WICED Studio Bluetooth

[MaMi_1707981]

Hello Cypress WICED Studio Support team,

I found unexpected behavior of WICED Studio 6.2.1.2.

At the following environment, SPP data transfer isn't completed.

[Environment]

       ClientControl----CYW20706  <----->  CYW20706---ClientControl

       CYW920706WCDEVAL or CYBT-343026-EVAL

       WICED Studio 6.2.1.2

       apps/demo/hci_audio_gateway

       SPP profile (hci_control_spp.c)

[Test procedure]

       1. build and download "apps/demo/hci_audio_gateway" on WICED Studio.

       2. Select SPP tab, "Connect" to peer and "Send" large file on Client Control.

[Attached file]

       log/   log files

           sent_data.txt

               -> sent data

           Orig_ClientControl_log_master.txt

           Orig_ClientControl_log_slave.txt

               -> log of ClientControl. Master device sent data to slave device.

           Orig_BTSpy.txt

               -> log of BTSpy

           Orig_received_data.txt

               -> received data. It should be same with sent_data.txt, but it is too short.

           Fixed_ClientControl_log_master.txt

           Fixed_ClientControl_log_slave.txt

           Fixed_received_data.txt

               -> log files when the following patch is applied.

       patch/  patch files

           client_control_spp_patch.txt

               -> patch file for spp.cpp in common/client_control/source/.

           hci_control_spp_patch.txt

               -> patch file for hci_control_spp.c in apps/demp/hci_audio_gateway/.

[My investigation]

    wiced_bt_rfcomm_write_data() returns WICED_BT_RFCOMM_CMD_PENDING.

    But hci_control_cpp and ClientControl couldn't handle it correctly.

    The WICED_BT_RFCOMM_CMD_PENDING is not a critical error.

    So, it would be better to be handled same as WICED_BT_RFCOMM_SUCCESS.

Could you double check my investigation ?

And, if it is correct, please fix this issue at the next release.

[AnjanaM_61]

Hi,

We have escalated this issue to our stack team.

Regards,
Anjana

[MaMi_1707981]

Hi Anjana,

Thank you for updating your status.

Please share the result when investigation will be done.

Best regards,

Mizuno

[AnjanaM_61]

Hello,

The problem was that the SPP app was entering flow control. The attached 2 files correctly handle flow control "on and off" when notified by the BT stack and properly notify "Clientcontrol" with the proper event.

Please replace the existing "hci_control.h" and "hci_control_spp.c" with the attached.

The folder is : "C...\WICED-Studio-6.21.2\common\apps\demo\hci_audio_gateway\"

Could you please test it at your side and let us know ?

Thanks,
Anjana

[MaMi_1707981]

Hello Anjana,

I found two issues about your patches.

1) Some data was lost.

   Please see the attached file, "log.zip".

       I sent "log.txt". The received data is log_recv.txt.

   When wiced_bt_rfcomm_write_data() returns WICED_BT_RFCOMM_CMD_PENDING, hci_spp_write() also returns WICED_BT_RFCOMM_CMD_PENDING.

   In this case,

       hci_spp_send()

       {

           result = hci_spp_write( handle, p_data, length, &bytes_written );  ---> "result == WICED_BT_RFCOMM_CMD_PENDING"

           ---> NOTE: At this point, some data might be queued but not all data. i.e. bytes_written < length.

           is_an_error = ! (( result == WICED_BT_RFCOMM_SUCCESS ) || ( result == WICED_BT_RFCOMM_NO_MEM ));  ---> "is_an_error == TRUE"

           if( ( bytes_written != length ) && ( !is_an_error ) )  ---> "goto else"

               if ( !is_an_error )  --->  "goto else"

                   p_scb->tx_complete_pending = 1;

       hci_control_port_event_cback()

           if ( ( event & PORT_EV_FCS ) && ( p_scb->flow_control ) )

               if (hci_control_cb.spp_scb.tx_complete_pending)

                   hci_spp_send_tx_complete( handle, WICED_BT_RFCOMM_SUCCESS );

                   ---> !!! Although some data was not queued, but hci_spp_send_tx_complete() returns SUCCESS.

                        i.e. DATA was LOST.

                        We have to store unsent data to p_scb->pending_buffer.

2) Sending didn't terminated even if SPP link was disconnected.

   Please see the followings.

  

   This is a case which SPP link was disconnected by a peer device during SPP sending.

   In this case, hci_spp_write() returns ERROR.

       is_an_error = ! (( result == WICED_BT_RFCOMM_SUCCESS ) || ( result == WICED_BT_RFCOMM_NO_MEM ));  ---> "is_an_error == TRUE"

       if ( !is_an_error )  --->  "goto else"

           p_scb->tx_complete_pending = 1;

   i.e. hci_spp_send_tx_complete() is never called. So, ClientControl shows "Wait failed".

Could you double check them ?

Best regards,

Mizuno

[AnjanaM_61]

Hello,

Thanks for your results. Will check it and update you.

Regards,
Anjana

[AnjanaM_61]

Hi,

Can you please try the attached 2 files. Let us know your results.

Thanks,
Anjana

[MaMi_1707981]

Hello Anjana,

Unfortunately. some data was lost.

Please see the attached logs.

   Sent data: log_sent.txt

   Received data: log_recv_test2.txt

[Environment]

   CYW920606EVAL(UART=3Mbps): sent data

   CYBT_343026_EVAL(UART=1Mbps): received data

Best regards,

Mizuno

[AnjanaM_61]

Can you confirm what application is on the "CYBT_343026_EVAL(UART=1Mbps): received data"

Is it the same application running in this environment? Client Control -> SPP -> Receive Data?

Does this application have fixes / changes to "hci_control_spp.c" and "hci_control.h"?

[MaMi_1707981]

Hello Anjana,

> Is it the same application running in this environment? Client Control -> SPP -> Receive Data?

Yes. I used ClientControl for both evaluation boards.

   I rebuilt ClientControl to support 1Mbps. (add 1000000 to as32BaudRate in device_manager.cpp)

   Client Control -> SPP -> Receive -> Receive to file.

> Does this application have fixes / changes to "hci_control_spp.c" and "hci_control.h"?

Yes. Both FW for CYW920706 and CYBT-343026 are applied your patches, hci_control_spp.c and hci_control.h.

Best regards,

Mizuno

[MaMi_1707981 · Accepted solution]

It took a long time.

I close this ticket.