Bug 1390297 - nouveau kms: nouveau should retry on display port link training errors
Summary: nouveau kms: nouveau should retry on display port link training errors
Keywords:
Status: CLOSED EOL
Alias: None
Product: Fedora
Classification: Fedora
Component: xorg-x11-drv-nouveau
Version: 25
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Ben Skeggs
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
: 1390274 (view as bug list)
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2016-10-31 16:04 UTC by Hans de Goede
Modified: 2017-12-12 10:44 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2017-12-12 10:44:21 UTC
Type: Bug


Attachments (Terms of Use)
0001-i2c-aux-g94-retry-transactions-after-hw-reports-an-e.patch (4.20 KB, patch)
2016-11-11 10:43 UTC, Hans de Goede
no flags Details | Diff
0002-i2c-gk110b-gm10x-use-the-correct-implementation.patch (1.63 KB, patch)
2016-11-11 10:43 UTC, Hans de Goede
no flags Details | Diff
0003-drm-nouveau-Intercept-ACPI_VIDEO_NOTIFY_PROBE.patch (5.44 KB, patch)
2016-11-11 10:43 UTC, Hans de Goede
no flags Details | Diff

Description Hans de Goede 2016-10-31 16:04:17 UTC
While testing display port (both the mini dp connector as well as dp over usb-c) on a Lenovo P50 I'm seeing DP training errors, these seem to happen when changing the mode on the monitor.

Initially plugging in the monitor works, then after a suspend resume I hit bug 1390274, and the switch to the mirror mode causes displaylink training errors, this might be related to the mode not being supported by the monitor ?

The training errors seem to cause monitor unplug + plug events, at which point the EDID race described in bug 1390274 hits and g-s-d goes into a loop. Sometimes during the loop it seems to get the right EDID and briefly the DP external monitor comes up in its native resolution, but the loop continues (I think g-s-d get behind processing events?).

If g-s-d manages to gets to a stable situation it is the mirror mode at a resolution not actually supported by the monitor (as described in bug 1390274). This happens sometimes (usually after a suspend resume) if I then try to manually fixup the config, it quickly goes into the loop again and typically it will not exit the reconfigure loop unless I unplug the monitor.

I think that nouveau needs to be changed to retry training a couple of times before reporting an unplug to userspace.

Comment 1 Hans de Goede 2016-11-11 10:30:48 UTC
*** Bug 1390274 has been marked as a duplicate of this bug. ***

Comment 2 Hans de Goede 2016-11-11 10:42:22 UTC
Ok, so the real issue here is that we are getting spurious KEY_SWITCHVIDEOMODE keypress events on hotplug and on modeset, causing the g-s-d loop.

So I started with a patch adding an ACPI notifier callback intercepting ACPI_VIDEO_NOTIFY_PROBE events stopping the KEY_SWITCHVIDEOMODE keypress events, breaking the loop, but that breaks hotplug detect
when the Nvidia GPU is runtime suspended.

There are 2 issues here:

1) Some bugs fixed in drm-next which are needed for hotplug to work on Maxwell GPUs in general
2) The GPU not seeing hotplug events when runtime suspended even with bugs from 1) fixed.

To fix 2. I added code to wake the GPU from runtime suspend on ACPI_VIDEO_NOTIFY_PROBE events, this does correctly wake the GPU up, but it does not see the hotplug events (likely because the interrupts triggered while it was suspended).

So next revision I added code to explicitly re-check the status of all connectors on ACPI_VIDEO_NOTIFY_PROBE events, which fixes the hotplug issues for real while also stopping the spurious KEY_SWITCHVIDEOMODE keypress events which were causing both the g-s-d loop as well as bug 1390274.

I'm attaching 3 patches for future reference here, 2 bug-fixes already in drm-next and the final version of the ACPI_VIDEO_NOTIFY_PROBE handling patch.

Comment 3 Hans de Goede 2016-11-11 10:43:04 UTC
Created attachment 1219700 [details]
0001-i2c-aux-g94-retry-transactions-after-hw-reports-an-e.patch

Comment 4 Hans de Goede 2016-11-11 10:43:32 UTC
Created attachment 1219701 [details]
0002-i2c-gk110b-gm10x-use-the-correct-implementation.patch

Comment 5 Hans de Goede 2016-11-11 10:43:49 UTC
Created attachment 1219702 [details]
0003-drm-nouveau-Intercept-ACPI_VIDEO_NOTIFY_PROBE.patch

Comment 6 Fedora End Of Life 2017-11-16 18:43:06 UTC
This message is a reminder that Fedora 25 is nearing its end of life.
Approximately 4 (four) weeks from now Fedora will stop maintaining
and issuing updates for Fedora 25. It is Fedora's policy to close all
bug reports from releases that are no longer maintained. At that time
this bug will be closed as EOL if it remains open with a Fedora  'version'
of '25'.

Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version'
to a later Fedora version.

Thank you for reporting this issue and we are sorry that we were not
able to fix it before Fedora 25 is end of life. If you would still like
to see this bug fixed and are able to reproduce it against a later version
of Fedora, you are encouraged  change the 'version' to a later Fedora
version prior this bug is closed as described in the policy above.

Although we aim to fix as many bugs as possible during every release's
lifetime, sometimes those efforts are overtaken by events. Often a
more recent Fedora release includes newer upstream software that fixes
bugs or makes them obsolete.

Comment 7 Fedora End Of Life 2017-12-12 10:44:21 UTC
Fedora 25 changed to end-of-life (EOL) status on 2017-12-12. Fedora 25 is
no longer maintained, which means that it will not receive any further
security or bug fix updates. As a result we are closing this bug.

If you can reproduce this bug against a currently maintained version of
Fedora please feel free to reopen this bug against that version. If you
are unable to reopen this bug, please file a new report against the
current release. If you experience problems, please add a comment to this
bug.

Thank you for reporting this bug and we are sorry it could not be fixed.


Note You need to log in before you can comment on or make changes to this bug.