DVI output not visible on LCD with DVI, VGA output works -- (MergedFB on Ubuntu Edgy, Powerbook G4 ATI radeon 9600 M10)

[Martin Langhoff]

On 1/14/07, Martin Langhoff <martin.langhoff at gmail.com> wrote:
> Thanks for the hints. One of the weird things I found on friday is
> that if I get x.org started with my work LCD over DVI, and then head
> home and wake the laptop up with the home LCD... DVI just works.

Ok - some more interesting info.

After a no-ext-monitor attached restart, X.org has all the metamodes
set, but when I attach the work monitor and switch modes with the
"usual trick" that is

   xrandr -s mymode
   radeontool regset FP_GEN_CNTL 0x000344cd

the _work_ monitor (the one that usually worked) doesn't come up. So
I've captured the registers, diff'd them against the known-good set,
and found that re-setting TMDS_PLL_CNTL and TMDS_TRANSMITTER_CNTL made
it work again.

Here's the interesting part of the diff -u badregs goodregs:
 RE_TOP_LEFT (26c0)     0x00000000
 RE_WIDTH_HEIGHT (1c44) 0x00000000
-LVDS_PLL_CNTL (02d4)   0x00011286
-TMDS_PLL_CNTL (02a8)   0x1fbb01d2
-TMDS_TRANSMITTER_CNTL (02a4)   0x10000043
+LVDS_PLL_CNTL (02d4)   0x00091286
+TMDS_PLL_CNTL (02a8)   0x1fbb0155
+TMDS_TRANSMITTER_CNTL (02a4)   0x10000040
 ISYNC_CNTL (1724)      0x00000033

Full disclosure: I did mess about with other regs. These are the ones
that seemed to fix it. Here's all I touched according to history |
grep regset

  473  sudo /usr/local/bin/radeontool regset FP_GEN_CNTL 0x000344cd
  474  sudo /usr/local/bin/radeontool regset LVDS_GEN_CNTL 0x000dcea1
  475  sudo /usr/local/bin/radeontool regset TMDS_PLL_CNTL 0x1fbb0155
  476  sudo /usr/local/bin/radeontool regset TMDS_PLL_CNTL    0x1fbb0155
  477  sudo /usr/local/bin/radeontool regset TMDS_TRANSMITTER_CNTL    0x10000040

Full registers diff attached too. I finally see the light (and the
monitor signal too!)


martin



>
> A couple of X.org restarts later, the home LCD doesn't read the DVI
> signal any more.
>
> Now that I've seen that, I'll capture the logs and registers next time
> I do it, and diff them.
>
> > reduced blanking may help. You might also experiment with some
> > slightly tweaked modelines.
>
> Reduced blanking is on... and yes, I think I'm going down the path of modelines.
>
>
>
> martin
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: registers-good-vs-bad.diff
Type: text/x-patch
Size: 6692 bytes
Desc: not available
Url : http://lists.x.org/archives/xorg-driver-ati/attachments/20070118/8b6bc3d9/registers-good-vs-bad.bin