patch set introducing the infrastructure for multi-advertising capability to the HCI core and mgmt API. v1 -> v2: - add missing braces in read_adv_features() v2 -> v3: - split change-set into several patches - replace err == 0 by !err - fix coding style problems v3 -> v4: publicly visible change: - when calling remove_advertising with an instance value of zero (i.e. remove all instances), the command response also returns an instance value of zero now as it doesn't make sense to return a single instance id when removing several instances refactoring and fixes (due to Arman's review): - splitting the change set into a much larger number of patches to facilitate review - use HCI_MAX_ADV_INSTANCES in the same patch that introduces it - use adv_info->hdev in the same patch that introduces it - make the logic of hci_find_adv_instance() more readable - make sure that hci_find_adv_instance() is called while hci_dev is locked - replace hci_num_adv_instances() by an instance counter in hci_dev - add inline comment in get_adv_instance_flags() explaining its return value in the error case - generate zero-length adv data if the current instance identifier is invalid - revert erroneous changes to the logic in clear_adv_instance() - use hci_adv_instances_clear() in clear_adv_instance() when removing all advertising instances also removing a reduntant error check - inserting TODO messages to make sure that advertising will not be switched off prematurely once we allow more than one advertising instance - inserting TODO messages to make sure that multiple advertising instances will be advertised in a round-robin fashion once we allow for more than one advertising instance - identify peding advertising instances (just added but not yet confirmed in add_advertising_complete) by a boolean flag in the adv_info struct so that we can identify and remove them even when the pending add_advertising command cannot be retrieved for some reason - makes sure that we do not leak advertising instances in this case - only send HCI commands to update advertising data when a new instance has actually been added v4 -> v5 (fixes due to Marcel's review): - split into separate change sets for kernel and userspace - do not trust the adv instance list length when compiling the adv features struct - introduce adv_info->pending in the same patch it's used for the first time Florian Grandel (16): Bluetooth: hci_core: Introduce multi-adv inst list Bluetooth: mgmt: dry update_scan_rsp_data() Bluetooth: mgmt: multi adv for read_adv_features() Bluetooth: mgmt: multi adv for get_current_adv_instance() Bluetooth: mgmt: multi adv for get_adv_instance_flags() Bluetooth: mgmt: improve get_adv_instance_flags() readability Bluetooth: mgmt: multi adv for enable_advertising() Bluetooth: mgmt: use current adv instance in set_advertising() Bluetooth: mgmt: multi adv for create_instance_scan_rsp_data() Bluetooth: mgmt: multi adv for create_instance_adv_data() Bluetooth: mgmt: refactor update_*_data() Bluetooth: mgmt: multi adv for set_advertising_complete() Bluetooth: mgmt: multi adv for add_advertising() Bluetooth: mgmt: multi adv for clear_adv_instances() Bluetooth: mgmt: multi adv for remove_advertising() Bluetooth: hci_core: remove obsolete adv_instance include/net/bluetooth/hci_core.h | 22 ++- net/bluetooth/hci_core.c | 113 ++++++++++++- net/bluetooth/mgmt.c | 353 ++++++++++++++++++++++----------------- 3 files changed, 331 insertions(+), 157 deletions(-) -- 1.9.1 -- 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
