AMD owns up to CPU bug

AMD owns up to CPU bug

Summary: Confirming the bug involved writing a custom fully bootable version of the Dragonfly OS.

TOPICS: Processors, Hardware

Updated: Statement from AMD.

AMD has confirmed that some of its processors contain a bug that could cause program errors under certain specific conditions.

The bug was initially discovered by Matt Dillon of OS developer DragonFly BSD, who outlined the issue in a newsgroup post late last year. AMD has now confirmed to Dillon that the bug does indeed exist in some of its CPUs.

"AMD has taken your example and also analyzed the segmentation fault and the fill_sons_in_loop code. We confirm that you have found an erratum with some AMD processor families. The specific compiled version of the fill_sons_in_loop code, through a very specific sequence of consecutive back-to-back pops and (near) return instructions, can create a condition where the process or incorrectly updates the stack pointer."

Dillon says that he was "pretty stoked," because, "it isn't every day that a guy like me gets to find an honest-to-god hardware bug in a major CPU." Confirming the bug was tricky and involved writing a custom fully bootable version of the Dragonfly OS.

AMD offered the following statement:

A program exception has been identified in previous generations of the AMD Opteron processor that occurs in certain environments that leverage a very specific GCC compiler build. A workaround has been identified for the small segment of customers this could potentially impact.

Also, this marginal erratum impacts the previous four generations of AMD Opteron processors which include the AMD Opteron 2300,8300 8300("Barcelona" and "Shanghai",) 2400, 8400 ("Istanbul",) and 4100, 6100 ("Lisbon" and "Magny-Cours") series processors.

It's worth pointing out that what's special about this story is not the fact that a bug was discovered in a CPU (errata for CPU bugs is common and nothing special or newsworthy most of the time) but the fact that it was discovered by a third-party software developer.

The news of this bug comes a day following the announcement that AMD was giving up its remaining 8.8 percent equity stake it held in GlobalFoundries after spinning off its manufacturing arm three years ago. This move will force the company to take a charge of $703 million in the first quarter but will allow it to be more flexible in choosing manufacturing partners, GlobalFoundries and TSMC, for its 28nm products.

Topics: Processors, Hardware

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • So the chances of this affecting someone in the real world is?

    Pretty much non-existent?
    • I'll remind you...

      ...that MacOS X is BSD-based and there has been talk about Apple using AMD processors.

      Besides, what guarantee do you have that it doesn't affect Windows and Windows applications as well?

      Reply to apetti:

      We're talking about a hardware bug. What would GCC have to do with anything?
      John L. Ries
      • I'll remind you...

        GCC processing, if I am not mistaken, cannot hafe any efferct on Windows, but as for OS X, I am not sure. GCC, GNU Compiler Collection would be Linux specific... I know OS X is built on a UNIX platform, but there may be some Linux attributes in place on OS X.

        That is a good question! I would imagine that an update to the chips is fairly simple... Lets see how this pans out!
      • RE: I'll Remind you

        1. Well this is for older generation processors
        2. It has very specific conditions to even get the bug to happen

        Don't get me wrong, it is a bug and while things could have been worse I do not think this is a huge deal by any means. I doubt it will affect Apple if any of those AMD rumors are true. Intel has had similar bugs that came up in rare and very specific circumstances. I wasn't trying to be a smart a$$ with my post it was truly a legit question about the chances of affecting the vast majority of users of these processors. I am fairly certain that 99.99% of the people and organizations using these processors will never have any issue.
      • Servers

        The Opterons are AMD's high end server chips. You're not going to find them in a Macintosh.
        Robert Hahn
      • Yes, OS X is BSD UNIX: Not Linux

        Apple does not have any AMD CPU server currently. If they ever build such, it will be with the latest generation CPUs.
        OS X does not use GCC. They have heavily invested in a more modern alternative: CLANG/LLVM. It is also not encumbered with GPL.
        The latest FreeBSD 9 also includes fully functional CLANG/LLVM compiler suite that can replace GCC for kernel/system/ports compilation. In later versions it will likely be the default compiler.
        There is likely a lot of Windows software that is compiled with GCC, but base Windows is not.
        Any of this does not mean that the same code may not be generated by any other compiler, of course. Now, that the bug is known, new compilers will simply make sure to not generate such code.

        Just like with the many Intel Pentium bugs, the required workarounds will be implemented either in GCC or the underlying OS.
  • Really!?!?!

    There is a "major" fault with this CPU. If it is in a geosynchronous orbit, while close the the moon, and a monkey is traveling at exactly 28.4 meters per second at the same time, there is a small chance that the monkey could impact the CPU causing a malfunction. I believe we should pull all these CPUs from the marketplace while we research a more durable chip.

    Pretty much the same thing except I'd pay for the newspaper that had a picture of that monkey.