This is the mail archive of the ecos-discuss@sourceware.cygnus.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: size sanity-check


Paul Kinzelman wrote:
> 
> I'm trying to evaluate the size of various components of ecos (on an
> ARM board). Does the following make sense...
> 
> I compiled the hello.c example but with no stdio and printf, downloaded
> it to the eval board, and gdb said the .text size was ~18K.
> Could I assume this is the size of the basic kernel (in the standard
> configuration) for gross approximation?

Not necessarily - you will still have other code in there, e.g. to support
device driver initialization, even though it may not be used. You would
want to remove the I/O libraries from your configuration to remove those
parts (and they are quite substantial relative to the size of the kernel
scheduler alone).

Also note the default configuration of the kernel has not been configured
for maximum ROM size efficiency, nor should it be. With so much potential
for configuration, measuring it may be meaningless.

> I tried compiling a few different clib calls and it looks like it links in
> only the functions it needs. Any way to find out how big the clib is
> approximately without compiling with all the function calls?

Unless you are going to be compiling an application that used every one of
those function calls, why would that be of use?

And it all may change as soon as you change just a few configuration
options, like the size of the malloc memory pool etc.

What are you trying to achieve?

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]