[PATCH v2 3/3] i2c: stm32: only send a STOP upon transfer completion

[Alain Volmat]

Current function stm32_i2c_message_xfer is sending a STOP
whatever the result of the transaction is.  This can cause issues
such as making the bus busy since the controller itself is already
sending automatically a STOP when a NACK is generated.  This can
be especially seen when the processing get slower (ex: enabling lots
of debug messages), ending up send 2 STOP (one automatically by the
controller and a 2nd one at the end of the stm32_i2c_message_xfer
function).

Thanks to Jorge Ramirez-Ortiz for diagnosing and proposing a first
fix for this. [1]

[1] https://lore.kernel.org/u-boot/20220815145211.31342-2-jorge@foundries.io/

Reported-by: Jorge Ramirez-Ortiz, Foundries <jorge at foundries.io>
Signed-off-by: Jorge Ramirez-Ortiz <jorge at foundries.io>
Signed-off-by: Alain Volmat <alain.volmat at foss.st.com>
---
 drivers/i2c/stm32f7_i2c.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/i2c/stm32f7_i2c.c b/drivers/i2c/stm32f7_i2c.c
index 0ec67b5c12..8803979d3e 100644
--- a/drivers/i2c/stm32f7_i2c.c
+++ b/drivers/i2c/stm32f7_i2c.c
@@ -477,16 +477,16 @@ static int stm32_i2c_message_xfer(struct stm32_i2c_priv *i2c_priv,
 			if (ret)
 				break;
 
+			/* End of transfer, send stop condition */
+			mask = STM32_I2C_CR2_STOP;
+			setbits_le32(&regs->cr2, mask);
+
 			if (!stop)
 				/* Message sent, new message has to be sent */
 				return 0;
 		}
 	}
 
-	/* End of transfer, send stop condition */
-	mask = STM32_I2C_CR2_STOP;
-	setbits_le32(&regs->cr2, mask);
-
 	return stm32_i2c_check_end_of_message(i2c_priv);
 }
 
-- 
2.25.1