[PATCH v4 0/5] Provide better MADT subtable sanity checks

Rafael J. Wysocki rafael at kernel.org
Mon Sep 28 14:13:01 PDT 2015


On Mon, Sep 28, 2015 at 10:17 PM, Al Stone <ahs3 at redhat.com> wrote:
> On 09/25/2015 05:29 PM, Rafael J. Wysocki wrote:
>> On Wednesday, September 16, 2015 05:26:40 PM Al Stone wrote:

[cut]

>> In particular, I'm not sure if we really need to return
>> -EINVAL from acpi_parse_entries_array() when we find a bad MADT entry or it
>> will be sufficient to simply go to the next entry in that case?
>>
>> Thanks,
>> Rafael
>
> I see there being two options: (1) return -EINVAL and indicate that the tables
> are incorrect, or (2) print a warning (or something more aggressive?), go to
> the next entry, and hope for the best with the remainder of the MADT subtables.
> The former is consistent with past behavior, I think, and the latter seems to
> me a bit of a gamble.  So, my vote is for (1), the current method; what are you
> thinking these days?

I would be for preserving the past behavior.

I'm a bit concerned that the new checks may trigger on systems where
the old ones didn't, but that is a separete problem.

Thanks,
Rafael



More information about the linux-arm-kernel mailing list