lwip-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[lwip-users] Re: [lwip] lwIP, TCP works now here. Revised cs8900a drive


From: Adam Dunkels
Subject: [lwip-users] Re: [lwip] lwIP, TCP works now here. Revised cs8900a driver released.
Date: Wed, 08 Jan 2003 23:14:33 -0000

Hi Leon!

On Tuesday 08 January 2002 11:22, you wrote:
> Yes. Could you describe what MEM_ALIGN should be set to,
> i.e. where exactly should it be dependent of?

It should reflect the alignment of the CPU. MEM_ALIGNMENT specifies how the 
memory returned by the mem.c, memp.c and pbuf.c modules should be aligned. If 
the CPU only allows 32-bit accesses to addresses that are divisible by 4, 
MEM_ALIGNMENT should be set to 4. If the CPU allows 32-bit accesses to 
addresses that are divisible by 2, MEM_ALIGNMENT should be 2. The x86 works 
faster with 4-byte alignment, so perhaps one should always set MEM_ALIGNMENT 
to either 2 or 4 (it wastes a bit of memory, though).

> Status on my side (a port to the Infineon C166 and Crystal
> Semiconductors CS8900A):
>
> - HTTP stops after some 1000 or so HTTP GET's, running from
> 3 parallel "wget" processes from a Linux box. I'm running
> tests again to see whether it stops exactly after a fixed
> number of requests.

Are you able to print out any debug output? If so, you could print out some 
of the statistics and the state of the TCP connections. It could give a clue 
as of what is happening.

/adam
-- 
Adam Dunkels <address@hidden>
http://www.sics.se/~adam
[This message was sent through the lwip discussion list.]




reply via email to

[Prev in Thread] Current Thread [Next in Thread]