Abstract: An error recovery process for a non-synchronous DASD control unit. A channel process within the control unit is able to operate on different fields from a device process within the control unit. Independent errors may be sensed by each process. The error recovery process provides for restarting only the trailing process and deferring handling of the error sensed by the leading process, whenever possible.

Abstract: A cached DASD peripheral subsystem is operated in a so-called non-synchronous mode, that is, device operations are not synchronous to the host processor operations. A rate-changing buffer is interposed between the channel attachment for the host processor and the DASD, preferably having a storage capacity for one full DASD track of data. The rotational position for the data records being transferred in either direction, i.e., write or read, are kept in the rate-changing buffer to be kept with the record being transferred through the rate-changing buffer. Each time the peripheral subsystem supplies ending status to the host processor and all records to be transferred are retentively stored in DASD, a record control field and the rotational position of the last record transferred in a chain of commands being transferred the rotational position of such last record, its type, and other control data are stored in a control store separate from any cache directory or track directory in cache.