suphp 0.6.2 backport request to dapper (0.6.1) and edgy (0.6.1) from feisty

Bug #73556 reported by Rouben
18
Affects Status Importance Assigned to Milestone
Dapper Backports
Invalid
Undecided
Unassigned
Edgy Backports
Invalid
Undecided
Unassigned
suphp (Debian)
Fix Released
Unknown
suphp (Ubuntu)
Fix Released
Undecided
Unassigned

Bug Description

Backport request: please backport suphp 0.6.2 to dapper and edgy from feisty.

Binary package hint: libapache2-mod-suphp

Whenever suphp refuses to run a script for any reason (e.g. UID/GID out of configured allowable range, wrong permissions, etc), it causes the following error messages to appear in the Apache error log:

---SNIP---
[Mon Nov 27 17:56:12 2006] [error] [client 142.150.160.59] Premature end of script headers: index.cgi
[Mon Nov 27 17:56:12 2006] [error] [client 142.150.160.59] SoftException in Application.cpp:193: Script "/var/www/index.cgi" resolving to "/var/www/index.cgi" not within configured docroot
[Mon Nov 27 17:56:12 2006] [error] [client 142.150.160.59] *** glibc detected *** double free or corruption (fasttop): 0x0806f990 ***
[Mon Nov 27 17:56:41 2006] [error] [client 142.150.160.59] Premature end of script headers: index.cgi
[Mon Nov 27 17:56:41 2006] [error] [client 142.150.160.59] SoftException in Application.cpp:291: UID of script "/var/www/index.cgi" is smaller than min_uid
[Mon Nov 27 17:56:41 2006] [error] [client 142.150.160.59] *** glibc detected *** double free or corruption (fasttop): 0x0806f9f8 ***
[Mon Nov 27 17:57:18 2006] [error] [client 142.150.160.59] Premature end of script headers: index.cgi
[Mon Nov 27 17:57:18 2006] [error] [client 142.150.160.59] SoftException in Application.cpp:472: Could not execute script "/var/www/index.cgi"
[Mon Nov 27 17:57:18 2006] [error] [client 142.150.160.59] Caused by SystemException in API_Linux.cpp:427: execve() for program "/var/www/index.cgi" failed: Permission denied
[Mon Nov 27 17:57:18 2006] [error] [client 142.150.160.59] *** glibc detected *** double free or corruption (fasttop): 0x0806f9f8 ***
---SNIP---

As you can see, the above are three distinct examples:

1. [Mon Nov 27 17:56:12 2006] was caused by the target script being outside of the allowable suphp docroot.
2. [Mon Nov 27 17:56:41 2006] was caused by wrong ownership: owner UID of the target script file was less than the allowable UID.
3. [Mon Nov 27 17:57:18 2006] was caused by wrong permissions (the www-data user/group has no read access to the script in question).

In all three cases, the last error message seen was always "*** glibc detected *** double free or corruption (fasttop): 0x0806f9f8 ***" which is a bit unnerving. I am not sure if this problem is potentially exploitable.

Note that this seems to be a known issue with suphp, and the latest release (0.6.2) seems to have addressed the issue according to the suphp homepage: http://www.suphp.org/

Changed in suphp:
status: Unknown → Unconfirmed
Rouben (rouben)
Changed in suphp:
assignee: nobody → rouben
status: Unconfirmed → Confirmed
Revision history for this message
Rouben (rouben) wrote :

This would appear to have been fixed in the latest unstable Debian version of the package. Thus this is a request to make the 0.6.1.20061108-1 version of suphp available in the Ubuntu package repositories (6.06 LTS and 6.10 at the very least).

Thank you!

Changed in suphp:
assignee: rouben → ubuntu-archive
Rouben (rouben)
Changed in suphp:
assignee: ubuntu-archive → ubuntu-backporters
Revision history for this message
John Dong (jdong) wrote :

Bugfix candidate for SRU. Please re-assign to Backporters if a developer asserts that a SRU would not be practical.

Changed in suphp:
assignee: ubuntu-backporters → nobody
Revision history for this message
Rouben (rouben) wrote :

As per John's suggestion, I'd like to submit this request for upgrade to the Ubuntu Stable Release Updates team. Thanks.

