[PATCH v2 0/6] spi: spi-sun4i: assorted fixes

This is a series of patches for the spi-sun4i driver, some of which have
been sitting around for a while.

1-4 are fixes and 5 is a small patch that introduces reporting of the
effective SPI speed.

Signed-off-by: Jonas Rebmann <redacted>
---
Changes in v2:
- Add Patch 5/6 to fix division by zero potential in 6/6
- 3/6: Correctly subtract 1 to gain the register value for SUN4I_CLK_CTL_CDR1
- Mitigate race condition potential in 4/6 by disabling and syncing
  interrupts before sun4i_spi_drain_fifo() in the timeout path
- Link to v1: https://patch.msgid.link/20260902-spi-sun4i-fixes-v1-0-19985ef75673@pengutronix.de

---
Jonas Rebmann (1):
      spi: spi-sun4i: sun4i_spi_transfer_one(): add error handling at clock setup

Marc Kleine-Budde (5):
      spi: spi-sun4i: sun4i_spi_transfer_one(): fix transfers for SUN4I_FIFO_DEPTH
      spi: spi-sun4i: sun4i_spi_fill_fifo(): always apply A10 FIFO fill level workaround
      spi: spi-sun4i: sun4i_spi_transfer_one(): fix setting of clock rate
      spi: spi-sun4i: sun4i_spi_handler(): fix race condition between transfer completion and RX FIFO interrupt
      spi: spi-sun4i: sun4i_spi_transfer_one(): report effectively used speed_hz of transfer

 drivers/spi/spi-sun4i.c | 68 +++++++++++++++++++++++++++++--------------------
 1 file changed, 40 insertions(+), 28 deletions(-)
---
base-commit: 89a312991dc6e638a36adc43ccb91dbc25504c04
change-id: 20260902-spi-sun4i-fixes-1a50e880dc7b

Best regards,
--  
Jonas Rebmann [off-list ref]

From: Marc Kleine-Budde <mkl@pengutronix.de>

In commit 6d9fe44bd73d ("spi: sun4i: fix FIFO limit"), the TX-FIFO is
filled max to SUN4I_FIFO_DEPTH - 1 (= 63) bytes to work around timeouts
observed on A10s SoCs.

Commit 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO
size") added support for transfers larger than the FIFO size. But this
commit only enabled the TX-FIFO empty interrupt for transfers larger
than the FIFO (= 64) bytes.

This breaks transfers with exactly 64 bytes: the TX-FIFO is only filled
with 63 bytes but the interrupt to refill the FIFO is not triggered. The
problem can be reproduced with the following command:

| spidev_test -D /dev/spidev0.1 -S 64 -s 20000000 -I 1
|
| [ 7797.548745] spi_master spi0: spi0.1: timeout transferring 64 bytes@20000000Hz for 110(100)ms
| [ 7797.557237] spidev spi0.1: SPI transfer failed: -110
| [ 7797.562308] spi_master spi0: failed to transfer one message from queue
| [ 7797.568936] spi_master spi0: noqueue transfer failed

To fix the problem enable the TX-FIFO interrupt if the total TX length
is larger than SUN4I_FIFO_DEPTH - 1 (= 63) bytes.

Fixes: 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO size")
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index d5c16392cd4d..2e2324453905 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -322,7 +322,7 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 	sun4i_spi_enable_interrupt(sspi, SUN4I_INT_CTL_TC |
 					 SUN4I_INT_CTL_RF_F34);
 	/* Only enable Tx FIFO interrupt if we really need it */
-	if (tx_len > SUN4I_FIFO_DEPTH)
+	if (tx_len > SUN4I_FIFO_DEPTH - 1)
 		sun4i_spi_enable_interrupt(sspi, SUN4I_INT_CTL_TF_E34);
 
 	/* Start the transfer */

-- 
2.55.0.123.gf60db8d575

From: Marc Kleine-Budde <mkl@pengutronix.de>

A SPI transfer defines the _maximum_ speed of the SPI transfer. However
the driver doesn't take into account that the clock divider is always
rounded down (due to integer arithmetic). This results in a too high
clock rate for the SPI transfer.

E.g.: with an mclk_rate of 24 MHz and an SPI transfer speed of 10 MHz,
the original code calculates a reg of "0", which results in an effective
divider of "2" and a 12 MHz clock for the SPI transfer.

