Date: Sat, 19 Mar 2005 13:40:50 -0800 From: "Ted Mittelstaedt" <tedm@toybox.placo.com> To: "Theo de Raadt" <deraadt@cvs.openbsd.org>, "Scott Long" <scottl@samsco.org> Cc: freebsd-questions@freebsd.org Subject: RE: Adaptec AAC raid support Message-ID: <LOBBIFDAGNMAMLGJJCKNAEMPFAAA.tedm@toybox.placo.com> In-Reply-To: <200503192035.j2JKZKOM018459@cvs.openbsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Theo and Scott, a few words: Theo: I happen to own a system that runs FreeBSD and uses an Adaptec AAA card, here's the dmesg: ahc0: <Adaptec aic7815 RAID memory controller> port 0x1000-0x10ff mem 0x40400000-0x407fffff,0x40100000-0x40100fff irq 11 at device 0.0 on pci2 RAID functionality unsupported device_probe_and_attach: ahc0 attach returned 6 ahc0: <Adaptec AAA-131 Ultra2 RAID adapter> port 0x1400-0x14ff mem 0x40000000-0x40000fff irq 11 at device 4.0 on pci2 aic7890/91: Ultra2 Wide Channel A, SCSI Id=7, 32/253 SCBs Adaptec has been unwilling to provide support for the parity calculator for this card, so it's RAID functionality is useless. The situation is very similar to the AAC card, worse in fact since there's not even a poor out-of-date binary substitute driver. HOWEVER, PLEASE keep in mind that I did NOT go out and buy this machine, did not direct money to be spent on this abomination. I inherited this machine which used to be a Windows server, and I have made the best use of what I had available. I HAVE other SCSI cards, but all of them aren't LVD SCSI, and the disk pack on this server is LVD SCSI drives. So to support these disks with a native RAID card I would have to go buy a new card - which isn't cost-justified for an old server. I understand your concern that leaving a half-a-driver in the OpenBSD GENERIC kernel might in fact encourage people to reward Adaptec by continuing to buy their products. I am a very strong supporter of the idea of people NOT purchasing products that don't have full FreeBSD support. But I must insist that this ISN'T true. If you want to discourage people from buying Adaptec products then PUT THAT RECOMMENDATION RIGHT IN THE HARDWARE COMPATIBILITY LIST. And make doubly damn sure that this list is at MOST, a SINGLE click away from the OpenBSD homepage. I assure you that whenever I purchase products or approve products for purchase at the ISP I work at, that I make sure they have native mode drivers for FreeBSD, OpenBSD, Solaris AND Linux, even if they are being purchased for use as Windows servers. You never know in a machine's lifespan what your going to end up using it for. Crippled drivers like the AAC driver, and the AAA shim support for my RAID card, fulfill a vital purpose, they facilitate migration from Windows platforms for older servers. Period. Nobody with any serious budget for a server who is setting out to build an OpenBSD or FreeBSD server, is going to go against the Hardware Compatibility List when selecting components. So, your decision to pull the AAC support does only one thing - it makes it so that people who want to use OpenBSD on Windows servers, cannot do it. Thus it delays the migration of those servers away from Windows. Now, for Scott: Scott, put a God-Damned sock in it. Adaptec has had a LONG history of non-cooperation with the Open Source community. Back when FreeBSD 1.x was going, Adaptec stalled for years with information with the 2740 card, and we were all stuck with using 1740 cards in our EISA servers. And finally someone had to port over the Linux driver for the 2740 which had been reverse-engineered as I recall, Adaptec never gave us any support for it until after the thing had been figured out. WHERE THE HELL is my AAA-131 driver if Adaptec as you assert they are so friendly? And Theo has done a case-winning explanation of how the AAC driver is crippled. YOU have not done a SINGLE thing to respond to ANY of the specific explanations of Theo's. If you are such a big man, then overlook his shrillness and provide a technical explanation of each of his technical points. I don't spec Adaptec SCSI parts that don't have native FreeBSD drivers, and right now, only the older Adaptec cards do. So, great, I can get my old SCSI cards off Ebay, out of the bargain bins at Goodwill, and the computer junk stores. Fat lot of good that does Adaptec. Money for NEW stuff ain't going to Adaptec. Instead their stonewalling simply makes their older shit more valuable, which undercuts their market for their new shit. And it's not like Adaptec isn't the market leader anyway, frankly if the SCSI market wasn't a niche market, they would be up against the Feds on anti-trust charges now. So who the HELL do they have to fear if they are giving out what they regard as proprietary data? NOBODY. Adaptec doesen't cooperate because their corporate culture is sick, it's because they are a bunch of power-grasping micromanagers that have a gigantic NIH attitude, and an inferiority complex a mile high. They don't release data because when they make mistakes they don't want people finding out about them. I've had inexplicible crashes on my Adaptec stuff in the past, very rarely that happens, but every once in a few years it does. And there's others that have posted here very recently, within the last few weeks in fact, who have had much worse trouble. You go ask Anthony Atkielski and Leroy van Logchem their opinions of how great their FreeBSD Adaptec drivers are performing on their hardware, you don't get this rosy picture that you seem to like to paint. Both of them posted problems in the FreeBSD questions forum recently and I didn't see you responding to either of them. It is inexcusable for a hardware vendor to not provide programming specs. PC hardware has always been sold in the past based on marketing and positioning, for all the bake-offs and such that are held claiming one chipset is better than another, the market leaders in hardware are the ones that cut OEM deal after OEM deal. That is how Adaptec got as big as they are, and there's been plenty of times in the past that their competitors have had faster and better product. And their competitors have spent the enormous sums to reverse-engineer the Adaptec products anyway to find out all the go-fast tricks. Denying programming specs does not in any way help to shield their hardware secrets from competitors, all it does it make it impossible to write open source drivers, and it hides any dumb mistakes they made in designing their hardware. Adaptec has a LONG way to go before they are a friend of the Open Source movement. Ted Mittelstaedt Author, The FreeBSD Corporate Networker's Guide http://www.freebsd-corp-net-guide.com/ > -----Original Message----- > From: owner-freebsd-questions@freebsd.org > [mailto:owner-freebsd-questions@freebsd.org]On Behalf Of Theo de Raadt > Sent: Saturday, March 19, 2005 12:35 PM > To: Scott Long > Cc: Jason Crawford; misc@openbsd.org; Adam; > freebsd-questions@freebsd.org > Subject: Re: Adaptec AAC raid support > > > > It's not a binary driver, it's a 2-clause BSD licensed driver that > > contains full source. You said that the OpenBSD driver is unstable, > > so I offered to help. That has nothing to do with binary apps. > > Deleting it from the OpenBSD tree is always an option, of course. > > The driver is free, but the tool is a binary. The interface "tunnel" > is coded in the driver, so that the "closed binary" tool can talk > through to the card. The messages exchanged are not documented, > either. > > Same thing. > > You are saying > > There are open bits > > and I am saying > > There are closed bits > > This whole thing is about the closed bits, not about the open bits. > > > Why do you keep apologizing for Adaptec, and attacking our efforts? > > _______________________________________________ > freebsd-questions@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-questions > To unsubscribe, send any mail to > "freebsd-questions-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?LOBBIFDAGNMAMLGJJCKNAEMPFAAA.tedm>