From: Jonas T <jonas.tarnstrom@re...> - 2004-09-27 17:34:52
I am having great progress with the 4-address support, stil some issues rem=
ain which I am unable to fix.
When encapping the frame in ieee80211_encap I do skb_push by 32 to get the =
data aligned for the hal.
In my logic this would also requier me to shorten the packet by two to get =
it the right length, so I do:
skb->len -=3D 2; (Should I do skb_trim (skb, skb->len-2) instead?)
I print the packet right before the encap function returns it and everythin=
g looks fine.
When radio sniffing the data, I notice that the last two bytes in the packe=
ts has been replaced by something else which causes the checksum for the =
ICMP to fail and the frame being dropped.
If I don't shorten the packet this doesn't happen but instead I get a "trai=
l" (called so by etherreal) of 2 bytes (0x00 0x00) which makes the fragme=
ntation (MTU) fail when a frame needs to be fragmented over the ethernet.=
It gets 1516 bytes and is therefor dropped by the bridge.
This seems to be the last major issue before I am done and any help on how =
the HAL/MAC requires the data to be aligned and other issues is appreciat=
Get latest updates about Open Source Projects, Conferences and News.