User Plea Means EISA Support Not Removed From Linux
jones_supa writes A patch was proposed to the Linux Kernel Mailing List to drop support for the old EISA bus. However a user chimed in: "Well, I'd like to keep my x86 box up and alive, to support EISA FDDI equipment I maintain if nothing else — which in particular means the current head version of Linux, not some ancient branch." Linus Torvalds was friendly about the case: "So if we actually have a user, and it works, then no, we're not removing EISA support. It's not like it hurts us or is in some way fundamentally broken, like the old i386 code was (i386 kernel page fault semantics really were broken, and the lack of some instructions made it more painful to maintain than needed — not like EISA at all, which is just a pure add-on on the side)."
In addition to Intel 80386, recent years have also seen MCA bus support being removed from the kernel. Linux generally strives to keep support even for crusty hardware if there provably is still user(s) of the particular gear.
I find it hard to believe that anyone is using EISA still. It got almost no traction in desktops and the only systems that ever had EISA slots were 386-486 era servers before the VL-BUS and PCI bus started to gain traction in late 486's.
If someone actually has a working EISA system, I'd like to see a photo. I had never managed to see more than one of these systems in my lifetime, and only saw one because it was being replaced in 1997 by a Pentium desktop.
I've actually seen more MCA systems than I've ever seen EISA.
"Linus Torvalds was friendly about the case"
They did it, they neutered him.
1Gbps and 10Gbps SFP+ cards are cheap. Upgrade Now!
Harrison's Postulate - "For every action there is an equal and opposite criticism"
Well that's the point. He's basically saying "It's no skin of our backs" to keep it supported, so he does.
I think that's kinda awesome where 1 guy says he still need it supported and it stays.
I was planning on using the lack of EISA support going forward to force an upgrade past the boss.
Now what excuse will I use with my wife? Somebody think of something or I'll have to deflate her!
Yeah and how reliable is that $20 PCI card compared to what they are currently running?
You generally do not replace a proven reliable system "just because".
By a strange coincidence “None at all” is exactly how much suspicion the ape-descendant Arthur Dent had that one of his closest friends was not descended from an ape, but was, in fact, from a small planet somewhere in the vicinity of Betelgeuse. Arthur Dent's failure to suspect this reflects the care with which his friend blended himself into human society - after a fairly shaky start. When he first arrived fifteen years ago, the minimal research he had done had suggested to him that the name ‘Ford Prefect' would be nicely inconspicuous. He will enter our story in thirty-five seconds and say “Hello, Arthur.” The ape-descendant will greet him in return, but in deference to a million years of evolution, he will not attempt to pick fleas off him; Earthmen are not proud of their ancestors and never invite them round to dinner.
Get free satoshi (Bitcoin) and Dogecoins
None at all.
"So if we actually have a user, and it works, then no, we're not removing EISA support. It's not like it hurts us or is in some way fundamentally broken." - Linus Torvalds
Get free satoshi (Bitcoin) and Dogecoins
If I've been using a piece of hardware from the 1980s and it still works perfectly for me, and it costs little or nothing at all to maintain it in the current kernel, I don't see why you wouldn't.
At some point, there will be a reckoning where they have to make a big change and the EISA stuff will not be reasonable to maintain any more. Until then, there's no reason to remove it just for the sake of "it's old". I presume once that point comes, Linus will ruthlessly shut down anyone who wants to keep it.
It just gets renewed, reused, and recycled, not only in more wealthy Western countries, and Third World countries, but by poorer people all over the world. (Hopefully with Linux and other FLOSS software installed!) There was an article in Slashdot in 2013 about an IBM 402 with punch cards, still in use!!! (I wonder if it could run Linux?) ;^)
But seriously, even though most of us are using computer less than 5 years old, a lot of old computers are still in use. What about 16 bit embedded systems, many running Linux! I have to agree with Linus, if the old technology in the kernel, does not adversely affect newer technologies, and people are still using it, then there is a legitimate reason for leaving it in the kernel. I trust his opinion.
