Bug 83175

Bug detailed description
On the Text mode,can't refresh ELD information on Bytrail. 
Connect a monitor with DP(HDMI) to check the ELD information,then unplug this monitor's DP(HDMI) cable and plug other monitor. But only display before monitor information.

Reproduce Steps
1.Connect Dell monitor with DP(HDMI) cable.
2.cat /proc/asound/card0/eld#0.1
3.Unplug DP(HDMI) cable to connect ASUS monitor
4.cat /proc/asound/card0/eld#0.1

Expected Result
After step4,should display ASUS monitor information.

Actual Result
After step4,display Dell monitor information.

Test Environment:
Hardware:
CRB:Wilson Beach SDS
Platform:BroadWell
CPU: Broadwell U D0
Chipset PCH: Wildcat Point –LP B0
Software:
BIOS: BDW-E1R1.86C.0074.R01.1405020115
Kernel: 3.16.0-rc7
Branch: git://git.kernel.org/pub/scm/linux/kernel/git/lrg/asoc.git 
intel/adsp-v3.16
Commit:0c8fd11d61db61ecf406e36adf54455eef3eb64e

update the Enviroment
Test Environment:
Hardware:
CRB:BAYLEY BAY
Platform: bty-m
CPU: Celeron(R) N2810
Audio card: ALC5640
Software:
BIOS: BBAYCRB1.X64.0078.R20.1401131845
Kernel: 3.16.0-rc5
Branch: git://anongit.freedesktop.org/drm-intel
drm-intel-nightly
tag:2014_07_20
Commit:873440

Mengdong, any ideas on this one on the audio side?

Please try drm-intel-nightly, provide dmesg with drm.debug=0xe module parameter, all the way from boot to reproducing the problem (i.e. steps in comment #1). Please also attach the /proc/asound/card0/eld#0.1 files with descriptive names.

Also, what do you mean exactly by "text mode"?

Hi Jani,

Actually, we find the bug happens on both text mode and GUI mode.

Thanks,
Libin

(In reply to comment #7)
> Actually, we find the bug happens on both text mode and GUI mode.

Okay, please provide the information requested in comment #5.

Hi Jani,

Please see the attachments for the details.

dmesg.init, eld.init, interrupts.init are the dmesg of bootup
dmesg.unplug, eld.unplug, interrupts.unplug are the dmesg after unplugging the HDMI monitor
dmesg.plug, eld.plug, interrutps.plug are the dmesg after plugging the HDMI monitor again.

If applying the test patch i915_hpd.patch, the eld information can be updated normally.

Thanks,
Libin

Created attachment 106642 [details]
dmesg.inig

Created attachment 106643 [details]
dmesg.plug

Created attachment 106644 [details]
dmesg.unplug

Created attachment 106645 [details]
eld.init

Created attachment 106646 [details]
eld.plug

Created attachment 106647 [details]
eld.unplug

Created attachment 106648 [details]
interrupts.init

Created attachment 106650 [details]
interrupts.plug

Created attachment 106651 [details]
interrupts.unplug

Created attachment 106652 [details] [review]
i915_hpd.patch

Created attachment 106729 [details] [review]
drm/i915: fix short vs. long hpd detection

Please try this patch on top of nightly.

*** Bug 84226 has been marked as a duplicate of this bug. ***

Libin, please try the patch from comment #20.

Hi Jani,
I added comment #20 patch to drm-intel/nightly kernel but ELD information still do not refresh on hot-plug events.
I had the same behavior with or without X started. However, I noticed that running $ xrandr is a workaround as everything is correct after this command (ELD and display).

Hi Jani,

Sorry for delay. We still haven't got the platform to test. The platform is occupied by another QA team for Media test.

We will test it and update the result ASAP.

(In reply to Jani Nikula from comment #22)
> Libin, please try the patch from comment #20.

Would be good to test the audio branch from http://cgit.freedesktop.org/~jani/drm/ also.

Hi Jani,
I build the kernel with the audio branch from http://cgit.freedesktop.org/~jani/drm/ and use the nightly config.
I verify the issue on the BSW connect HDMI cable with the kernel,but the platform can't boot up normally and the screen black all the time. 
I don't know if need the special config to build the branch kernel?


