Bug 10053 - System dead after resuming from S3, but not when docked - Dell Precision M20/ Latitude D610
Summary: System dead after resuming from S3, but not when docked - Dell Precision M20/...
Alias: None
Product: ACPI
Classification: Unclassified
Component: Power-Sleep-Wake (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Zhang Rui
Depends on:
Blocks: 7216
  Show dependency tree
Reported: 2008-02-20 10:14 UTC by Dawid Wr
Modified: 2008-06-11 02:11 UTC (History)
3 users (show)

See Also:
Kernel Version: 2.6.24
Regression: Yes
Bisected commit-id:

dmesg output taken after booting up (29.74 KB, application/octet-stream)
2008-02-21 06:48 UTC, Dawid Wr
acpidump output (80.20 KB, application/octet-stream)
2008-02-21 06:49 UTC, Dawid Wr
dmesg output (92.02 KB, application/octet-stream)
2008-02-27 15:52 UTC, Dawid Wr
Acpidump (104.77 KB, application/octet-stream)
2008-04-10 02:38 UTC, Mark Baas

Description Dawid Wr 2008-02-20 10:14:50 UTC
Latest working kernel version: 2.6.23
Earliest failing kernel version: 2.6.24
Distribution: Archlinux
Hardware Environment: Dell Precision M20 (almost the same as Dell Latitude D610), Pentium M 2.13, Sonoma i915, ati radeon x300 (using radeon driver)
Software Environment: 
Problem Description: System suspends properly after echo "mem" > /sys/power/state. After pressing the power button again, hardware wakes up but the system is not responding and neither the kernel magic keys work nor the LCD backlight. Dead. However if I suspend while being docked (dock module loaded) and then resume (still docked) it surprisingly will resume just fine. Tried many times and it just works that way. Please note that suspending when undocked and resuming when undocked and the vice versa will not work. 

2.6.23 worked without any problems, in all the situations described above.

Steps to reproduce:
As descirbed.
Comment 1 Dawid Wr 2008-02-20 10:16:40 UTC
*Please note that suspending when
docked and resuming when undocked and the vice versa will not work. 
Comment 2 Dawid Wr 2008-02-20 10:21:02 UTC
I also forgot to mention that nothing new is written to syslog after resuming, there is completely no trace of system activity.
Comment 3 Len Brown 2008-02-20 19:24:33 UTC
please attach that syslog
and the output from acpidump
Comment 4 ykzhao 2008-02-20 21:34:27 UTC
Hi, Dawid
    Will you please set "CONFIG_PM_TRACE , CONFIG_PM_TRACE_RTC "in kernel configuration and do the following tests? (The system should be booted without dock).
    a. boot the system with the option of "initcall_debug acpi_sleep=s3_bios"
    a. echo 1 > /sys/power/pm_trace
    b. echo mem > /sys/power/state
    c. press the power button and see whether the system can be resumed. If the system can't be resumed, please restart the system with the option of "initcall_debug" and attach the output of dmesg.
Comment 5 Dawid Wr 2008-02-20 23:57:05 UTC
Just to let you know: I will do the test next week.
Comment 6 Dawid Wr 2008-02-21 06:44:35 UTC
OK, I did the testing today. acpi_sleep=s3_bios causes immediate reboot here after pressing the power button to resume the system. It always used to be like that AFIR (did some testing some time ago). So I tried without this option and got some results. However dmesg output lacks initcall messages that I saw on the screen while booting, I am not sure if that's the way it's supposed to be.
Comment 7 Dawid Wr 2008-02-21 06:48:17 UTC
Created attachment 14931 [details]
dmesg output taken after booting up

I did:
- run with initcall_debug option
- echo 1 > /sys/power/pm_trace
- echo mem > /sys/power/state
- pressed power button, system hanged as described earlier
- rebooted with initcall_debug option
- dmesg >> dmesg_without_s3_bios
Comment 8 Dawid Wr 2008-02-21 06:49:10 UTC
Created attachment 14932 [details]
acpidump output
Comment 9 ykzhao 2008-02-24 17:11:35 UTC
Hi, Dawid
   Thanks for the info. But the output in comment #7 is not completed.
   Will you please upload the full dmesg output ? 
Comment 10 Dawid Wr 2008-02-26 04:58:26 UTC
Did you read my Comment #6? For some reason there are no initcall messages included in dmesg and I don't know why is that.
Comment 11 Dawid Wr 2008-02-26 04:58:39 UTC
Did you read my Comment #6? For some reason there are no initcall messages included in dmesg and I don't know why is that.
Comment 12 ykzhao 2008-02-26 22:31:07 UTC
HI, Dawid
    Maybe the reason about no initcall messages is that the kernel log buffer is very small.
    Will you please increase the CONFIG_LOG_BUF_SHIFT in kernel configuration and attach the full dmesg? (For example: 18/19).
    If boot option of "acpi_sleep=s3bios" is added, the system will immediately be rebooted after the power button is pressed. Please not add this option. 
    Will you please do the test again as in comment #7?
Comment 13 Dawid Wr 2008-02-27 15:52:43 UTC
Created attachment 15037 [details]
dmesg output

OK, that helped. Here's the proper dmesg output. I obviously didn't add acpi_sleep=s3bios option.
Comment 14 Dawid Wr 2008-03-10 13:04:37 UTC
Hi. I tried s2ram yesterday and... resuming works fine now. This is quite surprising, I didn't expect that. According to whitelist entry for my laptop model, all the s2ram does is saving and restoring vbe state.
Comment 15 Len Brown 2008-03-12 19:32:08 UTC
Is there still a failure here to debug, or is everything working?
Comment 16 Dawid Wr 2008-03-13 04:16:10 UTC
Len, I don't understand your question? Debugging is fine already, I attached the proper output. Not everything is working with S3 however, because with 2.6.23 resuming worked fine without any quirks and now I need to use s2ram, so there is some regression in this area definetely.
Comment 17 Zhang Rui 2008-03-13 20:05:14 UTC
Hmm, I think the question is:
Does the laptop still hang when resume? Or it's just a black screen issue?
Comment 18 Dawid Wr 2008-03-13 22:17:07 UTC
No, with s2ram tool everything works fine now.
Comment 19 Mark Baas 2008-04-10 02:36:40 UTC
In my case s2ram is not a solution. I have got the problem since i upgraded to ubuntu hardy which is using kernel 2.6.24. On the same system i have tried with 2.6.22 and it works, so i must be the kernel.
Does these options mean i have to rebuild the kernel: CONFIG_PM_TRACE , CONFIG_PM_TRACE_RTC
Comment 20 Mark Baas 2008-04-10 02:38:25 UTC
Created attachment 15711 [details]
Comment 21 Zhang Rui 2008-04-13 18:58:13 UTC
Mark, forst please make sure it's the same problem  as Dawid described in this bug report, i.e. the same symptom on the same/similar hardware.
If no, please open another bug report. :)
Comment 22 Shaohua 2008-06-03 20:38:46 UTC
Sounds like a black screen issue. Rui, should we close this one or move it to graphic catalog?
Comment 23 Zhang Rui 2008-06-11 02:11:38 UTC
This seems to be a black screen issue.
And this should be fixed in the latest kernel release, say 2.6.25, if it uses an Intel integrated graphics card.
Please give it a try, please re-open it if you still have any questions.

Note You need to log in before you can comment on or make changes to this bug.