What happens: if I send a serialization packet whose last byte is SLIP_END (0xC0) or SLIP_ESC (0xDB), the program crashes in an assert in ser_hal_transport.c:141.
Why: The event SER_PHY_EVT_TX_PKT_SENT is triggered two times, because ser_phy_hci_slip_cdc:tx_buf_put() is called two times with m_tx_phase == PHASE_ACK_END or m_tx_phase == PHASE_PACKET_END if the last byte needs to be escaped. This will cause two flushes and two events.
The bug might also be present in ser_phy_hci_slip.c because the code looks similar, but I haven't checked.
My fix: add is_escape argument to tx_buf_put, don't trigger events if it's true.
It works for me, but it feels a bit hacky, there's probably a better way by reworking the state machine.
142c142 < static bool tx_buf_put(uint8_t data_byte) --- > static bool tx_buf_put(uint8_t data_byte, bool is_escape) 150c150 < if (m_tx_phase == PHASE_ACK_END) --- > if (m_tx_phase == PHASE_ACK_END && !is_escape) 156c156 < else if (m_tx_phase == PHASE_PACKET_END) --- > else if (m_tx_phase == PHASE_PACKET_END && !is_escape) 205c205 < can_continue = tx_buf_put(tx_escaped_data); --- > can_continue = tx_buf_put(tx_escaped_data, true); 211c211 < can_continue = tx_buf_put(APP_SLIP_END); --- > can_continue = tx_buf_put(APP_SLIP_END, false); 220c220 < can_continue = tx_buf_put(APP_SLIP_END); --- > can_continue = tx_buf_put(APP_SLIP_END, false); 263c263 < can_continue = tx_buf_put(data); --- > can_continue = tx_buf_put(data, false);
sga
