Update: regression is found with drm-intel-testing-2016-02-29, the detailed results show as following:
Case 1: only eDP connected
S3 is normal for the first time, but system will hang there (resuming more than 10 mins) when do S3 again 
Case 2: only DP connected
Unable to resume from S3(resuming more than 10 mins).
Case 3: only HDMI connected 
Unable to resume from S3(resuming more than 10 mins).
Case 4: DP and HDMI connected 
Unable to resume from S3(resuming more than 10 mins).
Case 5: eDP and DP connected
System can resume normally from S3 for the first time, but DP stop working. System will hang there when do S3 again.
Case 6: eDP and HDMI connected
System can resume normally form S3 for the first 1-2 times, but HDMI stop working. System will hang there when do S3 again.
Case 7: eDP, DP and HDMI all connected
No output from DP and HDMI and the resuming time is more than 5 mins. After system resumed, the analog audio is works well. But system will hang there when do S3 again.

Test Environment:
Platform: Apollo lake RVP1a FabA
BIOS: APLKRVPA.X64.0119.R10.1601031622
KSC FW:1.05
BXT SOC : A0
GOP : 10.0.1022
Board ID: APL RVP 1A (01)
CPU Flavor: BXT Tablet (0)
FAB ID: FAB1
Total Memory : 8GB
Kernel:4.5.0-rc5+
Tree: git://anongit.freedesktop.org/drm-intel
Branch:drm-intel-testing
Commit: a27657a39741599ce9feef1ef39ae6654e9e6768
tag ID: drm-intel-testing-2016-02-29

Please boot with drm.debug=0xe, reproduce the bug, then attach the dmesg output here.

Since this is a regression, can you please bisect it and report which one is the bad commit?

Created attachment 122197 [details]
dmesg with drm.debug=0xe

Who can help do the bisect? gfx QA team?

With the latest configuration we can still observed this issue on BXT-P


Hardware information
=====================
Platform	BXT-P
Motherboard model	Broxton P
Motherboard type	NOTEBOOK Hand Held
Motherboard manufacturer	Intel Corp.
CPU family	Other
CPU information	06/5c
GPU Card	Intel Corporation Device 5a84 (rev 03) (prog-if 00 [VGA controller])

Software information
======================
Linux distribution : Ubuntu 15.10
Architecture       : 64-bit

 --> Component : drm 
	 url : http://cgit.freedesktop.org/mesa/drm 
	 tag : libdrm-2.4.67-5-gea07de9 
	 commit : ea07de9 
	 author : Rodrigo Vivi <rodrigo.vivi@intel.com> 
	 age : 13 days ago 
 --> Component : mesa 
	 url : http://cgit.freedesktop.org/mesa/mesa 
	 tag : mesa-11.1.2 
	 commit : 7bcd827 
	 author : Emil Velikov <emil.velikov@collabora.com> 
	 age : 5 weeks ago 
 --> Component : xf86-video-intel 
	 url : http://cgit.freedesktop.org/xorg/driver/xf86-video-intel 
	 tag : 2.99.917-560-gd167280 
	 commit : d167280 
	 author : Chris Wilson <chris@chris-wilson.co.uk> 
	 age : 3 weeks ago 
 --> Component : libva 
	 url : http://cgit.freedesktop.org/libva/ 
	 tag : libva-1.7.0.pre1 
	 commit : 35f1447 
	 author : Xiang Haihao <haihao.xiang@intel.com> 
	 age : 11 days ago 
 --> Component : vaapi (intel-driver) 
	 url : http://cgit.freedesktop.org/vaapi/intel-driver 
	 tag : 1.7.0.pre1 
	 commit : ab8e4e3 
	 author : Xiang Haihao <haihao.xiang@intel.com> 
	 age : 11 days ago 
 --> Component : cairo 
	 url : http://cgit.freedesktop.org/cairo 
	 tag : 1.15.2 
	 commit : db8a7f1 
	 author : Bryce Harrington <bryce@osg.samsung.com> 
	 age : 3 months ago 
 --> Component : xserver 
	 url :  http://cgit.freedesktop.org/xorg/xserver 
	 tag : xorg-server-1.18.1 
	 commit : 23e60f9 
	 author : Adam Jackson <ajax@redhat.com> 
	 age : 5 weeks ago 
 --> Component : intel-gpu-tools 
	 url : http://cgit.freedesktop.org/xorg/app/intel-gpu-tools 
	 tag : intel-gpu-tools-1.14-23-g364d378 
	 commit : 364d378 
	 author : Paulo Zanoni <paulo.r.zanoni@intel.com> 
	 age : 10 days ago 

