OpenBSD Clashes with Adaptec In Quest for Docs

TrumpetPower! writes "OpenBSD developers have been asking for documentation from Adaptec for over four months. Adaptec's response has been to deliberately misunderstand what is being asked of them. A former Adaptec employee admits that the hardware is buggy and tricky to get right. So, as a result, OpenBSD 3.7 will ship without Adaptec RAID support. Personally, I'm glad that Theo isn't resting on his laurels."

Why just OpenBSD?

It would be nice if more of the Linux big names would jump on the bangwagon and lobby with companies to get open source drivers for hardware.

Why just documentation?

Absolutely. Open source drivers would be a beautiful gift, in this case it's actually more than what is being asked for. Adaptec is asked to release specs on their raid controllers, they chose not to.

They are under an obligation to provide usefulness on legit architectures, but they aren't doing that. Adaptec should get over their shame of bugs, and allow the driver people at OpenBSD a chance at making things work.

There is no general fix for this problem, often specs are released way too late. On the other hand, releasing open source drivers will open specs for the same device. These specs aren't just trade secrets, they're actually necessary for building drivers.

Re:Why just documentation?

They are under an obligation to provide usefulness on legit architectures

Exactly what obligation does Adaptec have?

Re:Why just documentation?

None. Just as I have no obligation to ever buy an Adaptec piece of hardware again.

Re:Why just documentation?

That's not a "responsibility" or an "obligation". It is, however, an incentive, and should be quite a strong one at that.

Nowadays, I purchase equipment based more on its compatibility with FreeBSD (and occasionally OpenBSD) than any other factor (incl. performance and price), as that's what it's going to be used with.

As far as responsibility or obligation is concerned, Adaptec's got none to the Open Source community, unless you can consider it a direct failure of its responsibility to its shareholders. Just because Open Source is "fighting the good fight", doesn't mean anyone owes us anything.

There's an old saying

There's an old saying, which I think fits well here.

"Never ascribe to malice that which is adequately explained by incompetence." - Napolean

Re:There's an old saying

Ironically you've misunderstood irony...

I wonder how this will affect Adaptec?

I know that when I'm buying hardware, I first make sure that there's at least a reasonable chance that it will work in my operating system (Linux, by choice). So, in this case, if I was choosing a RAID card, and my system was BSD-based, then Adaptec would be down a few quid.

Re:I wonder how this will affect Adaptec?

well you can guess that this particular controller will be avoided by anyone with connections. Openbsd doesn't enjoy much use from desktop or developer users because it's too hard, and has few advantages.

The one advantage it does have is security, which is vital for running large scale servers. These servers have reliabilty as a high priority, so RAID is the norm.

Simple solution...

There's a very simple solution for this: Don't buy anything from Adaptec, ever. They'll be out of business; problem solved.

Just a note

Just a note; the "former Adaptec employee" is Scott Long of the FreeBSD project [freebsd.org].

I have not been using OpenBSD sice 1999, but hardware support was never its strong point... though what it supported was,like all the BSD's, supported extremely well.

It's a good call, in spirit of BSD. Scott's drivers are exellent and they just need to port those.

Re:Just a note

You misunderstand. OpenBSD already have a driver [openbsd.org]. They want documentation to improve that and more importantly implement a management program which can do critical stuff like check if any drives have failed.

The management utility in the FreeBSD ports tree is binary-only. OpenBSD refuse to accept binary only crap, which is why they want this documentation.

OpenBSD confirms it...

OpenBSD confirms it... Adaptec is dying!

this is a good solution

It's all about making sure the big shareholders know that the company's policies are costing them sales.

People say that Theo should stop being so annoying, but the only way shareholders find out is when it gets massively publicised like this.

It worked for the 802.11 drivers. It's worth a shot here.

Probably software raid

Like their old AAA ide raid controllers which was nothing more than IDE paddle boards with software raid logic..marketed as true hardware raid.. Documentation exposes the magic behind the illusion..(sometimes)

reminds me of Promise

Reminds me of Promise's definition of "Linux support" for a card I bought.

In the case of the SX-150 SATA raid card (which has a hardware XOR engine and whatnot), that meant "we have binary drivers for distributions which are several years old".

There is some source. Well, it's a 'wrapped' binary driver, and it's only available from "some guy" in Germany who begged Promise support long enough they gave it to him. You a)cannot compile it into the kernel b)cannot compile it for 2.6 because it simply isn't compatible. I sent numerous emails to Promise asking when a 2.6 driver would be available or if there was any updated source code. None were ever answered.

Same story with the tools- unless you're running Redhat 9.0 or some ancient version of Suse, forget ANY on-line monitoring.

Not that the customers are much better- one page I found about the card suggested that "software raid is faster anyway", which is an absurd proposition by itself. Regardless, why would you spend $100-200 more on a hardware-raid card complete with cache memory, and then just use the 2.6 SATA driver which only drives the SATA interfaces?

From what I understand, 3ware has better support for Linux, but that means I have to migrate a large amount of data off the old array..

Re:reminds me of Promise

Software RAID *is* very often faster, especially on a modern CPU paired with an older design -- you don't buy HW RAID because it is faster, you buy it for battery backup and offloading of low level operations to conserve CPU time and bus/memory bandwidth for user applications and so that if your OS or CPU/memory/whatever blows up, or you lose power, it won't corrupt the data on your disk array. Hardware RAID dedicated processors are simple, slow, "reliable" units -- not ultra-fast bleeding-edge dedicated units like you see on video cards.

Nothing ever changes at Adaptec

Theo says: "We are not asking for support. We are asking for documentation."

