92935 – Warnings on suspend/resume: CSR SSP Base Not fine, CSR HTP Not fine, DC6 already programmed to be disabled

Bug 92935 - Warnings on suspend/resume: CSR SSP Base Not fine, CSR HTP Not fine, DC6 already programmed to be disabled

Summary: Warnings on suspend/resume: CSR SSP Base Not fine, CSR HTP Not fine, DC6 alre...

Status:	CLOSED FIXED

Alias:	None

Product:	DRI
Classification:	Unclassified
Component:	DRM/Intel (show other bugs)
Version:	DRI git
Hardware:	Other All

Importance:	medium normal
Assignee:	Intel GFX Bugs mailing list
QA Contact:	Intel GFX Bugs mailing list

URL:
Whiteboard:
Keywords:

Depends on:
Blocks:

Reported:	2015-11-13 10:43 UTC by Timur Alperovich
Modified:	2017-07-24 22:44 UTC (History)
CC List:	1 user (show)

See Also:
i915 platform:	SKL
i915 features:	power/suspend-resume

Attachments
dmesg output (78.63 KB, application/x-gzip) 2015-11-13 10:43 UTC, Timur Alperovich	no flags	Details
fix incorrect assert about DC6 not being enabled (1.62 KB, patch) 2015-11-13 12:56 UTC, Imre Deak	no flags	Details \| Splinter Review
fix disable_power_well module option (4.41 KB, text/plain) 2015-11-13 12:56 UTC, Imre Deak	no flags	Details
don't enable DC6 during system suspend (1.21 KB, text/plain) 2015-11-13 12:57 UTC, Imre Deak	no flags	Details
fix disable_power_well module option v2 (4.90 KB, text/plain) 2015-11-13 13:01 UTC, Imre Deak	no flags	Details
dmesg output after applying the patches (45.56 KB, text/plain) 2015-11-13 19:36 UTC, Timur Alperovich	no flags	Details
dmesg output after applying the patches (45.56 KB, application/gzip) 2015-11-13 19:37 UTC, Timur Alperovich	no flags	Details
Show Obsolete (2) View All

Description Timur Alperovich 2015-11-13 10:43:53 UTC

Created attachment 119626 [details]
dmesg output

I noticed the following error in dmesg when suspending/resuming:
[  264.289328] Trying to free nonexistent resource <000000000000e000-000000000000e0ff>
[  264.289489] PM: suspend of devices complete after 163.064 msecs
[  264.289896] ------------[ cut here ]------------
[  264.289941] WARNING: CPU: 0 PID: 1814 at drivers/gpu/drm/i915/intel_runtime_pm.c:465 skl_enable_dc6+0x1dc/0x2d0 [i915]()
[  264.289943] CSR SSP Base Not fine
[  264.290316] WARNING: CPU: 0 PID: 1814 at drivers/gpu/drm/i915/intel_runtime_pm.c:466 skl_enable_dc6+0x20d/0x2d0 [i915]()
[  264.290317] CSR HTP Not fine
[  265.329897] WARNING: CPU: 1 PID: 1864 at drivers/gpu/drm/i915/intel_runtime_pm.c:558 skl_disable_dc6+0x93/0xc0 [i915]()
[  265.329898] DC6 already programmed to be disabled.

distribution: Debian (testing)
uname -m: x86_64
uname -r: drm-intel-nightly build; the last commit is: 4c2531304c0a
model: Dell XPS 9350 (i7-6500U)
no external display connected

dmesg debug output attached.

Comment 1 Imre Deak 2015-11-13 12:56:17 UTC

Created attachment 119635 [details] [review]
fix incorrect assert about DC6 not being enabled

Comment 2 Imre Deak 2015-11-13 12:56:48 UTC

Created attachment 119636 [details]
fix disable_power_well module option

Comment 3 Imre Deak 2015-11-13 12:57:24 UTC

Created attachment 119637 [details]
don't enable DC6 during system suspend

Comment 4 Imre Deak 2015-11-13 12:58:46 UTC

Could you try if the 3 attached patch fixes the problem? We are working on a final form of these changes to backport, but your feedback would be useful until that.

Comment 5 Imre Deak 2015-11-13 13:01:44 UTC

Created attachment 119638 [details]
fix disable_power_well module option v2

Comment 6 Timur Alperovich 2015-11-13 19:36:00 UTC

I applied the attached patches and the errors above do not appear on suspend/resume.

I did notice one more message on startup and after resume:
[   24.695870] i915 0000:00:02.0: BAR 6: [??? 0x00000000 flags 0x2] has bogus alignment
[...]
[   33.551325] i915 0000:00:02.0: BAR 6: [??? 0x00000000 flags 0x2] has bogus alignment

I don't know if this is related. The patches seem to address the errors reported in the initial bug report. I attached the dmesg output from suspend/resume with the applied patches in case that's useful.

Comment 7 Timur Alperovich 2015-11-13 19:36:30 UTC

Created attachment 119652 [details]
dmesg output after applying the patches

Comment 8 Timur Alperovich 2015-11-13 19:37:47 UTC

Created attachment 119653 [details]
dmesg output after applying the patches

Comment 9 Imre Deak 2015-11-13 21:30:30 UTC

Thanks for testing it. For drm-intel-nightly we are working on a more general solution to fix the display power well functionality which will also get rid of these warnings. For 4.3 we will follow-up with a backport fix based on the patches attached here.

The "PCI bar has bogus alignment" looks like a separate issue, something to do with PCI core resource enumeration. I found the following about it:
https://lkml.org/lkml/2015/4/21/581

Could you ask Bjorn about the status of this, or file a PCI core bug?

Comment 10 Timur Alperovich 2015-11-13 21:32:15 UTC

Thanks for the quick turnaround on this!

I will follow up with the PCI folks about the alignment issue. I did notice the same LKML thread when I was looking last night, but couldn't figure out where it lands. Thanks for pointing me in the right direction!

Comment 11 Imre Deak 2016-03-08 11:13:03 UTC

The fix was partly backported to 4.3, but the
"DC6 already programmed to be disabled"
WARN can still trigger incorrectly. Since the 4.3 stable kernel is closed already, please upgrade to 4.4 or newer if you want to get rid of this message.

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.