kernel
=========
drm-intel-testing / commit a27657a

commit a27657a39741599ce9feef1ef39ae6654e9e6768
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Mon Feb 29 09:42:22 2016 +0100

    drm-intel-nightly: 2016y-02m-29d-08h-41m-12s UTC integration manifest

Kernel version : 4.5.0-rc5
Architecture : source amd64 all

Was this reproducible only on BXT-P/APL platform? Or the issue appeared on SKL as well?

(In reply to Daniela Prodan from comment #6)
> Was this reproducible only on BXT-P/APL platform? Or the issue appeared on
> SKL as well?

It also appeared for me on SKL, related (I think) bug:
https://bugs.freedesktop.org/show_bug.cgi?id=91904

Hi Daniela, Mikolaj,

The issue is only reproduced on BXT-P/APL, not reproduced on SKL with recent commits.

tried with latest drm-nightly commit and get the same result, The issue is reproduced on BXT-P and not reproduced on SKL.

I tried to reproduce it on a SKL (Lenovo E31 laptop, i5-6200u) with latest on drm-intel-nightly.

I connected an HDMI monitor, the problem doesn't reproduce. Both screens show the correct output after resume.

The callstack in the dmesg is not related to this problem - it's a known issue (https://bugs.freedesktop.org/show_bug.cgi?id=89055) on SKL with multiple pipes. Anyway, that's a harmless warning and the fix is in progress.

I'm sorry, I don't have access to a BXT-P

Hi All,

The latest status is:
in GUI mode: it works
in text mode, display shows nothing

*** Bug 94772 has been marked as a duplicate of this bug. ***

Hi with the following kernel and the patch from http://patchwork.freedesktop.org/patch/72566/

kernel
=======
commit f5d413cccefa1f93d64c34f357151d42add63a84
Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Date:   Thu Mar 24 14:35:16 2016 +0000

    drm-intel-nightly: 2016y-03m-24d-14h-34m-29s UTC integration manifest

after resume from S3 DP issue is gone, but HDMI still continue

DP is recognized as : DP2
HDMI is recognized as : DP1


BXT configuration
=================
BIOS : 129.10
BXT Soc : A0
KSC : 1.12
Board ID : APL RVP 1A (01) FAB1

attachtments
============
 dmesg_HDMI_DP.log

Created attachment 122722 [details]
dmesg_HDMI_DP.log

Matt,
The patch you pointed in 93915 partially fixes this bug. 
I also add Bob Paauwe in cc.
Can the patch be improved to fully fix it?

The VBT solution is pushed to dinq

commit d252bf68b75792108ae2821c3a6e1cdc58e88cb9
Author: Shubhangi Shrivastava <shubhangi.shrivastava@intel.com>
Date:   Thu Mar 31 16:11:47 2016 +0530

    drm/i915: Set invert bit for hpd based on VBT


Can QA please verify whether this solves this problem

Humberto please use this new version (based on VBT) ; it fixed also #93915
Thanks

(In reply to cprigent from comment #15)
> The patch you pointed in 93915 partially fixes this bug. 

(In reply to yann from comment #17)
> it fixed also #93915

Side note, please reference other bugs by writing "bug 93915". Please reference comments by writing "comment #15". This will turn them to links. Thanks.

