Date: Tue, 28 Dec 1999 15:48:12 -0800 From: Lance Costanzo <lance@costanzo.net> To: freebsd-hardware@freebsd.org Subject: Re: ECC RAM useless with FreeBSD? Message-ID: <3.0.32.19991228154805.006cc640@costanzo.net>
next in thread | raw e-mail | index | archive | help
At 12:36 PM 12/28/1999 -0800, you wrote: >> I had thought that ECC RAM would enable recognition of memory >> failures (in addition to automatically fixing simple one-bit >> errors). I thought that the chipset would cause a CPU trap or >> something like that if an unrecoverable RAM error occurs, so >> the OS could take some appropriate action (at least produce >> a log message). Am I wrong? It seems to just fail silently, >> possibly even continuing to use wrong data in calculations, >> just as if there was only non-ECC RAM. >We don't, at this point in time, do anything with unrecoverable ECC >errors to the best of my knowledge. On HP 3000/9000 systems when a single bit errors occurs, the error is logged, the memory page is "scrubbed", and the memory page would be taken out of the pool if the scrubbing failed (hard single bit error). A double bit error causes an HPMC or LPMC (high/low priority machine check), and the system promptly halts. The older HP systems used proprietary memory, but many of the newer ones use 72pin ECC simms - I don't think there's anything special about them. An old PC (386) I used to have used 30-pin parity simms and would at least say "parity error" and crash. I guess the question is- Is not being able to log memory errors a problem with the PC hardware being inadequate, or feature missing from the operating system? Lance Costanzo http://www.webhighrise.com System Administrator Website and Virtual Domain Hosting lance@costanzo.net starting at $5/month, no setup fees To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3.0.32.19991228154805.006cc640>