Low Cost Array Bounds Checking for 64-bit Architectures
		 -------------------------------------------------------

				Abstract
				--------

Several programming languages guarantee that array subscripts are checked to
ensure they are within the bounds of the array. While this guarantee improves
the correctness and security of array-based code, it adds overhead to array
references. This performance limitation is a significant obstacle preventing
the scientific community from adopting compiler-enforced array bounds checks. 

To reduce the overhead, we have created and an abstraction called Index
Confinement Regions (ICRs).  The basic idea is to place an array into a
large virtual memory region, such that any reference to the array is
confined to the region. Only the portion of the ICR corresponding to the
array is permissible to access. ICRs require a large virtual address
space and a small page size to reduce fragmentation. Accordingly, we
modified the Linux operating system to provide this support. ICRs reduce
the number of necessary bounds checks for an n-dimensional array access
from 2n to 1 for C, and from n to 0 for Java, yielding a significant
reduction in execution time for array-intensive applications.