[solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

mr.MikyMaus · 2015-02-11 13:39:09

Hi,

EDIT: Workaround found - see my post #10

EDIT2: Solution for me was updating the BIOS - see my post #13

I got Lenovo Ideapad Flex, a Pentium N3520-based netbook with integrated Valley View graphics . As soon as KMS modules are loaded, the laptop display goes down - as in shut down, not just blank / backlight off. This is the same for late KMS start as well as early start.

The hardware in question is

0:02.0 VGA compatible controller: Intel Corporation Atom Processor Z36xxx/Z37xxx Series Graphics & Display (rev 0c) (prog-if 00 [VGA controller])
	Subsystem: Lenovo Device 3985
	Flags: bus master, fast devsel, latency 0, IRQ 104
	Memory at d0000000 (32-bit, non-prefetchable) [size=4M]
	Memory at c0000000 (32-bit, prefetchable) [size=256M]
	I/O ports at 1050 [size=8]
	Expansion ROM at <unassigned> [disabled]
	Capabilities: [d0] Power Management version 2
	Capabilities: [90] MSI: Enable+ Count=1/1 Maskable- 64bit-
	Capabilities: [b0] Vendor Specific Information: Len=07 <?>
	Kernel driver in use: i915

Using various kernel boot parameters I wasn't able to fix this. Notable tests:

- nomodeset, works, but no drm and X, obviously
- i915.modeset=0, same as above
- acpi_backlight=(various), backlight interface in /sys/class changed but display was still off
- acpi_os variations, no effect
- i915.(various), no effect
- video=(various):d, no effect except actually disabling other displays
- a few others I can't remember right now, will add

Physically there is only one HDMI port on the laptop (and likely an eDP port for the integrated display). However, there are five displays available according to drm (got these via ssh):

[root@gibon ~]# ls /sys/class/drm 
card0
card0-DP-1
card0-DP-2
card0-HDMI-A-1
card0-HDMI-A-2
card0-VGA-1
controlD64

Xrandr says:

[root@gibon ~]# DISPLAY=:0 xrandr
Screen 0: minimum 8 x 8, current 1366 x 768, maximum 32767 x 32767
DP1 disconnected (normal left inverted right x axis y axis)
DP2 connected 1366x768+0+0 (normal left inverted right x axis y axis) 222mm x 125mm
   1366x768      60.25*+
HDMI1 disconnected (normal left inverted right x axis y axis)
HDMI2 disconnected (normal left inverted right x axis y axis)
VGA1 disconnected (normal left inverted right x axis y axis)
VIRTUAL1 disconnected (normal left inverted right x axis y axis)

This is funny because in LMDE (Linux Mint) with an older 3.11-series debian kernel there are only four displays with slightly different names and this configuration works well enough.

Screen 0: minimum 320 x 200, current 1366 x 768, maximum 8192 x 8192
eDP1 connected 1366x768+0+0 (normal left inverted right x axis y axis) 222mm x 125mm
   1366x768       60.2*+
   1360x768       59.8     60.0  
   1024x768       60.0  
   800x600        60.3     56.2  
   640x480        59.9  
VGA1 disconnected (normal left inverted right x axis y axis)
HDMI1 disconnected (normal left inverted right x axis y axis)
DP1 disconnected (normal left inverted right x axis y axis)

The only boot parameter I'm using on Mint is acpi_backlight=vendor and it works without it as well.

Why I think this is not a backlight problem? Firstly, there is no eDP port in Arch Linux and the laptop display seems to be connected to what is referred to as DP2, which is wierd (as opposed to eDP1 in Linux Mint). Also, (in Arch Linux) using various backlight tweaks both in boot parameters and xorg.conf I was able to see that the /sys/class/backlight interface was up and running and parameters like bl_power or actual_brightness were changing with Fn+keys accordingly. Another reason why I think this is not a backlight problem is that little "feel" when the display goes down at Arch boot - there is this LCD "off effect" that also appears when I acctually power off the laptop but I cannot see the same effect when I press the Fn+(screen off) key in Mint, where the display just blanks.

