[4.4.y-cip,2/4] serial: sh-sci: Terminate TX DMA during buffer flushing

Commit Message

Lad Prabhakar Sept. 3, 2020, 2:27 p.m. UTC

From: Geert Uytterhoeven <[e-mail]>

commit 775b7ffd7d6d5db320d99b0a485c51e04dfcf9f1 upstream.

While the .flush_buffer() callback clears sci_port.tx_dma_len since
commit 1cf4a7efdc71cab8 ("serial: sh-sci: Fix race condition causing
garbage during shutdown"), it does not terminate a transmit DMA
operation that may be in progress.

Fix this by terminating any pending DMA operations, and resetting the
corresponding cookie.

Signed-off-by: Geert Uytterhoeven <[e-mail]>
Reviewed-by: Eugeniu Rosca <[e-mail]>
Tested-by: Eugeniu Rosca <[e-mail]>

Link: https://lore.kernel.org/r/[e-mail]
Signed-off-by: Greg Kroah-Hartman <[e-mail]>
Signed-off-by: Lad Prabhakar <[e-mail]>
---
 drivers/tty/serial/sh-sci.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

Comments

Pavel Machek Sept. 3, 2020, 8:46 p.m. UTC | #1

Hi!

> From: Geert Uytterhoeven <[e-mail]>
> 
> commit 775b7ffd7d6d5db320d99b0a485c51e04dfcf9f1 upstream.
> 
> While the .flush_buffer() callback clears sci_port.tx_dma_len since
> commit 1cf4a7efdc71cab8 ("serial: sh-sci: Fix race condition causing
> garbage during shutdown"), it does not terminate a transmit DMA
> operation that may be in progress.
> 
> Fix this by terminating any pending DMA operations, and resetting the
> corresponding cookie.

What is the impact of the bug? DMA running during reboot, but not
doing any harm?

Anyway, this looks okay to me.

Best regards,
								Pavel

Patch

diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
index 2b5b342b7462..166ca04aad4e 100644
--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -1475,11 +1475,18 @@  static void sci_free_dma(struct uart_port *port)
 
 static void sci_flush_buffer(struct uart_port *port)
 {
+	struct sci_port *s = to_sci_port(port);
+
 	/*
 	 * In uart_flush_buffer(), the xmit circular buffer has just been
-	 * cleared, so we have to reset tx_dma_len accordingly.
+	 * cleared, so we have to reset tx_dma_len accordingly, and stop any
+	 * pending transfers
 	 */
-	to_sci_port(port)->tx_dma_len = 0;
+	s->tx_dma_len = 0;
+	if (s->chan_tx) {
+		dmaengine_terminate_async(s->chan_tx);
+		s->cookie_tx = -EINVAL;
+	}
 }
 #else /* !CONFIG_SERIAL_SH_SCI_DMA */
 static inline void sci_request_dma(struct uart_port *port)