[ECOS] SLIP and the eCos net package

[ECOS] SLIP and the eCos net package

[Øyvind Harboe]

Is there a fundamental reason why SLIP is not supported by the eCos net
package? 


Having worked with lwIP for a little while, I find that the "net"
package w/eCos would probably work a lot smoother with other code out
there.

-- 

Øyvind Harboe
http://www.zylin.com




-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Andrew Lunn]

On Wed, Apr 14, 2004 at 03:49:41PM +0200, ?yvind Harboe wrote:
> Is there a fundamental reason why SLIP is not supported by the eCos net
> package? 

eCosCentric has a PPP implementation. SLIP seems to be mostly dead
with PPP taking its place. 

There is also another PPP implementation, look in the mail
archives. It did not make it into anoncvs because it appeared at the
time that eCosCentric code would get into anoncvs relatively
quickly. But that didn't happen at the time. I don't know what there
plans are now. Maybe Alex can comment.

      Andrew

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Jani Monoses]

> eCosCentric has a PPP implementation. SLIP seems to be mostly dead
> with PPP taking its place. 
 
Except on small targets with max 256K of RAM where even the relatively
small lwip ppp takes tens of Kilos or where you don't need the features
of PPP only a simple ip over serial.

> There is also another PPP implementation, look in the mail
> archives. It did not make it into anoncvs because it appeared at the

there are two more actually Mark Retallack's and the one in lwip.
I think the eCosCentric one is in the ecospro commercial kit only and
not(yet) intended for anoncvs.

Jani

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Alex Schuilenburg]

Andrew Lunn wrote:

> On Wed, Apr 14, 2004 at 03:49:41PM +0200, ?yvind Harboe wrote:
> 
>>Is there a fundamental reason why SLIP is not supported by the eCos net
>>package? 
> 
> 
> eCosCentric has a PPP implementation. SLIP seems to be mostly dead
> with PPP taking its place. 
> 
> There is also another PPP implementation, look in the mail
> archives. It did not make it into anoncvs because it appeared at the
> time that eCosCentric code would get into anoncvs relatively
> quickly. But that didn't happen at the time. I don't know what there
> plans are now. Maybe Alex can comment.

We had planned to contribute our eCosPro PPP stack by the end of last 
month but it failed release testing against anoncvs.  Hence it was 
passed back to Nick (who did the port).  The additional delay was 
because Nick was in the US giving an eCos course at ESCW so did not have 
time to investigate the problems.

Nick is now responsible for testing, fixing problems found so it should 
appear in anoncvs any day now.

-- Alex
Managing Director / CEO                              eCosCentric Limited
http://www.ecoscentric.com/                 The eCos and RedBoot experts



> 
>       Andrew
> 




-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Nick Garnett]

Alex Schuilenburg <alexs@ecoscentric.com> writes:

> Andrew Lunn wrote:
> 
> > On Wed, Apr 14, 2004 at 03:49:41PM +0200, ?yvind Harboe wrote:
> >
> >>Is there a fundamental reason why SLIP is not supported by the eCos net
> >> package?
> > eCosCentric has a PPP implementation. SLIP seems to be mostly dead
> > with PPP taking its place. There is also another PPP implementation,
> > look in the mail
> > archives. It did not make it into anoncvs because it appeared at the
> > time that eCosCentric code would get into anoncvs relatively
> > quickly. But that didn't happen at the time. I don't know what there
> > plans are now. Maybe Alex can comment.
> 
> We had planned to contribute our eCosPro PPP stack by the end of last
> month but it failed release testing against anoncvs.  Hence it was
> passed back to Nick (who did the port).  The additional delay was
> because Nick was in the US giving an eCos course at ESCW so did not
> have time to investigate the problems.

The testing failure was simply that the wrong serial port was being
used. PPP needs hardware flow control to work efficiently, and a port
without the necessary connections was being used.

> 
> Nick is now responsible for testing, fixing problems found so it
> should appear in anoncvs any day now.

That day is actually today, I was in the process of doing it when this
message came in. Keep an eye on the repository.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Øyvind Harboe]

> The testing failure was simply that the wrong serial port was being
> used. PPP needs hardware flow control to work efficiently, and a port
> without the necessary connections was being used.

Ouch!

The hardware I'm working with does not have HW flow control.

Is this a trainwrecker?


-- 

Øyvind Harboe
http://www.zylin.com




-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] SLIP and the eCos net package

[Nick Garnett]

Øyvind Harboe <oyvind.harboe@zylin.com> writes:

> > The testing failure was simply that the wrong serial port was being
> > used. PPP needs hardware flow control to work efficiently, and a port
> > without the necessary connections was being used.
> 
> Ouch!
> 
> The hardware I'm working with does not have HW flow control.
> 
> Is this a trainwrecker?

Nope. You can still use software flow control: XON/XOFF. It's just not
quite as efficient since it relies on in-band signalling on the serial
data stream and is therefore subject to things like interrupt latency
and FIFO depths that RTS/CTS is not.

The testing problem was simply that we were testing a system
configured to use RTS/CTS on hardware that didn't support it. So it
all just ground to a halt waiting for a CTS that never came. If we had
configured for XON/XOFF, it would have all worked perfectly.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss