Bug 93551 - Lenovo G570: no reaction to brighness keys after boot, works OK after suspend/resume
Summary: Lenovo G570: no reaction to brighness keys after boot, works OK after suspend...
Status: CLOSED INVALID
Alias: None
Product: ACPI
Classification: Unclassified
Component: EC (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: acpi_ec
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-02-20 10:32 UTC by Eugene A. Shatokhin
Modified: 2015-04-04 19:34 UTC (History)
2 users (show)

See Also:
Kernel Version: 3.19+ in linux-pm.git (linux-next branch, ), also - 3.14.x - 3.19 mainline
Subsystem:
Regression: No
Bisected commit-id:


Attachments
acpidump (247.62 KB, text/plain)
2015-02-20 10:32 UTC, Eugene A. Shatokhin
Details
dmidecode (15.32 KB, text/plain)
2015-02-20 10:34 UTC, Eugene A. Shatokhin
Details
dmesg after boot (303.00 KB, text/plain)
2015-02-20 10:36 UTC, Eugene A. Shatokhin
Details
dmesg after pressing brightness keys (303.08 KB, text/plain)
2015-02-20 10:39 UTC, Eugene A. Shatokhin
Details
dmesg after unpligging AC adapter and plugging it in again (328.42 KB, text/plain)
2015-02-20 10:42 UTC, Eugene A. Shatokhin
Details
dmesg after STR (534.69 KB, text/plain)
2015-02-20 10:44 UTC, Eugene A. Shatokhin
Details
dmesg after pressing brightness keys the second time (after STR) (571.96 KB, text/plain)
2015-02-20 10:47 UTC, Eugene A. Shatokhin
Details
dmesg after boot (the second boot) (706.22 KB, text/plain)
2015-02-20 10:56 UTC, Eugene A. Shatokhin
Details
/proc/interrupts before pressing brightness keys (2.18 KB, text/plain)
2015-02-20 10:57 UTC, Eugene A. Shatokhin
Details
/proc/interrupts after pressing brightness keys (2.18 KB, text/plain)
2015-02-20 10:58 UTC, Eugene A. Shatokhin
Details
dmesg before unplugging AC adapter (the second boot) (732.25 KB, text/plain)
2015-02-20 10:59 UTC, Eugene A. Shatokhin
Details
dmesg after unplugging AC adapter and plugging it in again (the second boot) (750.83 KB, text/plain)
2015-02-20 11:03 UTC, Eugene A. Shatokhin
Details

Description Eugene A. Shatokhin 2015-02-20 10:32:40 UTC
Created attachment 167641 [details]
acpidump

So far, I have observed this on two Lenovo G570 laptops we have in our test lab.

After boot, the system does not react to pressing brightness up/down keys (Fn-up/down). No visible reaction to AC plug and unplug either. After suspend-to-RAM and resume, everything starts working normally - until reboot. Then the problem shows up again.

The problem happens often, athough sometimes disappears for a long time.

OS: ROSA Linux R5 (KDE), x86-64

Kernel: the problem shows up at least with the mainline kernels 3.14.x, 3.15.x, 3.18.x, 3.19, as well as with the kernel I built from linux-next branch of linux-pm tree (HEAD: 52cba87a).

At first glance, the symptoms are similar to bug 82451, although the problem shows up in all kernels I have tested, so bisect is not an option.

Hardware: http://hw.rosalinux.ru/index.php?probe=c790ae5107&kind=debug

The output of acpidump is attached. More logs and data follow.
Comment 1 Eugene A. Shatokhin 2015-02-20 10:34:43 UTC
Created attachment 167651 [details]
dmidecode
Comment 2 Eugene A. Shatokhin 2015-02-20 10:36:56 UTC
Created attachment 167661 [details]
dmesg after boot

dmesg after boot. EC DEBUG messages are enabled.
Comment 3 Eugene A. Shatokhin 2015-02-20 10:39:09 UTC
Created attachment 167671 [details]
dmesg after pressing brightness keys
Comment 4 Eugene A. Shatokhin 2015-02-20 10:42:05 UTC
Created attachment 167681 [details]
dmesg after unpligging AC adapter and plugging it in again

As far as I can see, there are some new messages in dmesg this time, unlike when I pressed the brightness keys. Still, no visible reaction in the battery monitor, brightness level, etc.
Comment 5 Eugene A. Shatokhin 2015-02-20 10:44:13 UTC
Created attachment 167691 [details]
dmesg after STR

Then I put the system to sleep, then wake it up. Here is the output of dmesg after that.
Comment 6 Eugene A. Shatokhin 2015-02-20 10:47:34 UTC
Created attachment 167701 [details]
dmesg after pressing brightness keys the second time (after STR)

If I press the brightness up/down keys after the system resumed from suspend-to-RAM, there are some new messages in the log. The brightness changes OK as well this time - until I reboot the system, then the problem appears again.
Comment 7 Eugene A. Shatokhin 2015-02-20 10:52:31 UTC
Then I started the system again, this time - with acpi.debug_layer=0x6 acpi.debug_level=0x88000004. See second-boot-* logs below.
Comment 8 Eugene A. Shatokhin 2015-02-20 10:56:45 UTC
Created attachment 167711 [details]
dmesg after boot (the second boot)

No new messages in dmesg after I pressed brightness keys then. That log is the same, so I am attaching the log after boot only.
Comment 9 Eugene A. Shatokhin 2015-02-20 10:57:30 UTC
Created attachment 167721 [details]
/proc/interrupts before pressing brightness keys
Comment 10 Eugene A. Shatokhin 2015-02-20 10:58:09 UTC
Created attachment 167731 [details]
/proc/interrupts after pressing brightness keys
Comment 11 Eugene A. Shatokhin 2015-02-20 10:59:51 UTC
Created attachment 167741 [details]
dmesg before unplugging AC adapter (the second boot)
Comment 12 Eugene A. Shatokhin 2015-02-20 11:03:41 UTC
Created attachment 167751 [details]
dmesg after unplugging AC adapter and plugging it in again (the second boot)

As before, I see some new messages in dmesg when I unplug the AC adapter and plug it again. However, there is no visible indication of these events in the system.

Meanwhile, the system reacts to AC plug/unplug immediately after suspend/resume but not after boot.
Comment 13 Eugene A. Shatokhin 2015-03-03 16:04:04 UTC
Strange enough, it turned out that the root of the problem was probably not ACPI but rather vboxguest kernel module from VirtualBox installed on that system for some reason.

After I have removed vboxguest, I cannot reproduce the problem anymore. Same behaviour on both Lenovo G570 laptops.

That module tried to load at boot, did not detect VirtualBox and failed. And probably messed something up in the process.

I am closing the bug for now. If it is reproduced without VirtualBox, perhaps, the bug will be reopened.

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