[Sandybridge GT1] No display on dell all-in-one

[Robert Hooker]

-- CPU: Intel Core(TM) i3-2100 CPU @ 3.10GHz (Sandy Bridge)
-- GPU: Intel Corporation Sandy Bridge Integrated Graphics Controller
[8086:0102] (rev 09) (prog-if 00 [VGA controller])
-- system architecture: 32-bit
-- Linux distribution: Ubuntu 10.10 and 11.04
-- Package versions:
   libdrm - 2.4.21-1ubuntu2 and 2.4.23-1ubuntu6
   xf86-video-intel - 2:2.12.0-1ubuntu5.2 and 2:2.14.0-4ubuntu7
   kernel - 2.6.35-24.42-generic (2.6.35.8) through 2.6.39.1

This system is a desktop one with a eDP->LVDS converter (CH7511B) for the
all-in-one panel and it only works with VESA. As soon as KMS kicks in it goes
black and doesn't come back. External monitors still work fine, its just the
internal panel that is black. Ubuntu maverick and natty were tested, as well as
a 2.6.39.1 kernel. the 2.6.39.1 kernel actually hung with an external monitor,
with additional messages like these at the end

PCH HDCP audio interrupt

Followed by a two additional DRM messages:

<7> [ 17.668935 ] [drm.ironlake_dp_detect], DPCD: 0000
<7> [ 245.874118] [drm:i915_hotplug_work_func], running encoder hotplug
functions

[Robert Hooker (sarvatt)]

Created attachment 48233
drm.debug=0x0e dmesg from failed boot

-- CPU: Intel Core(TM) i3-2100 CPU @ 3.10GHz (Sandy Bridge)
-- GPU: Intel Corporation Sandy Bridge Integrated Graphics Controller [8086:0102] (rev 09) (prog-if 00 [VGA controller])
-- system architecture: 32-bit
-- Linux distribution: Ubuntu 10.10 and 11.04
-- Package versions:
   libdrm - 2.4.21-1ubuntu2 and 2.4.23-1ubuntu6
   xf86-video-intel - 2:2.12.0-1ubuntu5.2 and 2:2.14.0-4ubuntu7
   kernel - 2.6.35-24.42-generic (2.6.35.8) through 2.6.39.1

This system is a desktop one with a eDP->LVDS converter (CH7511B) for the all-in-one panel and it only works with VESA. As soon as KMS kicks in it goes black and doesn't come back. External monitors still work fine, its just the internal panel that is black. Ubuntu maverick and natty were tested, as well as a 2.6.39.1 kernel. the 2.6.39.1 kernel actually hung with an external monitor, with additional messages like these at the end

PCH HDCP audio interrupt

Followed by a two additional DRM messages:

<7> [ 17.668935 ] [drm.ironlake_dp_detect], DPCD: 0000
<7> [ 245.874118] [drm:i915_hotplug_work_func], running encoder hotplug functions

[Robert Hooker (sarvatt)]

Created attachment 48234
xrandr --verbose with external VGA plugged in

[Chris Wilson (ickle)]

Looks like this symptom again:

[ 10.878980] [drm:gen6_fdi_link_train], FDI_RX_IIR 0x100
[ 10.878983] [drm:gen6_fdi_link_train], FDI train 1 done.
[ 10.879632] [drm:gen6_fdi_link_train], FDI_RX_IIR 0x600
[ 10.879635] [drm:gen6_fdi_link_train], FDI train 2 done.
[ 10.879636] [drm:gen6_fdi_link_train], FDI train done.

Typically we see FDI_RX_IIR 0x700 for train 1, with circumstantial evidence suggesting that 0x100 (although seemingly correct wrt the specs) is incomplete.

