On Thu, Mar 26, 2015 at 2:29 AM, Samuel Ortiz <sameo@xxxxxxxxxxxxxxx> wrote: > Hi Robert, > > On Tue, Feb 24, 2015 at 12:01:48PM +0200, Robert Dolca wrote: >> This patch adds nci_request_driver and nci_req_complete_driver >> as a wrapper for __nci_request. When nci_req_complete_driver is >> called it also sets cmd_cnt to 1. This is done because the response is not >> sent to the NFC subsystem so cmd_cnt is not decremented there. >> >> nci_send_cmd was previously exported in order to send commands to >> the device from the driver. It shouldn't be used without >> nci_req_complete_driver because cmd_cnt will have the wrong value. > Any NCI packet should make it to the NCI core first, because the logic > and synchornization between Tx and Rx, commands and response is implemented > there. When exporting this kind of symbols, you're opening a can of worms by > potentially allowing each driver to implement the same kind of logic > that's already implemented in the NCI core. > > The solution I'd like to see implemented is the following one: Add 1 > additional nci_ops entry for handling proprietary packets on the Rx > path. From your driver perspective, you call nci_recv_frame for each and > every packet that you receive: No intercept, no trap. The core will call > your proprietary packets handler for each packet (NTF or RSP) with a > proprietary GID. > You should then be able to remove all the rx work, rx queue and > intercept logic from your driver. Thanks for the suggestion. Will do. Robert -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
