[PATCH v2] Input: ads7846 - don't use scratch for tx_buf when clearing register

[PATCH v2] Input: ads7846 - don't use scratch for tx_buf when clearing register

[Kris Bahnsen]

The workaround for XPT2046 clears the command register, giving the
touchscreen controller a NOP. The change incorrectly re-uses the
req->scratch variable which is used as rx_buf for xfer[5], so by
the time xfer[6] occurs, the contents of req->scratch may not be
0. It was found that the touchscreen controller can end up in
a completely unresponsive state due to it being given a command
the driver does not expect.

Instead, rely on the spi_transfer behavior of tx_buf being NULL to
transmit all 0 bits. Also set rx_buf to NULL because the value
returned does not matter. Thus moving the 3 byte pattern to clear
the command register to a single message.

This change was tested on real TSC2046 and ADS7843 controllers,
but not the XPT2046 the workaround was originally created for.
Confirming that the original modification to clear the command
register does not impact either real controller.

Fixes: 781a07da9bb94 ("Input: ads7846 - add dummy command register clearing cycle")
Cc: stable@vger.kernel.org
Co-developed-by: Mark Featherston <mark@embeddedTS.com>
Signed-off-by: Mark Featherston <mark@embeddedTS.com>
Signed-off-by: Kris Bahnsen <kris@embeddedTS.com>
---

V1 -> V2: Don't use rx_buf when clearing command reg

 drivers/input/touchscreen/ads7846.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)

diff --git a/drivers/input/touchscreen/ads7846.c b/drivers/input/touchscreen/ads7846.c
index 4b39f7212d35c..93edd2419db5b 100644
--- a/drivers/input/touchscreen/ads7846.c
+++ b/drivers/input/touchscreen/ads7846.c
@@ -327,7 +327,7 @@ struct ser_req {
 	u8			ref_off;
 	u16			scratch;
 	struct spi_message	msg;
-	struct spi_transfer	xfer[8];
+	struct spi_transfer	xfer[7];
 	/*
 	 * DMA (thus cache coherency maintenance) requires the
 	 * transfer buffers to live in their own cache lines.
@@ -403,16 +403,10 @@ static int ads7846_read12_ser(struct device *dev, unsigned command)
 	spi_message_add_tail(&req->xfer[5], &req->msg);
 
 	/* clear the command register */
-	req->scratch = 0;
-	req->xfer[6].tx_buf = &req->scratch;
-	req->xfer[6].len = 1;
+	req->xfer[6].len = 3;
+	CS_CHANGE(req->xfer[6]);
 	spi_message_add_tail(&req->xfer[6], &req->msg);
 
-	req->xfer[7].rx_buf = &req->scratch;
-	req->xfer[7].len = 2;
-	CS_CHANGE(req->xfer[7]);
-	spi_message_add_tail(&req->xfer[7], &req->msg);
-
 	scoped_guard(mutex, &ts->lock) {
 		ads7846_stop(ts);
 		status = spi_sync(spi, &req->msg);

base-commit: dd6c438c3e64a5ff0b5d7e78f7f9be547803ef1b
-- 
2.34.1

Re: [PATCH v2] Input: ads7846 - don't use scratch for tx_buf when clearing register

[Dmitry Torokhov]

On Mon, Apr 27, 2026 at 05:46:57PM +0000, Kris Bahnsen wrote:
> The workaround for XPT2046 clears the command register, giving the
> touchscreen controller a NOP. The change incorrectly re-uses the
> req->scratch variable which is used as rx_buf for xfer[5], so by
> the time xfer[6] occurs, the contents of req->scratch may not be
> 0. It was found that the touchscreen controller can end up in
> a completely unresponsive state due to it being given a command
> the driver does not expect.
> 
> Instead, rely on the spi_transfer behavior of tx_buf being NULL to
> transmit all 0 bits. Also set rx_buf to NULL because the value
> returned does not matter. Thus moving the 3 byte pattern to clear
> the command register to a single message.

Unfortunately my suggestion was flawed: I think this will flood the logs
with "Bufferless transfer has length %3". We need to have either tx or
rx buffer :(

Thanks.

-- 
Dmitry