general: 2.6.12-10 regressions from 2.6.12-9

[elijah]

Hibernation (suspend to disk) works fine with linux-image-2.6.12-9-k7, but will
not work with linux-image-2.6.12-10-k7.

When attempting to resume from disk with linux-image-2.6.12-10-k7, it starts to
resume but freezes halfway through the process.

When testing, I always use the same kernel to suspend as to resume.

My other hardware information is:

cpu: AMD Athlon(tm) XP 1700+, 1462 MHz
graphics card: nVidia GeForce2 MX/MX 400
storage CMD Silicon Image SiI 3112 SATARaid Controller
                           VIA VT82C586/B/VT82C686/A/B/VT8233/A/C/VT8235 PIPC
Bus Master IDE
disk: ATA SAMSUNG SP1213C
bridge: VIA VT8377 [KT400 AGP] Host Bridge
                           VIA VT8235 PCI Bridge
                           VIA VT8235 ISA Bridge

[Ben Collins (ben-collins)]

We've had varying reports of broken acpi from the 2.6.12-10 security package.
Have not yet determined the cause.

[Fabio Massimo Di Nitto (fabbione)]

Hi,

I am in the need of you to test 2 kernels for me. I cannot reproduce the problem
locally, so that doesn't help at all.

I did put 2 different kernels here:

http://people.ubuntu.com/~fabbione/security-kernels/

that reverts 2 specific patches of the last security update.

I need you to test them and tell me if any of them fix the problem.

NOTE: these are test kernels! if you are using binary drivers or compile your
own drivers, they might or might not work.
Also do NOT use for production machines since they do NOT cover all the security
updates.

Please report to me as soon as you can.

Thanks
Fabio

[elijah (elijah)]

The results of further testing are:

2.6.12-10-k7 no-CVE-2005-2709
reboots when waking from hibernate

2.6.12-10-k7 no-CVE-2005-3055
reboots when waking from hibernate

2.6.12-10-386
hangs when waking from hibernate

2.6.12-9-k7
works reliably

2.6.10-5-k7
hangs when waking from hibernate

I tested all kernels multiple times. It is very odd that only 2.6.12-9-k7 works,
and it works reliably while all the others fail reliably. It may be something
particular to my motherboard (tyan kt400). Thanks for looking into this.

[Ben Collins (ben-collins)]

*** Bug 27452 has been marked as a duplicate of this bug. ***

[Ben Collins (ben-collins)]

*** Bug 26463 has been marked as a duplicate of this bug. ***

[Ben Collins (ben-collins)]

*** Bug 26487 has been marked as a duplicate of this bug. ***

[Ben Collins (ben-collins)]

*** Bug 26497 has been marked as a duplicate of this bug. ***

[Ben Collins (ben-collins)]

*** Bug 26675 has been marked as a duplicate of this bug. ***

[Ben Collins (ben-collins)]

*** Bug 26943 has been marked as a duplicate of this bug. ***

[Fabio Massimo Di Nitto (fabbione)]

Hi again,

I am in the need of you guys to test some more kernels for. I cannot reproduce
the problem
locally, so that doesn't help at all.

I did put 2 different kernels (k7 and 686) here:

http://people.ubuntu.com/~fabbione/security-kernels/

there is a set of 6 subdirectories. Start from number 6 and go down to 1. You
can stop
as soon as the kernel stops working.

Please let me know from what version of these 6 you start experencing problems.

For amd64 people: the problematic patch is mostlikely to be the same. So we will
start with i386 testing.

NOTE: these are test kernels! if you are using binary drivers or compile your
own drivers, they might or might not work.
Also do NOT use for production machines since they do NOT cover all the security
updates.

Please report to me as soon as you can.

Thanks
Fabio

[ytene (ubuntu-ytene)]

Ben / Fabio,
Some more information for you. On Monday [19th] I tried another installation of
5.10, this time on a different AMD64 SMP architecture. My original post was
based on a struggle with a Tyan Thunder K8W running with two Opteron 250's. On
Monday I successfully completed an SMP installation on my other machine, an
A-Bit AV8 Motherboard with an Athlon64X2 4200 CPU. To my surprise this worked
perfectly and I write this to you know from this same hardware. I am running
with kernel image

2.6.12-10-amd64-k8-smp

and to my surprise and delight nothing else was necessary in order to get the
system to accept this image. This K8 kernel may still bork on my Opteron box
[will try at the weekend, will need a complete rebuild] but at least I have one
SMP system back ;o)

I'll try testing the other kernels and report that, too.

