Hi Marcel, >-----Original Message----- >From: Marcel Holtmann [mailto:marcel@xxxxxxxxxxxx] >Sent: Saturday, June 6, 2015 12:57 PM >To: Wang, Arron >Cc: linux-bluetooth@xxxxxxxxxxxxxxx >Subject: Re: [PATCH 2/2] Bluetooth: Add option to enable/disable A2MP module > >Hi Arron, > >>>> Make Bluetooth 3.0 HS feature configurable >>>> >>>> Signed-off-by: Arron Wang <arron.wang@xxxxxxxxx> >>>> --- >>>> net/bluetooth/Kconfig | 5 + >>>> net/bluetooth/Makefile | 3 +- >>>> net/bluetooth/a2mp.h | 19 +++ >>>> net/bluetooth/amp.h | 11 ++ >>>> net/bluetooth/hci_event.c | 260 +----------------------------------- >>>> net/bluetooth/hci_event_a2mp.c | 283 >>> ++++++++++++++++++++++++++++++++++++++++ >>>> net/bluetooth/hci_event_a2mp.h | 91 +++++++++++++ >>>> 7 files changed, 412 insertions(+), 260 deletions(-) create mode >>>> 100644 net/bluetooth/hci_event_a2mp.c create mode 100644 >>>> net/bluetooth/hci_event_a2mp.h >>> >>> I don’t think this is the solution here. I think A2MP support should >>> be self contained and utilises the synchronous HCI command framework >>> we have in place and also use in mgmt. >> >> Yes, create a separate file for hci a2mp event maybe not suitable, but there are >many hci ops/events for different Bluetooth controller BREDR/AMP/LE, add the >macro to separate them in one file may not easy to maintain, can we move these >AMP releated hci ops/event function to file amp.c? > >unless it has some complicated state machine impact, using the hci_request >infrastructure should be preferred compared to go through callback handlers in >hci_event.c. So first of all it needs to be investigated which of the A2MP HCI >functions can be turned into using hci_request. I found the current A2MP functions already use hci_request send command to hci subsystem, in hci_event.c they just invoke these functions, we can define these functions as inline but I not sure whether we also should put A2MP related events under BT_HS config option. Thanks Arron
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
