Memory reconfiguration method in a data processing system

Claim 1

Memory reconfiguration method in a data processing system wherein the memory comprises a variable number of modules (M0, M1, M2, M3) up to a maximum of P modules, each module including a plurality of addressable locations and partitioned in a plurality of memory blocks each having equal number of locations, said memory blocks being shared among said modules, the addressing of a memory location involving the identification and the selection of the module containing said location through the addressing of a directory (19) having a number of addressable locations at least equal to the possible maximum number of said blocks, each directory location being associated to a block and loaded with a block information representative of the belonging of said block to one of said modules, said block information, when read out, generating a signal selecting the related module, characterized in that memories composed of modules with variable numbers of blocks can be reconfigured by the following steps: a) storing a first block information representative of the belonging of the related blocks to a first module in all said directory locations, b) storing test codes in predetermined memory locations belonging to different blocks, c) checking by reading said predetermined locations whether said predetermined locations and therefore their related blocks belong to said first module, d) storing another block information representative of the belonging of blocks, not belonging to said first module, to a subsequent module in all other directory locations corresponding to said blocks identified as not belonging to said first module, e) storing test codes in other predetermined memory locations belonging to different blocks identified as not belonging to said first module, f) checking by reading said other predetermined locations whether said other predetermined locations and therefore their related blocks belong to said subsequent module, g) repeating in succession steps d) e) f) for all the possible subsequent modules, h) storing a further block information which, when read, generates an overflow signal in directory locations related to blocks identified as not belonging to any module.