|
While you can view his slides (you can find them in the
middle of picture
gallery and you can get the postscript version from
Geert's web site),
here are a few highlights from his talk:
- In 1993 the need first came up with the initial port to linux/68k and that's
when the frame buffer device was first written. Later, the same need came
up on other architectures linux was ported to, and which didn't have a text
mode.
- Some video cards on PC now don't support textmode, or some don't even
support VGA on PCs (like the Cyrix MediaGX)
- They have to use a trick to get output at the beginning of the boot process
because they can't display anything before the PCI bus is initialized, but
they have to display messages earlier than that. The trick is to write in a
dummy console and copy the result back to the virtual console once the card
has been initialized
- The little endian/big endian issues can be a pain when you have to swap
bytes around on a big endian machine with a little endian only PCI board
(some video cards support both BE and LE, but unfortunately others don't)
- While there have been some disagreements with the GGI folks, they were
credited for their work which helped a lot for the frame buffer
code on Intel.
|