On 2014-11-21 21:28, Anton Lundin wrote:
+static dc_status_t +hw_ostc3_device_upgrade_firmware (dc_device_t *abstract, unsigned int checksum) +{
- dc_status_t rc = DC_STATUS_SUCCESS;
- hw_ostc3_device_t *device = (hw_ostc3_device_t *) abstract;
- dc_context_t *context = (abstract ? abstract->context : NULL);
- unsigned char buffer[5];
- uint32_le_array(checksum, buffer);
- // Compute a one byte checksum, so the device can validate the
firmware image.
- buffer[4] = 0x55;
- buffer[4] ^= buffer[0];
- buffer[4] = (buffer[4]<<1 | buffer[4]>>7);
- buffer[4] ^= buffer[1];
- buffer[4] = (buffer[4]<<1 | buffer[4]>>7);
- buffer[4] ^= buffer[2];
- buffer[4] = (buffer[4]<<1 | buffer[4]>>7);
- buffer[4] ^= buffer[3];
- buffer[4] = (buffer[4]<<1 | buffer[4]>>7);
Can you change this into a loop? It's already more than cryptic enough. Any idea whether this is some known checksum?
- // Now the device resets, and if everything is well, it reprograms.
- serial_sleep (device->port, 500);
- // FIXME: How should we force the application to close the device
here?
Why do we need to wait here? If the device is rebooting, then the firmware update is successful, right? Since we won't get any confirmation from the ostc, what's the point of waiting here? It only blocks the caller from calling close. Or am I missing something else?
What happens at this point? Does the usb-serial device node (e.g. /dev/ttyUSBx) disappear? Setting the state to REBOOTING (as you changed the FIXME in one of the other patches) prevents to do anything, except for calling close. That's probably the right thing to do, but I just wonder what's going on under the hood.
This might be different for the ostc3 (usb-serial) and sport (bluetooth), because I suspect the sport will terminate the bluetooth during the reboot, while for the ostc3 the usb-serial chip is still there.
Jef