This is a multi-part message in MIME format.
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Another security issue has been found in LedgerSMB < 1.1.5 and all
versions of SQL-Ledger which allows an attacker to engage in directory
transversal, retrieval of sensitive information, user account
fabrication, or even arbitrary code execution. This was fixed in
LedgerSMB 1.1.5 and despite ample warning, the maintainer of SQL-Ledger
has not corrected the problem.
The problem occurs because the blacklisting functions for the text
editor strip out potentially dangerous targets rather than denying
access when a problem is detected. The stripping of such "dangerous"
elements involves first stripping the $userpath (usually users) and then
the $memberfile (by default users/members) and then opening the file
So, to go up two levels and open foo.txt, you could pass a url
containing the argument of file=.users./users/members./foo.txt to the
url for editing the template. After these are stripped out, you are
left with ../../foo.txt. You can also retrieve the memberfile by using
the path of file=useuserusers/memberssrs/members. Then by crafting a
similar URL or by altering the web page to post custom variables, you
can cause the application to overwrite this file, possibly deleting or
changing passwords, or adding user accounts.
This can also be used to cause arbitrary code to be executed as well.
SQL-Ledger and LedgerSMB < 1.2 rely on server-writable and executable
Perl scripts to store user preferences. These scripts are run at every
page load, are created on login, and destroyed at logout. Using the
same method, you can add arbitrary Perl code to the end of these files
causing that to be loaded the next time the target user loads a page.
Content-Type: text/x-vcard; charset=utf-8;