Re: [PATCH v4 1/4] firmware: ti_sci: Support transfers without response

From: Dhruva Gole
Date: Fri Dec 20 2024 - 06:41:47 EST


On Dec 19, 2024 at 21:02:12 +0100, Markus Schneider-Pargmann wrote:
> Check the header flags if an response is expected or not. If it is not
> expected skip the receive part of ti_sci_do_xfer(). This prepares the
> driver for one-way messages as prepare_sleep for Partial-IO.
>
> Signed-off-by: Markus Schneider-Pargmann <msp@xxxxxxxxxxxx>
> ---
> drivers/firmware/ti_sci.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/drivers/firmware/ti_sci.c b/drivers/firmware/ti_sci.c
> index 806a975fff22ae00ecb88587b2c47ba172120bc2..ec0c54935ac0d667323d98b86ac9d288b73be6aa 100644
> --- a/drivers/firmware/ti_sci.c
> +++ b/drivers/firmware/ti_sci.c
> @@ -398,10 +398,13 @@ static void ti_sci_put_one_xfer(struct ti_sci_xfers_info *minfo,
> static inline int ti_sci_do_xfer(struct ti_sci_info *info,
> struct ti_sci_xfer *xfer)
> {
> + struct ti_sci_msg_hdr *hdr = (struct ti_sci_msg_hdr *)xfer->tx_message.buf;

I think it's best to sort this in order? so it will come below struct
device?

> int ret;
> int timeout;
> struct device *dev = info->dev;
> bool done_state = true;
> + bool response_expected = !!(hdr->flags & (TI_SCI_FLAG_REQ_ACK_ON_PROCESSED |
> + TI_SCI_FLAG_REQ_ACK_ON_RECEIVED));
>
> ret = mbox_send_message(info->chan_tx, &xfer->tx_message);
> if (ret < 0)
> @@ -409,6 +412,9 @@ static inline int ti_sci_do_xfer(struct ti_sci_info *info,
>
> ret = 0;
>
> + if (!response_expected)
> + goto no_response;
> +

I am not a very big fan of using goto here, it feels like we should be
more clear in our implementation and make the else part below more
specific by using `else if` and checking for more specific SYSTEM_XX
states. Then based on POWEROFF, skip any polling or whatever wait for
response.
Infact, with that we may not even need the response_expected part, but I
am okay keeping both checks in place just for the clarity in design.

> if (system_state <= SYSTEM_RUNNING) {
> /* And we wait for the response. */
> timeout = msecs_to_jiffies(info->desc->max_rx_timeout_ms);
> @@ -429,6 +435,7 @@ static inline int ti_sci_do_xfer(struct ti_sci_info *info,
> dev_err(dev, "Mbox timedout in resp(caller: %pS)\n",
> (void *)_RET_IP_);
>

--
Best regards,
Dhruva Gole
Texas Instruments Incorporated