Dec 112017
Scans hard disk for soft errors. Detects possible problem spots. | |||
---|---|---|---|
File Name | File Size | Zip Size | Zip Type |
SOFSCAN.DOC | 7221 | 3190 | deflated |
SOFSCAN.EXE | 60949 | 23209 | deflated |
TPCREAD.ME | 199 | 165 | deflated |
Download File SOFSCAN.ZIP Here
Contents of the SOFSCAN.DOC file
RELEASE NOTES Pre-Release Version 0.5
September 25, 1990
OVERVIEW:
SofScan is a storage media laboratory test utility; one
you might expect to find in use by a certification/refurbishment facility.
SofScan tests DOS formatted media response under user controlled soft error
threshold levels, reporting the slightest discrepancy. Other disk test soft-
ware that relies on the BIOS for hardware status will usually locate only the
unrecoverable (hard) errors. The job of the disk hardware and BIOS is simply
"whatever it takes... get the data". Usually the BIOS design does not com-
plain to DOS or concern the user with the expected occasional recoverable
(soft) error. The hardware and BIOS may only, after exhaustive "grinding",
return an error status in lieu of the data. This good/bad method unfortunately
masks the marginal soft error media surface defects. Ignored, many of these
soft spots deteriorate and the user is informed (only after it is too late)
with the old familiar prompt "Abort, Retry, Fail?".
SofScan scrutinizes the disk read operation fully, thus flushing out the up
and coming media errors. Periodic use allows the discriminating PC user to
maintain an above average confidence level in his storage media. Many
previously unsalvageable hard disks have been placed back in service with the
help of SofScan.
FEATURE SUMMARY:
SofScan is a multi-step test process. A minimum of
four scans repeatedly READ the disk test area. The
final critical step, marking-BAD clusters, is turned
over to Norton Utilities DT-Disk Test. As many as 999 scans can be used to
certify continuous duty capable DOS media storage logical units. Any single
error incidence is trapped and reported. Files using error prone clusters are
reported by cluster number and file name. User settings customize the error
sensitivity level. A continuous dated record of cluster error history may be
maintained on the SofScan diskette. The error list is written as a batch file
named CLUSTERS.BAT on drive A:.
Norton's DT-Disk Test has an often overlooked command line option (/Cnnn) that
manually marks clusters BAD. Conversely, GOOD clusters which are marked BAD
may be returned to service with the switch /Cnnn-. DT verifies that the
to-be-marked cluster is not INUSE by a file.
Upon detection of an INUSE soft cluster, SofScan locates the complete path and
file name to include in the error report. This feature facilitates the manual
process of using DOS COPY to move each file to safety. The original must be
erased to release the INUSE soft cluster for marking BAD. Always mark any
UNUSED BAD clusters before moving the associated files of INUSE BAD clusters.
Of course you may mark the affected clusters by any other means available.
OPTIONS DESCRIPTION:
The Soft Error Threshold option directs SofScan to
lower or raise the error determination level. Level
[0] might cause SofScan to issue false error reports.
Alternately, Threshold Level [9] is capable of masking minor defects. By
definition, soft errors are often inconsistent. Continued testing is recom-
mended until new errors fail to appear. Using the Beginning and Ending Cluster
options you can limit testing to one area. Passes through the test clusters
may be set from [4] to [999] to facilitate overnight or even week-long testing
sessions. SofScan's unique process traps and reports any single cluster read
retry. The estimated test time is updated following any option selection
affecting the total run time.
Test Clusters previously marked BAD [ReTested] will verify marked-BAD
clusters. Thus allow the user to mark GOOD a cluster if it appears to be
mistakenly marked-BAD. Do not be whimsical about marking-GOOD a BAD cluster;
soft errors are inconsistent.
Caching controllers read the entire track when any cluster on the track is
requested. Therefore, all clusters on a track with a real error will, in
succession, report as BAD. If the Passes setting is less than the number of
clusters per track, additional sessions will continue to tag the remaining
clusters of the bad track. Disk Caching Software distorts the cluster
granularity even further. It must be removed to obtain valid test results.
Error level 32512 is forced for very bad clusters. This level may also be
reported erroneously if timer interrupts are masked by the Disk BIOS.
TESTING OPERATIONS - PHASE ONE:
Requires: SofScan.EXE in Diskette Drive A:
1. Remove all Disk Cache, TSRs, etc. The easy method is to boot DOS from a
floppy which bypasses the normal CONFIG.SYS and AUTOEXEC.BAT hard disk
startup routines. You may make a boot floppy entering: FORMAT A:/S
2. Log to the drive to be tested (SofScan only tests the default drive).
Insert the SofScan program and test report diskette in drive A: and enter:
A:SofScan
3. Select test options from the menu and press G to Go or Q to Quit.
4. As SofScan tests the disk the current pass and minutes required are shown.
The current test cluster displays at the right edge of the screen.
5. After the last pass, SofScan will pause to allow you to review the screen
error listing. A duplicate is saved to the output file CLUSTERS.BAT.
SofScan's command line options are /B,/E,/S,/P,/T for Beginning, Ending,
SoftErrorThreshold, Passes and BadTesting respectively.
MARKING OPERATIONS - PHASE TWO:
Requires: Norton Utils DT-Disk Test 4.00 DOS 3.30 dn
Norton Utils DT-Disk Test 4.50 DOS 3.31 up
A:clusters.BAT
1. Ensure that DT-Disk Test is on your hard disk and can be found by the PATH.
You may review A:CLUSTERS by entering: A:clusters [with no keyword]
2. Mark UNUSED BAD clusters, if any, first by entering: A:clusters MARKBAD
3. If INUSE bad clusters are reported move each file to a safe location using
DOS COPY if possible. (If not, STOP and seek professional assistance in
recovering critical data.) Erase each original file to release the INUSE
BAD clusters to be marked BAD. Enter: A:clusters FILESMOVED
4. If GOOD clusters are found to be marked-BAD and you are confident they
are dependable, enter the following: A:clusters MARKGOOD
Thank you for your support! If you have found SofScan useful, the nominal
licensing fee of $15 would, no doubt, be very appreciated. Don't hesitate
to call with questions, suggestions, or development project proposals.
Allen Swann, 4162 Buglers Rest, Casselberry, Florida 32707 (407) 695-8859
December 11, 2017
Add comments