Bug 106467

Summary: CPU pipe A FIFO underrun (was able to workaround with enable_rc6=0)
Product: DRI Reporter: Nick Lewycky <nicholas>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED DUPLICATE QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: normal    
Priority: medium CC: imre.deak, intel-gfx-bugs, jani.nikula
Version: XOrg git   
Hardware: Other   
OS: Linux (All)   
Whiteboard: Triaged, ReadyForDev
i915 platform: KBL i915 features: display/watermark
Bug Depends on:    
Bug Blocks: 105980    
Attachments:
Description Flags
intel_reg dump
none
lshw --sanitize --numeric
none
lspci -vvvv
none
dmesg | egrep -i '(intel|i915|drm)'
none
xrandr --verbose
none
full dmesg but not drm-tip
none
full dmesg running drm-tip none

Description Nick Lewycky 2018-05-10 20:26:23 UTC
Created attachment 139475 [details]
intel_reg dump

Hi! I have a Lenovo Yoga 910 which required option enable_rc6=0 in order to avoid screen corruption. With my distro's most recent update, we've picked up https://patchwork.kernel.org/patch/9644631/ and now I get screen corruption again. Here's a link to a 10-second video of what my laptop screen looks like: 
  https://drive.google.com/file/d/1Z7_Ce0NUKZhB9mIXpFFaeuMwXsmroV9m/view?usp=sharing

The screen corruption is intermittent. Having the screen updating, such as moving the mouse or playing a video on youtube, reduces the effect but it still flickers about once per second. I haven't seen any corruption when at a VT.

I've tried other settings that exist in the i915 driver but have been unable to find a workaround.

The link to intel_reg_dumper guide is 404 from https://01.org/linuxgraphics/documentation/how-report-bugs. I ran 'intel_reg dump' and attached that.

I was unable to get the video bios dump using the guide at https://01.org/linuxgraphics/documentation/development/how-dump-video-bios . The 'cat' step produces an 'input/output error' with a zero-byte file. (If I try cat after echo 0, I get an invalid argument error instead.) Let me know if you want an strace of the cat. This also produced errors in dmesg, the echo produced "Invalid PCI ROM header signature: expecting 0xaa55, got 0xffff" while the cat produced "Invalid mmio detected during user access".

Debian unstable/sid, xserver-xorg-video-intel package is version "2:2.99.917+git20171229-1", kernel is "4.16.5-1 (2018-04-29)".
Comment 1 Nick Lewycky 2018-05-10 20:26:42 UTC
Created attachment 139476 [details]
lshw --sanitize --numeric
Comment 2 Nick Lewycky 2018-05-10 20:26:56 UTC
Created attachment 139477 [details]
lspci -vvvv
Comment 3 Nick Lewycky 2018-05-10 20:27:19 UTC
Created attachment 139478 [details]
dmesg | egrep -i '(intel|i915|drm)'
Comment 4 Nick Lewycky 2018-05-10 20:27:37 UTC
Created attachment 139479 [details]
xrandr --verbose
Comment 5 Jani Saarinen 2018-05-11 04:56:02 UTC
Please send dmesg without any greps with drm.debug=0x1e log_buf_len=4M?.
Can you also try using latest drm-tip: https://cgit.freedesktop.org/drm-tip and send dmesg with that too?
Comment 6 Nick Lewycky 2018-05-11 05:04:50 UTC
Created attachment 139487 [details]
full dmesg but not drm-tip
Comment 7 Nick Lewycky 2018-05-11 06:31:56 UTC
Created attachment 139488 [details]
full dmesg running drm-tip
Comment 8 Jani Nikula 2018-05-11 07:58:03 UTC
rc6 is a red herring. Try e.g.

https://bugs.freedesktop.org/show_bug.cgi?id=106335#c3
Comment 9 Nick Lewycky 2018-05-11 17:54:22 UTC
I'm afraid the workaround has no effect on the corruption. Tried both on distro kernel and at drm-tip.

I also looked for other 'pm_qos_latency_tolerance_us' and found them under  00:15.0, 00:15.1 and 00:15.3 (all of them I2C, not related to nvme). They were set to 'auto', setting them to '0' had no effect on the corruption.
Comment 10 Martin Monperrus 2018-05-11 20:33:00 UTC
Duplicate of https://bugs.freedesktop.org/show_bug.cgi?id=103229 ?
Comment 11 Jani Saarinen 2018-05-14 06:25:33 UTC
Looks like a dup, reported, you agree even this from KBL and that referred one from SKL.
Comment 12 Francesco Balestrieri 2018-06-13 13:02:26 UTC
Closing as duplicate, comment if you disagree

*** This bug has been marked as a duplicate of bug 103229 ***

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.