Re: NMI errors in 2.0.30??

Richard B. Johnson (
Sun, 27 Apr 1997 21:58:09 -0400 (EDT)

On Sun, 27 Apr 1997, Leonard N. Zubkoff wrote:

> Date: Sat, 26 Apr 1997 21:40:44 -0400 (EDT)
> From: "Richard B. Johnson" <>
> The kernel doesn't "know" anything about an ECC mode in the BIOS. The
> kernel presumes that all RAM found is good and whatever is written to
> the RAM can be read back exactly as written.
> I believe that if you look into the Pentium and Pentium Pro Machine Check
> Architecture and the documentation for the 430HX and 440FX chipsets, you'll
> find that there is support for determining the addresses where parity errors
> and ECC errors occur. There are ways the system can recover if the memory
> that's suspect is holding "clean" data, for example a cached copy of disk data.
> In that case, re-reading the data from disk is an option.
> Leonard
Of course there is support to find the address (range) where the RAM errors
occur. Nice catch when these data are being DMAed to disk, when instructions
are being fetched, when one-time input data is involved, etc, and etc.

Ever see "unable to handle paging request at 0x8CAAxxxx"? Does the pager
assume that its a single-bit error and "fix" the address in the pointer?
If so, how does it know how to fix the value in the user's pointer it
got copied from? Note that you pass a COPY of something to a procedure, not
the "thing" itself (well in FORTRAN, it seems as though you are passing
an entire array... but) By the time a bad pointer gets to the kernel, it
doesn't "know" where to fix the real problem.

Dick Johnson
Richard B. Johnson
Project Engineer
Analogic Corporation
Voice : (508) 977-3000 ext. 3754
Fax : (508) 532-6097
Modem : (508) 977-6870
Ftp :
Email :,
Penguin : Linux version 2.1.35 on an i586 machine (66.15 BogoMips).
Warning : I read unsolicited mail for $350.00 per hour. Supply billing address.