FWTS - Fails on Whitley/Coyote Pass with Barlow Pass DIMM

[Craig Salmond]

While running the Validate PMTT table test on Whitley/Coyote Pass with Barlow Pass DIMM hardware, the test gets stuck in an endless loop writing to the log file (results.log) until the machine runs out of disk space, then hangs.

This message it repeatedly sends to results.log:

FAILED [HIGH] PMTTBadSubtableType: Test 1, PMTT Socket must have subtable with Type 1, got 0x0000 instead

[Colin Ian King (colin-king)]

I suspect the header->length field in pmtt_controller_test is zero and we don't progress through the table. Can you supply a dump of the ACPI tables using:

sudo acpidump

[Alex Hung (alexhung)]

a subtable has incorrectly zero length such as iasl reported as attached and summarized below:

[034h 0052 1] Subtable Type : 00 [Socket]
[035h 0053 1] Reserved : 00
[036h 0054 2] Length : 0000
[038h 0056 2] Flags (decoded below) : 00FF
        ¦ ¦ ¦ Top-level Device : 1
        ¦ ¦ ¦ Physical Element : 1
        ¦ ¦ ¦ ¦Memory Type : 3
[03Ah 0058 2] Reserved : 0020
Invalid zero length subtable

A patch was sent for review (see http://patchwork<email address hidden>/)

[Craig Salmond (csalmond)]

That has fixed the problem on at least one of the machines we saw this issue on. Waiting for others to verify.

[Alex Hung (alexhung)]

The patch in #3 has been released with fwts 21.01.00

[Jeff Lane  (bladernr)]

Hi, Alex,

The latest update in testing this is:

Here is some update after aligned with Lenovo ACPI engineer, I would like to double confirm with you on the fwts result, if it does not the blocker issue for Ubuntu Certification test?

It is the fwts 21.01.00-0ubuntu1 installed on current SUT due to some network issue, I failed to update it to the latest unstable version released on 1/26..

For PMTT: the current ACPI table used is current UEFI 6.4 VS the ACPI packaged in current fwts 6.3
For rsdp_sbbr: seems the failure is designed for ARM server as the following from weblink:

Server Base Boot Requirements (SBBR) specification is

intended for SBSA- compliant 64-bit ARMv8 servers.

fadt_sbbr: Don’t know if it is as the same as above defined for ARM server.

[Jeff Lane  (bladernr)]

This zip has some output from FWTS showing the current failures seen by Lenovo and relates to my comment above.

[Alex Hung (alexhung)]

@Jeff,

SBBR spec is for aarch64 server, and therefore sbbr tests do not apply to other architectures.

I usually recommend tests for x64 as listed in https://github.com/alexhungce/script-fwts/blob/master/recommend_fwts.sh, i.e. RECOMMEND_TESTS & HWE_TESTS

[Alex Hung (alexhung)]

Will you be able to upload acpidump (sudo acpidump > acpi.log) and I can look into PMTT table and tests.

[Jeff Lane  (bladernr)]

I'll ask, but won't expect a reply for a couple weeks since they should all be on holiday now. Thanks for getting back, I'll try to get them to reply directly.

[Craig Salmond (csalmond)]

It looks like the patch fixed the issue. Thank you!

[Craig Salmond (csalmond)]

I finally got my partner to respond with results from the latest build 21.03 and an updated acpidump. It is still seems to be getting the same errors. I will attach the two logs.

[Alex Hung (alexhung)]

@csalmond,

#13 is a different bug but it is addressed by an update to PMTT's revision 2 (patch is available @ http://patchwork<email address hidden>/).

I run fwts compiled from the latest source against #14 and the results.log is attached for your references.

The new version of fwts will be released this week.