Bug 58701

Summary: [ALL]I-G-T/testdisplay && interlaced mode showing not full (DP overscan)
Product: DRI Reporter: yanbing <bingx.a.yan>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED WONTFIX QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: enhancement    
Priority: medium CC: bingx.a.yan, christophe.prigent, jinxianx.guo, lei.a.liu, przanoni, qingshuai.tian
Version: XOrg git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features: display/DP
Attachments:
Description Flags
1920x1080i not full - dmesg
none
1080i-notfull1 photograph
none
1080i-notfuul2 photograph none

Description yanbing 2012-12-24 06:52:04 UTC
Created attachment 72053 [details]
1920x1080i not full - dmesg

Environment:
--------------------------
Kernel: (drm-intel-next-queued)c0c36b941b6f0be6ac74f340040cbb29d6a0b06c
Some additional commit info:
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Wed Dec 19 16:08:43 2012 +0000

    drm/i915: Return the real error code from intel_set_mode()

Steps:
---------------------------

./testdisplay -o 4   (or 5)


Result:
--------------------------
1.DP mode show wrong size on screen.
2.It's not full,you can see it on the attachment and the dmesg-log.
Comment 1 yanbing 2012-12-24 06:59:14 UTC
Created attachment 72054 [details]
1080i-notfull1 photograph

modes as below:

CRTS(3):[0]  1920x1080i 50 1920 2448 2492 2640 1080 1084 1094 1125 0x15 0x40 74250

CRTS(3):[0]  1920x1080i 60 1920 2008 2052 2200 1080 1084 1094 1125 0x15 0x40 74250

The two modes were wrong. And you can see it on the photos with the attachment.
Comment 2 yanbing 2012-12-24 07:00:31 UTC
Created attachment 72055 [details]
1080i-notfuul2 photograph
Comment 3 yanbing 2012-12-24 07:13:57 UTC
The fix branch had the same problem.
Comment 4 Daniel Vetter 2013-01-07 16:42:14 UTC
Overscan on DP, I go nuts. Is this a regression? Also, is 1080i the only modes where this shows?
Comment 5 Daniel Vetter 2013-01-07 16:43:09 UTC
Also, does this happen with this exact monitor on other platforms with the DP output in the same way, or is there a difference?
Comment 6 yanbing 2013-01-08 02:07:13 UTC
(In reply to comment #4)
> Overscan on DP, I go nuts. Is this a regression? Also, is 1080i the only
> modes where this shows?

1.It's not a regression,the problem did almost exist.
2.Yes,just 1080i mode show not full.
Comment 7 yanbing 2013-01-08 02:09:09 UTC
(In reply to comment #5)
> Also, does this happen with this exact monitor on other platforms with the
> DP output in the same way, or is there a difference?

1.I had try on IVB&HSW,all platforms had this problem.
2.I had try on other monitors,also exist.
Comment 8 yanbing 2013-01-08 05:32:39 UTC
ALL platform,DP && HDMI,1920x1080i mode all exist this problem.
Comment 9 Daniel Vetter 2013-01-08 08:06:25 UTC
(In reply to comment #8)
> ALL platform,DP && HDMI,1920x1080i mode all exist this problem.

To confirm: For DP you have a DP output, DP cable and DP input on the monitor? For HDMI overscan is somewhat expected (we don't yet set the right infoframe attributes ...), but for native DP this is rather puzzling.
Comment 10 yanbing 2013-01-08 08:27:28 UTC
(In reply to comment #9)
> (In reply to comment #8)
> > ALL platform,DP && HDMI,1920x1080i mode all exist this problem.
> 
> To confirm: For DP you have a DP output, DP cable and DP input on the
> monitor? For HDMI overscan is somewhat expected (we don't yet set the right
> infoframe attributes ...), but for native DP this is rather puzzling.

reply confirm:DP output,cable and input on the monitor.
Comment 11 shui yangwei 2013-01-21 07:18:27 UTC
Environment:
--------------------
Kernel: (drm-intel-testing)2945111e30137965700f061f6f4bda1bc3ab0ecd
Some additional commit info:
Merge: e5c6537 b514407
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sun Jan 20 13:23:53 2013 +0100

Description:
--------------------
This issue also exists on HSW machine.

Not only 1920x1080i, the mode 1440x576i shows not full also.
Comment 12 yanbing 2013-01-22 05:50:48 UTC
Environment:
--------------------
Kernel: (drm-intel-testing)2945111e30137965700f061f6f4bda1bc3ab0ecd
Some additional commit info:
Merge: e5c6537 b514407
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sun Jan 20 13:23:53 2013 +0100

Description:
--------------------

Add the mode 1440x480i also shows not full .


CRTS(3):[0]  1440x480i 60 1440 1478 1602 1716 480 488 494 525 0x101a 0x40 27000
Comment 13 Ville Syrjala 2013-02-15 20:06:41 UTC
All of the "bad" modes mentioned here are what CEA-861 call CE video formats (as opposed to IT video formats).

CEA-861 also states that in case the source doesn't transmit the overscan/underscan information, the sink should assume CE formats are overscanned and IT formats underscanned.

So it sounds like the display is just following CEA-861 to the letter. Whether that makes any sense for a DP display is another matter. DP spec doesn't mentione anything about overscan.

A bit of googling indicated that people are having similar issues with this display on some other operating systems as well.

So yeah, sounds like we'd need to add infoframe support for DP so that we could force the sink to use underscan.
Comment 14 Guo Jinxian 2013-10-12 08:40:37 UTC
Test on kernel-3.12.0-rc3_drm-intel-testing_e5c753_20131005
(commit e5c7535c5bd7c98878aba8965015105e131c86f6)

This issue also exists on HSW and ILK machine.

Not only 1920x1080i, the mode 1440x576i shows not full also.
Comment 15 Guo Jinxian 2013-10-23 08:54:39 UTC
Test on kernel(drm-intel-testing)d1b2b826f0969182f055d11c991f90fdc6a4924a

This issue also exists on IVB and ILK machine.

Not only 1920x1080i, the mode 1440x576i and 1440x480i shows not full also.
Comment 16 Daniel Vetter 2013-10-27 19:20:11 UTC
I agree with Ville's assessment in comment #13 hence downgrading to a feature request.
Comment 17 Guo Jinxian 2014-06-06 02:33:16 UTC
We found not only 1920x1080i has this issue, but other interlaced modes (like 1440x576i 1440x480i ) also able to reproduce this.
Comment 18 Elio 2015-08-12 14:33:52 UTC
Closed after more than one year of inactivity. Feel free to reopen if needed. Thanks.

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.