http://bugzilla.suse.com/show_bug.cgi?id=1170162 http://bugzilla.suse.com/show_bug.cgi?id=1170162#c1 --- Comment #1 from Simon Lees --- Upstream: https://phab.enlightenment.org/T8671 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=1170162 http://bugzilla.suse.com/show_bug.cgi?id=1170162#c3 Matthias Gerstner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |IN_PROGRESS --- Comment #3 from Matthias Gerstner --- The upstream commit should fix the issue. I would have liked to see some common logic between the different commands instead of coding this complex query again. It's very hard to read... -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=1170162 http://bugzilla.suse.com/show_bug.cgi?id=1170162#c5 --- Comment #5 from Simon Lees --- (In reply to Matthias Gerstner from comment #4)

The umount has another possible attack surface. Even if the program only allows removable devices to be mounted ... I could insert a removable device with a UNIX file system on it that contains symlinks.

Then I could point the umount command to /media/$user/somemount/somelink and the link target would be unmounted. The problem is that the `umount` program is used for unmounting, which follows symlinks.

It would be better to use the `umount2` system call and pass "UMOUNT_NOFOLLOW" to avoid symlinks being followed. Another approach could be to forbid slashes after /media/$user/somemount and making sure that /media and /media/$user aren't user controlled.

Can you approach upstream with this?

I have approached upstream with these, I suspect the reason for not using the umount2 system call is that the mount / unmount section of this helper binary is only intended to be used as a fallback for systems with no udisks2 support ie bsd's, I suspect the safest fix is to conditionally remove the support at compile time when udisks is detected. -- You are receiving this mail because: You are on the CC list for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1170162 https://bugzilla.suse.com/show_bug.cgi?id=1170162#c7 --- Comment #7 from Simon Lees --- Following up here again sorry for the delay. The response from upstream was this is a fallback for systems that don't have udisks which is present on most Linux systems, so generally this code was designed to run on BSD systems which enlightenment does support. I don't believe most BSD systems include the umount2 syscall so it isn't really a solution. This code won't run on openSUSE so I can patch it out if it makes you more comfortable. -- You are receiving this mail because: You are on the CC list for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1170162 https://bugzilla.suse.com/show_bug.cgi?id=1170162#c9 --- Comment #9 from Simon Lees --- Patch to remove the code is https://build.opensuse.org/package/view_file/X11:Enlightenment:Factory/enlig... sorry it took so long -- You are receiving this mail because: You are on the CC list for the bug.