lkml.org 
[lkml]   [2003]   [Apr]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    Subject(OT) md5sum proving to be an EXCELLENT memory test
    Because the subject of "bad RAM" comes up from time to time on this list, I 
    thought I would post an observation I made over the weekend. I was working
    with some server boxes with "generic" RAM that would mis-behave from time
    to time, and I really didn't understand why that would happen. The
    misbehavior included some OOPSes from time to time, but they would be
    random and so I didn't report them.

    Then I started to grab Red Hat Linux version 9, and because the server
    boxes were the only ones with enough open disk space I started to download
    the ISO images to them. Imagine my surprise when I went to run md5sum to
    check the download -- they would fail the test. Perform the test multiple
    times, and the failure pattern would CHANGE. (Copy the ISOs to another
    system via sftp, along with the MD5SUM, and they checked as
    perfect.) Perform md5sum on the files on the server and save the results,
    and the signatures would be different from run to run on the same files.

    Incompatible RAM.

    Proof: do a kernel computer, get signal 11. Put the "right" RAM in the
    boxes (in this case, a Viking PE8641U4SN3L-CL3 64-MB stick in an Intel
    CA810E motherboard) and both md5sum and a kernel compile worked just swell.

    By the way, md5sum on small files worked perfectly on these server boxes
    with IFA (Taiwan) 128-MB RAM installed, but on large files it would create
    the varied signatures.

    So, along with the test of compiling a kernel and seeing if you get signal
    11 messages, you can also load up some large files and run md5sum on them a
    couple of times. With six ISO images totalling 1.4 GB, it took a 500-MHz
    Pentium II about five minutes to run through the files and show the errors
    -- an order of magnitude faster than trying to do a kernel compile.

    FWIW

    Stephen Satchell

    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2005-03-22 13:34    [W:0.021 / U:1.048 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site