I also tried different kernels including -ck -ck-atom and -lts. No effect whatsoever. Microcode problem is ruled out, there is currently no microcode update for my hardware.

I'm baffled now and I have no idea what to make of this. What should be my next step? What I missed?

Thanks!

-miky

Last edited by mr.MikyMaus (2015-03-18 16:08:26)

boteium · 2015-02-11 17:21:01

it's a Z37xx platform. a lots of people have this problem as well.
try putting "xrandr --output DP2 --rotate right" in your .xinitrc and start xorg to see if you can unblank it.
of course you can "--rotate normal" afterwards .

I haven't solve this problem but this xrandr trick does work somehow.
right now I remove i915 from mkinitcpio.conf to prevent kms start too early so I can see system message during booting up.
of course there will be about 5 second blank screen during kms late start to Xorg desktop.
but at least it's less unbearable.

glenb77 · 2015-02-11 18:06:56

i have same issue on asus x200ma. i don't use a login manager and boot to prompt. i also make an alias in bashrc to suspend the machine with a single keystroke. at boot when screen is off. i login, suspend machine and wake it up iimmediately and i have the screen back. i use a single letter for login and for password. so basically it is three keystorkes and i have the screen back.

waking up from suspend turns the screen back on. if you have a keyboard fn key that works for that. that should also do it.

boteium · 2015-02-11 18:30:19

It's nice to hear another workaround that doesn't involve X.
unfortunately, suspend doesn't work my Z3735F.
Perhaps there are better ways to unblank it ?
I am hoping if there is something in /proc that can turn on/off the screen and have the same unblanking effect.

Last edited by boteium (2015-02-11 18:44:34)

mr.MikyMaus · 2015-02-11 21:56:56

Thanks for tips, I'll try them as soon as I get to the laptop. In the mean time I have another observation: when using Mint with Cinnamon, there was a seamless X solution there - exactly as boteium said - the screen went down at KMS load and back up with X. I was never able to track what was really going on there and quite frankly, I didn't care. It worked and that was what I needed back at the time. However while this was mentioned I realised this is no longer happening - the screen in Mint does not go down with KMS anymore. I suspect a kernel upgrade from 3.10 to 3.11 but it's just a guess. Any tips on how to track what's going on under the hood of Mint? Perhaps they found a solution...

Last edited by mr.MikyMaus (2015-02-11 21:57:38)

Los Frijoles · 2015-02-12 02:31:55

I'm also using an Asus K200MA and have the same issue. I managed to get into a console with nomodeset, but this is a real problem for me. I just ran updates as well with no effect.

Does anyone know what package is responsible for this? Is it the kernel or one of the display drivers?

If there is something in /proc or in /sys that could turn the screen back on that would be great. I'm going to look around and see what I can find. Since I've got terminal access I could try adding something that would run just after startup to turn the screen back on as a stopgap.

mr.MikyMaus · 2015-02-12 11:54:33

I've tried the rotating tip but to no avail, not even a blink I also noticed Mint uses uvesafb so I tried fiddling with this nice piece of tech but no results there either.

I've even tried a few other "triggers" with both xrandr and xorg.conf (--off and then re-enabling, "rescan" option, for instance) but still, no luck. The display is off. I've checked against xrandr output from Mint finding that there are different ID's (the hex number in xrandr --verbose's Identifier field) for the displays but EDID is the same for both eDP1 in Mint and DP2 in Arch so I guess this is just a labeling issue that has nothing to to with the real problem. Right now I'm compiling a 3.10-lts kernel to see if something changes. Hope I won't hit some systemd compatibility issues.

Will appreciate any tips on where to look, what to debug and what else can I read on the topic.

Los Frijoles · 2015-02-14 06:16:59

I got my screen back by downgrading to Linux 3.18.2-2. It's not a real fix, but it downgraded seemingly without issue.

So, somewhere between 3.18.2-2 and 3.18.5-1, something broke.

mr.MikyMaus · 2015-02-16 11:28:52

After compiling and installing a 3.10 kernel, namely 3.10.69-1-lts310-ck from aur package linux-lts310-ck, I got below mentioned error, no DRM, hence no KMS and no X.

