| | > The second issue is that the screen is cleared when wscons attaches.
| |
| | True. I couldn't try myself yet - is the screen useable now when booting
| | via serial console?
|
| No.  It seems to lock up hard when booting from the serial console:
|
| mach0 at pci0 dev 19 function 0: ATI Technologies Rage XL (rev. 0x27)
| mach0: 8 MB aperture at 0xe087a000, 4 KB registers at 0xe0879c00
| mach0: 8188 KB SDRAM running at 114.765 MHz, maximum RAMDAC clock 230 MHz
| 1280 3376 3440 1664 1024 1026 1034 1066
| mach0: initial resolution 1280x1024 at 8 bpp
|
| then nothing.

Apparently what's going on here is that the display is attaching and
unconditionally calling wsdisplay_cnattach(), which it should only do
if it really is the console.  It doesn't seem to be setting up the 
display correctly if the firmware failed to do its job, so I don't get
anything visible on the screen.

So it looks like mach64_get_modetiming() needs to be more intelligent
and determine whether it's been configured or not.  Alternatively, the
firmware can be queried to determine whether that display is currently
active and decide between mach64_get_modetiming() and using what's in
mach64_modetiming[] based on that information.

Eduardo