A Scheme to Execute the Dynamic Linker as a Non-Privileged Procedure

概要

論文の詳細を見る
Dynamic linking realizes flexible reference to the required information, and offers the direct addressing method independent of the physical location. It, however, has been considered to be difficult to implement because it has made the kernel size big. The dynamic linker can be removed from the security kernel, by executing it in the domain where the program executes. Here, the linker is divided into the segment linker and the entry linker. When a linker fault is caused, the segment linker is invoked in the faulted domain. It asks the file system to search for the target segment, and assigns a segment number. In case that an entry name is given, it sets the condition in the link to cause the second linkage fault. Here, control is returned to the faulted place. When the second linkage fault is caused, address formation proceeds to the stage to determine the target domain; and the entry linker is invoked in the target domain to refer to the symbol table of the target segment. By utilizing the domain switching mechanism of the system, the algorithm is made clear; and the linker is removed from the security kernel.
一般社団法人情報処理学会の論文
1978-04-20