firmware-testing-team team mailing list archive

[Colin Ian King <colin.king@xxxxxxxxxxxxx>]

Jeff,

I've updated the bug.  It looks like the _SRS control can be passed in 
two types of buffer (as created by the associated _CRS).  However, the 
_SRS assumes it has the the largest buffer when it creates the DRQL and 
DMAC byte fields, and hence it falls off the end of the smaller buffer.

I am not sure if the _SRS fails to execute the SIRQ (0x02, IRQL) when it 
is given the smaller buffer or not; I'd have to produce a debug kernel 
to trace the AML op code execution to figure this out.  As it stands, if 
this fails to execute then we may have some PCI misconfiguration going 
on.  If it does execute, we are OK, but we get this kernel error being 
reported on each S3 cycle.

As it stands, I think it would be prudent to see if the firmware can be 
fixed.  Realistically we can't quirk around this bug.

Colin

On 03/12/12 22:13, Jeff Lane wrote:
> Could someone take a look at this log from FWTS:
> https://launchpadlibrarian.net/122454479/results.log
>
> That is part of this bug:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1076979
>
> This was run on a system and while the system did indeed suspend and
> resume 30 times, the test overall failed because of a bunch of the
> following errors:
>
> FAILED [CRITICAL] KlogACPIErrorFieldexceedsBuffer: Test 1, CRITICAL
> Kernel message: [77191.628764]
> ACPI Error: Field [DRQL] at 144 exceeds Buffer [NULL] size 104 (bits)
> (20110623/dsopcode-236)
>
> ADVICE: The field exceeds the allocated buffer size. This can lead to
> unexpected results when
> fetching data outside this region.
>
> FAILED [HIGH] KlogACPIErrorMethodExecutionParse: Test 1, HIGH Kernel
> message: [77191.628769] ACPI
> Error: Method parse/execution failed [\_SB_.PCI0.SBRG.LPTE._SRS] (Node
> ffff88002cfc0488),
> AE_AML_BUFFER_LIMIT (20110623/psparse-536)
>
> ADVICE: Method failed: ResourceSourceIndex is present but ResourceSource
> is not.
>
> So ultimately, what I'm interested in is if this is a misconfiguration
> in Klog or something, or if this will legitimately impact the stability
> and usability of a system, thus failing certification.
>
> Thanks
> Jeff