Lucene search

K
githubGitHub Advisory DatabaseGHSA-FH74-HM69-RQJW
HistoryMay 27, 2021 - 6:41 p.m.

opencontainers runc contains procfs race condition with a shared volume mount

2021-05-2718:41:17
CWE-362
CWE-706
GitHub Advisory Database
github.com
88

7 High

CVSS3

Attack Vector

LOCAL

Attack Complexity

HIGH

Privileges Required

LOW

User Interaction

NONE

Scope

UNCHANGED

Confidentiality Impact

HIGH

Integrity Impact

HIGH

Availability Impact

HIGH

CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H

4.4 Medium

CVSS2

Access Vector

LOCAL

Access Complexity

MEDIUM

Authentication

NONE

Confidentiality Impact

PARTIAL

Integrity Impact

PARTIAL

Availability Impact

PARTIAL

AV:L/AC:M/Au:N/C:P/I:P/A:P

0.0005 Low

EPSS

Percentile

15.7%

Impact

By crafting a malicious root filesystem (with /proc being a symlink to a directory which was inside a volume shared with another running container), an attacker in control of both containers can trick runc into not correctly configuring the container’s security labels and not correctly masking paths inside /proc which contain potentially-sensitive information about the host (or even allow for direct attacks against the host).

In order to exploit this bug, an untrusted user must be able to spawn custom containers with custom mount configurations (such that a volume is shared between two containers). It should be noted that we consider this to be a fairly high level of access for an untrusted user – and we do not recommend allowing completely untrusted users to have such degrees of access without further restrictions.

Specific Go Package Affected

github.com/opencontainers/runc/libcontainer

Patches

This vulnerability has been fixed in 1.0.0-rc10. It should be noted that the current fix is effectively a hot-fix, and there are known ways for it to be worked around (such as making the entire root filesystem a shared volume controlled by another container). We recommend that users review their access policies to ensure that untrusted users do not have such high levels of controls over container mount configuration.

Workarounds

If you are not providing the ability for untrusted users to configure mountpoints for runc (or through a higher-level tool such as docker run -v) then you are not vulnerable to this issue. This exploit requires fairly complicated levels of access (which are available for some public clouds but are not necessarily available for all deployments).

Additionally, it appears as though it is not possible to exploit this vulnerability through Docker (due to the order of mounts Docker generates). However you should not depend on this, as it may be possible to work around this roadblock.

Credits

This vulnerability was discovered by Cure53, as part of a third-party security audit.

For more information

If you have any questions or comments about this advisory:

References

7 High

CVSS3

Attack Vector

LOCAL

Attack Complexity

HIGH

Privileges Required

LOW

User Interaction

NONE

Scope

UNCHANGED

Confidentiality Impact

HIGH

Integrity Impact

HIGH

Availability Impact

HIGH

CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H

4.4 Medium

CVSS2

Access Vector

LOCAL

Access Complexity

MEDIUM

Authentication

NONE

Confidentiality Impact

PARTIAL

Integrity Impact

PARTIAL

Availability Impact

PARTIAL

AV:L/AC:M/Au:N/C:P/I:P/A:P

0.0005 Low

EPSS

Percentile

15.7%