What is the best way from userspace to tell when ST kernel registration is complete with BT WILINK?

Hi,

Can you paste the logs for the below?
"If I try to open a Bluetooth socket using hci_route/hci_open_dev in libbluetooth (which use the standard ioctl interface), I receive errors from the ioctl with the kernel."

And also share the list of commands/API's that you execute?

The BT Turn on mechanism w.r.t the user space is the same irrespective of whether you use ST or not. The "hdev->open" call is what the user space invokes to turn on BT. This is blocking until BT is turned on or till BT_REGISTER_TIMEOUT ( = 6 seconds).

"I don’t believe that the ‘hci0’ messages are part of ST, but part of the kernel Bluetooth modules initialization process "
You are correct - these are the stack init commands.

Regards,
Gigi Joseph.

Thanks for the response.

Here are the requested logs and a test application/init script:

test_app:

#include <iostream>
#include <stdlib.h>
#include <errno.h>
#include <fstream>

#include <bluetooth/bluetooth.h>
#include <bluetooth/hci.h>
#include <bluetooth/hci_lib.h>
//#include <bluetooth/l2cap.h>
//#include <bluetooth/sdp.h>
//#include <bluetooth/sdp_lib.h>

using namespace std;

int main(int argc, char* argv[])
{
   int route = -1;
   int devid = -1;

   ofstream kmsg("/dev/kmsg");
   if (kmsg.is_open())
   {
      // Initialize Bluetooth protocol stack and driver
      int status   = system("/path/to/bluetooth.sh start");
      if (WEXITSTATUS(status) == 0)
      {
         // Obtain route and socket
         route = hci_get_route(nullptr);
         if (route < 0)
         {
            kmsg << "DEBUG: failed hci_get_route - errno = " << errno << endl;
         }

         devid = hci_open_dev(route);
         if (devid < 0)
         {
            kmsg << "DEBUG: failed hci_open_dev - errno = " << errno << endl;
         }

         // Print values to kmsg log
         kmsg << "DEBUG: route = " << route << endl;
         kmsg << "DEBUG: devid = " << devid << endl;
      }
      else
      {
         kmsg << "DEBUG: script failure = " << status << endl;
      }
   }
   else
   {
      cerr << "Unable to open /dev/kmsg" << endl;
   }

   // Close socket
   if (devid >= 0)
   {
      hci_close_dev(devid);
   }
	return 0;
}

bluetooth.sh:

# cat /path/to/bluetooth.sh 

uim="/usr/bin/uim"
uim_args="-f /sys/devices/platform/kim"
btdaemon="/usr/sbin/bluetoothd"
case "$1" in
	start)
		start-stop-daemon -S -p /var/run/uim.pid -m -b -x $uim -- $uim_args
		insmod /lib/modules/btwilink.ko
		#do something with the device so the kernel doesn't time out:
		hciconfig up > /dev/null
		start-stop-daemon -S -v -p /var/run/bluetoothd.pid -m -x $btdaemon
		;;
	stop)
		hciconfig hci0 down
		start-stop-daemon -K -p /var/run/uim.pid
		rmmod /lib/modules/btwilink.ko
		;;
	*)
		echo "Usage: /path/to/bluetooth.sh {start|stop}"
		exit 1
esac

echo "BT_MODULE_READY TO APP" > /dev/kmsg
exit 0

output:

# dmesg -c
...
# ./test_app
# dmesg
Bluetooth: Bluetooth Driver for TI WiLink - Version 1.0
hdev c20ab000
HCI device registered (hdev c20ab000)
hci0 c20ab000
(stc): st_register(4) 
(stc):  chnl_id list empty :4 
(stk) : st_kim_startBT_MODULE_READY TO APP
DEBUG: failed hci_get_route - errno = 19
DEBUG: route = -1
DEBUG: devid = 118
(stk) :ldisc_install = 1
(stc): st_tty_open 
(stk) :line discipline installed
(stk) :TIInit_8.6.15.bts
(stk) :ldisc_install = 2   #Note I added this in the kernel for debugging
(stc): add_channel_to_table: id 4
(stc): st_register(2) 
(stc): add_channel_to_table: id 2
(stc): st_register(3) 
(stc): add_channel_to_table: id 3
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 4
hci0: type 1 len 5
hci0: type 1 len 6
hci0: type 1 len 11
hci0: type 1 len 12
hci0: type 1 len 4
hci0: type 1 len 5
hci0: type 1 len 5
hci0: type 1 len 4
hci0: type 1 len 5
hci0: type 1 len 6
hci0: type 1 len 6
#

And as a clarification, I'm looking for an event that notifies me when the stack init commands are complete, so that hci_get_route will not fail.

Great. Thank you for your response.