Bug 80896

Summary: [BDW Regression]dmesg error "<3>[ 1.954836] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!" after boot
Product: DRI Reporter: Guo Jinxian <jinxianx.guo>
Component: DRM/IntelAssignee: Dima Ryazanov <dima>
Status: CLOSED FIXED QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: major    
Priority: low CC: christophe.prigent, huax.lu, hvbakel, intel-gfx-bugs, james.ausmus, mail, n.schnelle, przanoni, tobias.gerschner, tomboy64
Version: unspecifiedKeywords: bisected
Hardware: Other   
OS: All   
See Also: https://bugs.freedesktop.org/show_bug.cgi?id=90058
Whiteboard:
i915 platform: BDW i915 features: display/Other
Attachments:
Description Flags
dmesg
none
dmesg
none
dmesg(14bf99)
none
dmesg
none
dmesg output, post event
none
master control interrupt lied event
none
The master control interrupt lied none

Description Guo Jinxian 2014-07-04 06:45:35 UTC
Created attachment 102239 [details]
dmesg

==System Environment==
--------------------------
Regression: Yes. 
Good commit on -next-queued: b6fdd0f2b990006daba19eec676b632faa523fc8

Non-working platforms: BDW

==kernel==
--------------------------
origin/drm-intel-nightly: eb638c7fabe97a9df752aeb2f59a9463ce4aed8e(fails)
    drm-intel-nightly: 2014y-07m-03d-14h-50m-16s integration manifest
origin/drm-intel-next-queued: 5e59f7175f96550ede91f58d267d2b551cb6fbba(fails)
    drm/i915: Try harder to get FBC  
origin/drm-intel-fixes: 5549d25f642a7e6cfb8744d0031a9da404f696d6(works)
    drm/i915: Drop early VLV WA to fix Voltage not getting dropped to Vmin

==Bug detailed description==
dmesg error "<3>[    1.954836] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!" after boot.

Output:
[root@x-bdw01 ~]# dmesg -r | egrep "<[1-3]>" |grep drm
<3>[ 1.955683] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!
<3>[ 1.955848] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!
<3>[ 3.683649] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!
<3>[ 3.686870] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!


==Reproduce steps==
---------------------------- 
1.  boot
2.  dmesg -r | egrep "<[1-3]>" |grep drm
Comment 1 Paulo Zanoni 2014-07-04 12:09:10 UTC
Can you please bisect this? Which one is the bad commit?
Comment 2 Damien Lespiau 2014-07-04 15:43:29 UTC
No need to bisect, the error message was introduced by Oscar in

commit 38cc46d73ed99dd7002f1406002e52d7975d16cc
Author: Oscar Mateo <oscar.mateo@intel.com>
Date:   Mon Jun 16 16:10:59 2014 +0100

    drm/i915/bdw: Ack interrupts before handling them (GEN8)

And for those reason bit 23 of master ctl is set and the SDE IIR is 0.
Comment 3 Guo Jinxian 2014-07-14 01:40:02 UTC
Created attachment 102730 [details]
dmesg

(In reply to comment #2)
> No need to bisect, the error message was introduced by Oscar in
> 
> commit 38cc46d73ed99dd7002f1406002e52d7975d16cc
> Author: Oscar Mateo <oscar.mateo@intel.com>
> Date:   Mon Jun 16 16:10:59 2014 +0100
> 
>     drm/i915/bdw: Ack interrupts before handling them (GEN8)
> 
> And for those reason bit 23 of master ctl is set and the SDE IIR is 0.
Revert this commit on latest -next-queued´╝îerror "<3>[ 1.955683] [drm:gen8_irq_handler] *ERROR* The master control interrupt lied (SDE)!" unable to reproduce, but here has another error. Please check the output below. Need we report a new bug for this?

Output:
[root@x-bdw01 ~]# dmesg -r | egrep "<[1-3]>" |grep drm
<3>[    2.753243] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.768133] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.780219] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.792092] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.804271] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.813306] [drm:intel_dp_start_link_train] *ERROR* failed to enable link training
<3>[    2.831054] [drm:intel_dp_start_link_train] *ERROR* failed to update link training
<3>[    2.832254] [drm:intel_dp_complete_link_train] *ERROR* failed to train DP, aborting
Comment 4 Paulo Zanoni 2014-08-11 18:32:37 UTC
Hi

It seems you're booting with eDP+HDMI. I can't reproduce this with eDP+HDMI or just eDP. Can you please confirm whether this only happens when you boot with DP (or DP+something) attached?