Regards,

Ytene

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #11)
> Ben / Fabio,
> Some more information for you. On Monday [19th] I tried another installation of
> 5.10, this time on a different AMD64 SMP architecture. My original post was
> based on a struggle with a Tyan Thunder K8W running with two Opteron 250's. On
> Monday I successfully completed an SMP installation on my other machine, an
> A-Bit AV8 Motherboard with an Athlon64X2 4200 CPU. To my surprise this worked
> perfectly and I write this to you know from this same hardware. I am running
> with kernel image
>
> 2.6.12-10-amd64-k8-smp
>
> and to my surprise and delight nothing else was necessary in order to get the
> system to accept this image. This K8 kernel may still bork on my Opteron box
> [will try at the weekend, will need a complete rebuild] but at least I have one
> SMP system back ;o)
>
> I'll try testing the other kernels and report that, too.

OK! I will try to prepare the same 6 test kernels for amd64 too. The idea is that
i stripped the security patch with a specific order. Once we get to the kernel
number that will stop working, I will be able to prepare another test kernel
to isolate the fault to one patch (theoretically). There is also the possibility
that the interaction between 2 patches is making all this mess.
Also.. the funny part is that NONE of the patches are touching stuff like SMP
or ACPI or k7/k8 specific bits. so the hunt to the problem is complex.

We do really appreciate the help with your testing!

Fabio

[elijah (elijah)]

> there is a set of 6 subdirectories. Start from number 6 and go down to 1.
> You can stop as soon as the kernel stops working.

Resume from hibernate failed for me in #6 and #5. I stopped testing after that.

-elijah

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #13)
> > there is a set of 6 subdirectories. Start from number 6 and go down to 1.
> > You can stop as soon as the kernel stops working.
>
> Resume from hibernate failed for me in #6 and #5. I stopped testing after that.
>
> -elijah

Quite impressive! considering that number 6 is only a rebuild of 2.6.12-9.

I think the problems are not the patches, but something in the environment that did
build 2.6.12-9 and the manual builds i am doing.

Adam can you please provide me the build logs for these 2 security updates?

thanks
Fabio

[Francois Joubert (sommerfj)]

I have submitted bug 26487 and do not understand how you can resolve this bug as
a duplicate of 20771.

However I did download Kernel #1, needless to say ACPI still hangs on boot!

Francois

[ytene (ubuntu-ytene)]

Ben / Fabio,

I have just rebuilt my dual Opteron system. This time, of course, Synaptic
pulled in linux-image-2.6.12-10-amd64-k8-smp, as opposed to the -9 version. I am
delighted to be able to advise you that the -10 release seems to have solved the
problem. One [tiny] observation to make...

During the boot process on my 4200X2 system the ubuntu logo screen would appear
and then beneath a horizontal progress bar I would see a list of the sevices
being installed. With the SMP boot on the Opteron system the ubuntu logo appears
briefly, then disappears, then I see the VGA text-mode status of components
being initialised and launched, much as I used to see when running Mandrake or
Mandriva. I am not sure if this is intentional or just an odd event. I will
report back if anything unusual happens at boot [and keep a check of
../var/log/messages for a few days, to make sure that all is OK.

Thanks for all your help!

Right ... time to go and install nVidia support next!

Regards,

Clive

[Adam Conrad (adconrad)]

(In reply to comment #14)
>
> Adam can you please provide me the build logs for these 2 security updates?

In my home directory on chinstrap. Enjoy.

[ytene (ubuntu-ytene)]

Correction to previous post - though I may have seen a return of the issue, I
believe I might have more of a clue.