(In reply to Kimmo Nikkanen from comment #16)
> The VBT solution is pushed to dinq
> 
> commit d252bf68b75792108ae2821c3a6e1cdc58e88cb9
> Author: Shubhangi Shrivastava <shubhangi.shrivastava@intel.com>
> Date:   Thu Mar 31 16:11:47 2016 +0530
> 
>     drm/i915: Set invert bit for hpd based on VBT
> 
> 
> Can QA please verify whether this solves this problem

Hi : 

after test this commit we can see the same issue, DP works after S3 but HDMI not.



commit d252bf68b75792108ae2821c3a6e1cdc58e88cb9
Author: Shubhangi Shrivastava <shubhangi.shrivastava@intel.com>
Date:   Thu Mar 31 16:11:47 2016 +0530

    drm/i915: Set invert bit for hpd based on VBT

    This patch sets the invert bit for hpd detection for each port
    based on VBT configuration. Since each AOB can be designed to
    depend on invert bit or not, it is expected if an AOB requires
    invert bit, the user will set respective bit in VBT.

    v2: Separated VBT parsing from the rest of the logic. (Jani)

    v3: Moved setting invert bit logic to bxt_hpd_irq_setup()
        and changed its logic to avoid looping twice. (Ville)

    v4: Changed the logic to mask out the bits first and then
        set them to remove need of temporary variable. (Ville)

    v5: Moved defines to existing set of defines for the register
        and added required breaks. (Ville)

    Signed-off-by: Sivakumar Thulasimani <sivakumar.thulasimani@intel.com>
    Signed-off-by: Durgadoss R <durgadoss.r@intel.com>
    Signed-off-by: Shubhangi Shrivastava <shubhangi.shrivastava@intel.com>
    Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
    [Jani: fixed some checkpatch noise, added kernel-doc.]
    Signed-off-by: Jani Nikula <jani.nikula@intel.com>
    Link: http://patchwork.freedesktop.org/patch/msgid/1459420907-11383-2-git-send-email-shubhangi.shrivastava@intel.com


Architecture: source amd64 all
Version: 4.5.0-rc7

attached dmesg for the last comment "dmesg_HDMI.log"

Created attachment 122774 [details]
dmesg_HDMI.log

(In reply to Humberto Israel Perez Rodriguez from comment #21)
> Created attachment 122774 [details]
> dmesg_HDMI.log

Could you please provide the same log booting with drm.debug=14?

There are two blocks of stacked DDI/HDMI connectors next to each other on the board marked with DDI0 and DDI1. Are you connecting the HDMI display to the plug in the left or the right block?

Do you have the same problem with HDMI if you connect it to the other port?

(In reply to Imre Deak from comment #22)
> (In reply to Humberto Israel Perez Rodriguez from comment #21)
> > Created attachment 122774 [details]
> > dmesg_HDMI.log
> 
> Could you please provide the same log booting with drm.debug=14?
> 
> There are two blocks of stacked DDI/HDMI connectors next to each other on
> the board marked with DDI0 and DDI1. Are you connecting the HDMI display to
> the plug in the left or the right block?
> 
> Do you have the same problem with HDMI if you connect it to the other port?

Hi Deek :

the BXT-P platform the HDMI port is only enabled in the right port, in the left port is for DP onl.

after try for second time to go S3 with the DUT with the kernel 
commit : d252bf68b75792108ae2821c3a6e1cdc58e88cb9
branch : drm-intel-nightly

the DUT is not able to resume from S3 if is connected any external display (DP / HDMI)

(In reply to Humberto Israel Perez Rodriguez from comment #13)
> DP is recognized as : DP2
> HDMI is recognized as : DP1

Smells like lspcon.

The patch seems not reproduced after upgrade BIOS to V126, could anyone help to double check?

Platform: Apollo lake RVP1a FabA
BIOS: APLKRVPA.X64.0126.B51.1602260229
KSC FW:1.10
BXT SOC : B0
GOP : 10.0.1025
Board ID: APL RVP 1A (01)
CPU Flavor: BXT Tablet (0)
FAB ID: FAB1
Total Memory : 8GB
Kernel:4.6.0-rc2
Tree: git://anongit.freedesktop.org/drm-intel
Branch:drm-intel-nightly
Commit: 851708c7e97537ed618fadbe5d342eaf8fa5146d
HW-rework: F-28C, F-27

Humberto, please re-test using BIOS APLKRVPA.X64.0126.B51.1602260229

I believe that S3 failing in itself (so not the hotplug issue originally reported) is not a BIOS issue. I could reproduce it on my machine with a recent BIOS:
APLKRVPA.X64.0130.B11.1603210634

