home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!noc.near.net!hri.com!spool.mu.edu!agate!ucbvax!lrw.com!leichter
- From: leichter@lrw.com (Jerry Leichter)
- Newsgroups: comp.os.vms
- Subject: re: Re: Question about RMS and MSCP-pair
- Message-ID: <9301061230.AA17633@uu3.psi.com>
- Date: 6 Jan 93 11:30:25 GMT
- Sender: daemon@ucbvax.BERKELEY.EDU
- Organization: The Internet
- Lines: 38
-
-
- [I wrote:]
- > By the time you get a "forced error", the data is gone. The forced
- > error indicator is set on the REPLACEMENT block, not the original
- > one, and I don't believe the corrupted data, such as it was, is even
- > copied into the block.
-
- It's implementation-specific, according to the MSCP architecture spec.
- There is no reasonable reason to copy the data in the case of an
- uncorrectable error. However, some of the controller-based BBR
- implementations use a common code path with the correctable error /
- revector code and have a flag to say whether or not the forced error
- flag should be set. These implementations do in fact preserve the
- data, although it is certainly not the _correct_ data.
-
- Well, it's good to know my memory on this subject isn't COMPLETELY wrong!
-
- BTW, the "forced error flag" is set by computing the CRC
-
- I assume you mean ECC here.
-
- of the new
- block and writing it out complemented. Just a useless tidbit of trivia
- (unless you are going to physically talk to a drive with a controller
- of your own design 8-).
-
- This I didn't know. Interesting hack - with an interesting side-effect: If
- a replacement block with the "forced error flag" set gets hit by an error, it
- will no long appear to have the "forced error flag" set - it'll just appear
- wrong. Worse, there will be some class of hits which will look as if they are
- correctable errors, assuming the complemented ECC is the REAL ECC. In that
- case, a "forced error" block my spontaneously "correct itself" - but the data
- in it will be completely bogus. Presumably this will happen once on all the
- disks in the world within the next 10,000 years, or something like that - but
- it does tell you that you don't really want to keep files with forced errors
- in them around for long periods of time. If a forced error shows up in your
- backup runs - do something about the file.
- -- Jerry
-