At the time of my previous update I had a working SMP kernel on my machine,
running GNOME. Before updating to the nVidia drivers, I decided to pull in KDE.
To do this, I selected the following packages from synaptic. Please note that in
some cases the selection of one package automatically selects others. I
performed the selection in exactly the following sequence.

amarok
gtk2-engines-gtk-q2
kchart
kcontrol
kdeaddons
kdeaddons-kfile-plugins
kdeadmin-kfile-plugins
kdebase
kde-core
kdegraphics-kfile-plugins
kde-i18n-engb
kdemultimedia-kappfinder-data
kde-systemsettings
kdewebdev
kdm
kmoon
knetworkconf
koffice
ksystemlog
kubuntu-artwork-usersplash
openoffice.org2-kde

This time, when I performed a system reboot, I observed a similar hang as
before. Also, giving power to my speaker system, I noticed that instead of the
usual boot-up soundscape I would expect with KDE [and which I get with the
AMD64X2 4200 machine, all I can hear is square-wave white noise. I tried several
reboots and although I saw minor differences, I could not get the machine to
boot reliably back into an SMP kernel.

On the last reboot I returned to the uniprocessor linux-image-2.6.12-10-amd64-k8
and everything worked just fine. I now have KDE up and running on the
uniprocessor kernel.

I am going to rebuild from scratch, go to the SMP kernel under GNOME and then do
some basic testing.
I will then revert to the uniprocessor kernel and add KDE and do some more testing.
Lastly, I will boot back to the SMP kernel while staying with KDE and see what
happens.

Current view seems to be :-

UNI + GNOME = OK
UNI + KDE = OK
SMP + GNOME = OK
SMP + KDE = Not Working

Regards,

Clive

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #15)
> I have submitted bug 26487 and do not understand how you can resolve this bug as
> a duplicate of 20771.
>
> However I did download Kernel #1, needless to say ACPI still hangs on boot!
>
> Francois

Probably you want to start from #6 and go down to #1 like i asked to.

Thanks
Fabio

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #17)
> (In reply to comment #14)
> >
> > Adam can you please provide me the build logs for these 2 security updates?
>
> In my home directory on chinstrap. Enjoy.

Could you please add 10.24 and 9.23 please?

Thanks
Fabio

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #18)
>
> UNI + GNOME = OK
> UNI + KDE = OK
> SMP + GNOME = OK
> SMP + KDE = Not Working
>

Sorry Clive, just to make it more clear.. is this view true with both -9- and
-10- kernel?

Fabio

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #10)
> Hi again,
>
> I am in the need of you guys to test some more kernels for. I cannot reproduce
> the problem
> locally, so that doesn't help at all.
>
> I did put 2 different kernels (k7 and 686) here:
>
> http://people.ubuntu.com/~fabbione/security-kernels/
>
> there is a set of 6 subdirectories. Start from number 6 and go down to 1. You
> can stop
> as soon as the kernel stops working.
>
> Please let me know from what version of these 6 you start experencing problems.
>
> For amd64 people: the problematic patch is mostlikely to be the same. So we will
> start with i386 testing.

I added also amd64 kernels for testing.

Thanks
Fabio

[ytene (ubuntu-ytene)]

Fabio,

Confirmation that the results of my testing :

uni + GNOME = OK
uni + KDE = OK
SMP + GNOME = OK
SMP + KDE = Not Working

were all performed with 2.6.12-10-amd64-k8{-smp} kernels

I see in your note that you still need some testing done with the 12-9 release as well.
I'm going to be a little bit busy for the next day or so, but I will be happy to test
out the various kernel types probably on Thursday or Friday of this week.

Also, if my theory is correct, it is possible that some of the software that is
installed during the download of the listed KDE packages is responsible for breaking
the SMP solution. I suspect that the best way to track this down might be a bit lengthy
and will involve installing each of the packages, one at a time, and then re-testing. I
know this might be a lot of work, but I am happy to give this a try, too.

Will post results as soon as I have them.

Thanks for your continued support!

Clive

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #23)
> Fabio,
>
> Confirmation that the results of my testing :
>
> uni + GNOME = OK
> uni + KDE = OK
> SMP + GNOME = OK
> SMP + KDE = Not Working
>
> were all performed with 2.6.12-10-amd64-k8{-smp} kernels
>
> I see in your note that you still need some testing done with the 12-9 release
as well.
> I'm going to be a little bit busy for the next day or so, but I will be happy
to test
> out the various kernel types probably on Thursday or Friday of this week.
>

ok that's fine :)

> Also, if my theory is correct, it is possible that some of the software that is
> installed during the download of the listed KDE packages is responsible for
breaking
> the SMP solution.

that shouldn't but it can happen.

