Re: [Xen-devel] [xen-unstable bisection] complete test-amd64-i386-xl

[Keir Fraser <keir.xen@xxxxxxxxx>]

On 12/08/2011 14:03, "Liu, Jinsong" <jinsong.liu@xxxxxxxxx> wrote:

>> If it's really this changeset it's a bit weird. It would have to be
>> due to the more permissive table validation causing us to enable ACPI
>> ERST support based on a bad BIOS table, or a table which for some
>> reason we're not supporting properly, or a class of machines (e.g.,
>> AMD) for which our ERST logic is not currently implemented properly.
>> 
>> Does reverting just the change to erst_check_table() fix the
>> regression on the affected test boxes? What about the similar-looking
>> boot failure that you see, Jeremy?
>> 
>>  -- Keir
>> 
> 
> It looks strange to me. Native linux also update it to work well at different
> bios platform.
> We have tested it at our 'old' and a 'new' platform, it works well with the
> patch.
> I'm not sure why it cannot work at the machine you test.

It's obviously a latent bug in our handling of that table, which is
uncovered only when that table-validation check is relaxed to permit parsing
of the table on a much broader range of machines. Perhaps we can work with
you to run some out-of-tree patches to gather useful tracing information on
failing machines?

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel