Re: CKD commands and PDS writing
- From: DASDBill2@xxxxxxxxxxxx
- Date: 29 Jul 2005 03:44:59 -0700
In a message dated 7/29/2005 3:40:29 A.M. Central Daylight Time,
gah@xxxxxxxxxxxxxxxx writes:
>As I understand, the complement to Format Write is READ TRACK
>which will read all the data on the track, counts, keys, and
>data blocks. If the rest of the track isn't physically
>erased then I would have thought READ TRACK would find it.
I wouldn't call Format Write the complement of Read Track. Read Track is a
specific command, and Format Write is a process employed by several different
kinds of specific write commands. There is also a non-formatting write
process. Format Write means that data is written onto a track and then the rest
of the track is at least logically erased after the last byte of that data is
written, whereas the logical erasure does not occur with a non-formatting
write command. Write Track, a command which would write an entire track rather
than just one record (aka block - see below), is the complement of Read
Track. There is a Write Track command for the 2105, but it is not publicly
documented yet.
Read Track command (X'DE'), depending on how it is oriented, can read all
the data records on a track including R0, but not the Home Address. After the
last data record is read, an extra 8 bytes of X'FF' are sent into central
storage to signal that the last record on the track was read. Read Track reads
records that logically exist (as opposed to physically exist) on the track.
I make the distinction because years ago when a track or part of a track was
logically erased it was also physically erased by writing X'00' bytes on the
track. Today's controllers do not necessarily physically erase data when a
logical erasure takes place. If you do a formatting write command that ends
at position X on the track, all the track bytes after point X are logically
erased, which means they cannot be read by any of the publicly documented
commands in IBM's control unit reference manuals, including Read Track.
Writing an EOF record on a track requires using the Write Count Key and Data
command, which is a Formatting Write, and thus causes logical erasure of all
track data after the EOF record.
In IBM's control unit reference books, which document the CCW commands that
interact with DASD controllers, the word "record" is used to describe the unit
of stored data on a track which consists of a count field, key field, and
data field. IBM's software access methods call this storage unit a "block" and
have a different meaning for "record", which is an unfortunate source of
confusion. To be more precise, the data field of a record on a track is
exactly equal to the block that is transferred by access methods, which may then
interpret the block as having one or more logical records within it. A
physical record on a track as viewed by the commands described in control unit
reference manuals equates to a logical block as viewed by access methods.
Bill Fairchild
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to listserv@xxxxxxxxxxx with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
.
- Follow-Ups:
- Re: CKD commands and PDS writing
- From: Bruce Black
- Re: CKD commands and PDS writing
- Prev by Date: Re: The secret to multi-volume, non-SMS, DASD allocations?
- Next by Date: Re: How to detect a missing link list library?
- Previous by thread: RE: CKD commands and PDS writing
- Next by thread: Re: CKD commands and PDS writing
- Index(es):
Relevant Pages
|