> I suspect that the best way to track this down might be a bit lengthy
> and will involve installing each of the packages, one at a time, and then
re-testing. I
> know this might be a lot of work, but I am happy to give this a try, too.
>

Well it is still a problem for the kernel, in theory userland shouldn't make
such a difference
but it is a reality that it does.

Fabio

[Adam Conrad (adconrad)]

(In reply to comment #20)
>
> Could you please add 10.24 and 9.23 please?

Done. (Isn't holiday response time great?)

[ytene (ubuntu-ytene)]

Created an attachment (id=5503)
Screen image showing system hang with latest patch cluster

Fabio et al.
Built a brand new 5.10 environment this morning to continue testing kernels for
this regression/SMP issue. Got the "updates available" message as per usual
[first time entering GNOME]. Confirmed that it was OK for the patches to be
installed.

EXCEPT - one of them appears to be a new kernel image - 2.6.12.10-amd64-k8
(2.6.12.10-25) - and when the installer gets to this package it hangs. I gave
the installer a good 20 minutes to complete [no luck] then took the attached
png screen image.

So it looks as though the patch team are working on this too, but perhaps that
the latest image they are releasing carries the bug forward.

Will now try and switch back to an earlier, working kernel, and from there
carry on testing Fabio's kernel images. The only problem is that the updater
appears [screen scrape shows this] to be apply the updates in alphabetical
sequence based on name. No way of knowing if killing it at this point is going
to create broken dependencies.

and to think I decided to give ubuntu a try because I was getting tired of
Mandriva being so unreliable... ;o)

[ytene (ubuntu-ytene)]

Kernel Testing – First Report

1.Installation of the kernel image.
In keeping with instructions, first attempt was to deploy the kernel in
sub-directory 6.
Installation of the kernel generated an error at the first attempt:-

root@orac:/media/sdc1/Software/Linux/ubuntu/Kernels/6# dpkg -i
linux-image-2.6.12-10-amd64-k8-smp_2.6.12-10.24_amd64.deb
Selecting previously deselected package linux-image-2.6.12-10-amd64-k8-smp.
(Reading database ... 56684 files and directories currently installed.)
Unpacking linux-image-2.6.12-10-amd64-k8-smp (from
linux-image-2.6.12-10-amd64-k8-smp_2.6.12-10.24_amd64.deb) ...
Setting up linux-image-2.6.12-10-amd64-k8-smp (2.6.12-10.24) ...
Failed to create initrd image.
dpkg: error processing linux-image-2.6.12-10-amd64-k8-smp (--install):
 subprocess post-installation script returned error exit status 2
Errors were encountered while processing:
 linux-image-2.6.12-10-amd64-k8-smp
root@orac:/media/sdc1/Software/Linux/ubuntu/Kernels/6#

Checked /boot and found a softlink there with a zero byte length and a .new
suffix that had clearly failed to complete during the initial dpkg -i command.

Removed the .new file and re-issued the dpkg command. This completed
successfully at the second attempt.

Checked /boot/grub/menu.lst in an attempt to ensure that edits completed
successfully:-

>## ## End Default Options ##
>
>title Ubuntu, kernel 2.6.12-10-amd64-k8-smp Default
>root (hd0,0)
>kernel /vmlinuz root=/dev/sda5 ro quiet splash
>initrd /initrd.img
>savedefault
>boot
>
>title Ubuntu, kernel 2.6.12-10-amd64-k8-smp Default (recovery mode)
>root (hd0,0)
>kernel /vmlinuz root=/dev/sda5 ro single
>initrd /initrd.img
>boot
>
>title Ubuntu, kernel 2.6.12-9-amd64-k8 Previous
>root (hd0,0)
>kernel /vmlinuz.old root=/dev/sda5 ro quiet splash
>initrd /initrd.img.old
>savedefault
>boot
>
>title Ubuntu, kernel 2.6.12-9-amd64-k8 Previous (recovery mode)
>root (hd0,0)
>kernel /vmlinuz.old root=/dev/sda5 ro single
>initrd /initrd.img.old
>boot
>
>title Ubuntu, kernel 2.6.12-10-amd64-k8-smp
>root (hd0,0)
>kernel /vmlinuz-2.6.12-10-amd64-k8-smp root=/dev/sda5 ro quiet
>splash
>initrd /initrd.img-2.6.12-10-amd64-k8-smp
>savedefault
>boot
>
>title Ubuntu, kernel 2.6.12-10-amd64-k8-smp (recovery mode)
>root (hd0,0)
>kernel /vmlinuz-2.6.12-10-amd64-k8-smp root=/dev/sda5 ro single
>initrd /initrd.img-2.6.12-10-amd64-k8-smp
>boot
>
>title Ubuntu, kernel 2.6.12-9-amd64-k8
>root (hd0,0)
>kernel /vmlinuz-2.6.12-9-amd64-k8 root=/dev/sda5 ro quiet splash
>initrd /initrd.img-2.6.12-9-amd64-k8
>savedefault
>boot
>
>title Ubuntu, kernel 2.6.12-9-amd64-k8 (recovery mode)
>root (hd0,0)
>kernel /vmlinuz-2.6.12-9-amd64-k8 root=/dev/sda5 ro single
>initrd /initrd.img-2.6.12-9-amd64-k8
>boot
>
>title Ubuntu, memtest86+
>root (hd0,0)
>kernel /memtest86+.bin
>boot
>
>### END DEBIAN AUTOMAGIC KERNELS LIST

