On Fri, 26 Jan 2024 at 00:56, Joe Komlodi <komlodi@google.com> wrote:
>
> If a reset comes while the SMBus device is not in its idle state, it's
> possible for it to get confused on valid transactions post-reset.
>
> Signed-off-by: Joe Komlodi <komlodi@google.com>
> ---
> hw/i2c/smbus_slave.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> diff --git a/hw/i2c/smbus_slave.c b/hw/i2c/smbus_slave.c
> index e24a1ef472..58abde29de 100644
> --- a/hw/i2c/smbus_slave.c
> +++ b/hw/i2c/smbus_slave.c
> @@ -201,10 +201,19 @@ static int smbus_i2c_send(I2CSlave *s, uint8_t data)
> return 0;
> }
>
> +static void smbus_device_enter_reset(Object *obj, ResetType type)
> +{
> + SMBusDevice *dev = SMBUS_DEVICE(obj);
> + dev->mode = SMBUS_IDLE;
> + dev->data_len = 0;
> +}
> +
> static void smbus_device_class_init(ObjectClass *klass, void *data)
> {
> I2CSlaveClass *sc = I2C_SLAVE_CLASS(klass);
> + ResettableClass *rc = RESETTABLE_CLASS(klass);
>
> + rc->phases.enter = smbus_device_enter_reset;
> sc->event = smbus_i2c_event;
> sc->recv = smbus_i2c_recv;
> sc->send = smbus_i2c_send;
> -
This should probably be the 'hold' phase of reset, unless there's
a strong reason to do it in the 'enter' phase.
thanks
-- PMM