Re: [lwip-users] Delay during transmission

[JM]

I believe I figured it out.  I was calling tcp_write, then tcp_output, but since I was referencing the data to be sent by pointer, I wanted to wait until it was successfully transmitted before freeing that buffer.  So, I had something set up where it wasn't leaving the function (tcp_recv function) until it had successfully transmitted, or a timer expired, whichever happened first. 

The timer was 2s, and apparently lwIP wasn't sending the data until it left this function.  I thought it would do this in the interrupt (RAW mode). 

Long story short, I learned the correct way to set this up and now it's lightning fast.  From your response, it seemed like I was doing something fundamentally wrong in the way I was using these functions, so that's where I started looking.  Thanks for your help.


--- On Thu, 3/11/10, Kieran Mansley <address@hidden> wrote:

From: Kieran Mansley <address@hidden>
Subject: Re: [lwip-users] Delay during transmission
To: "Mailing list for lwIP users" <address@hidden>
Date: Thursday, March 11, 2010, 4:36 AM

On Wed, 2010-03-10 at 05:04 -0800, JM wrote:
> I should have attached that right away.  It's now attached, although
> this capture is apparently not the same as the one I used to extract
> the text summary.  In any case, the symptoms and lwIP code are the
> same.

That is quite odd.  I can't think of anything in lwIP that would exhibit
that behaviour.  Are you sure the application is calling lwIP with more
data to send?  It could just be that there is nothing in the send queue,
then 2 seconds later the app gives lwIP more data to send.  The window
update is usually also triggered by an application calling tcp_recved(),
so that would fit.

Did tcp_write() return an error at any time?

Kieran



_______________________________________________
lwip-users mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/lwip-users