⟦d6588c976⟧

TextFile

«ff»
*******************************************************

         R E L E A S E   D E S C R I P T I O N

*******************************************************




Module id number: CSS/719
--------------------------

Module name:      TEST-MAGTAPE
-----------------------------------

Actual release:   02.02            Release date: 840503
-----------------------            --------------------

Previous release: 02.01            Release date: 840102
-----------------------            --------------------




New facilities:
---------------

>> Release 0202 <<

 - A higher reportlevel(REPORT:3) gives the amount of output
   obtained by reportlevel 2 and the contents of the MTC
   commando buffer as soon as a module error is detected.

 - A 4k version only using the uppermost 4K words of the
   controller RAM is now available.
   The 16K version is of cause still available but the test tape
   format has been changed slightly to support a full 16K block
   which means that test tapes generated by older versions will
   cause an error during test 1.2 (SDR = Short Data Record) .


   ***** NOTE: *****

   A test tape generated by the command GENERATETAPE can
   only be used for test purposes with the same version (4K or
   normal) that generated it.


>> Release 0201 <<

   Also available for MX-AMOS.

 - The Magtape Controller RAM had to be placed in the topmost
   16K of page 3. Now the address must be specified by the user!
   An existing parameter DEVICE and a new parameter MEMORY are
   interpreted thus :

   DEVICE[15:4]  : crate no (relevant in MX)
   DEVICE[11:6]  : must be 0 (checked)
   DEVICE[ 5:6]  : device addr

   MEMORY[9:10]  : offset to crate start in kilowords
   MEMORY[15:6]  : must be 0 (checked)

>> Release 0103 <<


   A three letter mnemonic, and a 'M' followed by a number are
   used to designate the module errors.
   In case of errors during a SKIP operation, the number of
   records or filemarks not skipped is stated if reportlevel = 2.
   (CPECR 345).

   A multiskip of filemarks is implemented as a true multiskip,
   instead of one skip at a time.
   (CPECR 383).

   When using the REPEAT command, errors will be flagged with
   the current value of the loop counter.
   (CPECR 429).

Changes:
--------

>> Release 0202 <<

   The maximum block on a test tape is 32768 bytes and not 32767 bytes long.

   After receiving an interrupt the interrupt queue is emptied and eventual
   extra interrupts are flagged with the warning output when REPORTlevel=3:

   *** WARNING *** <n> REDUNDANT INTERRUPT(S) AFTER <m>
   TRANSFER STATUS <binary status> = <hexa status>

   This resolves MX PR 18

>> Release 0201 <<

   The PRIORITY parameter was not necessary and has been replaced by
   a useful command MEMORY. The interpretation of the new parameter
   is described above.

   The pseudo-random number generator was of poor quality (bit 15
   was never raised) and has been replaced by a better one.

   Test 6.2 (Erase with variable length) was sensible to the
   kind of tape station used. As no known S/W uses this option
   it has been decided to omit the test i.e. the test reports
   OK but cannot be trusted.
   (PR 541)

   OPERATION TIMEOUT is now considered a fatal error terminating
   a test sequence. The format of the error message is changed to

   OPERATION TIMEOUT AFTER <t> MSEC.

   (PR 542)

Errors corrected:
-----------------

>> Release 0202 <<

 - The decoding of errors took place even when the TCR bit was low.

 - The test of the TCR bit (READY test) was no good.

 - The program was sensitive to a parity error in the last byte of the
   MT RAM and would terminate with a run time error during test 1.2
   because the CPU does not allow you to read a byte if the other byte
   of the same word has bad parity. Even if the error stems from a HW
   pecularity it was decided to circumvent the error in SW by using
   only even byte lengths in the predefined tests !!

Reported errors, not corrected:
-------------------------------

CPECRs implemented in current release:
--------------------------------------

  MX PR 18

Comments:
---------
 «a5»
DataMuseum.dk

CR80 Hard and Floppy Disks

⟦d6588c976⟧ TextFile

Derivation

TextFile
