Fix: ROERR handling
On the AT32, an unhandled UART overrun (ROERR) latches: one overrun stops the USART feeding the DMA, every ack after it is lost, and only a re-init clears it. Signed-off-by: Niel Nielsen <nieldk@gmail.com>
This commit is contained in:
1 parent
9944e4d161
commit
73a3c85f2c
1 file changed
+10
@@ -150,6 +150,16 @@ int bwm_uart_write(const uint8_t *data, size_t len) {
|
||||
}
|
||||
|
||||
uint16_t bwm_uart_rx_available(void) {
|
||||
// An unhandled overrun (ROERR) latches on this USART and stops it feeding the
|
||||
// DMA - after one overrun every subsequent byte is lost until a re-init, which
|
||||
// is why a stalled OTA "recovers on re-run" but mostly fails in a session.
|
||||
// Clear it here so reception resumes on its own. ROERR clears by reading STS
|
||||
// then DT; we only do that when the flag is actually set - the DMA is stalled
|
||||
// then, so the byte we consume is the already-lost overrun byte.
|
||||
if (usart_flag_get(BWM_UART, USART_ROERR_FLAG) != RESET) {
|
||||
(void)BWM_UART->sts;
|
||||
(void)BWM_UART->dt;
|
||||
}
|
||||
return (uint16_t)((bwm_uart_rx_head() - s_rx_tail) & (BWM_RX_RING_SZ - 1));
|
||||
}
|
||||
|
||||
|
||||
Reference in new issue
Block a user