NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/comp/arch/12198

Internet Message Format | 1993-01-09 | 2.5 KB

Path: sparky!uunet!gatech!usenet.ins.cwru.edu!agate!doc.ic.ac.uk!uknet!mcsun!dxcern!dscomsa!zeus02.desy.de!hallam
From: [e-mail] (Phill Hallam-Baker)
Newsgroups: comp.arch
Subject: Re: AST on UNIX
Message-ID: <[e-mail]>
Date: 9 Jan 93 22:25:17 GMT
References: <[e-mail]> <[e-mail]>
Sender: [e-mail] (usenet)
Reply-To: [e-mail]
Organization: Deutsches Elektronen Synchrotron, Experiment ZEUS bei HERA
Lines: 47

In article <[e-mail]>, [e-mail]
(simon R) writes:

|>: >OK, try this. VMS has something called an AST (Asynchronous System Trap)
|>which
|>: >is similar to a user-mode interrupt. This may occur at any time: for
|>instance,
|>: >an asynchronous I/O can call an AST routine when the I/O completes. When
|>the
|>: >AST routine completes, control returns to the code that was executing at
|>the
|>: >time the AST occurred. (The analogy in NT is the APC, or Asynchronous
|>: >Procedure Call.)
|>: >
|>: >Now, much of the run-time stuff in UNIX is non-reentrant. In particular,
|>the
|>: >use of errno springs to mind. If the AST (or APC) was introduced into the
|>: >UNIX model, how can I use these non-reentrant routines safely in an AST
|>:
|>: While you are at it, don't forget to include a separate 32-bit value
|>: passed to the AST.  The AST is fully queue'd, and if you send 32 ASTs
|>: before the application executes it's handler the application will still
|>: see ALL the ASTs with their separate 32-bit value.
|>:
|>: If you can make this work, now port it to AIX, SunOS, Solaris, Ultrix,
|>: IRIX, and HP-UX.  Those are the only environments I currently would like
|>: it on, I'm sure there would be more within a few months.
|>:
|>Concurrent Computer (for whom I used to work) have had AST capability
|>in their real-time version of Unix (based on System V) since (as
|>MASSCOMP) their first product shipped some ten years ago.
|>As I recall, error status of the I/O operation was indeed passed as a
|>parameter
|>to the AST routine. It's major use was to overlap real time data input
|>with output to disk or whatever. ASTs could be prioritized, buffers
|>locked in memory, everything that you needed to do real-time data collection.
|>

The AST functionality is in any case to a large degree superceeded by POSIX
threads. The addition of threads to VMS has been one of the most useful of the
recent additions. Of course if you have ASTs then it is quite easy to add on the
threads.


--

Phill Hallam-Baker