Substitute "They" for "We" in that sentence and it could have been me speaking, when I was working at Adaptec and trying to release an in-house version of the starfire (a.k.a. "Duralan" ethernet MAC) driver. I hit that same brick wall over and over again while tying to get some chip specs and a linux driver released. Somehow, in their minds, "support" is translated into not releasing specs and drivers. Releasing such information, in contrast, is a failure to support customers. This wierd Orwellian doublethink seems to pervade the thinking of everyone connected with supporting Linux and other free OS's at Adaptec.

It's so amazing to see that nothing has changed at Adaptec in the last 7 years. My own driver episode was "resolved" (unsatisfactorily, for me) by Donald Becker agreeing to sign an NDA for the chip specs. Not to second guess Donald, but my thinking at the time was, "this just postpones the problem. Maybe it would be better just to boycott these imbeciles."

Not to close on a sour note, I should say that Adaptec was a great place to work in many ways, and I always viewed their attitude toward free software as an aberration. I still tend to do so, and perhaps that's wishful thinking on my part.

Threshold of complexity

Why adaptec isnt releasing detailed specs is obvious. If people had them they could better evaluate the product. Apparently the marketing dept. at adaptec fears transparency and complacency.

Look at the small and medium end raid market now. Theres not many players, Adaptec,promise,3ware and a bunch that adaptec bought up. Adaptec gains nothing by opening up itself to a point by point comparison with lesser competitors. Their name recognition is carrying them much the way IBM's used to. Further if the hardware is bugged and tricky and adaptec knew about it then they open themselves up to liability.

Their reasons are obvious keep the barriers high and keept those that can't climb them out.

interesting if not down right funny thread:

* To: Charles Swiger
* Subject: Re: Adaptec AAC raid support
* From: Bob Beck
* Date: Sat, 19 Mar 2005 13:56:41 -0700
* Cc: Theo de Raadt , Sean Hafeez , [e-mail], Scott Long , [e-mail]
* In-reply-to:
* Mail-followup-to: Charles Swiger , Theo de Raadt , Sean Hafeez , [e-mail], Scott Long , [e-mail]
* References:
* User-agent: Mutt/1.5.6i

> ...deliberately breaking OpenBSD's support for Adaptec hardware as some
> sort of ultimatum is a childish and self-destructive action. I hope
> the other OpenBSD committers veto any such action as being
> counterproductive and harmful to your users.

Horsecookies. What was done was remove AAC support from GENERIC,
because users know what is in GENERIC is supposed to be stable and a
good candidate for use. I've got AAC's. They aren't at the moment.
they die, and you can't do anything with the raid management without
rebooting, and Adaptec has shown no signs of releasing documentation
so that situation can be corrected.

Sure, there's a "free" driver, and a non-free management interface,
so it's only half a driver. Pretending to have a production system
using a raid card that with no supportable management interface so you
have to reboot to fix anything is like buying birth control pills in
packs of 20. Pretty soon you're going to take a good fucking on a day
you really can't afford it. Period.

As such AAC isnt' any more broken than it ever was. OpenBSD
just chooses not to encourage users to purchase a non-supportable
card by including support for it in the GENERIC kernel. Are you
saying it's more honest to leave unstable and incomplete support in
there? People who wish to use it anyway can always compile it in.

> Otherwise, you're likely to discover that most people choose to run an
> OS which works with the hardware they have, rather than sticking with
> OpenBSD.

Or choose to replace the hardware that isn't supportable by the
OS they want to run. Thank you LSI and Dell. LSI cards seem to work
fine.

-Bob

emphasis added by poster

Re:How many people...

Which make you think: "Why is OpenBSD doing this and not FreeBSD". I think it's sad that the FreeBSD developers don't seem to care that much about having free drivers (the AAC is free I believe, but the management interface is not).

Of cause what really annoys me is that the Linux developers seem to care even less. Why is it that the developers of free software can't stand togther and demand documentation? And why is it that it's the smallest team that must make these demands?

Re:How many people...

Yes, FreeBSD is the most popular BSD. But each BSD is it's own operating system, not a previous version of the same operating system like your analogy.

It did not start with BSD4.4-lite, go to 386BSD, move to NetBSD, then OpenBSD, then DragonFlyBSD and then FreeBSD. Each are their own system which split at one time or another from the same tree.

All four of those systems are maintained today and therefore it is not like Windows 9x complaining about hardware support. Windows does not maintain new versions of Windows 95.

OpenBSD is the extremely secure and extremely open of the BSDs and Unix-likes. OpenBSD refuses to have anything that isn't as Free and Open as their goals describe into their system. Linux and FreeBSD are more into the functionality over ideals idea. NetBSD I cannot speak for though as I don't really follow them.

Me

Have an old Ultra 1 doing firewall and light server duty for a DSL line. So far its had zero hardware issues and everything has worked. Wish I could have said the same for NetBSD. It locked up randomly on the same box.

I haven't used OpenBSD in a few years and was really impressed with their rewrite of packet filter. You linux folks should check it out.

Re:Tried e-mailing the guy....

Well, you see, they had been trying to switch over to BSD, they they had this driver problem...

Re:Tried e-mailing the guy....

"screw you, we'll ship with even less support for your product than we did before" dummy spits constitute "does not play nice with others" in my book.

This is not incompatible with 'a business-benefiting attitude' and 'acting in business-benefiting ways'. William Gates Jr has built a very successful business by acting exactly in this manner. If you think geeks have a bad attitude and businessmen do not, perhaps it's just because geeks publish their nastygram messages on the web and businesses keep them secret.

Trouble is, geeks carry no weight in business, and the businessfolks have all the money. It's up to us to decide if we want some of that money or not.

I think you have an incorrect assumption here. Theo de Raadt is not trying to get money. He is trying to improve a free operating system, OpenBSD.