Bug 17086

Summary: php-pear-Horde new security issue fixed upstream in 5.2.8
Product: Mageia Reporter: David Walser <luigiwalser>
Component: SecurityAssignee: QA Team <qa-bugs>
Status: RESOLVED INVALID QA Contact: Sec team <security>
Severity: normal    
Priority: Normal CC: lewyssmith, thomas
Version: 5   
Target Milestone: ---   
Hardware: i586   
OS: Linux   
URL: http://lwn.net/Vulnerabilities/663319/
Whiteboard: MGA5-64-OK
Source RPM: php-pear-Horde-5.1.4-8.mga5.src.rpm CVE:
Status comment:

Description David Walser 2015-11-04 19:46:23 CET
Debian has issued an advisory on November 3:
https://www.debian.org/security/2015/dsa-3391

The Debian bug has a link to the upstream commit that's supposed to contain the fix:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=803641

Reproducible: 

Steps to Reproduce:
Comment 1 Thomas Spuhler 2015-11-05 16:45:58 CET
Version 5.2.8 is the latest stable Horde

Status: NEW => ASSIGNED
Summary: php-pear-Horde new security issue fixed upstream in 5.2.11 => php-pear-Horde new security issue fixed upstream in 5.2.8

Comment 2 Thomas Spuhler 2015-11-05 17:02:30 CET
This bug has been resolved in mga5 and committed in cauldron. I will submit the build after php has been stabilized and php-pear has been updated.

The following packages are now in updates_testing:
php-pear-Horde-5.2.8-1.mga5.src.rpm
php-pear-Horde-5.2.8-1.mga5.noarch.rpm

CC: (none) => thomas
Assignee: thomas => qa-bugs
Whiteboard: (none) => CAULDRON TOO

Comment 3 David Walser 2015-11-05 18:21:17 CET
Thomas, php-pear-Horde still needs to be updated in Cauldron.

Are we sure that 5.2.8 has this commit?
https://github.com/horde/horde/commit/a199d74932c902844514b2a83d21e7e221257dae
David Walser 2015-11-05 18:21:32 CET

Whiteboard: CAULDRON TOO => (none)

Comment 4 David Walser 2015-11-05 18:22:32 CET
Oh, sorry I see that you already mentioned Cauldron.

The upstream advisory is confusing because of the different version numbers (I guess between the different components):
https://www.htbridge.com/advisory/HTB23272
Comment 5 David Walser 2015-11-05 21:38:59 CET
Fedora advisory confirming php-horde 5.2.8 is the right fix:
https://lists.fedoraproject.org/pipermail/package-announce/2015-November/170692.html

from http://lwn.net/Vulnerabilities/663319/
David Walser 2015-11-05 23:22:30 CET

URL: http://lwn.net/Vulnerabilities/663067/ => http://lwn.net/Vulnerabilities/663319/

Comment 6 Lewis Smith 2015-11-07 20:49:47 CET
Looking at this to discover what *is* php-pear-Horde, I am mystified by the description of it in Add/Remove Software (for version 5.1.4-8):
"This is now just an empty package that removes and obsoletes all php-pear-horde packages. They (php-pear-horde packages) can be installed individually by the user, using pear-install."

Some clarification, please.

CC: (none) => lewyssmith

Comment 7 Thomas Spuhler 2015-11-08 00:14:16 CET
yes, you are correct. We probably need to obsolete it in cauldron.
I already obsoleted all php-pear-horde packages earlier.
If you read on, the installation of the pear packages related to horde is problematic and in many cases breaks the installation. pear-install had all the good intention, but there was a reason why Fedora packaged them. Also, the pear folks didn't provide a lot of help making the rpm with correct deps. They wanted everybody to use pear-install.
So I would release this package to be safe as Fedora did.I don't think anybody is still using it.
Comment 8 Lewis Smith 2015-11-08 21:16:26 CET
Thanks Thomas.

Testing Mag5 x64
Just to see that the update happens without incident.

But the pre-update installation did not!
 # rpm -qa | grep php-pear-Horde
 #                               [i.e. Not installed]
 # urpmi php-pear-Horde
    $MIRRORLIST: media/core/release/php-pear-Horde-5.1.4-8.mga5.noarch.rpm
gosod php-pear-Horde-5.1.4-8.mga5.noarch.rpm o /var/cache/urpmi/rpms           
Paratoi...                       #############################################
      1/1: php-pear-Horde        #############################################
install failed
warning: %post(php-pear-Horde-5.1.4-8.mga5.noarch) scriptlet failed, exit status 1
ERROR: 'script' failed for php-pear-Horde-5.1.4-8.mga5
 # rpm -qa | grep php-pear-Horde
 php-pear-Horde-5.1.4-8.mga5
So, despite the warnings and 'failure', it seems to have been installed.

UPDATED from Update Testing repos, using MCC/System Update. No visible problem.
 # rpm -qa | grep php-pear-Horde
 php-pear-Horde-5.2.8-1.mga5
--------------------------------
As a cross-check, I tried un-installing it then re-installing directly from Updates Testing. Hmm... [tynnu = removing, pecyn = packet]
 # urpme php-pear-Horde
 tynnu php-pear-Horde-5.2.8-1.mga5.noarch
 tynnu pecyn php-pear-Horde-5.2.8-1.mga5.noarch.
      1/1: tynnu php-pear-Horde-5.2.8-1.mga5.noarch
                                 #############################################
 horde/horde not installed
[?]
 # urpmi php-pear-Horde
    $MIRRORLIST: media/core/updates_testing/php-pear-Horde-5.2.8-1.mga5.noarch.rpm
gosod php-pear-Horde-5.2.8-1.mga5.noarch.rpm o /var/cache/urpmi/rpms           
Paratoi...                       #############################################
      1/1: php-pear-Horde        #############################################
install failed
warning: %post(php-pear-Horde-5.2.8-1.mga5.noarch) scriptlet failed, exit status 1
ERROR: 'script' failed for php-pear-Horde-5.2.8-1.mga5
 # rpm -qa | grep php-pear-Horde
 php-pear-Horde-5.2.8-1.mga5
So it throws the same installation error as previously, but seems to get installed.

On the basis of "So I would release this package to be safe", I give it an OK which can be removed if any of the strange things I note matter.

Whiteboard: (none) => MGA5-64-OK

Comment 9 claire robinson 2015-11-10 11:02:35 CET
Thomas, any idea what is causing the %post failure?
Comment 10 Thomas Spuhler 2015-11-10 17:50:49 CET
yes, because it's an empty package. BTW, Fedora retired it early this year.
Comment 11 claire robinson 2015-11-10 17:55:03 CET
It sounds like the %post could maybe be removed then?
Comment 12 Thomas Spuhler 2015-11-10 17:59:07 CET
I guess, it could.
Actually, I wonder if we just should close this bug as invalid. The old version didn't have any files either.
Comment 13 David Walser 2015-11-10 18:01:34 CET
(In reply to Thomas Spuhler from comment #12)
> I guess, it could.
> Actually, I wonder if we just should close this bug as invalid. The old
> version didn't have any files either.

Yeah, if this update doesn't actually "fix" anything since it's a fake package, it should be INVALID.
Comment 14 Thomas Spuhler 2015-11-10 18:12:34 CET
Done. I just obsoleted it in cauldron (task-obsolete)
Sorry, I didn't remember this and made qa folks working on it.

Status: ASSIGNED => RESOLVED
Resolution: (none) => INVALID

Comment 15 claire robinson 2015-11-10 18:15:47 CET
No problem. We have improved Mageia quality :)