Date: Wed, 26 Feb 2003 06:54:14 -0700 (MST) From: "M. Warner Losh" <imp@bsdimp.com> To: marcel@xcllnt.net Cc: freebsd-arch@FreeBSD.ORG Subject: Re: [RFC] splitting of conf/NOTES Message-ID: <20030226.065414.31318249.imp@bsdimp.com> In-Reply-To: <20030224222205.GA1032@athlon.pn.xcllnt.net> References: <20030224204113.GC661@athlon.pn.xcllnt.net> <20030224.143957.17279856.imp@bsdimp.com> <20030224222205.GA1032@athlon.pn.xcllnt.net>
next in thread | previous in thread | raw e-mail | index | archive | help
In message: <20030224222205.GA1032@athlon.pn.xcllnt.net>
Marcel Moolenaar <marcel@xcllnt.net> writes:
: What I'm getting at is: will this only be i386 ROMs (ie with
: i386 code) or is it possible that when a machine has ISA
: emulation, one should be able to scan for ROMs in the ISA hole
: and find something that can actually be used. I'm inclined to
: assume that ROMs found in the ISA hole are i386 specific and
: ISA emulation is limited to I/O ports and registers only.
Typically, the ROMs that are found are used to create a map of
resources that are decoded by the bus but otherwise not used by other
drivers.
: Thus: even if ISA (sec) is not i386 specific, there may be ISA
: drivers that depend on ROM, BIOS and/or BIOS data (see sys/fb/vga.c)
: and consequently are not pure ISA drivers. They are PC drivers.
: This is the driver part I was referring to in my previous mail.
There are very few drivers in the tree that do this. However, the
whole console goo stuff is far too dependent on the PC architecture.
: On the ia64 branch I fiddled with sys/conf/files.* to make PC/ISA
: device drivers actually depend on the existence of "device isa",
: so that I could create an ISA-less kernel (needed for pluto{1|2}).
This should be a good thing. However, the purity of some of the older
files in the tree leaves something to be desired.
: This triggered some interesting bugs, such as the fact that the
: high-level syscons code (ie sc(4) itself) has been attached to
: the ISA bus and is identified/probed as an ISA device. I had to
: break the artificial dependency to get syscons working (with the
: supprt of a PCI VGA wannabe driver).
Yes. These are bugs in the tree.
Let's also not forget sio, which is mostly kinda sorta MI, but lacks
support for memory mapping and a number of other desirable things (but
marcel-sio seems to have fixed them :-).
Warner
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030226.065414.31318249.imp>