Syslinux - Aug 2009 - Need help tracking the spread of a BIOS bug

Help everyone... I need to track the spread of a BIOS bug.  One symptom
of this particular BIOS bug is that with pxelinux.0 from Syslinux
3.83-pre9, "localboot 0" works, but *if pxelinux.0 chainloads itself
first*, then "localboot 0" *does not work*.

I have reasons to believe this is a fairly widespread bug among newer
hardware (made in the last two years or so), and it is absolutely
critical that I get an idea for which systems are and are not affected.

If you can try the above experiment on as many systems as possible, and
email me the output of "dmidecode" and "lspci -vn" together
with an
"affected"/"not affected" that would be hugely appreciated.

This is just a small symptom of a much more serious bug which is just
waiting to bite, but I think I have a workaround for it -- but I need
your help in understanding when the workaround can be deployed.

Thanks,

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

On 26.08.2009, 06:25, H. Peter Anvin wrote:
> Help everyone... I need to track the spread of a BIOS bug.  One symptom
> of this particular BIOS bug is that with pxelinux.0 from Syslinux
> 3.83-pre9, "localboot 0" works, but *if pxelinux.0 chainloads
itself
> first*, then "localboot 0" *does not work*.
>
> I have reasons to believe this is a fairly widespread bug among newer
> hardware (made in the last two years or so), and it is absolutely critical
> that I get an idea for which systems are and are not affected.
>
> If you can try the above experiment on as many systems as possible, and
> email me the output of "dmidecode" and "lspci -vn"
together with an
> "affected"/"not affected" that would be hugely
appreciated.
>
>
> This is just a small symptom of a much more serious bug which is just
> waiting to bite, but I think I have a workaround for it -- but I need your
> help in understanding when the workaround can be deployed.
>
> Thanks,
Hi,

I'm interesting in testing (over 12 different motherboards, processors,
BIOSes), but could you please explain "pxelinux.0 chanloads itself"?
Maybe
a simple example?

Jack

Here are a few BIOS bug tests, attached.  Having a keyboard, mouse and
monitor switch would've been nice.  Heheh.  - Shao
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hpa_localboot_tests.zip
Type: application/x-zip-compressed
Size: 31684 bytes
Desc: hpa_localboot_tests.zip
URL:
<http://www.zytor.com/pipermail/syslinux/attachments/20090828/042e8933/attachment.bin>

On 26.08.2009, 17:54, H. Peter Anvin wrote:
> On 08/26/2009 02:40 AM, Jacek Kalinski wrote:
>
>>
>> Hi,
>>
>>
>> I'm interesting in testing (over 12 different motherboards,
processors,
>>  BIOSes), but could you please explain "pxelinux.0 chanloads
itself"?
>> Maybe
>> a simple example?
>>
>
> Absolutely.  Consider the following configuration file:
>
>
> prompt 1
>
> label localboot localboot 0
>
> label restart pxe pxelinux.0
>
>
> Entering "restart" should simply cause pxelinux.0 to reload
itself and
> display the boot prompt again.
>
> In the case of the BIOS bug in question, when used with 3.83-pre9,
> entering "localboot" will work right (boot from disk), but
entering
> "restart" followed by "localboot" will cause it to
fail.
>
>
> NOTE: make sure you're testing with 3.83-pre9.  3.82 has different
> behavior!
Here are the tests (with pxelinux.0 from 3.83-pre9).
In my network all computers works with 'localboot' as well 'restart
and
then localboot' (the same behavior).
I tested also "Shuttle MK35V" motherboard with BIOS 003 revision, but
I
was no able to gather dmidecode and lspci results (this workstation is not
able to boot from USB) ;)

I have found only 1 computer, which is not working with localboot at all.
I placed it in "non_working" directory with info.txt

Jack
-------------- next part --------------
A non-text attachment was scrubbed...
Name: localboot_tests.zip
Type: application/zip
Size: 71419 bytes
Desc: not available
URL:
<http://www.zytor.com/pipermail/syslinux/attachments/20090829/b85c4085/attachment.zip>