2.Testing of Kerne...

[Fabio Massimo Di Nitto (fabbione)]

(In reply to comment #27)
> Kernel Testing – First Report
>

>
> 2.Testing of Kernel Image
> Attempts to boot the /6 kernel image all reach the same point – grub selection
> completes, the screen is replaced with the default ubuntu wallpaper with the
> horizontal loading bar and the message “loading modules” and the system then
> hangs. Waiting for 5 minutes confirms system hang at that point.
>
> 3.Testing of Safe Mode Kernel Image
> Attempting to boot into the new Kernel in safe mode result in a similar
> end-point. However, after a brief delay the screen clears and the text-mode of a
> dump starts to scroll rapidly up the screen. This continues indefinitely [core
> dump maybe?] until I use the escape key to interrupt, where I get the following
> message displayed on my screen:-
>
> Kernel panic – not syncing – Aieee!!!
>
> [ 89.757625] Code: 80 3f 00 7e f9 e9 1d fd ff ff f3 90 80 3f 00 7e f9 e9 4a fd
> [ 89.757795] console shuts up...
>
>
> At which point I assume that this kernel [even though it is No. 6] has the sync
> problem that I observed in /var/log/messages when I reported this error initially.

SCORE! number 6 is just a rebuild of 2.6.12-9. *shrug*

> 4.Addenda relating to new Kernel Images
> To give maximum validity to test results, I decided to perform all these tests
> on a new, fresh, “vanilla” build. After the initial boot of the machine the
> “Update Manager” in GNOME reported new packages for downloading. The first time
> I saw this I accepted automatically without checking first. Turns out that there
> is a new kernel being shipped, 2.6.12.16.1. For what it's worth, this hangs my
> machine just as efficiently as all the rest!

don't mind it. that's only a meta package.

>
> 5.Final thought.
> I haven't tried this yet... but it so happens that I do have the i386 DVD for
> ubuntu 5.10 which I downloaded at the same time as my 64-bit version. Since I
> can't do anything useful with your test kernels [sorry, Fabio! I am happy to
> take your suggestions/directions on how best to proceed] I might just wipe away
> this 64-bit build and try a 32-bit version instead...
>
> Awaiting next instruction...

I guess the next step is to see why the build environment is making such a mess.

At this point given that number 6 is a rebuild of 2.6.12-9, i can't see any
reason why it should hang that way.

For now i will need to return from holidays, get access again to the build farm
and start poking bit by bit what hell is going on.

Thanks a lot for your tests
Fabio

[Ben Collins (ben-collins)]

*** Bug 27886 has been marked as a duplicate of this bug. ***

[ytene (ubuntu-ytene)]

Fabio,

Suggestion for you. Over the weekend I tried to deploy a revised version of
Mandriva Linux 2006, which they posted on their Members-only mailing list in
December. Unfortunately the original bugs are still there, so they have more
work to do. _However_, it did occur to me that you/we might benefit from the
fact that Mandriva 2006 does not display any problems with supporting the AMD64
chipset in an SMP environment. [Their bugs are with their own code. lol].

