SCIENTIFIC-LINUX-DEVEL Archives

July 2012

SCIENTIFIC-LINUX-DEVEL@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Dr Andrew C Aitchison <[log in to unmask]>
Reply To:
Dr Andrew C Aitchison <[log in to unmask]>
Date:
Sat, 14 Jul 2012 09:58:16 +0100
Content-Type:
TEXT/PLAIN
Parts/Attachments:
TEXT/PLAIN (54 lines)
On Fri, 13 Jul 2012, Pat Riehecky wrote:

> There have been a number of questions regarding problem with starting X after 
> the recent security update.  This email is an attempt to summarize the 
> situation, provide insight, demonstrate the process as a whole, and clear up 
> some confusion with the solution.

> Cause:
>
> At the root of the problem was the security update for xorg-x11-server 
> published on July 9th.  The problem was caused by a change in the module ABI 
> for loading xorg drivers that happened in SL6.2.  That change was a bugfix 
> and not a security update, so it was not pushed out to earlier releases.  The 
> security update for xorg-x11-server uses the new ABI. The package, however, 
> does not force the use of these newer drivers though any rpm dependency tools 
> - such as requires, conflicts, or obsoletes. Thus the xorg-x11-drv update can 
> be installed without incident, even though it will not work as expected.  Our 
> usual dependency resolution process is unable to locate and report 
> dependencies that are not listed by the package.  The change in the ABI 
> happened with SL6.2, but none of the xorg packages were security errata at 
> that time.  So, we didn't catch the problem then as we didn't have to 
> integrate the 6.2 ABI with the 6.1 ABI.
>
>
> On the period in testing:
>
> The xorg-x11-server package was in the sl-testing repository for 12 days 
> before the date of release.
> I received a few reports from people who tested the packages while they were 
> in testing.  It appears that those testers were on 6.2 and thus didn't run 
> into the incompatibility problem.  There were no problems reported with this 
> package during the testing window.

In that case I may have a different problem.
My two test machines which failed were running 6.2 with (to the best 
of my knowledge) all updates. They may have been updated from earlier
6.1 releases, but both have had sl-release-6.2-1.1 at least since May.

I guess that my problems are slightly different from others as
I can't get a console or ssh connection to the broken machines,
but others seem to be able to.

I now regret not posting a "my machine wont boot with the 6.3
security updates in testing for 6.2" without any useful information
about what or why, but until I had time to boot from a Live-DVD
I had no way of communicating with the machine.

Sorry if I'm a bit short, but in spite of probably a day's
work on this my laptop is still a brick.

-- 
Dr. Andrew C. Aitchison		Computer Officer, DPMMS, Cambridge
[log in to unmask]	http://www.dpmms.cam.ac.uk/~werdna

ATOM RSS1 RSS2