DASD cache block staging

Claim 3

A method according to claim 1, wherein: the cache directory includes DASD track and block descriptors in which the block descriptors are chained to form a circular list and further wherein each block has associated therewith a use bit; and the step of blocking and staging records to cache if the missed record spans more than a block extent includes the steps of (a) turning the use bit on when data in the block is referenced in the cache; (b) cycling a cursor through the set of block use bits in round robin order and turning the use bit off as the cursor advances; and (c) upon a preselected number of contiguous blocks being needed, selecting the first least recently used contiguous blocks whose use bits are marked off equivalent to the missed record spanning distance.

Claim 4

A method according to claim 1 in which the cache is either of the write-through or of the write-in types.

Claim 5

A new use of a data processing system comprising at least a first and second CPU; shared DASD devices; a plurality of sets of fan out paths (channels), each path set coupling a corresponding CPU as a source node; and a plurality of control nodes, each control node including means for accessing the devices and for selectively intercepting predetermined paths from either set in order to complete a single path connection to a single destination device; and a cache LRU managed through a directory coupling and operatively engaging said control nodes, said new use comprising a method for replacement of fixed block extents of information in the cache, each block either spanning at least one variable length record, or being one of several blocks spanning a single record, said method being invoked by reference by either CPU to an unavailable or missed record in the cache, comprising the steps of: blocking and staging records to the cache, said blocking and staging including comparison matching of the DASD and cache stored record identities; and copying onto the cache the DASD-recorded consecutive bytes of information starting at the DASD reference track from the missed record, and continuing until either a fixed block is spanned, the end of track is reached, or a record already in the cache is encountered such that: (a) if the missed record spans less than a block extent, ascertaining whether any record in the block has already been copied into the cache and deleting the first such record and all succeeding records from the block; or (b) if the missed record spans more than a block extent, allocating at least the missed record spanning equivalent of the least recently used blocks within the cache and copying said missed record into the cache from the DASD.

Claim 6

A new use of a data processing system comprising at least a first and second CPU; shared DASD devices; a plurality of sets of fan out paths (channels), each path set coupling a corresponding CPU as a source node; and a plurality of control nodes, each control node including means for accessing the devices and for selectively intercepting predetermined paths from either set in order to complete a single path connection to a single destination device; and a cache LRU managed through a directory coupling and operatively engaging said control nodes, said new use comprising a method for replacement of fixed block extents of information in the cache, each block either spanning at least one variable length record, or being one of several blocks spanning a single record, said method being invoked by reference by either CPU to an unavailable or missed record in the cache, comprising the steps of: blocking and staging records to the cache, said blocking and staging including comparison matching of the DASD and cache stored record identities; and copying onto the cache the DASD-recorded consecutive bytes of information starting at the DASD reference track from the missed record, and continuing until either a fixed block is spanned, the end of track is reached, or a record already in the cache is encountered such that: (a) if the missed record spans less than a block extent, ascertaining whether any record in the block has already been copied into the cache and deleting the first such record and all succeeding records from the block; or (b) if the missed record spans more than a block extent, allocating at least the missed record spanning equivalent of the least recently used blocks within the cache and copying said missed record into the cache from the DASD.

Claim 7

A new use according to claim 4, wherein the directory for each cache contains track and block descriptors in which the block descriptors are chained to form a circular list, and the steps of blocking and staging records to cache includes the steps of finding the track descriptor entry in the respective cache directory by using a requesting track identity by way of a hash table lookup.

Claim 8

A new use according to claim 4, wherein after a block has been staged from DASD, checking for duplicate records within the respective cache directory and deleting such records and all succeeding records from the block just staged.

Claim 9

A new use according to claim 4, wherein each control node includes a cache manager resident within and executable by the control nodes, said cache manager constituting a portion of an operating system equivalent for said nodes.