DebuggingKernelSuspend

Revision 35 as of 2012-04-14 01:32:35

Clear message

Debugging Central

This page is part of the debugging series — pages with debugging details for a variety of Ubuntu packages.

Introduction

This page describes how to debug Suspend to RAM/Resume problems on your computer. Do not confuse this with Suspend to disk (also known as hibernate). UnderstandingSuspend may have useful background information on where problems can occur.

Suspend and Resume use facilities within your BIOS called ACPI, or Advanced Configuration and Power Interface. Linux provides an ACPI subsystem that manages the suspend and resume process. Usually problems occur when resuming, and normally the culprit is a device driver that does not recover from a powered down state. If your computer successfully performs a suspend, then it is quite likely any resume problems are due to another device driver and not the ACPI subsystem.

The debugging procedure described below requires a linux kernel with the capability of "/sys/power/pm_trace". You can ensure that pm_trace'ing is possible by looking in the directory /sys/power/ (from a terminal type ls /sys/power/) to see if a file called pm_trace exists.

"resume-trace" debugging procedure for finding buggy drivers

Resume problems are difficult to debug. The approach used here needs to make notes on progress during resume and be able to recover them after a manual reboot. But there is no non-volatile storage available at the time resume is bringing up your computer. The only hardware on a PC motherboard that retains information across power cycles is the real time clock (RTC), so that is what is used. For those that want to know the details, read Documentation/power/s2ram.txt in your kernel sources. The implementation of suspend/resume debug trace is in drivers/base/power/trace.c.

Caveat Emptor: Using the following debug suggestions will radically change the values in your RTC chip, so much so that your file system will think it has been eons since the last fsck. You can avoid a long fsck delay by using 'tune2fs'. For example, 'tune2fs -i 0 /dev/sda1' disables fsck on boot. But first you'll want to use 'tune2fs -l <partition>' to find your current settings - look at the "Check interval" setting.

From a fresh install, the default time interval is 15552000, which is 6 months.

Remember that when you are all done, you'll want to set your RTC clock back. Note that NTP by itself is not sufficient to recover the correct time of day, since by default NTP is configured to do nothing if the time is off by more than 1000 seconds. You must first use the 'ntpdate' or 'date' command to get the clock close.

OK. With that understood.... In order to simulate your suspend/resume process, enter the following commands:

sudo sh -c "sync; echo 1 > /sys/power/pm_trace; pm-suspend"

At this point your computer should enter the suspend state within a few seconds. Usually the power LED will slowly flash when in the suspended state. When that has happened, initiate the resume process by pressing the power button. Observe closely if the disk light comes on briefly. This indicates that resume has begun. If resume fails to complete, then press the power button until the computer turns off. Power on your computer making sure that it loads the same kernel that exhibited the resume problem. You have about 3 minutes to start this boot process before the information saved in the RTC gets corrupted.

Start a console and enter:

dmesg > dmesg.txt

You can edit this file and find lines similar to these:

[   11.323206]   Magic number: 0:798:264
[   11.323257]   hash matches drivers/base/power/resume.c:46

There may well be another 'hash matches' line beyond that. If so, then you are in luck because the last one is the likely culprit. For example:

hash matches device i2c-9191

The only way to prove this is to remove the module prior to initiating suspend. Repeat as needed...

If you get a device number rather than name, lspci and /sys/devices/pci* are your friends.

Further hints

Detailed analysis of ACPI kernel code for debugging a suspend problem


Also see:


CategoryBugSquad CategoryDebugging