re: CVE request: lhasa: directory traversals

[Alexander Cherepanov <ch3root () openwall com>]

On 2015-01-14 20:33, Henri Salo wrote:
> On Tue, Jan 13, 2015 at 10:44:00PM +0300, Alexander Cherepanov wrote:
> > https://github.com/fragglet/lhasa/commit/64b96b5c1d08293b6c373f616b206d951ee358f7
> > https://github.com/fragglet/lhasa/commit/3bab39fd492a8924bdd25615ef40ca68c0c7ad0f
> > https://github.com/fragglet/lhasa/commit/adcd9912803e69ebeb000cc4c341fbc64820ed1f
> > https://github.com/fragglet/lhasa/commit/c26557dd1b2e640e9785686355c5a2945483460b
>
> All of these commits are only in versions 0.1.0 and 0.2.0 so no need for
> "incomplete fix for" CVE(s).

Yeah, all of them are between 0.0.7 and 0.1.0. But '..' and symlinks are 
usually seen as different issues.

> Use cases would have be nice.

An example for '..' is easy:

$ touch ../rel
$ jlha a test.lzh ../rel
$ rm ../rel
$ lhasa x test.lzh

But jlha front-end doesn't handle symlinks so you have to use non-free 
lha to archive symlinks. There are some examples in the lhasa repo though:

https://github.com/fragglet/lhasa/tree/master/test/archives/lha_unix114i

--
Alexander Cherepanov