Bug 1690104

Summary:	seccomp rules for personality(2) deny lots of normal stuff
Product:	[Fedora] Fedora	Reporter:	Peter Jones <pjones>
Component:	skopeo	Assignee:	Antonio Murdaca <amurdaca>
Status:	CLOSED ERRATA	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	29	CC:	amurdaca, bbaude, dwalsh, lsm5, nalin
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	skopeo-0.1.37-0.gite079f9d.fc30 skopeo-0.1.37-0.gite079f9d.fc29	Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2019-06-24 00:56:42 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Peter Jones 2019-03-18 19:29:29 UTC

Description of problem: seccomp denies every personality(2) flag except PER_LINUX, PER_LINUX32, and -1.  This makes containers with OS userlands exhibiting https://bugzilla.redhat.com/show_bug.cgi?id=1690102 (such as RHEL 7) fail in this awesome way:

[root@daf23c28d8db build]# setarch linux32 -B uname -m
x86_64
[root@daf23c28d8db build]# setarch linux32 uname -m
i686

Version-Release number of selected component (if applicable): 0.1.35-1.git404c5bd.fc29.x86_64

How reproducible: 100%


Steps to Reproduce:
1. build an image with a rhel 7 (or centos 7) root
2. run "setarch linux32 -B uname -m" in it
3. observe the wrong answer, which would be right if ADDR_LIMIT_3G worked

Actual results: personality(2) gives -EPERM instead of working


Expected results: personality(2) works normally without error


Additional info: I'm not really sure what the point of limiting personality(2) flags is - it could really just go on the allowed system calls list.

Comment 1 Daniel Walsh 2019-03-18 19:41:16 UTC

Peter what exactly does the personality call do?

If you add the syscall to /usr/share/containers/seccomp.json does the rest of the code work correctly?

Comment 2 Peter Jones 2019-03-22 13:44:07 UTC

It's the syscall that switches out (among other things) what uname returns, limits how much address space you have to 32-bits (or 3 gig in this case), and things of that nature.  If I add more values to its arguments in seccomp.json it does work, or if I add it to the list of calls that are just allowed, it also works.

Comment 3 Daniel Walsh 2019-03-25 10:52:28 UTC

Could you send me or attach your modified seccomp.json file.  Then I can make the modifications.  If you don't see this is a possible security escallation, then I am willing to take in the changes.

Comment 4 Fedora Update System 2019-06-15 14:12:11 UTC

FEDORA-2019-96f06abcec has been submitted as an update to Fedora 29. https://bodhi.fedoraproject.org/updates/FEDORA-2019-96f06abcec

Comment 5 Fedora Update System 2019-06-16 00:54:25 UTC

skopeo-0.1.37-0.gite079f9d.fc30 has been pushed to the Fedora 30 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2019-e2637b6f18

Comment 6 Fedora Update System 2019-06-16 01:14:44 UTC

skopeo-0.1.37-0.gite079f9d.fc29 has been pushed to the Fedora 29 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2019-96f06abcec

Comment 7 Fedora Update System 2019-06-24 00:56:42 UTC

skopeo-0.1.37-0.gite079f9d.fc30 has been pushed to the Fedora 30 stable repository. If problems still persist, please make note of it in this bug report.

Comment 8 Fedora Update System 2019-06-24 10:39:46 UTC

skopeo-0.1.37-0.gite079f9d.fc29 has been pushed to the Fedora 29 stable repository. If problems still persist, please make note of it in this bug report.