HP3000-L Archives

January 1999, Week 4

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jerry Fochtman <[log in to unmask]>
Reply To:
Jerry Fochtman <[log in to unmask]>
Date:
Tue, 26 Jan 1999 10:30:29 -0600
Content-Type:
text/plain
Parts/Attachments:
text/plain (56 lines)
At 08:10 AM 1/26/99 +0000, Dennis Handly wrote:
>Steve Widmar ([log in to unmask]) wrote:
>: How can we pull apart the active SL to determine the compiler build &
>: options?   Any ideas on general purpose comparison methods to confirm
>: source files match executables?
>: Steve
>
>Compiler options are not stored in the SL.
>
>You could decompile the two SL segments, remove/change the segment
>numbers, then try to diff the output.

In another note it was mentioned that the length of the modules is
not the same.  This implies that the emitted code is either different
or the source code has indeed changed.

Comparing the decompiled listing will tell you what instructions are
different.  One would still need to study the surrounding instructions
and processing flow and be familiar with the attributes of the language
to determine if the processing is basically the same.  For only a few
differences this may not be too bad. But if there are a lot of
differences this processes quickly gets more involved.

Updates to a compiler may emit different instructions for performing the
same task.  So the object file may be different anyway yet be from
identical sources.  While doing this is not out of the realm of
possibilities, it could prove challenging (got a CISC instruction
set book?) and not something I would recommend.  The last time I had
to do something like this I ended-up writing a program to compare the
object code associated with entry points, including handling the differences
in segment location, STT, etc. between the 2 objects. (No, I don't know
where the program is...).

The best option is probably execute both libraries against the same data
and compare the results.  The challenge here is to have a set of data which
exercises all the code paths/conditions in order to be comfortable that all
possible results have been compared.



/jf
                              _\\///_
                             (' o-o ')
___________________________ooOo_( )_OOoo____________________________________

                        Monday, January 25th

          Today in 1915 - Transcontinental telephone service
                          began in the U.S.

___________________________________Oooo_____________________________________
                            oooO  (    )
                           (    )  )  /
                            \  (   (_/
                             \_)

ATOM RSS1 RSS2