> I was wondering if anybody on the list has had any experience using the > IARVSERV macro for sharing data among address/data spaces. > > What is the overhead? Is anything going on "under the covers" beyond the > normal ALESERV and access register manipulations?
I am not sure what you mean by overhead. At a basic level the shared memory is being "stitched into" the translation table(s) of one or more other address spaces. Once they're there, the hardware translates the addresses the same as it would any other address. So there is zero overhead as far as execution speed is concerned. There is some extra control block storage for managing the shared areas of the paging tables and that could be considered "overhead" I suppose. Of course if you needed the function it provides, then calling it overhead leaves out the question "what was your other plan?" In other words, it costs what it costs. If you were going to use it, you would probably want to have a small number of larger shared areas than vice versa. > Would you say that the feature is primarily for the use of application > programs, or does it have a place in the system programmer's toolkit? It's a basic UNIX-ism. IARVSERV provides the underpinnings for the USS implementation of the UNIX shared memory services. Does it have a place in a system programmer's toolkit? Sure. It's a potentially useful tool. CC ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
