davids5 commented on code in PR #11020:
URL: https://github.com/apache/nuttx/pull/11020#discussion_r1370532073
##########
arch/arm/src/imxrt/imxrt_serial.c:
##########
@@ -2345,14 +2346,26 @@ static bool imxrt_rxflowcontrol(struct uart_dev_s *dev,
#ifdef SERIAL_HAVE_RXDMA
static int imxrt_dma_receive(struct uart_dev_s *dev, unsigned int *status)
{
- struct imxrt_uart_s *priv = (struct imxrt_uart_s *)dev;
- uint32_t nextrx = imxrt_dma_nextrx(priv);
- int c = 0;
+ struct imxrt_uart_s *priv = (struct imxrt_uart_s *)dev;
+ static uint32_t last_nextrx = -1;
+ uint32_t nextrx = imxrt_dma_nextrx(priv);
+ int c = 0;
/* Check if more data is available */
if (nextrx != priv->rxdmanext)
{
+ /* Now we must ensure the cache is updated if the DMA has
+ * updated again.
+ */
+
+ if (last_nextrx != nextrx)
+ {
+ up_invalidate_dcache((uintptr_t)priv->rxfifo,
+ (uintptr_t)priv->rxfifo + RXDMA_BUFFER_SIZE);
+ last_nextrx = nextrx;
+ }
Review Comment:
Yes. We always have: The Rx is used in circular mode. Looping the data in
the the "fifo". If you invalidate in the callback only that is not enough.
The above code detect that the DMA moved and re-invalidates. It saves about
14/15 invalidates.
I think we need this fix across all the arches that use this "rxfifo" and
have a dcache
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscr...@nuttx.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
