Bug#397438: Panel stays blank in MergedFB mode

[Bryan Stillwell]

On Fri, Dec 22, 2006 at 07:46:02AM +0100, Michel Dänzer wrote:
>Int10 is mostly handled outside of the driver. BTW, I'm not worrying
>about the checksum - that warning seems pretty common without causing
>problems - but about the fact that at least some registers seem to get
>clobbered with bogus values.

I believe I was putting the 'Option "NoInt10" "True"' in the wrong
location inside of xorg.conf before.  This time I put it in the screen
section and received some interesting results (I'm still using the
debian unstable radeon_drv and ati_drv on both distros):

On Debian unstable:
* No more "Bad V_BIOS checksum" message
* Only uses one monitor still
* A new message in Xorg.0.log when using NoInt10:

(II) Attempted to read BIOS 64KB from /sys/bus/pci/devices/0000:00:00.0/rom: got 44KB


On SLES10:
* Only one monitor when using NoInt10 unless X was previously started
  without NoInt10.
* A new message in Xorg.0.log when using NoInt10:

(II) Truncating PCI BIOS Length to 45056


These results lead me to believe that the problem is in the int10 code.
So the next thing I did was to try and use the libint10.so from sles10
in Debian unstable, and it worked!  I now have a working dual-head setup
in Debian!  It still took the -ignoreABI option though.

I would like to use git bisect to locate the commit that broke
dual-head, but at first glance it doesn't seem to be as easy to do with
int10 as it was with the ati video driver.  Any pointers?

Thanks,
Bryan