[Bug 100215] [SNB][IVB][BAT] kms_cursor_legacy at basic-busy-flip-before-cursor-atomic fails in CI

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=100215

--- Comment #9 from krisman at collabora.co.uk ---
(In reply to Chris Wilson from comment #2)
> Both are due to a prolonged hpd event interrupting userspace, e.g.:
> 
> [  337.504999] [drm:intel_dp_aux_ch [i915]] dp_aux_ch timeout status
> 0x7145003f
> [  337.505013] [drm:drm_dp_dpcd_access] Too many retries, giving up. First
> error: -110

I think this happens because more than one connector is sharing the same HPD
line with the DP port.  In the system where I could reproduce the issue,
whenever I connect an HDMI cable, it got successfully detected, but the kernel
also tried to run the DP detection code in sequence, triggering the timeouts. 
Since HDMI and DP were both initialized to port B, the HPD pin is raised by
connecting the HDMI cable, and intel_digital_port_connected() doesn't have a
chance to abort the DP detection attempt.

My patch aborts the hotplug detection of further connectors when the first
attached connector of a given port is detected.  This prevents the DP timeout I
mentioned when plugging HDMI and I think is backed up by the specification[1],
in page 86, since it says that no devices sharing the same pins can be enabled
simultaneously.

While this prevents the timeouts, I am not sure it fixes the issue 100%,
because it is hard to reproduce in my system.  I'll keep trying and in the
meantime, I'll send the patch as an RFC to the list.

[1]
https://01.org/sites/default/files/documentation/snb_ihd_os_vol3_part3_0.pdf

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20170418/6046fa16/attachment.html>