Re: FW: HOT Plug and Expansion ROMs

["Monish Shah" <monish@hpfcmss.fc.hp.com>]

On Apr 10,  9:37am, Tony Goodfellow wrote:
> Subject: RE: FW: HOT Plug and Expansion ROMs

Old stuff deleted.  If you don't remember, you probably don't care.  :-)

> The point I am making is that the current hot Plug proposal PROHIBITS the
> use of ROM code.

Yes, that does prevent you from doing what you want to do.

> The proposal is that IF a ROM is present AND it is marked
> as HOT PLUG aware then the O/S SHALL use it.

This, however, prevents us from doing what we want to do.  For example, HP
might want to build hot plug systems running HP-UX (HP's UNIX) on PA-RISC
boxes.  These systems have no concept of Openboot and HP probably doesn't
want to invest in that.  We'd rather have drivers written to do full
initialization.  BTW, these systems don't understand BIOS either - we have
our own firmware architecture.

The above is my opinion of what HP position is likely to be, not
necessarily an official HP position.

Let me summarize.

The question is: should drivers do full initialization, or can they assume
that card firmware always did appropriate initialization?

Possible answers are:

A.  Driver should do full initialization.  This is the current assumption
    of the hot-plug spec.

B.  Card firmware, if any, must be run.  Driver need not do full
    initialization.  This doesn't work if firmware is BIOS.  So, this
    solution forces you to use Openboot.

C.  Let the system designers choose.  I.e., a particular combination of H/W
    and OS chooses one of the answers above.  A different combination can
    choose a different answer.

I would not object to changing the hot-plug spec so that it changes from
option A to option C.  I strongly disagree with switching to option B.

> Tony Goodfellow

Monish Shah
Hewlett Packard
"Œz