Re: questions about undi patch

[Thomas Reifferscheid]

> BTW, I think your observations are a good point to go on. Until now we
> assume, that it is a memory problem. Could you trace it a bit more.
> Possibly you could put in a delay into pxegrub. Maybe that fix it.

I consider putting a delay as a workaround, but not solving
the problem.

Currently I'm into undi.c  undi_poll()
where the problem occurs if
eb_pxenv_undi_isr() get's called twice.

Hang on to more information.