[Bug 69960] New: Implement "fast" and "no" link training protocols for eDP/DP

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Mon Sep 30 05:18:24 PDT 2013


https://bugs.freedesktop.org/show_bug.cgi?id=69960

          Priority: medium
            Bug ID: 69960
                CC: intel-gfx-bugs at lists.freedesktop.org
          Assignee: intel-gfx-bugs at lists.freedesktop.org
           Summary: Implement "fast" and "no" link training protocols for
                    eDP/DP
        QA Contact: intel-gfx-bugs at lists.freedesktop.org
          Severity: enhancement
    Classification: Unclassified
                OS: Linux (All)
          Reporter: gt6 at gmx.net
          Hardware: Other
            Status: NEW
           Version: XOrg CVS
         Component: DRM/Intel
           Product: DRI

I own an Asus UX32VD which uses eDP for its panel and it takes Linux some 3-6
seconds to show anything on screen when booting, resuming from suspend or
changing resolutions. I describe the problem in bug 56086. Apparently, the
delay is due to problems with eDP link training.

However, especially for laptop panels which are always connected to the machine
and never change with regard to cable length and other variables, doing "full
link training" should not be necessary every time. Two alternatives, "fast link
training" and "no link training" are mentioned in many places on the web
(although a proper specification eludes me). Specifically, they usually say:

"System can be configured to use fast link training (simplified protocol) or no
link training (no training protocol, designed for fastest display re-enable
time)"

Apparently, it should be possible to use full link training only *once*,
remembering the values for amplitude/preemphasis/whatever else is involved, and
then re-using those values upon re-establishing the link on later occasions.

I know for a fact that on my exact laptop model, when resuming from suspend,
Windows 8.0 re-establishes the link and shows the desktop in less than 1
second, while my Linux takes about 3-6 seconds, and this is not due to X or
other factors, but likely due to slow link training.

I don't know specifics about the protocol, but apparently this shouldn't be too
difficult. After a "full" link training has been performed, the settings should
be stored and then tried when booting or resuming the next time the same
resolution is used on an internal panel.

It would be great if "fast" and "no" link training could be implemented because
at the moment, Linux falls behind other operating systems on laptops, which
resume almost instantly from suspend and offer a much nicer user experience in
this regard.

Thank you for your consideration.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20130930/79fcbc8d/attachment.html>


More information about the intel-gfx-bugs mailing list