In message <[EMAIL PROTECTED]>,
Rik van Riel writes:
>THIS is the real reason for preferring source code support drivers.
>Not even the usually higher quality of the open source drivers or
>the faster availability of the manufacturer's drivers change this
>situation.
As a nice concrete example, TekRam recently updated the hardware of the
DC390-F. It still works fine with NetBSD, and presumably also with FreeBSD.
But it no longer works with the most recent SRM ROM available for Alpha
PC164 systems, so such systems cannot be booted from new SCSI hardware
anymore. Since the ROMs are closed (and not merely a little closed, but
mildly protected - the ROM images are compressed in some unknown way),
it's not possible to just set a bit and make it work with, say, a 3C875J
card, or find out why the newer DC390-F doesn't work.
Sooner or later, the smallest IDE drive sold will be too large for the
PC164 BIOS to handle. Then the machine becomes a doorstop.[*]
That said, I *do* see reasons for people to, in some cases, want to distribute
systems with binary drivers. I can understand Dennis's concern. It's a shame
that there are no BSD-type systems that allow vendors to provide binary
drivers which ship with the system, or kernel makefiles that can handle
binary-only modules. If such a system existed - perhaps, even, with some kind
of commercial support network - it would be a godsend for people in his
position. Alas.
-s
[*] Ignoring, for now, the question of whether the PC164 is *already* a
doorstop in a world of dual and quad processor 1Ghz machines.
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
