Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 21 Jul 1997 22:56:33 +0200
From:      Stefan Esser <se@FreeBSD.ORG>
To:        Terry Lambert <terry@lambert.org>
Cc:        Shimon@i-connect.net, freebsd-hackers@FreeBSD.ORG
Subject:   Re: pcireg.h lost children... ?
Message-ID:  <19970721225633.48908@mi.uni-koeln.de>
In-Reply-To: <199707202320.QAA09987@phaeton.artisoft.com>; from Terry Lambert on Sun, Jul 20, 1997 at 04:20:36PM -0700
References:  <19970718202558.63332@mi.uni-koeln.de> <199707202320.QAA09987@phaeton.artisoft.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Jul 20, Terry Lambert <terry@lambert.org> wrote:
> > One of the changes to the PCI code, that might follow (I have
> > not completely made up my mind) is, that the probe will just
> > associate a driver with a list of vendor/device IDs, and there
> > will no longer be a xxx_probe() function in the drivers.
> 
> How does a PCI driver for, for instance, a DEC chip based ethernet
> board find boards for which a vendor/driver assignment has not yet
> been made?

Well, if the device ID has not been assigned, 
then the chip most probably does not exist :)

If you mean that the list of IDs available to
the driver is incomplete: well how is that 
different from the current behaviour, where
a list of device IDs is just compiled into
the driver ?

> Currently, the probe code will find the board, and it will just work.

Hmmm ? I'm not sure I understand what you mean.
I surely won't try to attach a board that does
only match the vendor ID and the class code.
This has been done by NCR in their early SDMS
code, and they noticed it was wrong, when it 
caused boot failures on systems with new chips,
that could not be driven by that version of the
SCSI BIOS.

> One big problem I can see this causing (though I admit it cures
> one of my loudest complaints: destructive probing) is that the
> vendor/device/driver association list could quicly get too large
> to fit on the same floppy as the BSD kernel.  8-(.

Hmmm ? The list could for example contain one
vendor ID and possibly multiple device IDs on a
line starting with the driver name. This is some
10 bytes per unique ID, and well, if you want to
support some 100 PCI IDs, may add up to 1KB of 
highly compressible data :)

Regards, STefan



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970721225633.48908>