Date: Fri, 17 Jun 2005 11:25:34 +0200 From: Marius Strobl <marius@alchemy.franken.de> To: "David O'Brien" <obrien@freebsd.org> Cc: freebsd-sparc64@freebsd.org Subject: Re: SUCCESS: Sun Blade 100 with Firewire-attached HDD (IEEE 1394) Message-ID: <20050617112534.U80563@newtrinity.zeist.de> In-Reply-To: <20050617080938.GD1485@dragon.NUXI.org>; from obrien@freebsd.org on Fri, Jun 17, 2005 at 01:09:38AM -0700 References: <42A4C2E0.9010007@cdc.informatik.tu-darmstadt.de> <20050607000820.B80563@newtrinity.zeist.de> <20050617080938.GD1485@dragon.NUXI.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Jun 17, 2005 at 01:09:38AM -0700, David O'Brien wrote: > On Tue, Jun 07, 2005 at 12:08:20AM +0200, Marius Strobl wrote: > > On Mon, Jun 06, 2005 at 11:40:48PM +0200, Ralf-Philipp Weinmann wrote: > > > The FreeBSD/sparc64 5.4R Hardware Notes list this machine as "fully > > > supported" (section 2.1). This seems odd to me, since it is clearly NOT > > > the case. Without this patch, the controller simply wasn't initialized > > > correctly. > > > > Yeah, the sparc64 hardware notes have to be read in the context of > > the time when they where written which was mainly when e.g. the > > MD PCI code was fixed to work on stuff like E450 and some of the > > weirder USII[e,i] based systems etc. From a today's perspective they > > sound more like propaganda... > > I feel the language is fine. For entire existence of the FreeBSD/sparc64 > I've never missed firewire or floppy on my SunBlade 100. When we wrote > the notes, "fully supported" mean the CPU, on-board disk controller (ATA > or SCSI), NIC, and sio(4) were fully supported. For typical development > and server deployment that is all that was needed. > > There are devices found on some i386 machines that we don't support > (wireless NIC and modem, usb2 for a long time), yet most would consider > them "fully supported". > Personally I agree but the "fully supported" in the sparc64 hardware notes seems to regularly trick people into thinking that FreeBSD is a drop-in replacement for Solaris in all aspects which simply isn't true. FYI, even sio(4) was never fully supported by the in-tree sources, in order to use the NS16650 with sio(4) as a console on sparc64 one had to apply the hacks from people.freebsd.org/~tmm. AFAICT for the entire existence of FreeBSD/sparc64 the ALI ATA controllers used on-board in Blade 100 etc. also were only guaranteed to work at UDMA33, regardless whether used with original Sun drives or not (f.e. search the freebsd-sparc64 archives for tmm@ talking about data corruption at UDMA66). Given such limitations im not sure whether it was and is advisable to describe e.g. the Blade 100 as "fully supported", even by your definition. Ultra models from Ultra 2 and onwards however probably can be described as "fully supported" when using your definition. Anyway, the hardware notes still should outline what's really meant by "fully supported" in order to not cause confusion and frustration. Marius
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050617112534.U80563>