[SRU] Revert 'e1000e: add handshake with the Intel CSME to support S0ix'

Bug #1949353 reported by Bin Li
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OEM Priority Project
New
Critical
Unassigned
linux (Ubuntu)
Incomplete
Undecided
Unassigned

Bug Description

[SRU Justification]

https://patchwork.ozlabs.org/project/ubuntu-kernel/list/?series=258514

From the 5.13.0-1012-oem kernel we found NO Ethernet connecting after resume from suspend, on 5.13.0-1010-oem we don't met this issue, so this is a regression on I219 in lots of platforms.
And this blocked our OEM projects, could we revert the patches in short term? When the firmware or the related patches are ready, we merge it again.

Oct 25 22:45:44 p15s-ThinkPad-P15s-Gen-2i kernel: e1000e 0000:00:1f.6 enp0s31f6: Detected Hardware Unit Hang:
                                                    TDH <3>
                                                    TDT <7>
                                                    next_to_use <7>
                                                    next_to_clean <3>
                                                  buffer_info[next_to_clean]:
                                                    time_stamp <ffff5cf8>
                                                    next_to_watch <3>
                                                    jiffies <ffff6090>
                                                    next_to_watch.status <0>
                                                  MAC Status <40080283>
                                                  PHY Status <796d>
                                                  PHY 1000BASE-T Status <3800>
                                                  PHY Extended Status <3000>
                                                  PCI Status <10>

[Impact]

These doesn't improve power consumption on some e1000e platforms.

[Fix]

I tried to revert these 3 patches on 5.13.0-1019.23, it works fine.

commit f65d71bdcb52e8d82dd0a3e71ae4359c0972fe3b
Author: Sasha Neftin <email address hidden>
Date: Wed Aug 18 15:10:49 2021 +0800

    e1000e: Additional PHY power saving in S0ix

    BugLink: https://bugs.launchpad.net/bugs/1937252

    After transferring the MAC-PHY interface to the SMBus set the PHY
    to S0ix low power idle mode

commit c97933898708a37678ba1fc6f1f5704fbe92774c
Author: Sasha Neftin <email address hidden>
Date: Wed Aug 18 15:10:48 2021 +0800

    e1000e: Add polling mechanism to indicate CSME DPG exit

    BugLink: https://bugs.launchpad.net/bugs/1937252

    Per guidance from the CSME architecture team, it may take
    up to 1 second for unconfiguring dynamic power gating mode.
    Practically it can take more time. Wait up to 2.5 seconds to indicate
    dynamic power gating exit from the S0ix configuration. Detect
    scenarios that take more than 1 second but less than 2.5 seconds
    will emit warning message.

commit 66069cb4ebaf0eccfdec934ebfd92d3fdb136adc
Author: Sasha Neftin <email address hidden>
Date: Wed Aug 18 15:10:47 2021 +0800

    e1000e: Add handshake with the CSME to support S0ix

    BugLink: https://bugs.launchpad.net/bugs/1937252

    On the corporate system, the driver will ask from the CSME
    (manageability engine) to perform device settings are required
    to allow S0ix residency.
    This patch provides initial support.

[Test Case]
Do suspend and resume 10 times, it works fine.

for i in {1..10}; do
        echo $i
        rtcwake -m no -s 20 && echo mem > /sys/power/state
        if dmesg | grep -q "Detected Hardware Unit Hang"; then
                exit
        fi
        sleep 15
done

[Where problems could occur]

00:1f.6 Ethernet controller [0200]: Intel Corporation Ethernet Connection (13) I219-LM [8086:15fb] (rev 20)
CPU: 11th Gen Intel(R) Core(TM) i7-1185G7 @ 3.00GHz

Bin Li (binli)
tags: added: oem-priority originate-from-1948738 sutton
Bin Li (binli)
Changed in oem-priority:
importance: Undecided → Critical
Revision history for this message
Ubuntu Kernel Bot (ubuntu-kernel-bot) wrote : Missing required logs.

This bug is missing log files that will aid in diagnosing the problem. While running an Ubuntu kernel (not a mainline or third-party kernel) please enter the following command in a terminal window:

apport-collect 1949353

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

Changed in linux (Ubuntu):
status: New → Incomplete
tags: added: impish
Revision history for this message
Bin Li (binli) wrote :

https://lists.ubuntu.com/archives/kernel-team/2021-November/125883.html

BugLink: https://bugs.launchpad.net/bugs/1951861

[Impact]
e1000e devices stop working after s2idle resume.

[Fix]
Revert the offending patch series, which is to bring some minor power
saving.

[Test]
Vendor did extensive tests and didn't find any regression, and the power
consumption remains the same.

[Where problems could occur]
Maybe on some system the series can save some energy, so reverting will
cause some power consumption increase.

Kai-Heng Feng (3):
  Revert "e1000e: Additional PHY power saving in S0ix"
  Revert "e1000e: Add polling mechanism to indicate CSME DPG exit"
  Revert "e1000e: Add handshake with the CSME to support S0ix"

 drivers/net/ethernet/intel/e1000e/ich8lan.h | 3 -
 drivers/net/ethernet/intel/e1000e/netdev.c | 358 +++++++++-----------
 drivers/net/ethernet/intel/e1000e/regs.h | 1 -
 3 files changed, 154 insertions(+), 208 deletions(-)

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.