Hi Alex, > Hi, > > this contains at first some small fix for the netdev notifier which I deteced > while hacking 6lowpan stateful compression support. > > I rework the debugfs entry for the stateful compression. Now each context > entry for a 6LoWPAN contains has his own subdirectory (named as his CID > value). > In this subdirectories there exists the following files for manipulating the > context entry: > > - active: If the context belongs to the table or not. > - compression: the compression flag according rfc6775. > - prefix: ipv6 prefix in xxxx:xxxx:....:xxxx form. > - prefix_lenght: the length of the ipv6 prefix. > > This offers a setting per property instead doing remember the full context and > doing echo "$FULL_CONTEXT_INFORMATION" > ctx_table. > > These file should be prepared to move into the sysfs directory when it seems > to running fine. > > There exist one file "show" inside $LOWPAN_IFACE/contexts/. This file is for a > "pretty-printout" for the current context table. I use the same format-string > like RIOT-OS, example: > > cid|prefix |C > ------------------------------------------------- > 0| 2001::/64 |1 > 1| 2001::1/128|1 > 2| 2001::5/128|1 > 7| 2002::/64 |1 > > > difference here is that RIOT shows the lifetime, but I currently see no sense to > handle the lifetime in kernelspace. > > Also I added support to clean the context table when doing a "ifdown". The > reason is that IPv6 interface will also drop all address information while doing > a ifdown and a 6LoWPAN should not use previous context information after > ifup again. > > - Alex > > Alexander Aring (2): > 6lowpan: iphc: add support for stateful compression > ieee802154: 6lowpan: fix return of netdev notifier > > include/net/6lowpan.h | 31 ++++ > net/6lowpan/core.c | 39 +++- > net/6lowpan/debugfs.c | 247 +++++++++++++++++++++++++ > net/6lowpan/iphc.c | 420 > ++++++++++++++++++++++++++++++++++++------ > net/ieee802154/6lowpan/core.c | 7 +- > 5 files changed, 682 insertions(+), 62 deletions(-) > > -- > 2.6.1 Thanks for the great work you have done! I have tested the latest stateful compression module, and it works very well! One note is that the "C" flag, in the debugfs entry "show", list active state instead of compression state. However this is a minor and I will highlight it in the PATCH 1/2 mail. The context table was set up with 2 entries, using CID 0 and 15: cid|prefix |C ------------------------------------------------- 0| 2001:db8::/64 |1 15| 2001:db8::1/128|1 Device setup: Linux computer (bt0) - Address: 2001:db8::1/64 Device A - CID 0, compression ON - CID 15, compression OFF - Address: 2001:db8::MAC/64 Device B - CID 0, compression ON - CID 15, compression OFF - Address: 2001:db8::MAC/64 Device C - No CIDs - Address: 2001:db8::MAC/64 Scenario 1: ping6 <A> -s 0 -c 1 TX: Full compression of source and dest RX: Full compression of source and dest Scenario 2: ping6 <B> -s 0 -c 1 TX: Full compression of source and dest RX: Compression of source, half compression of dest (bt0) Scenario 3: - Disable CID echo 0 > /sys/kernel/debug/6lowpan/bt0/contexts/0/active echo 0 > /sys/kernel/debug/6lowpan/bt0/contexts/15/active ping6 <C> -s 0 -c 1 TX: No compression RX: No compression Scenario 4: - Adjust prefix length on CID 0 to prefix_len 32. echo 32 > /sys/kernel/debug/6lowpan/bt0/contexts/0/prefix_len echo 1 > /sys/kernel/debug/6lowpan/bt0/contexts/0/active echo 1 > /sys/kernel/debug/6lowpan/bt0/contexts/15/active ping6 <A> -s 0 -c 1 TX: Full compression of source and dest RX: Full compression of source and dest Regards, Lukasz -- To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
