Method and means for CPU recovery of non-logged data from a storage subsystem subject to selective resets

Claim 1

A method for the automatic recovery of data from a defaulting one of a pair of failure independent control units, each control unit having means for switchably interconnecting channel to device interfaces, means for the registration of its error state, and a control unit/control unit communications interface, each interface being of the asynchronous demand/response type, the method steps comprising: notifying a channel accessing the defaulting control unit of its unavailability responsive to a trouble signal originating within said defaulting control unit concurrent with notifying the companion control unit of this status; peer coupling data from the defaulting control unit to a channel interface coupling the companion control unit through and under the control of the companion control unit over the communications interface responsive to the unavailability status indication; and buffering any accessing channel response at the defaulting control unit to the unavailability notification until said data has been offloaded subsequent to which the channel response may be executed.

Claim 2

A method for the automatic recovery of data from a defaulting one of a pair of failure independent control units, each control unit having means for switchably interconnecting channel to device interfaces, means for the registration of its error state, and a control unit/control unit communications interface, each interface being of the asynchronous demand/response type, the method steps comprising: notifying a channel accessing the defaulting control unit of its unavailability responsive to a trouble signal originating within said defaulting control unit concurrent with notifying the companion control unit of this status; peer coupling data from the defaulting control unit to a channel interface coupling the companion control unit through and under the control of the companion control unit over the communications interface responsive to the unavailability status indication; and buffering any accessing channel response at the defaulting control unit to the unavailability notification until said data has been offloaded subsequent to which the channel response may be executed.

Claim 7

A method for the recovery of non-logged error data by a CPU/channel from a storage subsystem, which data would normally be overwritten upon the subsystem or portion thereof being RESET by the CPU/channel, said subsystem comprising a pair of failure independent control units, each control unit having means for switchably interconnecting channel to device interfaces, means for the registration of its error state, and a control unit/control unit communications interface, each interface being of the asynchronous demand/response type, the method steps comprising notifying a CPU/channel accessing the defaulting one of a control unit pair of its unavailability responsive to a trouble signal originating within said defaulting control unit concurrent with notifying the companion control unit of this status, said step of concurrent notification further including the steps of initiating a DISCONNECT IN sequence to the accessing channel and setting an appropriate status bit in the error registration means of the companion control unit; peer coupling data from the defaulting control unit to a CPU/channel interface coupling the companion control unit through and under the control of the companion unit over the communications interface upon the companion control unit polling the status bit; and buffering any accessing channel response at the defaulting control unit to the unavailability notification until said data has been offloaded, subsequent to which, the channel response may be executed such that the accessing channel responsive to the DISCONNECT IN sequence with a SELECTIVE RESET having said reset honored after the offloading by erasing all data in the defaulting control unit.

Claim 8

A method for the recovery of non-logged error data by a CPU/channel from a storage subsystem, which data would normally be overwritten upon the subsystem or portion thereof being RESET by the CPU/channel, said subsystem comprising a pair of failure independent control units, each control unit having means for switchably interconnecting channel to device interfaces, means for the registration of its error state, and a control unit/control unit communications interface, each interface being of the asynchronous demand/response type, the method steps comprising notifying a CPU/channel accessing the defaulting one of a control unit pair of its unavailability responsive to a trouble signal originating within said defaulting control unit concurrent with notifying the companion control unit of this status, said step of concurrent notification further including the steps of initiating a DISCONNECT IN sequence to the accessing channel and setting an appropriate status bit in the error registration means of the companion control unit; peer coupling data from the defaulting control unit to a CPU/channel interface coupling the companion control unit through and under the control of the companion unit over the communications interface upon the companion control unit polling the status bit; and buffering any accessing channel response at the defaulting control unit to the unavailability notification until said data has been offloaded, subsequent to which, the channel response may be executed such that the accessing channel responsive to the DISCONNECT IN sequence with a SELECTIVE RESET having said reset honored after the offloading by erasing all data in the defaulting control unit.