home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!math.fu-berlin.de!uni-paderborn.de!urmel.informatik.rwth-aachen.de!cip-s03!dak
- From: dak@cip-s03.informatik.rwth-aachen.de (David Kastrup)
- Newsgroups: comp.std.c++
- Subject: Re: Use of nested functions
- Date: 28 Jan 93 12:53:25 GMT
- Organization: Rechnerbetrieb Informatik - RWTH Aachen
- Lines: 12
- Distribution: world
- Message-ID: <dak.728225605@cip-s03>
- References: <1993Jan13.174051.21288@ucc.su.OZ.AU> <9302002.3172@mulga.cs.mu.OZ.AU> <1993Jan22.081555.12027@us-es.sel.de> <1993Jan26.224107.9187@ucc.su.OZ.AU> <9302813.18482@mulga.cs.mu.OZ.AU>
- NNTP-Posting-Host: cip-s03.informatik.rwth-aachen.de
-
- One problem with trampolines and self-modifying code is that there
- is no longer a separate, read only code space. This can complicate
- code sharing by demand paging (self-mod code), and be influenced by
- the processor's caches and prefetch queues, if self-modifying code is
- expected to work under all circumstances.
- I know for a fact that some CPU testers find out about perfectly
- object-code compatible CPUs by finding out, under which circumstances
- in self-modifying code is the old code executed instead of the
- modified one.
-
- The real problem is dealing with separate code space cache controllers,
- which are likely not to be designed for that purpose.
-