Show description

A corresponding value table containing values for scalars and codewords for arrays named in the symbol table. 27 Early Descriptor Architectures 28 The remainder of memory is allocated dynamically to user programs and data, including those addressed through the value table. Figure 2-4 shows the structure of a Rice University Computer sample procedure. Procedure instructions can address variables within the procedure segment without reference to codewords (that is, relative to the program counter).

Access to objects outside the domain can occur only through invocation of a C-list capability; the possessor of a capability invokes an operation on the object it addresses by specifying the capability, the operation to be performed, and other optional parameters. A process is the execution entity of a domain, and its C-list may contain capabilities for other subordinate processes over which it exercises control. Capabilities in the CAL-TSS system have three components: • a typefieldthat specifies the nature of the object addressed, • an option bitsfieldthat indicates operations which can be performed by the possessor of the capability, and • a valuefieldthat identifies the object and contains a pointer to the object.

One of the more important gains from the use of descriptors is the protection of procedures. If procedures can be invoked only by referencing a descriptor, then two benefits are realized. First, a procedure can only be invoked at its entry point contained in the descriptor; it cannot be entered at a random point. Second, procedure code is protected from accidental or deliberate modification. Despite their differences, all of these machines have a com- 37 Early Descriptor Architectures mon link to capability architectures: they all use descriptors to name programming objects.