In fact, there are [at least] 3 things that work with Mandriva that do not work
with ubuntu 5.10 :-

1. The AMD64 SMP kernel [actually any SMP-kernel] is automatically loaded by the
installer program [since it detects the presence of an SMP-based machine]. This
would be a nice-to-have for ubuntu.

2. The installation process also automatically detects the presence of an
nVidia-based graphics card system and then performs some kind of kernel
module-patching exercise, during installation, to patch the deployed kernel with
the appropriate nVidia modules. The technique used is extremely reliable [I've
not had a single problem with GeForce2, 3, or 4 cards] and completely
effortless/transparent to the user.

3. The sound system works perfectly with Mandriva. With ubuntu, any use of the
sound system generates white noise. See Bug #23731.

This is a [very cheeky] suggestion, but do you think it would be possible to
examine the Mandriva build and see how they have managed to avoid these
problems? I am sure that these are just simple configuration issues...

If it helps [maybe I could use modprobe or something to get more details of a
working kernel] then please feel free to tell me what needs to be done.

I am very sorry if this sounds like a rant, but it's a bit frustrating when you
realise that others in the GNU/Linux community have encountered and solved a
technical problem that is baffling us and that we seem unable or unwilling to
learn from their experience to resolve it! I do appreciate the differences
between .deb and .rpm distros, and of course that fully-built distros are quite
complex, but we're all here to share and enjoy Linux, not re-invent wheels!
Sorry, sermon over. ;o)

Seriously, there must be a way that we can leverage the work of others, rather
than force you to go through some trial-and-error routine with the build farm,
just to churn out a bunch of kernels for me to test...

What do you think?

Happy New Year to all

Clive

[Ben Collins (ben-collins)]

(In reply to comment #30)
> Fabio,
>
> Suggestion for you. Over the weekend I tried to deploy a revised version of
> Mandriva Linux 2006, which they posted on their Members-only mailing list in
> December. Unfortunately the original bugs are still there, so they have more
> work to do. _However_, it did occur to me that you/we might benefit from the
> fact that Mandriva 2006 does not display any problems with supporting the AMD64
> chipset in an SMP environment. [Their bugs are with their own code. lol].

This bug is long enough as it is. Let's keep it on topic. None of your
suggestions have anything to do with the problem this bug report is meant to
help fix. The problem, as Fabio has almost concluded is likely a regression in
binutils. Figuring that out had nothing to do with what other distributions are
doing, since we did not have this problem just a couple of kernel revisions ago.
Looking at the other distros wouldn't have helped us figure this out either.

As for the detection issues, those are installer problems, and not anything to
do with amd64 specifically, and way off topic for this bug report.

[ytene (ubuntu-ytene)]

Fabio,

Not sure if this adds any value to your research, but I have just completed
another ubuntu build on my Thunder K8W system. This time I used the i386 DVD. So
far, everything is going well. Sequence:

1. Base system build.
2. System patch based on recommendations from GNOME's Update Manager [complete set].
3. Installation of the i386 SMP kernel [that's 2.6.12-10-686-smp][tested and
verified working perfectly].
4. Installation of KDE packages. - Note that I applied *EXACTLY* the same set as
described in my update posting of 2005-12-27.
5. Reboot into SMP+KDE environment.

This i386 environment appears to be working exactly as I would have hoped from
the AMD64 environment. Of course, it is now 10 days since that test, so it is
possible that some of the downloaded packages may have had incremental fixes
applied since my last AMD64 test.

To validate this theory I will try another AMD64 build tomorrow.

However, other users tracking this thread might be encouraged to try a 386
installation on their AMD64 environments, at least as a short-term work-around.
We know that the AMD64 core processor design executes i386 instructions even
more efficiently than Pentium family chips and more effectively than 32-bit
Athlon chips of the same clock speed, so in theory we should all still see good
performance, if not full 64-bit computing.

I hope that your testing is making progress, and perhaps even this update
provides you with some useful background information.

Thanks and Regards,

Clive

[Ben Collins (ben-collins)]

*** Bug 28222 has been marked as a duplicate of this bug. ***

[Fabio Massimo Di Nitto (fabbione)]

Hi everybody,

I have another 2 kernels for you to test here (amd64-k8 and k8-smp)

http://people.ubuntu.com/~fabbione/security-kernels/binutils/

in the subdir you will find old/ that is a rebuild of 2.6.12-9 as it was builded from buildd (theoretically) and new is a rebuild of 2.6.12-9 with a version of binutils that have been upload after the working 2.6.12-9 hitted the archive.

