fixes for the SELinux policy #1236
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
We don't use any of the SELinux controls for network access, such as packet or peer labeling. This sets permissive initial SIDs for these entities to avoid unexpected restrictions. Clean up the treatment of other initial SIDs at the same time. Many of the unused ones have network-related names, and the fact that they are treated differently can lead to confusion. Signed-off-by: Ben Cressey <[email protected]>
The intent was for type attributes like "trusted_s" to only contain subject types, but they were defined by negation and also included all the object types. By using xor, we end up with the set members we expect. Since objects are not typically the actor in an SELinux access check, this did not result in any additional permissions being granted, with one exception: the ability to add a file to a filesystem is governed by the "associate" action on the "filesystem" class, where the actor is the file rather than the process. Removing objects from the set of trusted users caused the system to fail to boot, since files could no longer be created or relabeled. To fix this, the "associate" action has been removed from the set of "relabel" actions, and replaced with rules that permit the expected behavior. Signed-off-by: Ben Cressey <[email protected]>
These actions were originally restricted because fanotify can be used to block access to files via permission events, which could disrupt the operation of the host. However, fanotify doesn't deliver events for accesses from outside the mount namespace, so it's unlikely that containers depend on this functionality. Meanwhile, the restrictions on file and directory watches interfered with containers that tried to monitor the journal file in order to collect logs. The updated policy allows broad use of inotify and dnotify watches, while preventing access to fanotify's permission events. Signed-off-by: Ben Cressey <[email protected]>
tjkirch
approved these changes
Dec 11, 2020
jamieand
approved these changes
Dec 15, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue number:
Fixes #1151
Description of changes:
Revise access rules for inotify, dnotify, and fanotify, as the policy previously blocked pods from collecting journal logs.
Adjust initial SIDs for network entities, as a preliminary step toward blocking network access from containers to the host.
Fix logic error in sets of subject types, which I uncovered while adding audit rules to log access to host sockets.
Testing done:
Ran the sonobuoy tests for the aws-k8s-1.18 variant (x86_64, aarch64). Test suite passed. Tested Cilium, Falco, EBS CSI driver, and the local storage provisioner on x86_64. Confirmed that the Cloudwatch Container Insights agent can collect journal logs. Ran a task for the aws-ecs-1 variant using the ECS CLI.
Verified in all cases that no new AVC denials were logged in
dmesgor in the journal.Terms of contribution:
By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.