Use DIV_ROUND_UP() instead of a plain integer division to fix the
problem.

While there simplify the divider calculation for the CDR1 case, use
order_base_2() instead of two ilog2() calculations.

Fixes: b5f6517948cc ("spi: sunxi: Add Allwinner A10 SPI controller driver")
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 16 +++++++---------
 1 file changed, 7 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index 3649bcabcc9a..8a9dcd3b6b8f 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -212,7 +212,7 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 				  struct spi_transfer *tfr)
 {
 	struct sun4i_spi *sspi = spi_controller_get_devdata(host);
-	unsigned int mclk_rate, div;
+	unsigned int mclk_rate, div, div_cdr1, div_cdr2;
 	unsigned long time_left;
 	unsigned int start, end, tx_time;
 	unsigned int tx_len = 0;

@@ -296,15 +296,13 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 	 * First try CDR2, and if we can't reach the expected
 	 * frequency, fall back to CDR1.
 	 */
-	div = mclk_rate / (2 * tfr->speed_hz);
-	if (div <= (SUN4I_CLK_CTL_CDR2_MASK + 1)) {
-		if (div > 0)
-			div--;
-
-		reg = SUN4I_CLK_CTL_CDR2(div) | SUN4I_CLK_CTL_DRS;
+	div_cdr1 = DIV_ROUND_UP(mclk_rate, tfr->speed_hz);
+	div_cdr2 = DIV_ROUND_UP(div_cdr1, 2);
+	if (div_cdr2 <= (SUN4I_CLK_CTL_CDR2_MASK + 1)) {
+		reg = SUN4I_CLK_CTL_CDR2(div_cdr2 - 1) | SUN4I_CLK_CTL_DRS;
 	} else {
-		div = ilog2(mclk_rate) - ilog2(tfr->speed_hz);
-		reg = SUN4I_CLK_CTL_CDR1(div);
+		div = min(SUN4I_CLK_CTL_CDR1_MASK + 1, order_base_2(div_cdr1));
+		reg = SUN4I_CLK_CTL_CDR1(div - 1);
 	}
 
 	sun4i_spi_write(sspi, SUN4I_CLK_CTL_REG, reg);

-- 
2.55.0.123.gf60db8d575

Bail out if clk_get_rate() or clk_set_rate() fail during
sun4i_spi_transfer_one(). This might happen if the clock isn't known yet
or fails to set the requested rate.

This also ensures that mclk_rate is nonzero and can be safely divided by
in a future commit.

Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index ea8be0170fbf..94e1c7bbd632 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -279,9 +279,17 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 
 	/* Ensure that we have a parent clock fast enough */
 	mclk_rate = clk_get_rate(sspi->mclk);
+	if (!mclk_rate)
+		return -EINVAL;
+
 	if (mclk_rate < (2 * tfr->speed_hz)) {
-		clk_set_rate(sspi->mclk, 2 * tfr->speed_hz);
+		ret = clk_set_rate(sspi->mclk, 2 * tfr->speed_hz);
+		if (ret)
+			return ret;
+
 		mclk_rate = clk_get_rate(sspi->mclk);
+		if (!mclk_rate)
+			return -EINVAL;
 	}
 
 	/*

-- 
2.55.0.123.gf60db8d575

From: Marc Kleine-Budde <mkl@pengutronix.de>

In commit 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO
size"), support for transfers larger than the FIFO size was added.

This commit moves the draining of the RX-FIFO from
sun4i_spi_transfer_one() (after completion of the transfer) to the IRQ
handler when the IRQ "transfer complete" is handled. In addition, the
IRQ "RX-FIFO 3/4 full" is activated for all transfers.

However, this does not take into account that the RX-IRQ for transfers
that exceed 3/4 of the FIFO size is still pending after the IRQ
"transfer complete" has been processed. All interrupt sources are only
deactivated after the wait_for_completion_timeout() in
sun4i_spi_transfer_one().

