R devel - May 2005 - memory problem

Hello!

 

I am undertaking my first attempt at using compiled C code within R and
have run into this lovely problem when I call my C code (using .C).

 

The instruction at "0x10001d1b" referenced memory at
"0x01fa2000". The
memory could not be "written".

 

Now this doesn't happen every time and furthermore I've tested the code
in plain C and it all works fine leading me to think has been caused by
an insidious misuse of pointers somewhere or other.  I'm guessing I am
overwriting some bit of memory that actually belongs to R.  However,
before I start checking and rechecking all my memory management I just
wanted to see if there is are any other paths I should be investigating.

 

Regards,

 

Jared

 

 


	[[alternative HTML version deleted]]

On Tue, 3 May 2005 Jared.O'Connell@csiro.au wrote:
> I am undertaking my first attempt at using compiled C code within R and
> have run into this lovely problem when I call my C code (using .C).
>
> The instruction at "0x10001d1b" referenced memory at
"0x01fa2000". The
> memory could not be "written".
>
> Now this doesn't happen every time and furthermore I've tested the
code
> in plain C and it all works fine leading me to think has been caused by
> an insidious misuse of pointers somewhere or other.  I'm guessing I am
> overwriting some bit of memory that actually belongs to R.  However,
> before I start checking and rechecking all my memory management I just
> wanted to see if there is are any other paths I should be investigating.
If you run R under a debugger such as gdb you will get much more useful 
information.  You haven't even told us your OS: for UNIX-alikes using

 	R -d gdb

is the most common way.

-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Sorry - I *was* using Windows.  Rebooted in Linux and ran

R -d gdb

subsequently found and fixed problem in less than 5 minutes (runaway pointer).
Lesson learned.

Thanks,

Jared


-----Original Message-----
From:	Wolfgang Huber [mailto:huber@ebi.ac.uk]
Sent:	Tue 5/3/2005 6:54 PM
To:	Prof Brian Ripley
Cc:	O'Connell, Jared (CMIS, Floreat); r-devel@stat.math.ethz.ch
Subject:	Re: [Rd] memory problem
Prof Brian Ripley wrote:
> 
> If you run R under a debugger such as gdb you will get much more useful 
> information.  You haven't even told us your OS: for UNIX-alikes using
> 
>     R -d gdb
> 
> is the most common way.
And

R -d ddd

(set the option "Run in separate execution window" before starting R)
offers a nice GUI for the debugger. Obviously, you need to have 
installed ddd: http://www.gnu.org/software/ddd/

Best regards
   Wolfgang

-------------------------------------
Wolfgang Huber
European Bioinformatics Institute
European Molecular Biology Laboratory
Cambridge CB10 1SD
England
Phone: +44 1223 494642
Fax:   +44 1223 494486
Http:  www.ebi.ac.uk/huber