Thanks,
Paulo
Comment 5 Paulo Zanoni 2014-08-11 18:37:42 UTC
(In reply to comment #4)
> Hi
> 
> It seems you're booting with eDP+HDMI. I can't reproduce this with eDP+HDMI

What I wanted to say was: "It seems you're booting with eDP+DP."

> or just eDP. Can you please confirm whether this only happens when you boot
> with DP (or DP+something) attached?
> 
> Thanks,
> Paulo
Comment 6 lu hua 2014-08-12 08:30:06 UTC
Created attachment 104474 [details]
dmesg(14bf99)

It doesn't happen on latest -queued and -nightly kernel.I guess someone fixed it.
Comment 7 Paulo Zanoni 2014-08-12 12:37:49 UTC
(In reply to comment #6)
> Created attachment 104474 [details]
> dmesg(14bf99)
> 
> It doesn't happen on latest -queued and -nightly kernel.I guess someone
> fixed it.

Closing bug then. If it still happens, please reopen.
Comment 8 Guo Jinxian 2014-08-13 06:11:22 UTC
Verified on latest -nightly(723dbbfb6ce02b5662c0c4cac9be3bcdfbb11d35)

[root@x-bdw01 ~]# dmesg -r | egrep "<[1-3]>" |grep drm
[root@x-bdw01 ~]#
Comment 9 Dima Ryazanov 2015-05-19 20:33:49 UTC
Created attachment 115902 [details]
dmesg

I just saw this error while running the 4.0.3 kernel (4.0.3-300.fc22.x86_64).
Comment 10 Tobias Gerschner 2015-05-28 19:26:45 UTC
Created attachment 116126 [details]
dmesg output, post event
Comment 11 Tobias Gerschner 2015-05-28 19:30:33 UTC
Comment on attachment 116126 [details]
dmesg output, post event

This happens on a new NUC, with

CPU0: Intel(R) Core(TM) i7-5557U CPU @ 3.10GHz (fam: 06, model: 3d, stepping: 04)

I'm running a 4K display off the miniDP port.

It happens almost every single time I walk away from the computer within 10-15 minutes.
Comment 12 Jani Nikula 2015-08-18 14:18:51 UTC
*** Bug 90058 has been marked as a duplicate of this bug. ***
Comment 13 M.B. 2015-08-22 10:09:42 UTC
Created attachment 117855 [details]
master control  interrupt lied event

Not exactly sure what I'm looking at here. Using live infrastructure (libdrm, mesa, xorg-server, xf86-video-intel, all -9999) on gentoo.

Linux hatshepsut 4.1.6-gentoo #1 SMP Thu Aug 20 13:12:11 CEST 2015 x86_64 Intel(R) Core(TM) i5-5200U CPU @ 2.20GHz GenuineIntel GNU/Linux

If you need additional info, please let me know.
Comment 14 Kim Carlsen 2015-10-12 10:22:37 UTC
Created attachment 118836 [details]
The master control interrupt lied

kernel output on a 
Lenovo X1 Carbon (3rd gen) connected to a lenovo onelink pro dock
Comment 15 Kim Carlsen 2015-10-12 10:28:29 UTC
(In reply to Kim Carlsen from comment #14)
> Created attachment 118836 [details]
> The master control interrupt lied
> 
> kernel output on a 
> Lenovo X1 Carbon (3rd gen) connected to a lenovo onelink pro dock

Outputting 2560x1440@60 works fine, outputting 3840x2160@60 produces no picture on external monitor. 

When removing the onelink prodock and connecting the dp cable directly to the laptop then all resolutions works (3840x2160@60).

Whenever running xrandr it adds 36 lines to kernel log of 'The master control interrupt lied'
Comment 16 Kim Carlsen 2015-10-12 10:32:51 UTC
When 4k resolution is working, these error message is still present in kernel.log. Maybe resolution issue is unrelated to IRQ issue
Comment 17 Jani Nikula 2015-10-12 14:24:58 UTC
(In reply to Kim Carlsen from comment #16)
> When 4k resolution is working, these error message is still present in
> kernel.log. Maybe resolution issue is unrelated to IRQ issue

Please file that as a separate bug to not conflate this one, thanks.
Comment 18 Daniel Vetter 2015-10-23 13:27:39 UTC
commit 97e5ed1111dcc5300a0f59a55248cd243937a8ab
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Fri Oct 23 10:56:12 2015 +0200

    drm/i915: shut up gen8+ SDE irq dmesg noise

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.