Dan

I tested audio branch from http://cgit.freedesktop.org/~jani/drm/ yesterday, hadn't the time to post my results which are not very different from Dan's.
HDMI is not detected, eDP works as normal. I used byt-m default config "config-bytm-drm" attached

Created attachment 107653 [details]
bytm default config

Hi Jani,

We found the latest kernel has fixed this bug. There are a lot of patches since we reported this bug. Do you know which patches fixed this bug?

Thanks,
Libin

By "latest kernel" do you refer to Linus' tree or one of the drm-intel branches?

Here's a list of recent HD audio related commits in drm-intel-next-queued, from newest to oldest. Please try 7c10a2b5876e and 
28855d2ac349, and you can reverse bisect which commit fixed the bug.

28855d2ac349 drm/i915/audio: add DOC comment describing HDA over HDMI/DP
82910ac6d575 drm/i915: make pipe/port based audio valid accessors easier to use
d5ee08de1bbe drm/i915/audio: add audio codec enable debug log for g4x
76d8d3e5b57d drm/i915/audio: add audio codec disable on g4x
c1dec79aaee6 drm/i915: enable audio codec after port
495a5bb81d34 drm/i915/audio: add vlv/chv/gen5-7 audio codec disable sequence
c6bde93b9225 drm/i915/audio: rewrite vlv/chv and gen 5-7 audio codec enable sequence
5fad84a7530f drm/i915: rewrite hsw/bdw audio codec enable/disable sequences
c46f111f5171 drm/i915: clean up and clarify audio related register defines
d806fbf5d38f drm/i915/audio: remove misleading checks for !eld[0]
69bfe1a9b4df drm/i915: introduce intel_audio_codec_{enable, disable}
2aa0de39fa07 drm/i915/ddi: write ELD where it's supposed to be done
6189b0369c67 drm/i915/audio: set ELD Conn_Type at one place
820d2d774828 drm/i915/audio: pass intel_encoder on to platform specific ELD functions
33d1e7c6f416 drm/i915: pass intel_encoder to intel_write_eld
f9f682ae3530 drm/i915/audio: beat some sense into the variable types and names
87fcb2ad4582 drm/i915/audio: constify hdmi audio clock struct
7c10a2b5876e drm/i915: add new intel audio file to group DP/HDMI audio

Since it's fixed now, I'm closing the bug.

Thanks, Yani. This is fixed on drm-intel kenrel, BTW.

I verify the issue on the latest kernel 3.18.0-rc4, also can reproduce it.

The environment:
Hardware
CRB:Braswell RVP
Platform: Braswell 
CPU: Braswell B1
Audio card: ALC282

Software
BIOS: BRASWEL1.X64.0041.R00.1410131600
Kernel: 3.18.0-rc4 (2014-11-17)
Branch: git://anongit.freedesktop.org/drm-intel
drm-intel-nightly
Commit:e49ebf

(In reply to zhaodan from comment #33)
> I verify the issue on the latest kernel 3.18.0-rc4, also can reproduce it.
...
> Kernel: 3.18.0-rc4 (2014-11-17)

Please do not refer to drm-intel-nightly kernels with "3.18.0-rc4" or similar. It is confusing. If that is from 'uname -a' or dmesg, it depends on your config options. The -nightly kernel has tons of patches on top of v3.18-rc4.

> Branch: git://anongit.freedesktop.org/drm-intel
> drm-intel-nightly
> Commit:e49ebf

This is better, but five digits of commit sha1 is insufficient. There will be collisions making the commit ambiguous. 12 digits is recommended to avoid collisions in a git project the size of the kernel.


Anyway, please try the current drm-intel-nightly, make sure it has

commit eb45fa0b93e03b03848cd048dcc57648409c8125
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Tue Nov 18 12:11:29 2014 +0200

    drm/i915/audio: fix monitor presence indication after disable

I suspect this will fix it.

Let's just presume this is fixed in -nightly. Please reopen if that's not the case.

Hi Jani,
I had verify the issue on the nightly kernel 3.18.0-rc5,and can't reproduce the issue.so I change the status as verified.

The environment:
Hardware
CRB:Braswell RVP
Platform: Braswell 
CPU: Braswell B1
Audio card: ALC282

Software
BIOS: BRASWEL1.X64.0041.R00.1410131600
Kernel: 3.18.0-rc5 (2014-11-21)
Branch: git://anongit.freedesktop.org/drm-intel
drm-intel-nightly
Commit:cdb41f

Closing old verified.