Bug 111623 - No package C-states on Dell XPS 15 9570 after resume from deep sleep
Summary: No package C-states on Dell XPS 15 9570 after resume from deep sleep
Status: NEEDINFO
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: DRI git
Hardware: x86-64 (AMD64) Linux (All)
: not set not set
Assignee: sujaritha.sundaresan
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-09-10 05:25 UTC by s.zharkoff
Modified: 2019-09-17 05:39 UTC (History)
1 user (show)

See Also:
i915 platform: KBL
i915 features: firmware/guc, firmware/huc, power/runtime PM, power/suspend-resume


Attachments
dmesg before resume (114.36 KB, text/plain)
2019-09-10 06:17 UTC, s.zharkoff
no flags Details
dmesg after resume (114.06 KB, text/plain)
2019-09-10 06:17 UTC, s.zharkoff
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description s.zharkoff 2019-09-10 05:25:28 UTC

    
Comment 1 s.zharkoff 2019-09-10 05:29:49 UTC
Movig the bug from kernel bugzilla
https://bugzilla.kernel.org/show_bug.cgi?id=204751

Kernel 5.2 series - suspend, resume - package enter c-states up to C10.
Kernel 5.3 starting from 5.3-rc1 till latest drm-tip builds - states up to C10 before suspend, no package C-states after resume. 

Zhang Rui:

before suspend
CPU%c7	CoreTmp	PkgTmp	GFX%rc6	GFXMHz	Totl%C0	Any%C0	GFX%C0
99.40	38	38	99.85	350	1.29	0.79	0.14
after resume
CPU%c7	CoreTmp	PkgTmp	GFX%rc6	GFXMHz	Totl%C0	Any%C0	GFX%C0
99.49	39	39	1.32	350	1.21	1.00	95.04

It is graphics that blocks cpu from entering package c-state.
Please file a bug at freedesktop.org and attach the full dmesg output with kernel parameter drm.debug=0xe, after the problem reproduced.
Comment 2 Lakshmi 2019-09-10 05:52:16 UTC
Is it a duplicate of Bug 111594?
Comment 3 s.zharkoff 2019-09-10 06:17:12 UTC
Created attachment 145312 [details]
dmesg before resume
Comment 4 s.zharkoff 2019-09-10 06:17:41 UTC
Created attachment 145313 [details]
dmesg after resume
Comment 5 s.zharkoff 2019-09-10 06:27:54 UTC
(In reply to Lakshmi from comment #2)
> Is it a duplicate of Bug 111594?

could be, have not tried to do dpms off - for me it is on suspend resume. (In reply to Lakshmi from comment #2)
> Is it a duplicate of Bug 111594?

For me xset dpms force off works fine, C-states are still available until suspend-resume.
Comment 6 Jon Ewins 2019-09-16 15:44:11 UTC
Can you please retry with enable_guc=0 set in the module parameters.  That will disable the use of the GuC and HuC units.  That was seen to help in the case of Bug 111594.
Comment 7 s.zharkoff 2019-09-17 05:39:41 UTC
I've tried guc=0 for 5.3 release. Yes - the c-states are there, but before suspend my laptop spending about 70% in C10. After resume C8 is the lowest state it can reach.  So seems that guc is involved in this, but it is not a root cause...


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.