Okay. I have spent the morning hacking with the mda. I downloaded test5-pre2
and set up a fresh kernel tree. Compiled it with my normal .config and but
with all the fbdev turned off.
1. Unpatched kernel.
The machine boots it fine, this is test5-pre2 running with the VGA and MDA
working. On bootup the cursor is in the top left corner until the kernel
initialises the MDA, when the cursor gets turned off. No problems for me, but
this apparently gives false positives for detecting an MDA when it is not
there, so this is not acceptable.
2. Patched kernel.
I applied the patch supplied by you to the list. The patch uncomments the
code I had commented that was messing up my cursor, and adds a save and
restore to the mda register test function. When I boot the kernel the machine
crashes. I get to the point where the MDA is initialised, and the MDA flashes,
and the kernel stops booting. No error messages appear on either screen.
3. Patch kernel with bits commented.
I applied your patch. I left the MDA detection code uncommented, but
commented out all the modifications to the mda register test function. On
bootup it works, the cursor first jumps to an odd place on the screen, and
then a fraction of a second later is turned off by the initialisation code.
This is probably the best of bother worlds, with the MDA detection not giving
false negatives, but with the cursor being in the wrong place for only a short
period of time. If you were were not watching for the cursor on the MDA you
would not see it move.
So basically what happened was I commented out some MDA detection code because
it was moving the cursor, then added a line to turn the cursor off, and
supplied the patch to Alan. This looks nice on my machine, but breaks on other
machines. By uncommenting the code that I commented it will work on other
machine and look less nice for people using the same MDA as me for a moment or
two at boot-up.
My suggestion is to just uncomment the code that I commented, that seems to
fix it.
I could try changing the udelay lengths in the patch you wrote. See if it
makes a difference. It is on the write of save back to the card that it fails.
I tried writing the mda_index_port out before writing save, but that did not
help either.
-- Don't worry -- shop.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Jul 23 2000 - 21:00:12 EST