With a few fixes S3 works now for me and those fixes could be also relevant here.
So Humberto, besides trying to upgrade the BIOS as Yann suggested could you also try the following branch:
https://github.com/ideak/linux/commits/bxt-rpm

(In reply to Imre Deak from comment #27)
> I believe that S3 failing in itself (so not the hotplug issue originally
> reported) is not a BIOS issue. I could reproduce it on my machine with a
> recent BIOS:
> APLKRVPA.X64.0130.B11.1603210634
> 
> With a few fixes S3 works now for me and those fixes could be also relevant
> here.
> So Humberto, besides trying to upgrade the BIOS as Yann suggested could you
> also try the following branch:
> https://github.com/ideak/linux/commits/bxt-rpm

Hi Imre Deak :

i tried with a newer bios 131.10 and the branch : https://github.com/ideak/linux/commits/bxt-rpm, but is the same issue, after return from S3 the HDMI is not seen and xrandr also does not recognize it

kernel info
==================
commit 068ab4d97212c00defebf525821001785c7caf2e
Author: Imre Deak <imre.deak@intel.com>
Date:   Fri Apr 8 20:40:16 2016 +0300

    drm/i915/bxt: Sanitize DC state tracking after suspend-to-ram and -disk

    After suspend-to-ram or -disk we don't know what power state the display
    is, DC0 or DC9 are both possible states, so reset the DC software tracking
    in these cases. This gets rid of 'DC state mismatch' error messages
    during suspend to ram and disk which is a false positive based on the
    above.

    Signed-off-by: Imre Deak <imre.deak@intel.com>

branch = bxt-rmp
kernel = 4.6.0-rc2

Created attachment 122827 [details]
dmesg_bios_131.1

(In reply to Humberto Israel Perez Rodriguez from comment #28)
> (In reply to Imre Deak from comment #27)
> > I believe that S3 failing in itself (so not the hotplug issue originally
> > reported) is not a BIOS issue. I could reproduce it on my machine with a
> > recent BIOS:
> > APLKRVPA.X64.0130.B11.1603210634
> > 
> > With a few fixes S3 works now for me and those fixes could be also relevant
> > here.
> > So Humberto, besides trying to upgrade the BIOS as Yann suggested could you
> > also try the following branch:
> > https://github.com/ideak/linux/commits/bxt-rpm
> 
> Hi Imre Deak :
> 
> i tried with a newer bios 131.10 and the branch :
> https://github.com/ideak/linux/commits/bxt-rpm, but is the same issue, after
> return from S3 the HDMI is not seen and xrandr also does not recognize it
> 
> kernel info
> ==================
> commit 068ab4d97212c00defebf525821001785c7caf2e
> Author: Imre Deak <imre.deak@intel.com>
> Date:   Fri Apr 8 20:40:16 2016 +0300
> 
>     drm/i915/bxt: Sanitize DC state tracking after suspend-to-ram and -disk
> 
>     After suspend-to-ram or -disk we don't know what power state the display
>     is, DC0 or DC9 are both possible states, so reset the DC software
> tracking
>     in these cases. This gets rid of 'DC state mismatch' error messages
>     during suspend to ram and disk which is a false positive based on the
>     above.
> 
>     Signed-off-by: Imre Deak <imre.deak@intel.com>
> 
> branch = bxt-rmp
> kernel = 4.6.0-rc2

Well, in comment#23 you said:
"the DUT is not able to resume from S3 if is connected any external display (DP / HDMI)"

This means to me that not only that HDMI is not detected after S3 resume, but S3 doesn't work at all, that is the machine hangs or similar generic issue. Could you clarify what did you mean in comment#23? That is does S3 work at all for you with current -nightly (is the system still alive, can you SSH in)? With the above branch do you have working eDP/DP after S3 suspend/resume?

Not detecting HDMI after S3 is expected. The HDMI port is routed through the LSPCON protocol converter and the driver doesn't have support for that (patches are under review). Since BIOS configures LSPCON you will have a working output right after booting, but that configuration will be lost after S3 suspend/resume.

