Hi Thomas,
I will suppose you use the last CVS head. The
difference between your both tests is that in the first one, ARP is used to
found your target MAC/IP entry, but in the second one, ARP is used to found your
gateway entry (so, if your gateway exists, but not your target, ARP will
first find your gateway MAX address, and next, lwIP send a packet to
(gateway MAC/target IP).
The assertion you give us is in etharp_query, if mem_malloc(MEMP_ARP_QUEUE) failed when
ARP_QUEUEING=1. If you remove the assert, you got a crash. So, we need to
change code to be able to work even if we don't have any
MEMP_ARP_QUEUE.
I will take a look
----- Original Message -----
Sent: Friday, November 09, 2007 11:48
PM
Subject: [lwip-users] arp failure results
in newEntry != NULL assertion
I'm able to readily reproduce the following, and am wondering
if other folks have seen similar behavior, or if it's just a problem with my
port.
On my target running lwIP, I while(1) sendto() to a non-existent
target on the ***LOCAL*** subnet. Result: newEntry!=NULL
assertion from etharp.c. lwIP failed to allocate a memp so that it could
cache the outgoing packet. Shouldn't really be an assertion, as there
needs to be error handling code in place for that, but that's another issue.
On my target running lwIP, I while(1) sendto() to a non-existent
target on a ***REMOTE*** subnet. Result: code works
fine
_______________________________________________ lwip-users mailing
list address@hidden http://lists.nongnu.org/mailman/listinfo/lwip-users
|