[En-Nut-Discussion] 1 second delay in TCP handshake

Lars Andersson lakab at telia.com
Sun May 11 18:03:40 CEST 2003


I am experiencing a long delay in Nut 3.2.1 HTTP sequence that I did not
see before in Nut 2.
Has anyone some ideas on what to look for here? Any thoughts are
appreciated.

My environment is this:
   Ethernut 1.3 RevC, Atmega103, Nut 3.2.1, WinAVR(GCC) 3.3 
   DEFS = -DNUT_CPU_FREQ=3686400 is set.


What could cause the delay between frame 17-18 and 32-33 in the Ethereal
printout below? 
Ethernut is doing something at high priority here, the idle loop at prio
254 is frozen 
for over 1 second.
It does not matter which type HTTP page is requested it can be CGI or
UROM, the delay 
is long and stable at 1.1 sec or a little more. The Watchdog is disabled
so it does 
not interfere.  

Nut_B is the Ethernet board, p3 is a PC
DHCP is not used, Address is set with NutNetIfConfig() 

No.  Time        Source    Dest   Protocol Info
16   10.878573   p3        Nut_B  HTTP     GET /cgi-bin/lyse.cgi
HTTP/1.1
17   10.889113   Nut_B     p3     TCP      http > 1169 [ACK] Seq=1000001
Ack=2989106957 Win=2793 Len=0
18   12.040077   Nut_B     p3     HTTP     HTTP/1.0 200 Ok 
19   12.043278   Nut_B     p3     HTTP     Continuation
20   12.043407   p3        Nut_B  TCP      1169 > http [ACK]
Seq=2989106957 Ack=1000255 Win=64066 Len=0
21   12.055865   Nut_B     p3     HTTP     Continuation

31   28.475614   p3        Nut_B  HTTP     GET /index.html HTTP/1.1
32   28.486730   Nut_B     p3     TCP      http > 1142 [ACK] Seq=1000001
Ack=2414421563 Win=2818 Len=0
33   29.604773   Nut_B     p3     HTTP     HTTP/1.0 200 Ok
34   29.614407   Nut_B     p3     HTTP     Continuation
35   29.614532   p3        Nut_B  TCP      1142 > http [ACK]
Seq=2414421563 Ack=1000619 Win=64320 Len=0
36   29.620254   Nut_B     p3     HTTP     Continuation
37   29.623011   Nut_B     p3     TCP      http > 1142 [FIN, ACK]
Seq=1000817 Ack=2414421563 Win=3216 Len=0
38   29.623185   p3        Nut_B  TCP      1142 > http [ACK]
Seq=2414421563 Ack=1000818 Win=64122 Len=0


Regards,
Lars H. Andersson





More information about the En-Nut-Discussion mailing list