Date: Fri, 18 Nov 2005 00:51:46 -0700 From: Scott Long <scottl@samsco.org> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: freebsd-current@freebsd.org Subject: Re: Side note on Shuttle XPC / AMD X2 (SN95G5V3) (Re: [PATCH] nve(4) locking cleanup) Message-ID: <437D8812.40605@samsco.org> In-Reply-To: <200511172330.jAHNUhgV007502@apollo.backplane.com> References: <20051117010651.97608.qmail@web50303.mail.yahoo.com> <200511171915.jAHJF0L6006229@apollo.backplane.com> <200511171639.00807.jhb@freebsd.org> <200511171702.20956.jhb@freebsd.org> <200511172330.jAHNUhgV007502@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Matthew Dillon wrote: > :I found someone. :) They don't have an X2 processor, just a 6400, and it does > :seem that the MP Table in there case does have some busted entries. However, > :the link devices used for APIC routing do work ok so with ACPI enabled 6.0 > :booted up ok out of the box with APIC enabled. Probably they (Shuttle) got > :by with a busted MP Table because Windows probably just uses ACPI and worked > :fine out of the box. > : > :-- > :John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ > > Buy an AMD Athlon X2 for that person so you can get SMP working, and > see if you hit the IRQ 7 problem. If the system has to reroute anything > through the 8259 (e.g. like the clock interrupt), you will probably > see spurious irq 7's. If it disables the 8259 entirely it should be ok. > I suspect that if the ACPI link entries are good, that will be the > only problem you face getting SMP working. > Luckily, we no longer have to worry about routing the 8254 through the AT-PIC anymore in FreeBSD. More and more chipsets and motherboards are axeing the logic to make this work, and it's a loosing battle to keep on trying to work around it. It wouldn't be too surprising if the 8259 goes away entirely in the next 2-3 years on Athlon and Opteron designs. > Note that I did a followup commit to the one I posted to clean > up my programming option for the hypertransport configuration. The > one I posted was just an initial workaround. Look at the CVS logs > for more info. Those configuration registers are insane and the > Shuttle BIOS clearly misprogrammed the one I was finally able to > track down. > > I intend to implement the link entry stuff for DragonFly after our > next release. For this release we already have too many feature items > and not enough time to stabilize yet another one... so that will be > post-release work. > > I have to say that the new Shuttle XPC populated with an Athlon X2 > is a really nifty machine. It's ultra fast, SMP, yet much, much quieter > then older slower shuttles. It runs nice and cool. It is clearly a > bug winner for Shuttle and AMD. It does look nice. Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?437D8812.40605>