IMHO, I think the FLOSS community has an obligation to continue to support older hardware & technologies that certain other proprietary O/S manufactures have long ago abandoned. Isn't that one of the reasons the Free Software and Open Source Software communities, and software were created in the first place?
So why is it reasonable for this user insist on running a trunk kernel, instead of an old branch that only gets security updates?
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
Because he's the maintainer.
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/MAINTAINERS#n2998
DEFXX FDDI NETWORK DRIVER
M: "Maciej W. Rozycki"
S: Maintained
F: drivers/net/fddi/defxx.*
He's supposed to run the latest kernel, and keep this driver working...
It is obvious that nobody would install a brand new system with such old technology. It should be equally obvious, though, that just as one expects old buildings to "just stay up" (with a little maintenance), there are plenty of old technology systems still up and running just fine for whatever they do. Lots of people in big cities have 75-year-old telephone wiring which works fine for what it always did (though it can't handle DSL), and the same thing will happen to the brandy-newest fiber optic cable when someone comes up with an LED laser frequency that needs a different glass with different chromatic aberration. There are lots of industrial and scientific devices out there that never got updated drivers past (whatever release of whatever system), and they cost a lot of money at the time, and they still work. (Though I admit that, while they may be worth maintaining, at some point one has to give up on trying to update them.)
I was assuming that EISA was just a special case inside of the same code as ISA and that what was proposed was to remove all ISA support. Is that what was going to happen?
ISA is old but I am sure there is quite a bit more than just one person out there with some sort of legacy hardware using it. I have a little bit of ISA hardware myself that I would like to use but not quite enough to build up a legacy PC. Every now and then I search the internet for ISA to USB adapters. There actually IS one company selling such a beast but it is way to expensive to be worthwile for me. But.. if I had some expensive piece of lab equipment or something like that with a proprietary ISA adapter... it would make sense.
I remember reading (back in Pentium I days) that NASA still used 386s for a lot of things that went into space. The larger transistors inside made the chip more radiation resistant.
I wonder if they ever found a way to make modern processors more radiation resistant or if they just added more shielding or maybe even still use the old stuff...
There is really no reason you need kernel 3.x 1980s hardware x86 HW. None at all.
Have you actually enumerated the individual computing needs of a hundred million people? It's a big world out there, and tunnel vision biases will cripple you.
Maybe, but I would hazard a guess that where this old tech is being used at all is likely to be in a resource poor environment.
Alternatively, one with long term support requirements and onerous change control: e.g. military, nuclear, medical.
In unrelated news, Mozilla cuts off digital signatures. Even if they have millions of users.
I have a bunch of these, and some top of th line IBM servers (c1991) to run them.
They are better than everything newer except the Aquaris... :)
Truth isn't Truth - Guliani
recent years have also seen MCA bus support being removed from the kernel
Just for reference, here's also the original discussion on MCA support removal from 2012 in LKML.
Get off our lawn.
I remember that every time I changed a card out the machine took 30 minutes to reconfigure itself, because some doochebag of a programmer wrote the #$%#$% configurator that all the vendors used. An operation that could have been done in 5 seconds if written properly. That was the first ... and last EISA machine I ever bought.
-Matt
I saw a lot of EISA systems. It was a reasonable performer and physically robust (not as sensitive as PCI cards to positioning in slots, etc.). I'd say that EISA hardware was generally of very good quality, but high-end enough that most consumers wouldn't run into it despite being a commodity standard, sort of like PCI-X.
The systems I had experience with were running Linux, even then. :-)
STOP . AMERICA . NOW
Didn't your momma teach you not to pick scabs?
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
perfect for a small, optional, add-on to systemd that you dont have to use if you dont want
I've got a bunch of PCI DVB/capture cards that are in the same boat. They *could* be useful if I had drivers for them, but alas, they do not.
My personal disfavorite is software that depends on dongles which have OS-specific drivers. The software *WOULD* work on the newer OS if the dongle had a driver that allowed it to authenticate (of course, dongles suck in general).
