Bug 903306

Summary:	haproxy: Fails to properly drop supplementary groups after setuid / setgid calls
Product:	OKD	Reporter:	Jan Lieskovsky <jlieskov>
Component:	Containers	Assignee:	Rob Millner <rmillner>
Status:	CLOSED UPSTREAM	QA Contact:	libra bugs <libra-bugs>
Severity:	low	Docs Contact:
Priority:	low
Version:	1.x	CC:	jeremy, jhonce, mfisher, robinlee.sysu, rohara, tkramer
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
URL:	http://www.openwall.com/lists/oss-security/2013/01/23/7
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:	903293	Environment:
Last Closed:	2013-04-16 22:33:36 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:
Bug Depends On:	903293, 903307
Bug Blocks:	903295, 903301, 903303

Description Jan Lieskovsky 2013-01-23 16:58:39 UTC

+++ This bug was initially created as a clone of Bug #903293 +++

Description of problem:
As noted in bug #894626 and in:
  [1] http://www.openwall.com/lists/oss-security/2013/01/23/7

haproxy previously failed to drop supplementary groups properly when trying to drop root privileges.

By itself this problem is not a security flaw, but still serious enough the upstream fix:
  [2] git.1wt.eu/web?p=haproxy.git;a=commitdiff;h=ab012dd3

to be backported into all of the affected versions.

Version-Release number of selected component (if applicable):
haproxy-1.4.19-1.el6_3

How reproducible:
Always

Steps to Reproduce:
1. See https://bugzilla.redhat.com/show_bug.cgi?id=894626#c0 for further reproducer details
  
Actual results:
Supplementary groups are not dropped properly after setuid / setgid calls.

Expected results:
(All) Supplementary groups should be dropped when dropping root privileges.

Comment 1 Jhon Honce 2013-01-23 19:07:38 UTC

Schedule haproxy update when this patch is available upstream.

Comment 2 Rob Millner 2013-02-27 19:22:24 UTC

OpenShift is not exposed to this issue, reducing severity.

With a both the system haproxy and a scalable app:

[root@ip-10-77-75-81 ~]# ps auxwwwww |grep haproxy
haproxy   9205  0.0  0.0  23412  3836 ?        Ss   14:18   0:00 /usr/sbin/haproxy -D -f /etc/openshift/port-proxy.cfg -p /var/run/openshift-port-proxy.pid -sf 1839
501       9502  0.0  0.0  16440  1312 ?        Ss   14:18   0:00 /usr/sbin/haproxy -f /var/lib/openshift/512e5bf6ef805c16c4000006//haproxy-1.4/conf/haproxy.cfg -sf 9373
501       9552  0.0  0.2 111032 22420 ?        Sl   14:18   0:00 haproxy_ctld.rb                                                                                  
root     10159  0.0  0.0 103236   876 pts/0    S+   14:19   0:00 grep haproxy

[root@ip-10-77-75-81 ~]# grep Group /proc/9205/status
Groups:	

[root@ip-10-77-75-81 ~]# grep Group /proc/9502/status
Groups:	501

Comment 3 Rob Millner 2013-04-16 22:33:36 UTC

Closing this bug since there's nothing for OpenShift to do but wait for upstream.