Bug 17125 - PLLs chosen that cause flicker on dual-link DVI monitor.
Summary: PLLs chosen that cause flicker on dual-link DVI monitor.
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-08-13 16:17 UTC by Chris Ball
Modified: 2009-04-14 07:45 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments

Description Chris Ball 2008-08-13 16:17:32 UTC
Hi,

Radeon commit 72feaa37ea07620f5f2ead438dbc72a1c8883cd3 ("prefer lower ref dividers..") causes flicker here, on a Radeon HD 3450 connected to an Apple 30" cinema display.  The PLL settings are different before and after the patch, so that's a likely culprit.

Before (working):  

http://dev.laptop.org/~cjb/xorg-noflicker.log

After (broken):

http://dev.laptop.org/~cjb/xorg-flicker.log

In particular, before:

best_freq: 268000000
best_feedback_div: 1072
best_ref_div: 27
best_post_div: 4
(II) RADEON(0): crtc(0) PLL  : refdiv 27, fbdiv 0x430(1072), pdiv 4

and after:

best_freq: 268000000
best_feedback_div: 268
best_ref_div: 9
best_post_div: 3
(II) RADEON(0): crtc(0) PLL  : refdiv 9, fbdiv 0x10C(268), pdiv 3
Comment 1 Chris Ball 2008-08-13 16:58:53 UTC
Alex and I tried some PLL changes that had different results.

Broken (post-patch):  268/9/3
Fixed (pre-patch):  1072/27/4
Almost working, still occasional flicker:  804/27/3

The 804/27/3 was reached via:

<agd54> try changing line 232 of radeon_crtc.c
<agd5f> change the '<' to '>'

} else if ((flags & RADEON_PLL_PREFER_LOW_REF_DIV) && (ref_div < best_ref_div)) {
Comment 2 Alex Deucher 2008-08-14 11:52:26 UTC
fixed in git:
92ee21df344a989778e37369c7beb3904a00ead6
Comment 3 Chris Ball 2008-08-14 12:59:51 UTC
> fixed in git:
> 92ee21df344a989778e37369c7beb3904a00ead6

This gets me back to 804/27/3, which means that I still have the flicker regression behavior.  Alex says he'll fix up the patch, and I'll retest then.
Comment 4 Chris Ball 2008-08-14 13:47:42 UTC
Mea culpa, stale driver.  This correctly finds 1072/27/4, closing.
Comment 5 Ian Pilcher 2009-04-14 06:18:04 UTC
I'm seeing something that may be the "occasional flicker" behavior.  I
suspect it may be PLL-related, because I am able to fix it with the
radeonhd driver by using the "AtomBIOS" "pll=on" option.

How can I determine if the driver version in Fedora 10 includes this
fix?

Thanks!
Comment 6 Alex Deucher 2009-04-14 07:45:16 UTC
(In reply to comment #5)
> I'm seeing something that may be the "occasional flicker" behavior.  I
> suspect it may be PLL-related, because I am able to fix it with the
> radeonhd driver by using the "AtomBIOS" "pll=on" option.
> 
> How can I determine if the driver version in Fedora 10 includes this
> fix?

Do you have a 2560x1600 monitor?  This bug and fix really only relates to high dot clocks.  If you are using a different monitor please file a new bug and attach your xorg log and config.


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.