Some adapters need to be prepared/unprepared before bitbanging the bus.
Do this for the initial STOP, too.
Signed-off-by: Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx>
---
Ok, another idea to fix the regression. I'm not 100% sure if the placement is
perfect, but it should serve well enough as a proof of concept to see if this
is the actual issue. It looks promising, though.
drivers/i2c/algos/i2c-algo-bit.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/drivers/i2c/algos/i2c-algo-bit.c b/drivers/i2c/algos/i2c-algo-bit.c
index 4a34f311e1ff..fab4437ecbb5 100644
--- a/drivers/i2c/algos/i2c-algo-bit.c
+++ b/drivers/i2c/algos/i2c-algo-bit.c
@@ -648,10 +648,19 @@ static int __i2c_bit_add_bus(struct i2c_adapter *adap,
adap->quirks = &i2c_bit_quirk_no_clk_stretch;
/* Bring bus to a known state. Looks like STOP if bus is not free yet */
+ if (bit_adap->pre_xfer) {
+ ret = bit_adap->pre_xfer(adap);
+ if (ret < 0)
+ return ret;
+ }
+
setscl(bit_adap, 1);
udelay(bit_adap->udelay);
setsda(bit_adap, 1);
+ if (bit_adap->post_xfer)
+ bit_adap->post_xfer(adap);
+
ret = add_adapter(adap);
if (ret < 0)
return ret;
--
2.11.0