This opens a race window for "RX-FIFO 3/4 full" interrupts to come.
The sequence is as follows:

| sun4i_spi_transfer_one()
|     sun4i_spi_fill_fifo()       // fill TX-FIFO with 48 bytes
|     // enable RX-FIFO 3/4 full IRQ
|     wait_for_completion_timeout();
|
| // SPI controller transfers 48 bytes
| // SPI controller issues "transfer complete" and "RX-FIFO 3/4 full" IRQ
|
| // IRQ handler start
|     sun4i_spi_handler()
|         // ACK "transfer complete" IRQ
|         sun4i_spi_drain_fifo();
|         complete();                         ----.
|         return IRQ_HANDLED;                      \
| // IRQ handler end                                \__ race
|                                                   /   window
|     // wait_for_completion_timeout() continues   /
|     // disable all IRQ sources              ----'

Avoid the race condition by disabling all interrupts when handling the
"transfer complete" IRQ and before calling complete(). Also move the
draining of the RX-FIFO back into sun4i_spi_transfer_one() where it
was before commit 196737912da5 ("spi: sun4i: Allow transfers larger than
FIFO size").

This has the added benefit of spending a little less time in the IRQ
handler.

Cc: Tobias Schramm <t.schramm@manjaro.org>
Fixes: 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO size")
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index 8a9dcd3b6b8f..ea8be0170fbf 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -82,6 +82,8 @@ struct sun4i_spi {
 
 	struct completion	done;
 
+	int			irq;
+
 	const u8		*tx_buf;
 	u8			*rx_buf;
 	int			len;

@@ -333,6 +335,7 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 	start = jiffies;
 	time_left = wait_for_completion_timeout(&sspi->done,
 						msecs_to_jiffies(tx_time));
+
 	end = jiffies;
 	if (!time_left) {
 		dev_warn(&host->dev,

@@ -340,12 +343,11 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 			 dev_name(&spi->dev), tfr->len, tfr->speed_hz,
 			 jiffies_to_msecs(end - start), tx_time);
 		ret = -ETIMEDOUT;
-		goto out;
+		sun4i_spi_write(sspi, SUN4I_INT_CTL_REG, 0);
+		synchronize_irq(sspi->irq);
 	}
 
-
-out:
-	sun4i_spi_write(sspi, SUN4I_INT_CTL_REG, 0);
+	sun4i_spi_drain_fifo(sspi, SUN4I_FIFO_DEPTH);
 
 	return ret;
 }

@@ -357,8 +359,7 @@ static irqreturn_t sun4i_spi_handler(int irq, void *dev_id)
 
 	/* Transfer complete */
 	if (status & SUN4I_INT_CTL_TC) {
-		sun4i_spi_write(sspi, SUN4I_INT_STA_REG, SUN4I_INT_CTL_TC);
-		sun4i_spi_drain_fifo(sspi, SUN4I_FIFO_DEPTH);
+		sun4i_spi_write(sspi, SUN4I_INT_CTL_REG, 0);
 		complete(&sspi->done);
 		return IRQ_HANDLED;
 	}

@@ -456,6 +457,7 @@ static int sun4i_spi_probe(struct platform_device *pdev)
 		return ret;
 	}
 
+	sspi->irq = irq;
 	sspi->host = host;
 	host->max_speed_hz = 100 * 1000 * 1000;
 	host->min_speed_hz = 3 * 1000;

-- 
2.55.0.123.gf60db8d575

From: Marc Kleine-Budde <mkl@pengutronix.de>

Report the effectively used speed_hz for the transfer by setting
tfr->effective_speed_hz.

For more information see commit 5d7e2b5ed585 ("spi: core: allow
reporting the effectivly used speed_hz for a transfer"), which adds this
feature to the SPI core.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index 94e1c7bbd632..59941b38b757 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -310,9 +310,11 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 	div_cdr2 = DIV_ROUND_UP(div_cdr1, 2);
 	if (div_cdr2 <= (SUN4I_CLK_CTL_CDR2_MASK + 1)) {
 		reg = SUN4I_CLK_CTL_CDR2(div_cdr2 - 1) | SUN4I_CLK_CTL_DRS;
+		tfr->effective_speed_hz = mclk_rate / (2 * div_cdr2);
 	} else {
 		div = min(SUN4I_CLK_CTL_CDR1_MASK + 1, order_base_2(div_cdr1));
 		reg = SUN4I_CLK_CTL_CDR1(div - 1);
+		tfr->effective_speed_hz = mclk_rate / (1 << div);
 	}
 
 	sun4i_spi_write(sspi, SUN4I_CLK_CTL_REG, reg);

-- 
2.55.0.123.gf60db8d575

From: Marc Kleine-Budde <mkl@pengutronix.de>

In commit 6d9fe44bd73d ("spi: sun4i: fix FIFO limit"), the TX FIFO is
filled max to SUN4I_FIFO_DEPTH - 1 (= 63) bytes to work around
timeouts observed on A10s SoCs.

