Hello, Here are the first patches of the big scan support series. Only passive scan is present in these patches, we must agree on this first step before moving forward with active scans, beaconing, associations and so on. There are many other patches to come. All of this is based on the initial work from David Girault and Romuald Despres, they are often credited as Co-developpers. Thanks, Miquèl Changes in v2 (some of them apply to other parts of the series not sent today): * Initially tried to create a new (empty) path for COORD packets. As the processing, for now, is exactly the same as for NODE packets, I decided to go for a switch case as well as a comment indicating that the filtering should be enhanced with time. * Reworked the rx path to clarify the ongoing filtering. * Added an enum with the different filtering levels. * Updated hwsim to reflect the right filtering level. * Reworked again the rx path to include more filtering, in particular I added a real promiscuous mode as well as a level 4 filter (frame fields validation) which is enabled if the xceiver does not do it already. * Set all the drivers (but the ca8210) able to create coordinators interfaces in theory. * Prevented drivers with no promiscuous support to create COORD interfaces in the first place. * Dropped the frame type enumeration because it was actually unused. * The code already checked if the promiscuous mode was enabled, but it was not clear enough apparently, so I reorganized the logic a little bit in order to not hide this check. * What was missing about the promiscuous mode however, was to remember in which state we found the PHY to avoid disabling the promiscuous mode if it was not enabled when we started scanning. * Created a second workqueue, which would not be flushed in the close path because it does not need to be just for the mac commands. This allows acquiring the rtnl within scan_lock without issues. * Thanks to the above change, I was able to drop the two patches reducing the use of the rtnl (in the tx path and in hwsim:change_channel()) but definitely not the change in the MLME Tx code which is really painful. David Girault (1): net: ieee802154: Trace the registration of new PANs Miquel Raynal (10): net: mac802154: Introduce filtering levels net: mac802154: Drop IEEE802154_HW_RX_DROP_BAD_CKSUM net: mac802154: Allow the creation of coordinator interfaces net: ieee802154: Advertize coordinators discovery net: ieee802154: Handle coordinators discovery net: ieee802154: Add support for user scanning requests net: ieee802154: Define a beacon frame header net: mac802154: Prepare forcing specific symbol duration net: mac802154: Introduce a global device lock net: mac802154: Handle passive scanning drivers/net/ieee802154/mac802154_hwsim.c | 10 +- include/linux/ieee802154.h | 7 + include/net/cfg802154.h | 74 ++++- include/net/ieee802154_netdev.h | 36 +++ include/net/mac802154.h | 28 +- include/net/nl802154.h | 91 +++++++ net/ieee802154/Makefile | 2 +- net/ieee802154/core.c | 2 + net/ieee802154/nl802154.c | 331 +++++++++++++++++++++++ net/ieee802154/nl802154.h | 9 + net/ieee802154/pan.c | 115 ++++++++ net/ieee802154/rdev-ops.h | 28 ++ net/ieee802154/trace.h | 65 +++++ net/mac802154/Makefile | 2 +- net/mac802154/cfg.c | 41 ++- net/mac802154/ieee802154_i.h | 38 ++- net/mac802154/iface.c | 26 +- net/mac802154/main.c | 40 ++- net/mac802154/rx.c | 63 ++++- net/mac802154/scan.c | 291 ++++++++++++++++++++ net/mac802154/tx.c | 12 +- 21 files changed, 1269 insertions(+), 42 deletions(-) create mode 100644 net/ieee802154/pan.c create mode 100644 net/mac802154/scan.c -- 2.34.1
