Hi Mukesh,
On 3/11/2024 8:56 AM, Hans Hu wrote:
This is a bug that was accidentally introduced when
adjusting the wmt driver. Now fix it
what exactly is the bug which you are fixing here ?
This bug was introduced by myself in a recent commit,
id: 4b0c0569f03261aa4c10c8f5b24df6c1ca27f889
https://patchwork.ozlabs.org/project/linux-i2c/patch/20240306212413.1850236-5-andi.shyti@xxxxxxxxxx/
The function viai2c_irq_xfer() is working in the interrupt context,
if it returns a non-0 value indicating that the current msg access
has ended, otherwise it has not ended.
For the access that msg->len is 0, when the interruption occurs,
it means that the access has ended, it should return 1;
Otherwise wait_for_completion_timeout() will timeout.
Signed-off-by: Hans Hu <hanshu-oc@xxxxxxxxxxx>
---
drivers/i2c/busses/i2c-viai2c-common.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/i2c/busses/i2c-viai2c-common.c
b/drivers/i2c/busses/i2c-viai2c-common.c
index 4c208b3a509e..894d24c6b4d3 100644
--- a/drivers/i2c/busses/i2c-viai2c-common.c
+++ b/drivers/i2c/busses/i2c-viai2c-common.c
@@ -145,7 +145,7 @@ static int viai2c_irq_xfer(struct viai2c *i2c)
if (msg->len == 0) {
val = VIAI2C_CR_TX_END | VIAI2C_CR_CPU_RDY |
VIAI2C_CR_ENABLE;
writew(val, base + VIAI2C_REG_CR);
- return 0;
+ return 1;
Question: Do you really need to do anything when no data is there to
transfer ? I am not sure what's the strategy adopted here.
This is to be consistent with former i2c-wmt.c:
https://elixir.bootlin.com/linux/v6.8/source/drivers/i2c/busses/i2c-wmt.c#L175
Hans,
Thanks
