The RTL9300 has some limitations on the maximum DMA transfers possible.
For reads this is 2080 bytes (520*4) for writes this is 520 bytes. Deal
with this by splitting transfers into appropriately sized parts.

Fixes: 42d20a6a61b8 ("spi: spi-mem: Add Realtek SPI-NAND controller")
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
---

Notes:
    Changes in v2:
    - Drop unecessary "also" change. data.in and data.out point to the same
      memory but the latter is marked as const which causes some compiler
      warnings when we're trying to get a dma mapping for it.

 drivers/spi/spi-realtek-rtl-snand.c | 46 +++++++++++++++++++----------
 1 file changed, 30 insertions(+), 16 deletions(-)

diff --git a/drivers/spi/spi-realtek-rtl-snand.c b/drivers/spi/spi-realtek-rtl-snand.c
index 23c42c8469e4..cd0484041147 100644
--- a/drivers/spi/spi-realtek-rtl-snand.c
+++ b/drivers/spi/spi-realtek-rtl-snand.c
@@ -231,19 +231,22 @@ static int rtl_snand_xfer(struct rtl_snand *snand, int cs, const struct spi_mem_
 
 static int rtl_snand_dma_xfer(struct rtl_snand *snand, int cs, const struct spi_mem_op *op)
 {
+	unsigned int pos, nbytes;
 	int ret;
 	dma_addr_t buf_dma;
 	enum dma_data_direction dir;
-	u32 trig;
+	u32 trig, len, maxlen;
 
 	ret = rtl_snand_xfer_head(snand, cs, op);
 	if (ret)
 		goto out_deselect;
 
 	if (op->data.dir == SPI_MEM_DATA_IN) {
+		maxlen = 2080;
 		dir = DMA_FROM_DEVICE;
 		trig = 0;
 	} else if (op->data.dir == SPI_MEM_DATA_OUT) {
+		maxlen = 520;
 		dir = DMA_TO_DEVICE;
 		trig = 1;
 	} else {
@@ -264,26 +267,37 @@ static int rtl_snand_dma_xfer(struct rtl_snand *snand, int cs, const struct spi_
 	if (ret)
 		goto out_unmap;
 
-	reinit_completion(&snand->comp);
+	pos = 0;
+	len = op->data.nbytes;
 
-	ret = regmap_write(snand->regmap, SNAFDRSAR, buf_dma);
-	if (ret)
-		goto out_disable_int;
+	while (pos < len) {
+		nbytes = len - pos;
+		if (nbytes > maxlen)
+			nbytes = maxlen;
 
-	ret = regmap_write(snand->regmap, SNAFDLR,
-			   CMR_WID(op->data.buswidth) | (op->data.nbytes & 0xffff));
-	if (ret)
-		goto out_disable_int;
+		reinit_completion(&snand->comp);
 
-	ret = regmap_write(snand->regmap, SNAFDTR, trig);
-	if (ret)
-		goto out_disable_int;
+		ret = regmap_write(snand->regmap, SNAFDRSAR, buf_dma + pos);
+		if (ret)
+			goto out_disable_int;
 
-	if (!wait_for_completion_timeout(&snand->comp, usecs_to_jiffies(20000)))
-		ret = -ETIMEDOUT;
+		pos += nbytes;
 
-	if (ret)
-		goto out_disable_int;
+		ret = regmap_write(snand->regmap, SNAFDLR,
+				CMR_WID(op->data.buswidth) | nbytes);
+		if (ret)
+			goto out_disable_int;
+
+		ret = regmap_write(snand->regmap, SNAFDTR, trig);
+		if (ret)
+			goto out_disable_int;
+
+		if (!wait_for_completion_timeout(&snand->comp, usecs_to_jiffies(20000)))
+			ret = -ETIMEDOUT;
+
+		if (ret)
+			goto out_disable_int;
+	}
 
 out_disable_int:
 	regmap_update_bits(snand->regmap, SNAFCFR, SNAFCFR_DMA_IE, 0);
-- 
2.47.0

On Thu, 31 Oct 2024 08:49:20 +1300, Chris Packham wrote:
> The RTL9300 has some limitations on the maximum DMA transfers possible.
> For reads this is 2080 bytes (520*4) for writes this is 520 bytes. Deal
> with this by splitting transfers into appropriately sized parts.
> 
> 

Applied to

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-next

Thanks!

[1/1] spi: spi-mem: rtl-snand: Correctly handle DMA transfers
      commit: 25d284715845a465a1a3693a09cf8b6ab8bd9caf

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark