This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: Comms Stacks
- From: Jani Monoses <jani at iv dot ro>
- To: "Geoff Patch" <grp at cea dot com dot au>
- Cc: ecos-discuss at sources dot redhat dot com
- Date: Wed, 25 Sep 2002 09:58:55 +0000
- Subject: Re: [ECOS] Comms Stacks
- References: <000201c2641c$e3304ad0$220210ac@cea.com.au>
>
> 1. If we started using lwip would we need to make changes to any of our
> application software that uses the normal socket level API calls such as
> socket(), bind() listen() etc?
lwIP does not have select().It's easy to add though.
>
> 2. Are there any quantifiable benefits to using lwip in terms of reduced
> footprint or reduced CPU usage?
We had a lwIP based ecos app running in 128K RAM and 128 K flash of which
ecos/lwip was about half.It only uses the ecos kernel and the memory allocator.
It can be tuned for size or performance.
Don't know about CPU usage.So it can fit in 60K flash and use 40-50K RAM on an ARM.
>
> 3. Are there any limitations inherent to using lwip?
Yes, it is a lot less mature than the other two stacks.The main reason for this that
it is only semi-maintained upstream and has incomparably small peer review.Most of its users
are doing microcontrollers and 16 bit chips.An exposure to the ecos community (judging by
the traffic on the mailing list alone) would easily triple the number of users and enhancers.
>
> 4. Does anyone have any experiences with lwip that they'd like to pass on,
> either good or bad?
So our exeprience here is good mainly because we were mostlyt interested in small footprint
I have no benchmarks wrt performance because haven't tried the other stacks.
--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss