One of the problems with the whole virtual memory overcommit issue is the difficulty of working out how much memorytruly is committed. For example, consider all of the ways that Unixkernels can commit themselves to holding down memory on behalf of a userprocess. In addition to obvious things like System V shared memory, thisincludes (but isn't limited to):

• pipe data buffers, including unused ones; theoretically every pipepair created commits the kernel to at least 4K of buffer or so.
• socket send and