[ECOS] May be problem with TCP/IP stack.

[ECOS] May be problem with TCP/IP stack.

[Tihomir Borovski]

Hi Everybody,
Strange problem came up when integrating GoAhead web server into eCos:
Server behaves stable (of course after workarounding the memory leak in
socketSelect()),
The problem occurs when trying to reload multiple times even very simple
html page, the
first 13 reloads are OK, but the 14th reload can't make it, from the client
side (browser)
server seems stuck for about 40-50 sec, and after that period everything is
ok for the
next 13 reloads and the same problem with the 14th one.
At the server side the main loop of the server is still rolling on but it
seems that
the request for the page is not 'heard' (because the timeout of socketSelect
expires)

P.S. I have telnet server on the same target which behaves in the same way:
every 14th
subsequent request for connection is not sensed by accept() in a period of
about 45 secs.

Target details:
CPU: Samsung S3C4530A (ARM core with ethernet)
RAM: 8MB
FLASH 4MB
OS: eCos 2.0

Any ideas?


-- 
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] May be problem with TCP/IP stack.

[sebastien Couret]

On Thursday 02 September 2004 10:45, Tihomir Borovski wrote:
> Hi Everybody,
> Strange problem came up when integrating GoAhead web server into eCos:
> Server behaves stable (of course after workarounding the memory leak in
> socketSelect()),
> The problem occurs when trying to reload multiple times even very simple
> html page, the
> first 13 reloads are OK, but the 14th reload can't make it, from the client
> side (browser)
> server seems stuck for about 40-50 sec, and after that period everything is
> ok for the
> next 13 reloads and the same problem with the 14th one.
> At the server side the main loop of the server is still rolling on but it
> seems that
> the request for the page is not 'heard' (because the timeout of
> socketSelect expires)
>
> P.S. I have telnet server on the same target which behaves in the same way:
> every 14th
> subsequent request for connection is not sensed by accept() in a period of
> about 45 secs.
>
> Target details:
> CPU: Samsung S3C4530A (ARM core with ethernet)
> RAM: 8MB
> FLASH 4MB
> OS: eCos 2.0
>
> Any ideas?

Have a look at the mailing list, I have submitted an answer to what should be 
the same problem.

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