Make hci_uart_register_device() and hci_uart_unregister_device() call serdev_device_close()/open() themselves instead of relying on the various hci_uart drivers to do this for them. Besides reducing code complexity, this also ensures correct error checking of serdev_device_open(), which was missing in a few drivers. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
The HCILL or eHCILL protocol from TI is actually an H:4 protocol with a few extra events and thus can also use the h4_recv_buf helper. Instead of open coding the same funtionality add the extra events to the packet description table and use h4_recv_buf. Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Tested-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The struct hcill_cmd to create an skb with a single u8 is pointless. So just use skb_put_u8 instead. Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
download_firmware() is never called from atomic context. It is only called by ll_setup() that is called only via function pointer "->setup" used in hci_uart_setup() in drivers/bluetooth/hci_serdev.c and drivers/bluetooth/hci_ldisc.c. hci_uart_setup() is called only via function pointer "->setup" used in hci_dev_do_open() in net/bluetooth/hci_core.c. All of the above functions do not enter atomic context. Besides, ll_setup() calls msleep() and hci_dev_do_open calls mutex_lock(). So it indicates that all the above functions call functions that can sleep. Despite never getting called from atomic context, download_firmware() calls mdelay() for busy wait. That is not necessary and can be replaced with msleep to avoid busy wait. This is found by a static analysis tool named DCNS written by myself. Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
This adds the "ti,cc2560" compatible string for a TI CC2560 chip. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
This adds an optional nvmem consumer to get a BD address from an external source. The BD address is then set in the Bluetooth chip after the firmware has been loaded. This has been tested working with a TI CC2560A chip (in a LEGO MINDSTORMS EV3). Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
This adds support for setting the public address on Texas Instruments
Bluetooth chips using a vendor-specific command.
This has been tested on a CC2560A chip. The TI wiki also indicates that
this command should work on TI WL17xx/WL18xx Bluetooth chips.
During review, there was some question as to the correctness of the byte
swapping since TI's documentation is not clear on this matter. This can
be tested with the btmgmt utility from bluez. The adapter must be powered
off to change the address. If the baswap() is omitted, address is reversed.
In case there is a issue in the future, here is the output of btmon during
the command `btmgmt public-addr 00:11:22:33:44:55`:
Bluetooth monitor ver 5.43
= Note: Linux version 4.15.0-rc2-08561-gcb132a1-dirty (armv5tejl) 0.707043
= Note: Bluetooth subsystem version 2.22 0.707091
= New Index: 00:17:E7:BD:1C:8E (Primary,UART,hci0) [hci0] 0.707106
@ MGMT Open: btmgmt (privileged) version 1.14 {0x0002} 0.707124
@ MGMT Open: bluetoothd (privileged) version 1.14 {0x0001} 0.707137
@ MGMT Open: btmon (privileged) version 1.14 {0x0003} 0.707540
@ MGMT Command: Set Public Address (0x0039) plen 6 {0x0002} [hci0] 11.167991
Address: 00:11:22:33:44:55 (CIMSYS Inc)
@ MGMT Event: Command Complete (0x0001) plen 7 {0x0002} [hci0] 11.175681
Set Public Address (0x0039) plen 4
Status: Success (0x00)
Missing options: 0x00000000
@ MGMT Event: Index Removed (0x0005) plen 0 {0x0003} [hci0] 11.175757
@ MGMT Event: Index Removed (0x0005) plen 0 {0x0002} [hci0] 11.175757
@ MGMT Event: Index Removed (0x0005) plen 0 {0x0001} [hci0] 11.175757
= Open Index: 00:17:E7:BD:1C:8E [hci0] 11.176807
< HCI Command: Vendor (0x3f|0x0006) plen 6 [hci0] 11.176975
00 11 22 33 44 55 .."3DU
> HCI Event: Command Complete (0x0e) plen 4 [hci0] 11.188260
Vendor (0x3f|0x0006) ncmd 1
Status: Success (0x00)
...
< HCI Command: Read Local Version Info.. (0x04|0x0001) plen 0 [hci0] 11.189859
> HCI Event: Command Complete (0x0e) plen 12 [hci0] 11.190732
Read Local Version Information (0x04|0x0001) ncmd 1
Status: Success (0x00)
HCI version: Bluetooth 2.1 (0x04) - Revision 0 (0x0000)
LMP version: Bluetooth 2.1 (0x04) - Subversion 6431 (0x191f)
Manufacturer: Texas Instruments Inc. (13)
< HCI Command: Read BD ADDR (0x04|0x0009) plen 0 [hci0] 11.191027
> HCI Event: Command Complete (0x0e) plen 10 [hci0] 11.192101
Read BD ADDR (0x04|0x0009) ncmd 1
Status: Success (0x00)
Address: 00:11:22:33:44:55 (CIMSYS Inc)
...
Signed-off-by: David Lechner <david@lechnology.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
This adds an endianness conversion when setting the baudrate using a vendor-specific command. Otherwise, bad things might happen on a big- endian system. Suggested-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
This adds a #define for the vendor-specific HCI command to set the baudrate instead of using the bare 0xff36 multiple times. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
When a TI Bluetooth chip is reset, it takes about 100ms for the RTS line of the chip to deassert. For my use case with a TI CC2560A chip, this delay was not long enough and caused the local UART to never transmit at all (TI AM1808 SoC UART2). We can wait for the CTS signal using serdev_device_wait_for_cts() instead of trying to guess using msleep(). Also changed the comment to be more informative while we are touching this code. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>