Expected results: old/ should work just fine. new/ should show breakage.

Please test and let me know.

Thanks
Fabio

[ytene (ubuntu-ytene)]

Fabio, Ben,

I've just completed some [very preliminary] testing with Dapper. So far, I can advise that [on the Thunder K8W dual-Opteron board, at least] this bug appears to have been eliminated. [I do appreciate that Dapper uses a much later kernel, but hey, this is progress...] Basic testing shows that the installer auto-detects the 2nd [SMP] processor and the machine is configured to handle this from the outset. Intrestingly, neither uname -r, nor Synaptic list -SMP specific kernels for this release, even though it's clearly an SMP kernel, as ably demonstrated by, for example, GNOME System Monitor. I must go read the release notes in more detail...

Obviously Dapper is still at Flight-6 alpha release stage, but on the basis of a couple of hours basic GNOME/KDE app/functionality testing this is looking mightily impressive.

Am currently running the Breezy i386-SMP kernel quite comfortably on the K8W, but now looking to jump to Dapper in June, once the release is declared stable.

Not sure if either of you have been involved in the Dapper kernel development, by my thanks and appreciation to all that have!

Regards,

Clive

[Fabio Massimo Di Nitto (fabbione)]

Hi everybody again,

I still have 2 kernels for you to test here (amd64-k8 and k8-smp)
and nobody did asnwer my request for information. Should I assume this problem is gone automatically?

please test and let me know.

Thanks
Fabio

[ryan (ryanvw1st)]

Hai Fabio,

I don't know about the K8 troubles, however I do know
about the K7 troubles with 2.6.12-10(k7) and with
2.6.12-10(386).

I have been installing more stuff (also Sun Java,
Konquerer, Krusader, Opera, Realplayer10, etc) on my
ubuntu (Breezy) AthlonXP 2600+ ws and I still will
have system freezing/system hangs events whenever I
run a 2.6.12-10 kernel!

However on kernel 2.6.12-9(k7) it runs stable and
without problems, which I definitely prefer !

my Breezy Badger was updated online from Hoary
Hedgehog, some small Dapper tools are in there too. I
take all the automatic updates that come by..

I use this system as an actual ws so I rather don't do
'experimental stuff' with it, other than updates and
installs that I need.

I will however try any reworked 2.6.12-10 kernel to
see if it is stable when it comes available for
apt-get..

ofcourse soon now, Dapper will be out and I will then
upgrade to the 2.6.15-6? kernel. Hope it works ;)

Thanks for your efforts!
Ryan van Wijngaarden
netherlands

--- Fabio Massimo Di Nitto <email address hidden>
wrote:

>
>
> Hi everybody again,
>
> I still have 2 kernels for you to test here
> (amd64-k8 and k8-smp)
> and nobody did asnwer my request for information.
> Should I assume this problem is gone automatically?
>
> please test and let me know.
>
> Thanks
> Fabio
>
> --
> general: 2.6.12-10 regressions from 2.6.12-9
> https://launchpad.net/malone/bugs/26859
>

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

[David Taylor (me-davidandrewtaylor)]

Since changing PC's I have not seen the problem - so can't help as I no longer have the affected machine. It was a HP Vectra VEI8 if any one has one to test.

[Fabio Massimo Di Nitto (fabbione)]

ryan wrote:
> Hai Fabio,
>
> I don't know about the K8 troubles, however I do know
> about the K7 troubles with 2.6.12-10(k7) and with
> 2.6.12-10(386).

[SNIP]

I did ask specifically to test one kernel since the problem
is not the kernel. it's binutils that generates the code.

> I will however try any reworked 2.6.12-10 kernel to
> see if it is stable when it comes available for
> apt-get..
>

There are none available for i386. I did rebuild only
amd64 and there were tons of reports with amd64.

One of them could please test?

Fabio

--
I'm going to make him an offer he can't refuse.

[Fabio Massimo Di Nitto (fabbione)]

No info have been provided on the test packages. This makes impossible for me to track down the issue.

If somebody is willing to test, please reopen the bug and add the info requested.

Fabio