Re: how does purify work?

nsrcccw@leonis.nus.edu.sg (Chea Chee Weng)
5 Aug 2000 21:35:51 -0400

          From comp.compilers

Related articles
how does purify work? toddhoff@my-deja.com (2000-08-04)
Re: how does purify work? pfaffben@msu.edu (Ben Pfaff) (2000-08-05)
Re: how does purify work? cfc@world.std.com (Chris F Clark) (2000-08-05)
Re: how does purify work? plakal@cs.wisc.edu (2000-08-05)
Re: how does purify work? Martin.Raabe@WindRiver.com (Martin Raabe) (2000-08-05)
Re: how does purify work? nsrcccw@leonis.nus.edu.sg (2000-08-05)
Re: how does purify work? ralph@inputplus.demon.co.uk (2000-08-10)
Re: how does purify work? gvmt@bgl.vsnl.net.in (Venkatesha Murthy G.) (2000-08-10)
Re: how does purify work? cfc@world.std.com (Chris F Clark) (2000-08-13)
| List of all articles for this month |

From: nsrcccw@leonis.nus.edu.sg (Chea Chee Weng)
Newsgroups: comp.os.vxworks,comp.compilers,comp.programming
Followup-To: comp.os.vxworks,comp.compilers,comp.programming
Date: 5 Aug 2000 21:35:51 -0400
Organization: National University of Singapore
References: 00-08-027
Keywords: debug

toddhoff@my-deja.com wrote:
: Once upon a time Purify did work for vxworks. It would be a very nice
: tool to have and it seems very odd that there's no product with
: equivalent functionality.


: If one wanted to build a Purify like tool, does anyone know how it
: would be done? I don't really understand what it does and haven't
: found much related info.


You can find this using a patent search - use either the one at ibm or
the www.uspto.gov. The "purify" method was patented but at this point
i am unable to remember the "keywords" to find it. Anyway you may find
a bunch of ways people did this either search "parasoft" for their
insure++ tools. (for memory overrun and others). or search for
"Detecting concurrency errors in multi-threaded programs" 6,009,269 to
see how one method to find race condition detecting is made.


Roughly, all will augment/instrument the source or object code to do
this (otherwise some will have a modified VM like the JavaVM concept
to monitor violation).


For instrumenting code (or injecting more functions) to an existing
binary (without the source code), you could see one Microsoft tool
with source (Gee can't remember now but it is in microsoft research
and free to download or try "DynInst" that works for non-Intel
machines (See for it - is was part of Paradyn tools)).


With all these tools and info, it does not seem hard to make your own
but i have a hunch that there may be "special" condition cases in all
the above methods (patents) that one has to hack to get things working


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.