The wait_event() makes the application hang for ever in the following case:
[1] the hardware flow control is enabled.
[2] the other end (or the remote end) is terminated, and the TX is still
waiting for the hardware flow control signal to become asserted.
This patch fixes it by changing the wait_event to wait_event_interruptible.
Signed-off-by: Huang Shijie <shijie8@gmail.com>
Signed-off-by: Fugang Duan <B38611@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
unsigned long flags;
if (sport->dma_is_enabled) {
+ int ret;
+
/* We have to wait for the DMA to finish. */
- wait_event(sport->dma_wait,
+ ret = wait_event_interruptible(sport->dma_wait,
!sport->dma_is_rxing && !sport->dma_is_txing);
+ if (ret != 0) {
+ sport->dma_is_rxing = 0;
+ sport->dma_is_txing = 0;
+ dmaengine_terminate_all(sport->dma_chan_tx);
+ dmaengine_terminate_all(sport->dma_chan_rx);
+ }
+ imx_stop_tx(port);
imx_stop_rx(port);
imx_disable_dma(sport);
imx_uart_dma_exit(sport);