Re: [RFC PATCHv2 1/7] HSI: Introducing HSI framework

[Sebastien Jan <s-jan@ti.com>]

Hi Carlos,

After review, I do not have many comments on the interface, as we already 
aligned on most of it.

Please see my comments inlined below.

On Friday 07 May 2010 17:18:31 Carlos Chinea wrote:
[strip]
> diff --git a/include/linux/hsi/hsi.h b/include/linux/hsi/hsi.h
[strip]
> +/**
> + * hsi_start_tx - Signal the port that the client wants to start a TX
> + * @cl: Pointer to the HSI client
> + *
> + * Return -errno on failure, 0 on success
> + */
> +static inline int hsi_start_tx(struct hsi_client *cl)
> +{
> +       if (!hsi_port_claimed(cl))
> +               return -EACCES;
> +       return hsi_get_port(cl)->start_tx(cl);
> +}
> +
> +/**
> + * hsi_stop_tx - Signal the port that the client no longer wants to
>  transmit + * @cl: Pointer to the HSI client
> + *
> + * Return -errno on failure, 0 on success
> + */
> +static inline int hsi_stop_tx(struct hsi_client *cl)
> +{
> +       if (!hsi_port_claimed(cl))
> +               return -EACCES;
> +       return hsi_get_port(cl)->stop_tx(cl);
> +}

As I can see, these two I/F functions are the way an HSI protocol layer can 
play with Tx_wake lines if it has to, right? 
I suppose it allows more flexibility with regards to 3/4 wires HSI flavors 
management and avoids additional callbacks to Tx_wake related events?