Browse Prior Art Database

Address Protection in User Level Process

IP.com Disclosure Number: IPCOM000104487D
Original Publication Date: 1993-Apr-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 1 page(s) / 44K

Publishing Venue

IBM

Related People

Albaugh, VA: AUTHOR [+2]

Abstract

Disclosed is a method to provide data protection between trusted and untrusted code segments in a single process model.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 75% of the total text.

Address Protection in User Level Process

      Disclosed is a method to provide data protection between
trusted and untrusted code segments in a single process model.

      Some applications are written using a two process model to
provide address protection.  In this model, a trusted program is
given full addressability to critical data objects ( such as a data
base) and acts as a "server" to a untrusted "client" processes.  The
client process gains access to the critical data only indirectly by
sending requests and receiving replies from the "server" process.
This permits each request to be validated by the server.  This multi
process implementation does restrict data access to trusted code, but
incurs significant communication and process switching overhead.

      The process switch and interprocess communication can be
avoided while still maintaining protection.  This can be done by
registering a text address range as trusted rather than an entire
process as trusted.  System services can be provided that grant and
remove addressabilty to sensitive data.  The system services that
grant addressability verify that control is being returned to a
trusted section of code.  And the trusted section of code always
removes addressability before returning to the untrusted code
section.  The trusted code section could exist as a prebuilt
subroutine library.

      Enforcing protection on address ranges rather than processes
permits the trusted and non-trus...