Date: Sat, 10 May 2003 10:30:18 -0700 From: Marcel Moolenaar <marcel@xcllnt.net> To: "Bruce A. Mah" <bmah@freebsd.org> Cc: freebsd-ia64@freebsd.org Subject: Re: Request for Content: FreeBSD/ia64 5.1-RELEASE Hardware Notes Message-ID: <20030510173018.GB563@dhcp01.pn.xcllnt.net> In-Reply-To: <20030510160857.GA24789@intruder.bmah.org> References: <20030510000205.GC7976@sixshooter.v6.thrupoint.net> <20030510020137.GA2188@athlon.pn.xcllnt.net> <20030510044232.GA8378@sixshooter.v6.thrupoint.net> <20030510160857.GA24789@intruder.bmah.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, May 10, 2003 at 09:08:57AM -0700, Bruce A. Mah wrote: > > > > 1 Introduction > > > > > > It's probably a nice gesture to link to the platform home pages. > > I guess we could link directly to the ia64 platform page, rather than > the top-level page for all platforms. Yes, of course. That was what I meant. > > > > 2 Supported Processors and Motherboards > > > > > > Processors: > > > Itanium & Itanium 2 > > Other things users might want to know: > > SMP vs. UP? Both :-) SMP is even more unstable than UP. I'm trying to squeeze some UP related fixes in to improve stability, but there are high priority jobs. > What types of consoles are supported (e.g. VGA, serial)? Yes. Both syscons and sio(4) work. > > > Motherboards: > > > It's probably a good idea to focus on chipset for now: > > > chipsets: > > > HP zx1 (only on development branch) > > "On development branch" means the P4 branch, right? If so, I'm am > tempted to leave it out of the hardware notes for CVS HEAD, or at > least maybe we should say "Perforce repository". Is this situation > going to change as a result of your blanket commit approval? Yes, development branch means perforce. We have told people about this because the HP boxes are in use a lot of the time. The situation is not going to change soon. Maybe 5.2, but more likely later. > > > > 3.1 Disk Controllers > > > > > > ATA: Except for the PC chipset embedded ATA controllers, most should > > > work OOTB. > > I wish I knew which ones those were, so we could turn them off. The best thing is to not sum up the device we know exist on i386, but instead mention that all ia64 compatible ATA controllers are expected to be supported. > > > > 3.8 Serial Interfaces > > > > > > Diva Serial [GSP] Multiport UART (development branch only [puc(4)] > > > > > I may have turned sio(4) by mistake in the original version of this > document. If so, we should turn it off now. sio(4) works on Intel legacy ia64 machines. sio(4) does not work on PA legacy ia64 machines. This is the crux of the problem on ia64. All devices on HP machines are memory mapped and there's *no* ISA device support other than the PCI dictated VGA legacy. As long as the box has an Intel legacy, most devices work (sio(4), syscons(4), pcm(4), atkbd(4) psm(4)...) Each of these fail on non-Intel legacy machines because the driver makes ISA specific assumptions that do not hold: sio: no support for memoru mapped I/O syscons: Expect BIOS, VGA probes BIOS etc etc pcm: Probes MSS ISA ports ad nauseam atkbd: Fixed ISA port psm: See atkbd > > > The UHCI on the "BigSur" seems to have > > > a weird hub attached on both ports which generally makes USB unusable > > > I don't think it's specific to ia64. > > What's "BigSur"? Can we maybe put this in the chipset/motherboard > notes? BigSur is the code name for the Itanium prototype. Some products are based on this (such as the HP i2000). I'm not sure we should talk in code names in the docs... I think, BTW, it's not UHCI itself that doesn't work. I think our hub code does something the Microsystems hubs do not like. The hangs are caused while resetting and querying the hubs. Since all devices are at least one hub away from the host controller, this makes USB useless... -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030510173018.GB563>