Bug 199772 - LSPP: RHEL5 Alpha1 installs both 32- and 64- bit packages on ppc64
Summary: LSPP: RHEL5 Alpha1 installs both 32- and 64- bit packages on ppc64
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: anaconda
Version: 5.0
Hardware: powerpc
OS: Linux
medium
medium
Target Milestone: ---
: ---
Assignee: Anaconda Maintenance Team
QA Contact: Mike McLean
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-07-21 21:02 UTC by IBM Bug Proxy
Modified: 2007-11-30 22:07 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2006-10-06 15:41:11 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Description IBM Bug Proxy 2006-07-21 21:02:57 UTC
LTC Owner is: gjlynx.com
LTC Originator is: gcwilson.com


Contact Information = gcwilson.com or mcthomps.com
 
---Additional Hardware Info---
ppc64 LPAR
 
Machine Type = HV4 ppc64
 
---Debugger---
A debugger is not configured
 
---Steps to Reproduce---
Perform a minimal RHEL5 install on a ppc64 LPAR.  You will see a mixture of 32-
and 64- bit packages are installed.  
 
---Problem Description---
Both 32- & 64- bit versions of some packages are installed on 64-bit machines.
This can cause problems.  For example, if you have the 32-bit version of PAM
installed and install the 64-bit version, the 64-bit version of the helper
binaries silently overwrite the 32-bit versions.
 
---Installation and Packaging Component Data--- 
Install disk info:
  /dev/sda
 
Install method: Netboot, NFS
 
Install ISO information: RHEL5 alpha1

Userspace tool common name: rpm

The userspace tool has the following bit modes: Both

Userspace rpm: rpm
 
*Additional Instructions:

Comment 1 RHEL Program Management 2006-09-22 21:06:08 UTC
This request was evaluated by Red Hat Product Management for inclusion in a Red
Hat Enterprise Linux release.  Product Management has requested further review
of this request by Red Hat Engineering.  This request is not yet committed for
inclusion in release.

Comment 2 Suzanne Logcher 2006-10-03 21:36:52 UTC
Please retest on Beta 1 and provide package list duplicates.

Comment 3 Pete Graner 2006-10-05 18:40:43 UTC
Reassigning to tmaraz since this is a PAM issue. I think this is fixed I just
can't find the bz right now. Thomas can you pls look and make a dup or if not
then pls look into this ASAP. I'm marking it as a B2 blocker just to make sure
we don't loose it.

Thanks

Comment 4 Tomas Mraz 2006-10-06 08:23:17 UTC
Pete, why this should be a PAM issue? PAM and many other multilib packages
contain not only libraries but also binaries. It would be a big packaging change
to require all multilib packages to contain just libraries and split everything
else to a different subpackages. This problem definitely doesn't affect only PAM
and should be probably eventually resolved somehow but I don't think it is RHEL5
material.

Also note that the request is to not install both 32 bit and 64 bit packages by
default and not to modify packaging of PAM and everything else multilib so
probably anaconda is better component.

Comment 8 Tomas Mraz 2006-10-06 15:37:00 UTC
That minimal install of PPC RHEL 5 Alpha1 installed both ppc and ppc64 PAM and
other libraries although for LSPP eval they need just ppc libraries. (The same
for x86_64 where minimal install should install only x86_64 libraries and so on.)

I don't know if it has been resolved for RHEL5 Beta 1 or recent FC6Tests. I
didn't try minimal install recently. If it has been already resolved put it to
MODIFIED please.


Comment 9 Jeremy Katz 2006-10-06 15:41:11 UTC
This isn't a bug, it's by design so that people can have a _consistent_ and
_predictable_ experience on multilib systems.

Comment 10 Klaus Weidner 2006-10-09 23:08:40 UTC
The behavior we saw was that when installing the 64bit PAM library on the ppc64
platform, the working 32bit pam_tally executable binary from the 32bit PAM
library was silently overwritten with a nonworking 64bit version. (The login
programs use the 32bit versions of the pam_tally.so PAM module, and the data
format it uses in /var/log/faillog is incompatible with the 64bit pam_tally
program.) The same problem occured on x86_64.

That's not what I would call a consistent and predictable experience...

Anyway, this issue is moot now, the new pam_tally2 system claims to use a new
and word size independent storage format which would fix this issue.


Note You need to log in before you can comment on or make changes to this bug.