(In reply to Imre Deak from comment #30)
> (In reply to Humberto Israel Perez Rodriguez from comment #28)
> > (In reply to Imre Deak from comment #27)
> > > I believe that S3 failing in itself (so not the hotplug issue originally
> > > reported) is not a BIOS issue. I could reproduce it on my machine with a
> > > recent BIOS:
> > > APLKRVPA.X64.0130.B11.1603210634
> > > 
> > > With a few fixes S3 works now for me and those fixes could be also relevant
> > > here.
> > > So Humberto, besides trying to upgrade the BIOS as Yann suggested could you
> > > also try the following branch:
> > > https://github.com/ideak/linux/commits/bxt-rpm
> > 
> > Hi Imre Deak :
> > 
> > i tried with a newer bios 131.10 and the branch :
> > https://github.com/ideak/linux/commits/bxt-rpm, but is the same issue, after
> > return from S3 the HDMI is not seen and xrandr also does not recognize it
> > 
> > kernel info
> > ==================
> > commit 068ab4d97212c00defebf525821001785c7caf2e
> > Author: Imre Deak <imre.deak@intel.com>
> > Date:   Fri Apr 8 20:40:16 2016 +0300
> > 
> >     drm/i915/bxt: Sanitize DC state tracking after suspend-to-ram and -disk
> > 
> >     After suspend-to-ram or -disk we don't know what power state the display
> >     is, DC0 or DC9 are both possible states, so reset the DC software
> > tracking
> >     in these cases. This gets rid of 'DC state mismatch' error messages
> >     during suspend to ram and disk which is a false positive based on the
> >     above.
> > 
> >     Signed-off-by: Imre Deak <imre.deak@intel.com>
> > 
> > branch = bxt-rmp
> > kernel = 4.6.0-rc2
> 
> Well, in comment#23 you said:
> "the DUT is not able to resume from S3 if is connected any external display
> (DP / HDMI)"
> 
> This means to me that not only that HDMI is not detected after S3 resume,
> but S3 doesn't work at all, that is the machine hangs or similar generic
> issue. Could you clarify what did you mean in comment#23? That is does S3
> work at all for you with current -nightly (is the system still alive, can
> you SSH in)? With the above branch do you have working eDP/DP after S3
> suspend/resume?
> 
> Not detecting HDMI after S3 is expected. The HDMI port is routed through the
> LSPCON protocol converter and the driver doesn't have support for that
> (patches are under review). Since BIOS configures LSPCON you will have a
> working output right after booting, but that configuration will be lost
> after S3 suspend/resume.

In the comment 23, i said that only with the commit d252bf68b75792108ae2821c3a6e1cdc58e88cb9 i was not able to resume by second time , but with any other commit from drm-intel-nightly s3 works without issue.

with the branch https://github.com/ideak/linux/commits/bxt-rpm, i can resume from S3 with eDP, i did not try with DP because right now our other BXT-P is busy, in my current BXT only has eDP + HDMI (DP port is not working)

