Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1511225 > unrolled thread

Re: [PATCH] x86/AMD: Apply erratum 688 on machines without a BIOS fix

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH] x86/AMD: Apply erratum 688 on machines without a BIOS fix Borislav Petkov <bp@alien8.de> - 2016-10-28 18:30 +0200
    Re: [PATCH] x86/AMD: Apply erratum 688 on machines without a BIOS fix sonofagun@openmailbox.org - 2016-10-31 23:00 +0100
      Re: [PATCH] x86/AMD: Apply erratum 688 on machines without a BIOS fix Borislav Petkov <bp@alien8.de> - 2016-11-01 00:00 +0100

#1511225 — Re: [PATCH] x86/AMD: Apply erratum 688 on machines without a BIOS fix

On Mon, Oct 24, 2016 at 07:14:50PM +0200, Borislav Petkov wrote:
> > Yes, using Ubuntu 16.04 will just crash everything! For example I had
> > crashes with the software updater program. Moreover firefox would become
> > unresponsive even with one tab.
> 
> Ok, lemme install 16.04 on that box and see if I can reproduce.

Ok, Ubuntu 16.04.1 is running on the box now, no issues so far. Any
special workload I should run?

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.

[toc] | [next] | [standalone]

#1512934

> Ok, Ubuntu 16.04.1 is running on the box now, no issues so far. Any
> special workload I should run?
No, will simply crash without running something special! If you get no 
issues then that is bad news. I get frequent and repeatable crashes. I 
forgot to mention that all those crashes occur at program launch. If the 
program launches, it does not crash. Unfortunately with Ubuntu it is not 
possible to keep oopses from the error report program. On the laptop  we 
have a Debian installation, I will switch to it and get crash 
information there so that we figure out why it behaves that way. Besides 
that, I have some more tests to do but I am running out of ideas so I 
might not be able to help you more on it as my laptop appears to be 
really broken! Poor performance might be the result here and not the 
cause of my issues. The 688 fix just makes the system respond better. As 
far as I am concerned, I do not wish any module options for turning on 
the fix. I would prefer to use DMI maching for this specific machine and 
thus having the fix automatically.

All subsystems of the laptop appear to be good(RAM has been tested and 
the HDD has passed our test). There is a sound issue on the HDA but such 
issues are common on most laptops and will be dealt soon.

> Is it a desktop system or a laptop?
I got no reply on this question so I suppose you have a desktop. Since 
my APU is installed on a laptop, I expect different behaviour.
There were many intel based laptops that had fewer lanes on the DMI 
interconnect bridging northbridge and southbridge. Maybe my laptop has 
the A-Link in reduced mode. That could explain my performance issues. It 
must be easy to verify that as all documents are available for its 
northbridge and southbridge. I will check the settings of both chips 
thoroughly.

My brother has spotted a C70 board. Normally I would not buy it but it 
has a better but slower(without CPB) F14 CPU and I am curious if it will 
behave better like your board does. I might order it if it is still 
available.

[toc] | [prev] | [next] | [standalone]

#1512980

On Mon, Oct 31, 2016 at 11:54:44PM +0200, sonofagun@openmailbox.org wrote:
> No, will simply crash without running something special! If you get no
> issues then that is bad news. I get frequent and repeatable crashes. I
> forgot to mention that all those crashes occur at program launch. If the
> program launches, it does not crash. Unfortunately with Ubuntu it is not
> possible to keep oopses from the error report program.

Can't you enable crash dumps?

$ ulimit -c unlimited

That should reenable core dumping which can then be examined with gdb.
You could put the program executable and the core somewhere on the web
so that I can take a look...

In any case, I'd like to see what those crashes look like. Can you send
dmesg, does it even say something in dmesg related to those crashes?

> On the laptop we have a Debian installation, I will switch to it and
> get crash information there so that we figure out why it behaves that
> way. Besides that, I have some more tests to do but I am running out
> of ideas so I might not be able to help you more on it as my laptop
> appears to be really broken!

Maybe a hw issue? RAM broken, cooling failing...

> Poor performance might be the result here and not the cause of my
> issues. The 688 fix just makes the system respond better. As far as I
> am concerned, I do not wish any module options for turning on the fix.
> I would prefer to use DMI maching for this specific machine and thus
> having the fix automatically.

The problem with DMI strings is that then we have to always go and
update them. And that's always a PITA.

I'm just trying to avoid an unnecessary performance penalty to users
with the erratum workaround where the erratum itself didn't even occur
in the first place. Like in my case, for example. I've never had any
issues with that machine for the time I've been using it.

> All subsystems of the laptop appear to be good(RAM has been tested and
> the HDD has passed our test). There is a sound issue on the HDA but
> such issues are common on most laptops and will be dealt soon.

If by "tested" you mean, you ran memtest on it, memtest is notorious for
not always catching faulty DIMMs.

> > Is it a desktop system or a laptop?
> I got no reply on this question so I suppose you have a desktop.

Oh sorry, I must've missed that question. No, the Ontario I have is a
laptop, something like this one:

https://support.lenovo.com/de/en/documents/pd015763

x121e with an AMD CPU, I *think* it is E-350.

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web