> - IOC3 networking works, with 2MB/s maximum for a 30MB http transfer
This is strange. Ask Ralf.
> - Serial console works, but start of output is delayed. The driver
> probably lacks transfer bootstrap, and starts transmisssion only when
> the first buffer is full.
Yuck. What should be done about it?
> - Rebooting for SMP works, but this patch apparently broke arcload (hangs
> after "Entering Kernel").
This was an unrelated problem, fixed.
> - Serial console and framebuffer initializations seem to be mutually
> exclusive, that's impractical if you plan to route console (debug)
> output over serial while working on the graphics console.
Heh, I work on the graphics console using the graphics console. It's more
funny.
> - 'echo "foo" >/dev/fb0' kills the machine.
Oops.
I return -EINVAL in read and write on ImpactSR. Sick. Do you know where it
fails, exactly?
Stanislaw
|