[Chris Wilson (ickle)]

Created attachment 48245
Clear FDI IIR before use

This patch should have no effect... But who knows and if you have a reproducible system it would be good to test. :)

[Robert Hooker (sarvatt)]

(In reply to comment #3)
> Created an attachment (id=48245) [details]
> Clear FDI IIR before use
>
> This patch should have no effect... But who knows and if you have a
> reproducible system it would be good to test. :)

It didn't have any effect for the bug and still went blank after grub. unfortunately this was all of the log I could get from the person with the machine

http://sarvatt.com/kern.log

(too big to attach)

[Chris Wilson (ickle)]

Oh boy. Is it seriously in a hotplug loop?

Jesse, doesn't this look suspicious:

Jun 23 15:58:56 u-23-AIO kernel: [ 140.240475] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.240476] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.240478] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.246637] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.246638] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.246640] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.254675] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.254677] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.254678] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.262625] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.262626] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.262628] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.270662] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.270663] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.270664] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.278615] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.278616] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.278618] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.286653] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.286655] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.286656] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.294601] [drm:pch_irq_handler],
Jun 23 15:58:56 u-23-AIO kernel: [ 140.294602] [drm:i915_hotplug_work_func], running encoder hotplug functions
Jun 23 15:58:56 u-23-AIO kernel: [ 140.294604] PCH HDCP audio interrupt
Jun 23 15:58:56 u-23-AIO kernel: [ 140.302616] [drm:output_poll_execute], [CONNECTOR:21:HDMI-A-3] status updated from 2 to 2
Jun 23 15:58:56 u-23-AIO kernel: [ 140.303126] [drm:intel_dp_aux_ch], dp_aux_ch timeout status 0x5143003e
Jun 23 15:58:56 u-23-AIO kernel: [ 140.303127] [drm:ironlake_dp_detect], DPCD: 0000
Jun 23 15:58:56 u-23-AIO kernel: [ 140.303129] [drm:output_poll_execute], [CONNECTOR:23:DP-2] status updated from 2 to 2
Jun 23 15:58:56 u-23-AIO kernel: [ 140.304159] [drm:i2c_algo_dp_aux_xfer], dp_aux_xfer return 2
Jun 23 15:58:56 u-23-AIO kernel: [ 140.331223] [drm:i2c_algo_dp_aux_xfer], dp_aux_xfer return 2
Jun 23 15:58:56 u-23-AIO kernel: [ 140.331224] [drm:output_poll_execute], [CONNECTOR:5:eDP-1] status updated from 1 to 1
Jun 23 15:58:56 u-23-AIO kernel: [ 140.331228] [drm:intel_ironlake_crt_detect_hotplug], ironlake hotplug adpa=0xf40000, result 0
Jun 23 15:58:56 u-23-AIO kernel: [ 140.331229] [drm:intel_crt_detect], CRT not detected via hotplug
Jun 23 15:58:56 u-23-AIO kernel: [ 140.331231] [drm:output_poll_execute], ...

[Chris Wilson (ickle)]

Created attachment 48459
Pass pch_iir to its handler

The useful part is the extra debug line to try and clarify what's being set in that hotplug loop. If it actually fixes the issue, I'd be amazed ;-)

[Chris Wilson (ickle)]

It would be worthwhile to check with drm-intel-fixes for the DP hotplug fixes (along with the conformance fixes). However, I believe that we need to mask the audio bits in the interrupt mask...

[Robert Hooker (sarvatt)]

Sorry this one is taking so long to reply to, I had to go through 3 layers of people to get info out of it at the time and those machines are gone. I should be able to get some hands on time with it in the next month or two without ancient versions. It might be a good idea to just close it and I'll reopen/make a new one with better info then.

[Keng-Yu Lin (lexical)]

Lenovo AIO C320 (SandyBridge Celeron G530) has this bug too.

[Gordon Jin (gordon-jin)]

Keng-Yu Lin has the hardware and could be able to test the latest code. The current drm-intel-next still doesn't work for the Lenovo AIO C320.

I'm promoting this bug to P1.

[Keith Packard (keithp)]

I'll need an updated kernel log from the current drm-intel-next kernel with drm.debug=0x0e from the Dell AIO machine.

