-
-
Notifications
You must be signed in to change notification settings - Fork 48
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Reduce leakage of disposable VM content and history into dom0 filesystem #4972
Comments
Thanks for the research! Some of those indeed needs improvements. But note that majority of this points in practice only give information about the name of disposable VM, which isn't really much. It will leak how much you use disposable VMs, but not really what you do inside. And since pool of available dispXXX names is relatively small (10k), those names aren't really unique to the specific qubes instance. |
I'm personally quite bothered by the ctrl-shift-c copy-paste mechanism writing to disk :-/
|
On Sat, Apr 13, 2019 at 03:55:40AM -0700, cfcs wrote:
I'm personally quite bothered by the ctrl-shift-c copy-paste mechanism writing to disk :-/
eeks :( where to exactly?
(and thank you for pointing this out.)
…--
tschau,
Holger
-------------------------------------------------------------------------------
holger@(debian|reproducible-builds|layer-acht).org
PGP fingerprint: B8BF 5413 7B09 D35C F026 FE9D 091A B856 069A AA1C
|
Not really. It is in /var/run/qubes, which is on tmpfs. With a little caveat: swap |
Ah, fair enough! I was just getting errors that "out of disk space," but that must have been from the |
On Sat, Apr 13, 2019 at 04:40:37AM -0700, Marek Marczykowski-Górecki wrote:
Not really. It is in /var/run/qubes, which is on tmpfs. With a little caveat: [swap](#977)
thanks for clarifying!
…--
tschau,
Holger
-------------------------------------------------------------------------------
holger@(debian|reproducible-builds|layer-acht).org
PGP fingerprint: B8BF 5413 7B09 D35C F026 FE9D 091A B856 069A AA1C
|
@andrewdavidwong - FYI, I don't think this one sub-item should be tagged minor: "xfwm4*.state files in /home/admin/.cache/sessions/ contains many references to disposable VM WINDOW TITLES (application names, web site names, etc.) since the installation of the Qubes install (as well as current session)." |
Sure, but what do you mean by "sub-item"? |
For what it's worth, I dont think that KDE stores anything like the same amount of state information as Xfce. |
There were 22 items in the original issue (10 in section 1, 12 in section 2). In particular, Section 2 (contents of files), Item 2 (xfwm4*.state files). Search for xfwm4. Brendan |
Yes, but the priority applies to the entire issue. We don't have priorities for proper parts of issues. I guess what you meant is that the presence of that particular item merits increasing the priority of the entire issue above "minor," which is fine. |
@brendanhoar Other than the Xfce specific issue, do you think any of the others rank above minor? |
However, I posit that the forensic value of knowing each and every disposable VM invocation from the date of installation (e.g. October 2018 on the installation I tested) may bump some of the other items out of the minor status even if they do not contain contents of disposable VM sessions. Perhaps I should break these out into separate issues grouped in some way? In my first post on this issue, I asked the qubes developers team to perform triage, but perhaps that is on me as the reporting user? At first glance, a good portion of the reported items related to logs/dmesg/console outputs could be mitigated either by via reduction of logging or by date-limited log rotation w/ scrub/erasure of files > n days old...or both. Thanks, |
There are two kinds of triage:
It sounds like you're thinking of the second one. Ultimately, we (not you) are responsible for both of these, but issue reporters can always help us with these by following the issue reporting guidelines -- in particular, filing separate issues separately. |
@andrewdavidwong - Thanks for the very clear response! Would you recommend that I spin-off a new separate issue for dispVM "content" remaining on the dom0 filesystem (I feel these are less likely to be considered "minor") and leave only the dispVM "usage history" items here? I would also modify the original post to note what was split out so as to not confuse people reading the issue thread later. B |
Sure, that sounds good. Thanks! |
Data logged to
I know this is an edge case, but I think that more warnings are better than none. |
QubesOS/qubes-desktop-linux-xfce4#28 should take care of the xfwm4-*.state files |
Xfwm4 keeps state files in `~/.cache/sessions/` even when the `Save session for future login` option is unchecked. This isn't a problem in most cases, but it can leave unnecessary forensic artifacts on the disk which can be a problem for systems like QubesOS: QubesOS/qubes-issues#4972
The problem you're addressing (if any)
Considerable information about the history of disposable VM usage, as well as some contents of data from inside disposable VM leaks into the filesystem of dom0 and in most cases, survives reboots.
In particular, the leakage into the xfwm4-*.state files (and lack of cleanup) is particular disturbing.
e.g. in dom0
rm disp_files.txt # if exists
rm disp_contents.txt # if exists
sudo grep -rn '/' -e 'disp[0-9]' --exclude-dir={dev,proc,sys} | sort > disp_contents.txt
sudo find / -name *disp[0-9]* | sort > disp_files.txt # matches references to disposable VM names in filenames
Some findings on files related to disposable VMs (via disp_files.txt):
Some finding related to files containing references to disposable VMs in dom0 (via disp_contents.txt):
Describe the solution you'd like
Where is the value to a user, and who might that user be?
All users who may want the historical content of the disposable VM usage not to be memorialized. Journalists, security researchers, etc.
Describe alternatives you've considered
None appear to cover 100% of use cases. Perhaps TAILS in an HVM might reduce it the most.
Additional context
Relevant documentation you've consulted
Related, non-duplicate issues
#3504 (similar, but does not provide a focus on specific leakage into dom0)
#4408 (anti-forensics on swap files only)
#1819 (anti-forensics request only covers the block devices for the VM)
#1293 (similar to 1819)
#3360 (specific to dom0 logging only)
#2024 (emulating Tails' approach, dom0 leakage still possible)
The text was updated successfully, but these errors were encountered: