Bad items from Powertop on oem kernel
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OEM Priority Project |
Triaged
|
High
|
Bin Li | ||
| linux-oem (Ubuntu) |
New
|
Undecided
|
koba | ||
Bug Description
On Jaguar2, we found some items are bad from powertop, could we set them 'Good' by default from oem kernel, so that we could save more power consumption.
System info
====
"ThinkPad X1 Yoga Gen 7", "BIOS": "N3AET52W (1.17 ) 02/21/2022",
CPU": "12th Gen Intel(R) Core(TM) i5-1235U", VGA: [8086:46a8],
Kernel: 5.14.0-1024-oem
Power Consumption
===
The idle-screen-
Default value and the value after set all ‘Good’ in powertop
ShortIdle: 5.5W, 5W
LongIdle: 1.8W, 2W
S2Idle: 0.6W, 0.6W
Bad items
===
* VM writeback timeout
(echo ‘1500’ > /proc/sys/
* Should we turn off the NMI watchdog for the OEM kernel? It prompt“NMI watchdog should be turned off” (/proc/
* Runtime PM for I2C Adapter i2c-[2-10] (i915 gmbus tc[1-6]/dp[a-c])?
By default these values are ‘on’, should we set it to ‘auto’? (/sys/bus/
* Runtime PM for PCI Device Intel Corporation Device [4601,46a8,
Host bridge,VGA controller,Signal processing controller,System peripheral,Serial bus controller,Network Controller,ISA bridge,USB controller,RAM memory,Serial controller.
(/sys/bus/
* Runtime PM for PCI Device MEDIATEK Corp. Device 4d75
* Runtime PM for PCI Device Samsung Electronics Co Ltd Device a80a
* Autosuspend for USB device USB Bridge [MCHP]?
(/sys/bus/
| Bin Li (binli) wrote | #1 |
| Changed in oem-priority: | |
| assignee: | nobody → Bin Li (binli) |
| importance: | Undecided → High |
| tags: | added: hwe-powerconsumption |
| Changed in linux-oem (Ubuntu): | |
| assignee: | nobody → AaronMa (mapengyu) |
| assignee: | AaronMa (mapengyu) → koba (kobako) |
| tags: | added: oem-priority |
| AceLan Kao (acelankao) wrote | #2 |
| Bin Li (binli) wrote | #3 |
| Changed in oem-priority: | |
| status: | New → Triaged |
For NMI watchdog, I found some explanation from google.
What could happen if I disable nmi_watchdog permanently?
Imagine your system locks up. There's 2 possibilities now:
1) You have a hardware NMI button (some servers do). Push it, then the kernel (if properly configured) dumps a trace to console and reboots.
2) Your kernel reaches a halting state that can't be interrupted by any other method. In this case, the watchdog can reboot the machine.