Assigned to me to provide logs and details of investigation

Can you try to capture logs either with netconsole or a serial console?  Also is it confirmed that i915 is the culprit here?  Does the failure to resume also occur without i915 loaded or compiled in?

Please try downgrading your DMC firmware to version 1.05 to see if that makes a difference; my team has seen some suspend/resume failures that only happen with the 1.06 DMC.

Also, as Jesse mentioned, ensure that i915 is the culprit by trying to suspend/resume without it in the picture.  Some of our early silicon has suspend/resume issues that are unrelated to software stack.

In addition to what Matt said. With version 1.06, the system hangs when entering suspend.  With version 1.05, the system will suspend and resume, but hangs shortly after resume.  

It looks like a combination of commit eade444272c303cac771bf4500b5b40ca119ad31  and the 1.06 firmware causes the hang on suspend.  We haven't tracked down the hang shortly after resume yet.

Also, was this captured by the BAT CI tests?  If not, we need to figure out why...

Assigned to Humberto:
Please try without i915 and also with DMC 1.05

Hi :

With DMC 1.06 for BXT-P and kernel drm-intel-testing with i915 loaded, after send the platform to S3 with the next command (echo mem > /sys/power/state) I can confirm that the system enters to S3 across postcode (0003) but when I try to resume after some time it seems that the system never resume from s3 instead looks like that the system reboots, the same behavior occurs with the kernel without i915 loaded and dmc 1.05 from (https://01.org/linuxgraphics/downloads/bxtdmcver105) also I have to comment that with i915 loaded and the below configuration the performance is affected in graphical environment as well in ttys, but it seems to improve without i915.

attached : Xorg.log and dmesg.log


Graphic stack versions  :
-----------------------------------------
cairo version: 1.15.2 / commit :  db8a7f1 
drm version :  libdrm-2.4.66  / commit : b38a4b2 
intel-driver : 1.6.2 / commit: 683edee
libva version : libva-1.6.2 / commit : 304bc13
mesa version : mesa-11.0.8 / commit : 261daab 
xf86-video-intel version : 2.99.917  / commit : baec802 
xserver version :xorg-server-1.18.0 / commit :7921764 

kernel drm-intel-testing:
------------------------------------------
commit 91587c722c28c4116dedbfbf08aa874377bc76f8
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Fri Dec 4 17:35:54 2015 +0100

    drm-intel-nightly: 2015y-12m-04d-16h-35m-07s UTC integration manifest

kernel version : 4.4.0-rc3
git url        : git://anongit.freedesktop.org/drm-intel
git branch     : drm-intel-testing
git describe   : drm-intel-next-2015-11-20-rebased-13721-g91587c7


Kernel without i915 loaded:
---------------------------------------------

commit b54e0f487f15213e85ca80c9ae9b953eb617cb42
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Thu Jan 7 11:52:35 2016 +0200
drm-intel-nightly: 2016y-01m-07d-09h-52m-06s UTC integration manifest

Created attachment 120911 [details]
dmesg.log

Created attachment 120912 [details]
Xorg.log

Ok if this occurs w/o i915 loaded we have a separate issue to fix up first before we can debug anything i915 specific like performance.

could you please help test S3 in a minimal Linux kernel(w/o graphic loaded), how about
append ‘init=/bin/bash nomodeset text’ in command line and test 
/sys/power/pm_test with (from right to left respectively):

[none] core processors platform devices freezer

thanks

(In reply to yu.c.chen@intel.com from comment #11)
> could you please help test S3 in a minimal Linux kernel(w/o graphic loaded),
> how about
> append ‘init=/bin/bash nomodeset text’ in command line and test 
> /sys/power/pm_test with (from right to left respectively):
> 
> [none] core processors platform devices freezer
> 
> thanks

Hi Yu :

I was not able to boot the BXT-P with the option "init=/bin/bash" because simply the DUT does not boot, it always stop in initramfs prompt, on the other hand i was able to boot with "nomodeset and text" options in the grub.
So once there i did one by one command and then i send the DUT to sleep (echo mem > /sys/power/state), and after each command the DUT returns (please see dmesg.log and test.log)

As well i have to comment that i've update the BIOS version and KSC of my BXT-P

Ifwi       : 119.10
Bxt SOC    : A0
KSC        : 1.06
GOP        : 10.0.1022
Board ID   : APL RVP 1A

as well i tried with the following configuration :

with default cmdline (without "text nomodeset")

Command  : echo freezer > /sys/power/pm_test
Behavior : after type the below command and send the DUT to S3 state with the command (echo mem > /sys/powe/state) looks like that the screen is frozen for a couple of seconds and then the screen come back normal, also i have to comment that the postcode remains the same '00AD'

Command  : echo devices > /sys/power/pm_test
Behavior : after type the below command and send the DUT to S3 state with the command (echo mem > /sys/powe/state) the screen goes black and returns after a few seconds and then come back normal, also i have to comment that the postcode remains the same '00AD'. 

Command  : echo platform > /sys/power/pm_test
Behavior : after type the below command and send the DUT to S3 state with the command (echo mem > /sys/powe/state) the DUT change its postcode to "0003" and the screen goes back for a few seconds and then come back normal.

Command  : echo processors > /sys/power/pm_test
Behavior : after type the below command and send the DUT to S3 state with the command (echo mem > /sys/powe/state) the screen goes black and returns after a few seconds and then come back normal, also the postcode remains the same as "0003"

Command  : echo core> /sys/power/pm_test
Behavior : after type the below command and send the DUT to S3 state with the command (echo mem > /sys/powe/state) the screen goes black and returns after a few seconds and then come back normal, also the postcode remains the same as "0003"

Please see the attachements :
dmesg2.log

Created attachment 121044 [details]
dmesg2.log

Let's close this for now here while we aren't sure it is a graphics related issue since we've seen it without i915 loaded.

*** This bug has been marked as a duplicate of bug 93684 ***