[Bug 60685] radeonfb fails to probe EDID on a PowerBook G4 (TiBook)

[Heikki Lindholm]

Public bug reported:

I have a 667 MHz PowerBook3,4, which has a "Radeon Mobility M7 LW
[Radeon Mobility 7500]." The default boot-up on Dapper, where radeonfb
is used, fails very often and it _seems_ even more so after the
2.6.15-26.47 (after more than 20 boots I finally got a working console.)
The failure manifests itself as a non-synced display, eg. the display
shows a grey bar or random flickering trash.

Now, I compared the dmesg output of a failed boot attempt (ssh'ing into the machine) and a successful boot attempt, and there is a difference:
Successfull boot:
[   28.799644] radeonfb: Monitor 1 type LCD found
[   28.799652] radeonfb: EDID probed
[   28.799657] radeonfb: Monitor 2 type no found
[   28.799674] radeonfb: Using Firmware dividers 0x0002008e from PPLL 0
[   28.958642] radeonfb: Dynamic Clock Power Management enabled
Failed boot:
[   27.818747] radeonfb: Monitor 1 type LCD found
[   27.818756] radeonfb: Monitor 2 type no found
[   27.818767] radeonfb: Using Firmware dividers 0x0002008e from PPLL 0
[   27.818776] radeonfb: Assuming panel size 8x1
[   27.977745] radeonfb: Dynamic Clock Power Management enabled
Notice that the failed boot finds the LCD monitor, but does not, for whatever reason, probe the EDID and goes on to assume a grand 8x1 panel!

Additionally, any similar issue has NEVER happened when booting to Mac
OS X and this problem has been around from at least 2.6.9 kernel, which
I remember using with Debian before switching to Dapper. Usually booting
between OS X and Linux corrects the problem, whereas successive attempts
to boot Linux just fail in a row.

** Affects: linux-source-2.6.15 (Ubuntu)
     Importance: Untriaged
         Status: Unconfirmed

-- 
radeonfb fails to probe EDID on a PowerBook G4 (TiBook)
https://launchpad.net/bugs/60685