There is one design problem in using the 'etherboot' driver in
GRUB (although I really like etherboot !!!)
In the build process of rtherboot, the different parts of c files
and driver files are compile with options for the 'one' ethernet card,
not for all. Let's say in other words:
When a boot ROM for the NE2000 is built, then a set of
macros are defined for NE2000. If a boot ROM for a CS8900 is
built, than other macros are defined.
'etherboot' does a probing but only instrumented for the
one card, on which the boot ROM is used.
In GRUB the users want to enable a set of ethernet cards, but I have
problems with this. Often the probing of ethernet interfaces hangs, if
the card is not found.
And this leads me to the bug report for etherboot:
Especially the NE*000 and the NE2000/PCI hangs in the first pio_write
function in the last 'while-loop', if there is no NE card. Further
my Rtl8029 (OvisLink NE2000 PCI) hangs in the NE2000/PCI probe on this
position (code was instrumented with 'printf ()' to debug it.
I also got the problem under GRUB, that on a machine having NE2000/ISA
the CS8900 probe hangs (both interfaces were compiled in into GRUB).
Is this a problem I detected alone, or also have other users simialar
problems to them ... ?
With friendly regards
Christoph Plattner