It's nice to see that Linux is still being developed by the people, for the people.
Political correctness is really just herd psychology pushed by insecure people who desperately seek social conformity.
You really don't want CA that aren't trusted in your browser... even if it breaks some sites... a broken CA can break ALL your https sites
We should have killed SSL3 years ago, but to avoid breaking some sites, we found later that SSL3 bug could break all sites, even if they had already TLS1.2
When something should die, let it die or you will pay for it later
Higuita
Probably not much time at all, if any.
It's just along for the ride now. Total economics fail would be spending time ripping it out.
Note that the '386 code was demanding actual effort to maintain so it's gone.
And maybe the users of these cards just have very high reliability needs and do not want to buy a new card that has a 10% chance to crap out in the first year. There may be a lot of other valid reasons. "New" does not imply "good". And if it works, do not break it.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
There is really no reason you need kernel 3.x 1980s hardware x86 HW. None at all.
I found it sad to lose 386 support because there are a bunch of cheap embedded boards with 386s on them still floating around out there. But not sad enough to do anything about it... Regardless, if I can't have a modern kernel on something then I mostly won't mess with it. It's just easier to not have to think your way around multiple versions.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
EISA was also used in the earliest RISC workstations that supported Windows NT 3.1. There was the MIPS Magnum R4000, and the DECpc AXP 150, both of which had EISA buses for NT, and equivalent Turbochannel based workstations for UNIX (Irix/RISC-OS for the Magnum R3000 and Ultrix for the DEC 3000 AXP). MIPS abandoned the model after its foray into the NT market ended quickly, while DEC replaced both the above models w/ the PCI based AlphaStations and AlphaServers.
You have many motherboards options with latest gen hardware and a parallel port, still.
That may work if your software is so backwards it needs to think the parallel port is attached to the ISA bus.
http://www.asrock.com/mb/Intel...
http://www.asrock.com/mb/AMD/A...
http://www.asrock.com/mb/Intel...
Hello, Arthur.
Tedious Bloggy Stuff - hooray?
So two thoughts come to me after reading the summary and the comments
1/ This clearly demonstrates that Linux is the right technology to use when supporting business syetms with legacy hardware (I'll second the opinions that enterprise keeps hardware around the longest - I supported users on obsolete AS/400s which, whilst not as old as what some people here talk about, still mean we're frequently learning old technology in Support); and the point that the leadership (Linus right now, hopefully the same with whomever comes to replace him eventually!) can be more ameanable to keeping up support for old hardware is great. I dream of desktop uptake, but enterprise and research are where it's at.
2/ However I also wonder - isn't this an offshoot problem of the fact that Linux is a monolothic kernel? Can this kind of interface-specific support not be modularized? Say, an API/ABI (in-kernel)standard that allows the kernel to plough on with currently evolving requirements, whilst maintaining a stable interface for previously integrated kernel features that have been split off into modules...?
(and no, I'm not at all familiar with the ins and outs of kernel development and architecture - I just read newsposts and Wikipedia ...)
-- "Simplicity is prerequisite for reliability." --Dijkstra
Isn't EISA (or ISA) still used on CURRENT industrial embedded systems like PC104 ? I'll soon know, I've just been asked to work on some... In that case it's very good if a modern kernel can support it.
Non-Linux Penguins ?
yes. At the same time, the real question is, why do you use a modern kernel? If you have hardware from the 1980s, why can't you also use software from the 1980s?
they still do actually(I think they are up to p1/p2 era for space). That said, a modern Linux kernel is not really the optimal choice, try something like an embedded OS.
It's good to know that the developers are listening to the community. Not sure how this particular user knew of EISA support being dropped but good that we can make sure the Linux kernel and GNU/Linux operating system survives as community-driven software which will always outdo proprietary software where it is driven by a single company's wishes. GNU/Linux is excellent at supporting new and old hardware and retaining backwards compatibility. Thanks again to all of the developers and for making GNU/Linux such a huge success.