Bug 6421

Summary: XF86VidModeGetModeLine reports bogus dot clock which breaks mythtv
Product: xorg Reporter: greg <gsstark>
Component: Driver/RadeonAssignee: Xorg Project Team <xorg-team>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: high CC: dns
Version: 6.9.0   
Hardware: Other   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Test program none

Description greg 2006-03-29 03:47:06 UTC 
The symptom is that mythtv can't play video smoothly, instead it plays for a 
few frames then stops for about a second. Then resumes and then stops...

The cause I believe is that it's trying to sync to a bogus refresh rate it's 
calculating based on bogus data returned by XF86VidModeGetModeLine.

Also, in XVidTune it says "Pixel Clock (Mhz): 1976225.70" which of course is 
ridiculous. If my graphics card could handle a 1.9 petahertz signal it would be 
doing something more exciting than driving my monitors.

I wrote a small C program to test the output of XF86VidModeGetModeLine and this 
is the result it output:

dot_clock: 1976225695
 hdisplay: 3520
 hsyncstart: 3568
 hsyncend: 3600
 htotal: 3680
 vdisplay: 1200
 vsyncstart: 1203
 vsyncend: 1209
 vtotal: 1235
 flags: 80000000
 privsize: 0
rate: 434832.268747


And mythtv has trouble keeping up with a 434khz frame rate...
Comment 1 greg 2006-03-29 03:47:53 UTC 
Created attachment 5105 [details]
Test program
Comment 2 Adam Jackson 2006-04-04 06:23:03 UTC 
it looks like the modeline struct is just never getting filled in properly.  can
you attach your config file and log so I can figure out what path this is hitting?
Comment 3 Michel Dänzer 2006-04-04 16:02:46 UTC 
I think this is caused by MergedFB, see corresponding thread on the xorg list.
Comment 4 Roland Scheidegger 2006-12-02 17:28:23 UTC 
This should be fixed by e742aeb28c7d9d6e75932c408bcc7c44af52e303 I assume. If
not reopen.

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.