Commit 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO
size") added support for transfers larger than the FIFO size. This
commit did not consider the A10 workaround (limit the TX FIFO fill
size to SUN4I_FIFO_DEPTH - 1 (= 63) bytes) for refilling the TX FIFO
in the IRQ handler.

To apply the workaround independent from where sun4i_spi_fill_fifo()
is called, remove the length argument from the function and directly
take the max fill level of SUN4I_FIFO_DEPTH - 1 into account when
calculating the free space in the FIFO.

Due to the lack of HW this patch has not been tested on an A10 SoC,
but on an A20 SoC. It was not possible to reproduce the timeout on the
A20.

Fixes: 196737912da5 ("spi: sun4i: Allow transfers larger than FIFO size")
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Jonas Rebmann <redacted>
---
 drivers/spi/spi-sun4i.c | 24 +++++++++++++-----------
 1 file changed, 13 insertions(+), 11 deletions(-)

diff --git a/drivers/spi/spi-sun4i.c b/drivers/spi/spi-sun4i.c
index 2e2324453905..3649bcabcc9a 100644
--- a/drivers/spi/spi-sun4i.c
+++ b/drivers/spi/spi-sun4i.c

@@ -142,15 +142,21 @@ static inline void sun4i_spi_drain_fifo(struct sun4i_spi *sspi, int len)
 	}
 }
 
-static inline void sun4i_spi_fill_fifo(struct sun4i_spi *sspi, int len)
+static inline void sun4i_spi_fill_fifo(struct sun4i_spi *sspi)
 {
 	u32 cnt;
+	int len;
 	u8 byte;
 
-	/* See how much data we can fit */
-	cnt = SUN4I_FIFO_DEPTH - sun4i_spi_get_tx_fifo_count(sspi);
+	/*
+	 * See how much data we can fit
+	 *
+	 * Filling the FIFO fully causes timeout for some reason
+	 * at least on spi2 on A10s
+	 */
+	cnt = SUN4I_FIFO_DEPTH - 1 - sun4i_spi_get_tx_fifo_count(sspi);
 
-	len = min3(len, (int)cnt, sspi->len);
+	len = min_t(int, cnt, sspi->len);
 
 	while (len--) {
 		byte = sspi->tx_buf ? *sspi->tx_buf++ : 0;

@@ -311,12 +317,8 @@ static int sun4i_spi_transfer_one(struct spi_controller *host,
 	sun4i_spi_write(sspi, SUN4I_BURST_CNT_REG, SUN4I_BURST_CNT(tfr->len));
 	sun4i_spi_write(sspi, SUN4I_XMIT_CNT_REG, SUN4I_XMIT_CNT(tx_len));
 
-	/*
-	 * Fill the TX FIFO
-	 * Filling the FIFO fully causes timeout for some reason
-	 * at least on spi2 on A10s
-	 */
-	sun4i_spi_fill_fifo(sspi, SUN4I_FIFO_DEPTH - 1);
+	/* Fill the TX FIFO */
+	sun4i_spi_fill_fifo(sspi);
 
 	/* Enable the interrupts */
 	sun4i_spi_enable_interrupt(sspi, SUN4I_INT_CTL_TC |

@@ -373,7 +375,7 @@ static irqreturn_t sun4i_spi_handler(int irq, void *dev_id)
 
 	/* Transmit FIFO 3/4 empty */
 	if (status & SUN4I_INT_CTL_TF_E34) {
-		sun4i_spi_fill_fifo(sspi, SUN4I_FIFO_DEPTH);
+		sun4i_spi_fill_fifo(sspi);
 
 		if (!sspi->len)
 			/* nothing left to transmit */

-- 
2.55.0.123.gf60db8d575