Changed in suphp:
assignee: nobody → ubuntu-sru
Changed in suphp:
status: Unconfirmed → Fix Released
Revision history for this message
Johan Christiansen (johandc) wrote :

Bump: (request for SRU or backport)

suPHP 0.6.2 has been released.
The following problems have been fixed with this release:

    * Double free() problem with certain versions of GCC

Features / improvements:

    * Apache 2.2 compatibility (Second reason why we need at least a backport)

Revision history for this message
Rouben (rouben) wrote :

Fix released in Feisty.

Changed in suphp:
assignee: ubuntu-sru → nobody
status: Confirmed → Fix Released
Revision history for this message
Rouben (rouben) wrote :

Requesting backports to dapper and edgy. Desired release already available in feisty.

Compatibility should not be an issue, since this is a fairly minor bugfix release (0.6.1 -> 0.6.2).

As reported by Johan already:
suPHP 0.6.2 has been released.
The following problems have been fixed with this release:

    * Double free() problem with certain versions of GCC

Features / improvements:

    * Apache 2.2 compatibility (Second reason why we need at least a backport)

Rouben (rouben)
description: updated
Revision history for this message
Rouben (rouben) wrote :

Builds OK on edgy using prevu as per the instructions here: http://ubuntuforums.org/showthread.php?t=268687

Minor tweaks needed in suphp-0.6.2/debian/control file to override dependencies on apache-2.2. Patch for contol file attached. Apply patch as usual:
patch <targetfile> <patchfile>

Revision history for this message
Rouben (rouben) wrote :

Builds OK on dapper as well (build host is edgy, but target is dapper) using prevu. The patch I attached with my previous comment also worked for the dapper build (please refer to my previous comment).

Revision history for this message
zenrox (kergan) wrote :

b/i/r(id assume dont know how to test)
+1 for edgy using proper patch

Changed in edgy-backports:
status: Unconfirmed → Confirmed
Changed in suphp:
status: Unknown → Fix Released
Revision history for this message
Rouben (rouben) wrote :

Can someone please confirm this on dapper as well? Thanks!

Revision history for this message
John Dong (jdong) wrote :

Please attempt this as a SRU first, or have an MOTU reply with a reason why a SRU is not appropriate.

Changed in dapper-backports:
status: Unconfirmed → Needs Info
Changed in edgy-backports:
status: Confirmed → Needs Info
Revision history for this message
Rouben (rouben) wrote :

This is a request for a MOTU SRU for the suphp package for both Edgy and Dapper as per the backporting team's request. This package has already been updated in Feisty.

Due to the nature of the problem (see bug description), this could possibly be a serious bug resulting in loss of user data or possibly crashes (memory management issue).

Changed in suphp:
assignee: nobody → motu-sru
status: Fix Released → Unconfirmed
Changed in dapper-backports:
assignee: nobody → snowmaninva66
status: Needs Info → In Progress
Revision history for this message
StefanPotyra (sistpoty) wrote :

Hi,

yes, this should be a SRU, however please follow https://wiki.ubuntu.com/MOTU/SRU. In particular I'd like to see a debdiff relative to the dapper and edgy source package and the description updated with what's actually the cause of the error and how it is resolved with an upgrade.

Thanks,
    Stefan.

Revision history for this message
StefanPotyra (sistpoty) wrote :

-> unassigning motu-sru. Please *subscribe* us, once you've got a debdiff to review. Otherwise it won't show up in our worklist.

Thanks,
     Stefan.

Changed in suphp:
assignee: motu-sru → nobody
Revision history for this message
Martin Pitt (pitti) wrote :

Closing Feisty task as per comment 5.

Changed in suphp:
status: Unconfirmed → Fix Released
Revision history for this message
Martin Pitt (pitti) wrote :

I do not see an approval from a backporters member here, so setting back to needsinfo.

Changed in dapper-backports:
status: In Progress → Needs Info
Revision history for this message
Rouben (rouben) wrote :

b/i/r +1 edgy
b/i/r +1 dapper

The patch mentioned above works!

Changed in dapper-backports:
assignee: snowmaninva66 → nobody
status: Needs Info → Confirmed
Changed in edgy-backports:
status: Needs Info → Confirmed
Revision history for this message
John Dong (jdong) wrote :

Backports Team does not (and is not authorized) to handle bugfix-only updates, unless the SRU path has no chance. I have not seen this the case, so as of now, the backports request is rejected. Reopen if SRU fails.

Changed in dapper-backports:
status: Confirmed → Rejected
Changed in edgy-backports:
status: Confirmed → Rejected
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.