[drm:i915_pci_probe] *ERROR* Preliminary hardware support disabled

My guess would be that the Intel driver in 3.10 does not support my hardware yet but I don't really understand why since I'm pretty sure it worked on 3.10 Debian kernel for which I didn't find any extra patches regarding Intel driver.

The next step is kernel version 3.11 but I'm still accepting any and all tips on what to do or where to look and what to look for

mr.MikyMaus · 2015-02-19 11:29:21

A new observation: with kernel 3.12-ck the behaviour is slighltly different. Upon i915 module load display turns native until some point in late boot when it goes down again (not due to X though, I disabled graphical.target and display manager). I've tried both early and late KMS, same results. When using early the display remains native indefinitely when the init process is paused (dm-crypt askpass in my case).

I've tried a few boot options mentioned in OP again, no difference.

There are a few errors on early boot (not verbatim):

[drm:i915_stolen_to_physical] *ERROR* conflict detected with stolen region [0xbb000000 - 0xbf000000]
[drm:valleyview_eable_rps] *ERROR* GT fifo had a previous error 10
[drm:intel_pipe_config_compare] *ERROR* mismatch in clock (expected 70000, found 162000)
[drm:intel_pipe_config_compare] *ERROR* mismatch in clock (expected 70000, found 162000)

Disabling systemd-backlight service made things work!

Also display names are properly enumerated (by both xrandr and /sys/class/drm): there is no DP2 and the "connected" display is listed as eDP1. Also, the VGA1 display says it's connected and up even though there is no physical VGA port.

However, I lost the ability to control backlight, save for on/off using hardware hotkeys. Changing intensity using hotkeys does nothing at all. This is an off-topic problem here in this thread though.

I'll stick with 3.12 kernel for the time being and see if any new kernel upgrade makes things better or not. If I'll have some spare time I might go through kernel changelogs to see what might have happened.

What about those errors? Should I care?

Last edited by mr.MikyMaus (2015-02-19 11:42:53)

glenb77 · 2015-02-28 19:46:36

the problem has gone away using repo-ck latest kernel 3.19-1. this on an asus x200ma. hopefully it resolves in 3.19 kernel series for others

Pelzflorian · 2015-03-09 07:33:30

Gone away for me too on an Asus F200MA with linux-libre 3.19.1. I didn't test the non-libre Linux, but I assume it's the same. Glad to be back on a non-LTS kernel.

mr.MikyMaus · 2015-03-18 16:06:41

Linux 3.19.1 did not solve the issue for me. At least not on it's own. Some short time after updating to 3.19.1 I also had a chance to update BIOS via Lenovo's proprietary tool through Windows. The BIOS update did the trick for me. I did not want to go there since the BIOS update changelog wasn't very specific about the update (it only said some minor changes were present) so, the lesson learned is - don't trust proprietary updates and changelogs - you never know what else is there. But, well, do them anyway. You don't really have a choice there.

Last edited by mr.MikyMaus (2015-03-18 16:11:15)

mitvix · 2015-11-09 12:21:16

For me the firmware update doesn´t work on DELL Latitude E5440. The solution come after change to Ubuntu distro :-[

Debian 8.1 kernel 3.x same issue
Debian 8.2 kernel 3.x or 4.x same issue
Linux Mint 17.2 kernel 3.16 same issue
Ubuntu 15.x problem solved :-(

I don´t know if this workaround is a good or bad thing :-(.

Maybe installing official intel drivers can help.

Arch Linux

#1 2015-02-11 13:39:09

[solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#2 2015-02-11 17:21:01

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#3 2015-02-11 18:06:56

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#4 2015-02-11 18:30:19

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#5 2015-02-11 21:56:56

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#6 2015-02-12 02:31:55

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#7 2015-02-12 11:54:33

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#8 2015-02-14 06:16:59

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#9 2015-02-16 11:28:52

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#10 2015-02-19 11:29:21

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#11 2015-02-28 19:46:36

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#12 2015-03-09 07:33:30

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#13 2015-03-18 16:06:41

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

#14 2015-11-09 12:21:16

Re: [solved] KMS / Intel turns display off (black screen, Z37xxx, Lenovo)

Board footer