Bug 94706

Summary: [APL] UI is very slow
Product: DRI Reporter: cprigent <christophe.prigent>
Component: DRM/IntelAssignee: Imre Deak <imre.deak>
Status: CLOSED NOTOURBUG QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: blocker    
Priority: highest CC: elio.martinez.monroy, intel-gfx-bugs
Version: unspecified   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
i915 platform: BXT i915 features: display/Other
Description Flags
kern.log when booting with acpi off
kern.log when booting
config-4.5.0-nightly+ none

Description cprigent 2016-03-25 18:09:03 UTC
Created attachment 122567 [details]
kern.log when booting with acpi off

Platform: Broxton P A0 Platform 
CPU Name : Intel(R) @ 1.2 GHz (family: 6, model: 92, stepping: 8) – 4 cores
CRB : Apollo Lake RVPC1 Fab1
Mandatory Reworks : R14 , R06C and R16
KSC: 1.05
Linux distribution: Ubuntu 15.10 64 bits
Kernel: drm-intel-nightly 4.5.0 e7a7673 from http://cgit.freedesktop.org/drm-intel/
  commit e7a7673e9840fe8b50a5a2894c75565ec7858a00
  Author: Daniel Vetter <daniel.vetter@ffwll.ch>
  Date:   Sat Mar 19 11:10:45 2016 +0100
  drm-intel-nightly: 2016y-03m-19d-10h-09m-53s UTC integration manifest
libdrm-2.4.67-7 49041c3 from git://anongit.freedesktop.org/mesa/drm
mesa: mesa-11.1.2 7bcd827 from git://anongit.freedesktop.org/mesa/mesa
cairo: 1.15.2 db8a7f1 from git://anongit.freedesktop.org/cairo
waffle: bb29b2a from git://github.com/waffle-gl/waffle
util-macros-1.19.0-2 d7acec2 from git://git.freedesktop.org/git/xorg/util/macros
xorg-server-1.18.1 23e60f9 from git://git.freedesktop.org/git/xorg/xserver
xf86-video-intel 2.99.917-560 d167280 from git://anongit.freedesktop.org/xorg/driver/xf86-video-intel
libva-1.7.0.pre1 35f1447 from git://git.freedesktop.org/git/vaapi/libva
vaapi-intel-driver 1.7.0.pre1 ab8e4e3 from git://anongit.freedesktop.org/vaapi/intel-driver
DMC 1.4 from: https://01.org/linuxgraphics/intel-linux-graphics-firmwares
intel-gpu-tools-1.14 c19b049 from http://anongit.freedesktop.org/git/xorg/app/intel-gpu-tools.git

1. Boot in text mode
2. Execute some basic commands like change directory with tabulation

Actual result:
2. Keyboard input needs delay and auto-completion needs sometimes 2 seconds

Expected result:
2. Console respond normally

We have abnormal numbers of failures in IGT (150 regressions), some basic tests frequently fails when executed in a loop
Reproduced via ssh (login timeouts or needs also delay)
Not reproduced with kernel boot command line: acpi=off
Comment 1 cprigent 2016-03-25 18:17:46 UTC
Created attachment 122568 [details]
kern.log when booting

Boot is abnormally long
UI is very slow
Comment 2 cprigent 2016-03-25 18:23:38 UTC
Actual result when launching glxgears:
Comment 3 cprigent 2016-03-29 09:15:42 UTC
Created attachment 122603 [details]
Comment 4 cprigent 2016-03-29 09:15:58 UTC
Created attachment 122604 [details]
Comment 5 cprigent 2016-03-29 09:16:23 UTC
Created attachment 122605 [details]
Comment 6 cprigent 2016-03-29 10:06:31 UTC
Reproduced with kernel 4.5.0 f5d413c from nightly:

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

Set priority and severity to maxium
Comment 7 Imre Deak 2016-03-30 21:30:07 UTC
*** Bug 93797 has been marked as a duplicate of this bug. ***
Comment 8 Imre Deak 2016-03-30 21:37:31 UTC
See bug 93797 for a workaround for this. I don't think this is a graphics issue, but a problem related to system power states, since there is a bad interrupt latency affecting other devices too (keyboard, network). Based on this I'd like someone from the core team to look at this.
Comment 9 cprigent 2016-03-31 09:42:28 UTC
I confirm it is not reproduced with kernel boot command line: processor.max_cstate=2
Comment 10 Imre Deak 2016-04-19 20:30:56 UTC
This issue has been confirmed to be a problem on early steppings to be fixed in C stepping. There is a workaround patch available and on A stepping at least we can boot limiting CPU c states to work around the problem. As this isn't a graphics related issue and the relevant people are already aware of it, I'm closing this ticket.

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.