Linux kernel 7.0 finally abandons the 28-year-old Intel 440BX chipset's EDAC driver — removal marks goodbye to the legendary motherboard chipset

Intel MMX
(Image credit: Getty Images)

Intel's 440BX motherboard chipset was arguably the finest of such specimens to have ever existed. Nothing lasts forever, though, and the upcoming Linux kernel 7.0 is now dropping support for the chipset's EDAC driver (hat tip to Phoronix).

The code hasn't been functional since 2007 due to incompatibilities with the more widely used Intel AGP driver. The lack of EDAC meant that 440BX machines with ECC RAM would still fix errant bit flips, but without software-side notifications. The Intel AGP driver, on the other hand, is used by dozens of older chipsets. Now, 440BX EDAC support has been officially removed, not just disabled.

The arrival of the 440BX was a breath of fresh air in one fell swoop, solving both equations at once — and then some. It was stable as a rock, with comparatively few incompatibilities, would handle most out-of-spec hardware with aplomb, and was fast as heck. That was enough to immediately grant it the king's crown, but then there was the matter of overclocking.

Google Preferred Source

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

TOPICS
Bruno Ferreira
Contributor

Bruno Ferreira is a contributing writer for Tom's Hardware. He has decades of experience with PC hardware and assorted sundries, alongside a career as a developer. He's obsessed with detail and has a tendency to ramble on the topics he loves. When not doing that, he's usually playing games, or at live music shows and festivals.

  • kneekoo
    Good thing my motherboard has a 400LX chipset. :LOL:(y)
    Reply
  • avatar4886
    Admin said:
    Linux kernel 7.0 does away with 440BX EDAC driver

    Linux kernel 7.0 finally abandons the 28-year-old Intel 440BX chipset — driver removal marks goodbye to the legendary motherboard chipset : Read more
    Fix the cpu picture at the top of the article. The 440bx chips was for pentium 2 and 3 not the socket 7 based pentium mmx
    Reply
  • bit_user
    The title said:
    Linux kernel 7.0 finally abandons the 28-year-old Intel 440BX chipset — driver removal marks goodbye to the legendary motherboard chipset
    I rarely find fault with Bruno's articles, but this title is just outright wrong. They didn't drop support for the chipset driver. All they did was remove the code needed to support its EDAC functionality.

    The article touches on the point about the EDAC driver and what it does, but fails to mention that its EDAC driver has been broken for 19 years, without anyone noticing or bothering to fix it. Also, fails to clearly state that 440BX-based machines will still continue to work in other respects.

    In Linux, it's customary to drop code that no one is willing to step up and actively maintain. When a driver is dropped, anyone still using it has the option to stay on a LTS kernel release prior to when it was dropped. In this case, even if they stayed on a prior LTS kernel, they'd still need to devise a fix for it and patch their kernel with it.

    Finally, it's worth remembering that 440BX is so old that it supports neither SATA nor PCIe. Of course, you can use a PCI SATA controller card, if you wanted to use newer storage devices with it. But, since the supply of IDE drives has long since dried up, it's likely that most such systems were taken out of service after their HDD died and couldn't easily be replaced.
    Reply
  • bit_user
    Also, I've got to point out that ECC support for many Alder Lake and Raptor Lake CPUs is broken in the current Intel EDAC driver. The maintainer of that driver accepted at least one patch to fix a couple of those models, without even lifting a finger to deal with the others.
    Reply
  • mitch074
    bit_user said:
    I rarely find fault with Bruno's articles, but this title is just outright wrong. They didn't drop support for the chipset driver. All they did was remove the code needed to support its EDAC functionality.

    The article touches on the point about the EDAC driver and what it does, but fails to mention that its EDAC driver has been broken for 19 years, without anyone noticing or bothering to fix it. Also, fails to clearly state that 440BX-based machines will still continue to work in other respects.

    In Linux, it's customary to drop code that no one is willing to step up and actively maintain. When a driver is dropped, anyone still using it has the option to stay on a LTS kernel release prior to when it was dropped. In this case, even if they stayed on a prior LTS kernel, they'd still need to devise a fix for it and patch their kernel with it.

    Finally, it's worth remembering that 440BX is so old that it supports neither SATA nor PCIe. Of course, you can use a PCI SATA controller card, if you wanted to use newer storage devices with it. But, since the supply of IDE drives has long since dried up, it's likely that most such systems were taken out of service after their HDD died and couldn't easily be replaced.
    Well, we're talking a 25+ years old chipset, which is still used as the basis for many VM. It was also the best chipset Intel offered before they went nuts on RAMBUS and Netburst for almost half a decade, and it was supremely stable - even without ECC, you could throw pretty much any stick of RAM at it and it would run it at max clock speed with ridiculous latency - Abit B*6 motherboards were well-known for that.
    I knew the chipset supported ECC on server boards (it actually supports dual CPU setups and was the basis for many a cheap servers using hacked Celerons to run at 2*450 MHz), but even without it (and let's be frank, nobody missed it in the 19 years the driver has been broken), you could still run a nice server off of it using a couple GHz P-III processors and a SCSI controller card.
    As for the lack of IDE drives for boot, there's actually a very easy workaround : CompactFlash cards and a dumb adapter. 100% compatible, all brand new, quite cheap, no latency and can saturate the UDMA 33 IDE controller's bandwidth easily - perfect retrogaming machine.
    Last time I tinkered with one, the biggest problem I had was having the AGP port work at full speed.
    Reply
  • Tbonius
    This brings back memories. I had an ASUS CUBX-E motherboard and with a coppermine processor and micron ram, I had the front bus clocked to 150. Man, that thing was a go'er.
    Reply
  • Bruno Ferreira
    @bit_user thanks for the feedback. The title was tweaked shortly after publication to clarify; it was an internal miscommunication.
    Reply
  • bit_user
    mitch074 said:
    Well, we're talking a 25+ years old chipset, which is still used as the basis for many VM.
    The VM thing is not relevant to the EDAC support. The hypervisor will log and report ECC errors, so it doesn't matter if the guest OS supports it or not.
    Reply
  • mitch074
    bit_user said:
    The VM thing is not relevant to the EDAC support. The hypervisor will log and report ECC errors, so it doesn't matter if the guest OS supports it or not.
    Yeah, you don't really want to emulate ECC support. No-one cared for ECC support for this chipset in Linux for 19 years now either, so, frankly, it's moot.
    Reply
  • bit_user
    mitch074 said:
    No one cared for ECC support for this cipset in Linux for 19 years now either, so, frankly, it's moot.
    Well, part of the problem is that most people don't know how to check whether it's enabled or not. They just put ECC RAM in a system + CPU that supports it and assume it's working.

    So, while I agree that ECC support for this system is no longer an issue today, I think we certainly can't say that was true 19 years ago!!
    Reply