Also, please create a separate bug for the Lenovo AIO, also attaching a kernel log.

[AceLan Kao (acelankao)]

Created attachment 54050
dmesg with new drm-intel-next kernel

With the new drm-intel-next kernel[1], the internal display works well.

1. http://kernel.ubuntu.com/~kernel-ppa/mainline/drm-intel-next/2011-11-17-oneiric/

[Keng-Yu Lin (lexical)]

For the record of the Lenovo AIO hardware, the bug is gone with BIOS Version E4KT15AUS and Ubuntu drm-intel-next build version 3.1.0-997-generic #201111170422.

[Cwm20000 (cwm20000)]

Dear Keng-Yu Lin,
    There's the same issue happened in our internal ,the motherboard design is followed intel AIO mini ITX standard design,also eDP==>>LVDS model fail ,external HDMI OK, could you kindly let me know what codes revised for the BIOS "E4KT15AUS" to slove this issue? tks

[AceLan Kao (acelankao)]

I built a DKMS package that packed the work in drm-intel-next tree, and it works on 3.0.0-13-generic kernel.

[AceLan Kao (acelankao)]

This one will fix the display problem after resuming from S3.
All the code in the dkms package is in our precise(12.04) kernel now, so there is no upstream work remains.

[Doublenian-xie (doublenian-xie)]

  My All-In-One have the similar problem with the internal displayer .But ,it go black not very frequently. it happened at before login-in ,login-in and after login-in some seconds .when I execute "xrandr" or "get-edid" at the terminal,the screen will go black and return back instantly,but sometimes ,it never return back . But when I ssh to this All-in-on from other computer,and execute "xrandr" or "get-edid" again ,then the screen light on and everything is OK .

[Doublenian-xie (doublenian-xie)]

Created attachment 56358
dmesg with drm.debug=0x0e from suceess boot

[Doublenian-xie (doublenian-xie)]

Created attachment 56359
drm.debug=0x0e dmesg form execute "xrandr" and never come back .

compare to the "dmesg with drm.debug=0x0e from sucess boot" ,The dmesg almost the same ,and I can't figure out what happened .
--CPU Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz

--GPU VGA compatible controller [0300]: Intel Corporation Sandy Bridge Integrated Graphics Controller [8086:0102] (rev 09)

-- kernel version : Debian kernel :linux-image-3.2.0-rc7-686-pae

eDP===>LVDS :Manufacturer: CHR Model: 7511 Serial#: 21501

[Chris Wilson (ickle)]

Jesse, I think this is another Vostro with the same hotplug symptoms, and I know how much you like them... :)

[Jesse Barnes (jbarnes-virtuousgeek)]

Please see the update to 42278, there's a patch there and instructions on how to avoid output detection. It may be the same issue.

[Jesse Barnes (jbarnes-virtuousgeek)]

No update, but I assume this is fixed by:

commit 46c325a871d50096b312e0f7f48182f2152ee836
Author: Jesse Barnes <email address hidden>
Date: Mon Jun 11 14:39:56 2012 -0400

    drm/i915: don't enumerate HDMID if an eDP panel is already active on the por

in drm-intel-fixes.

[Neogw (neogw)]

(In reply to comment #20)
> No update, but I assume this is fixed by:

commit
> 46c325a871d50096b312e0f7f48182f2152ee836
Author: Jesse Barnes
> <email address hidden>
Date: Mon Jun 11 14:39:56 2012 -0400

> drm/i915: don't enumerate HDMID if an eDP panel is already active on the por
> in drm-intel-fixes.

could you advise where can we locate this patch?

[Jesse Barnes (jbarnes-virtuousgeek)]

It's in Linus's tree:

commit b708a1d5ea7880b399dbd45cacafff6ae8d73171
Author: Jesse Barnes <email address hidden>
Date: Mon Jun 11 14:39:56 2012 -0400

    drm/i915: don't enumerate HDMID if an eDP panel is already active on the por