(In reply to Humberto Israel Perez Rodriguez from comment #31)
> (In reply to Imre Deak from comment #30)
> > (In reply to Humberto Israel Perez Rodriguez from comment #28)
> > > (In reply to Imre Deak from comment #27)
> > > > I believe that S3 failing in itself (so not the hotplug issue originally
> > > > reported) is not a BIOS issue. I could reproduce it on my machine with a
> > > > recent BIOS:
> > > > APLKRVPA.X64.0130.B11.1603210634
> > > > 
> > > > With a few fixes S3 works now for me and those fixes could be also relevant
> > > > here.
> > > > So Humberto, besides trying to upgrade the BIOS as Yann suggested could you
> > > > also try the following branch:
> > > > https://github.com/ideak/linux/commits/bxt-rpm
> > > 
> > > Hi Imre Deak :
> > > 
> > > i tried with a newer bios 131.10 and the branch :
> > > https://github.com/ideak/linux/commits/bxt-rpm, but is the same issue, after
> > > return from S3 the HDMI is not seen and xrandr also does not recognize it
> > > 
> > > kernel info
> > > ==================
> > > commit 068ab4d97212c00defebf525821001785c7caf2e
> > > Author: Imre Deak <imre.deak@intel.com>
> > > Date:   Fri Apr 8 20:40:16 2016 +0300
> > > 
> > >     drm/i915/bxt: Sanitize DC state tracking after suspend-to-ram and -disk
> > > 
> > >     After suspend-to-ram or -disk we don't know what power state the display
> > >     is, DC0 or DC9 are both possible states, so reset the DC software
> > > tracking
> > >     in these cases. This gets rid of 'DC state mismatch' error messages
> > >     during suspend to ram and disk which is a false positive based on the
> > >     above.
> > > 
> > >     Signed-off-by: Imre Deak <imre.deak@intel.com>
> > > 
> > > branch = bxt-rmp
> > > kernel = 4.6.0-rc2
> > 
> > Well, in comment#23 you said:
> > "the DUT is not able to resume from S3 if is connected any external display
> > (DP / HDMI)"
> > 
> > This means to me that not only that HDMI is not detected after S3 resume,
> > but S3 doesn't work at all, that is the machine hangs or similar generic
> > issue. Could you clarify what did you mean in comment#23? That is does S3
> > work at all for you with current -nightly (is the system still alive, can
> > you SSH in)? With the above branch do you have working eDP/DP after S3
> > suspend/resume?
> > 
> > Not detecting HDMI after S3 is expected. The HDMI port is routed through the
> > LSPCON protocol converter and the driver doesn't have support for that
> > (patches are under review). Since BIOS configures LSPCON you will have a
> > working output right after booting, but that configuration will be lost
> > after S3 suspend/resume.
> 
> In the comment 23, i said that only with the commit
> d252bf68b75792108ae2821c3a6e1cdc58e88cb9 i was not able to resume by second
> time ,

So by doing two consecutive S3 suspend/resumes? And how did things fail, did the system become totally unresponsive?

That commit alone shouldn't cause any generic S3 failures, it should only affect whether your external DP will be detected or not.

> but with any other commit from drm-intel-nightly s3 works without issue.

You mean without d252bf68b75792108ae2821c3a6e1cdc58e88cb9, I assume. But with that commit DP detection started to work for you, except that for the second time something failed. I believe that this failure is still there in nightly and I hope that the bxt-rpm branch would fix it for you.

> with the branch https://github.com/ideak/linux/commits/bxt-rpm, i can resume
> from S3 with eDP, i did not try with DP because right now our other BXT-P is
> busy, in my current BXT only has eDP + HDMI (DP port is not working)

Is your current BXT a BXT-P? Do you know why DP port doesn't work on it?

Please test with the BXT-P machine you originally reported the bug on if eDP + external DP work on both current -nightly and the bxt-rpm branch after a few S3 suspend/resume cycles.

Hi :

i've tested in the original BXT where i saw this issue, please see my comments below :

with drm-intel-nightly
=========================
after try with nightly and several s3 suspend/resume iterations HDMI stop after resume from s3, please see the attached dmesg_nightly.log with drm.debug=14 as parameter

commit c0a9d3a8bcf7f049a391a601578054d96708667e
Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date:   Thu Apr 14 15:22:39 2016 +0300

    drm-intel-nightly: 2016y-04m-14d-12h-22m-13s UTC integration manifest


Kernel version : 4.6.0-rc3
Architecture : source amd64 all

with branch bxt-rpm from : https://github.com/ideak/linux
===========================================================
after try with bxt-rpm branch and several s3 suspend/resume iterations HDMI stop after resume from s3, please see the attached dmesg_bxt_rpm.log with drm.debug=14 as parameter

commit 72c0a995bd9c6e0d9ac8517180f305b1aae3ffa5
Author: Imre Deak <imre.deak@intel.com>
Date:   Wed Apr 13 17:35:40 2016 +0300

    drm/i915/gen9: Put init power domains during suspend if DMC not loaded

    Signed-off-by: Imre Deak <imre.deak@intel.com>

branch : bxt-rpm


Note : for both kernels eDP + DP works after several S3 iterations without any issues

Created attachment 122940 [details]
dmesg_bxt_rpm.log

Created attachment 122941 [details]
dmesg_nightly.log

HDMI is not working when resuming form S3 due to LSPCON issue, this is tracked by: *https://bugs.freedesktop.org/show_bug.cgi?id=94558
As it works for eDP + DP, I propose we close it and track 94558.