Re: [PATCH 03/13] scpi: Add legacy send, prepare and handle remote functions

From: Neil Armstrong
Date: Tue Aug 23 2016 - 04:20:56 EST


On 08/19/2016 06:13 PM, Sudeep Holla wrote:
>
>
> On 18/08/16 11:10, Neil Armstrong wrote:
>> In order to support the legacy SCPI procotol, add specific message_send,
>> prepare_tx and handle_remote functions since the legacy procotol
>> do not support message queuing and does not store the command word in the
>> tx_payload data.
>>
>> Signed-off-by: Neil Armstrong <narmstrong@xxxxxxxxxxxx>
>> ---
>> drivers/firmware/arm_scpi.c | 69 +++++++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 69 insertions(+)
>>
>> diff --git a/drivers/firmware/arm_scpi.c b/drivers/firmware/arm_scpi.c
>> index 0bb6134..50b1297 100644
>> --- a/drivers/firmware/arm_scpi.c
>> +++ b/drivers/firmware/arm_scpi.c
>> @@ -202,6 +202,7 @@ struct scpi_chan {
>> spinlock_t rx_lock; /* locking for the rx pending list */
>> struct mutex xfers_lock;
>> u8 token;
>> + struct scpi_xfer *t;
>> };
>>
>> struct scpi_drvinfo {
>> @@ -364,6 +365,23 @@ static void scpi_handle_remote_msg(struct mbox_client *c, void *msg)
>> scpi_process_cmd(ch, cmd);
>> }
>>
>> +static void legacy_scpi_handle_remote_msg(struct mbox_client *c, void *__msg)
>> +{
>> + struct scpi_chan *ch =
>> + container_of(c, struct scpi_chan, cl);
>> + struct legacy_scpi_shared_mem *mem = ch->rx_payload;
>> + unsigned int len;
>> +
>> + len = ch->t->rx_len;
>> +
>> + ch->t->status = le32_to_cpu(mem->status);
>> +
>> + if (len)
>> + memcpy_fromio(ch->t->rx_buf, mem->payload, len);
>> +
>> + complete(&ch->t->done);
>> +}
>> +
>> static void scpi_tx_prepare(struct mbox_client *c, void *msg)
>> {
>> unsigned long flags;
>> @@ -384,6 +402,15 @@ static void scpi_tx_prepare(struct mbox_client *c, void *msg)
>> mem->command = cpu_to_le32(t->cmd);
>> }
>>
>> +static void legacy_scpi_tx_prepare(struct mbox_client *c, void *__msg)
>> +{
>> + struct scpi_chan *ch =
>> + container_of(c, struct scpi_chan, cl);
>> +
>> + if (ch->t->tx_buf && ch->t->tx_len)
>> + memcpy_toio(ch->tx_payload, ch->t->tx_buf, ch->t->tx_len);
>
>
> I see that you are not using the list. Any particular reason for that ?
>
> IMO, that *might* help to reuse more code, but I may be wrong. Let's see
> Some commands like DVFS take more time compared to simple query type of
> commands. Queuing does help there instead of blocking the channel until
> the receipt of response.

I'll like to use the list, but, the "cmd" value is not stored in the shared tx
memory, so we cannot recover the original tranfer from reading the tx memory cmd.

This is why I added a "struct scpi_xfer *t;" in the scpi_chan structure to store
the current transfer.

>
>> +}
>> +
>> static int legacy_high_priority_cmds[] = {
>> LEGACY_SCPI_CMD_GET_CSS_PWR_STATE,
>> LEGACY_SCPI_CMD_CFG_PWR_STATE_STAT,
>> @@ -434,6 +461,48 @@ static void put_scpi_xfer(struct scpi_xfer *t, struct scpi_chan *ch)
>> mutex_unlock(&ch->xfers_lock);
>> }
>>
>> +static int legacy_scpi_send_message(u8 cmd, void *tx_buf, unsigned int tx_len,
>> + void *rx_buf, unsigned int rx_len)
>> +{
>> + int ret;
>> + u8 chan;
>> + struct scpi_xfer *msg;
>> + struct scpi_chan *scpi_chan;
>> +
>> + chan = legacy_scpi_get_chan(cmd);
>> + scpi_chan = scpi_info->channels + chan;
>> +
>> + msg = get_scpi_xfer(scpi_chan);
>> + if (!msg)
>> + return -ENOMEM;
>> +
>> + mutex_lock(&scpi_chan->xfers_lock);
>> +
>
> May be you can copy msg->cmd to msg->slot and that may help to reuse
> more code or worst-case keep them aligned.

Yes, it could be. But since the msg is not reused bu the tx_prepare and handle_response,
we can pass anything here.
And for the rockchip case, we must pass an xfer unrelated pointer here since they need
a specially crafted memory structure for the mailbox.

>
>> + msg->cmd = PACK_LEGACY_SCPI_CMD(cmd, tx_len);
>> + msg->tx_buf = tx_buf;
>> + msg->tx_len = tx_len;
>> + msg->rx_buf = rx_buf;
>> + msg->rx_len = rx_len;
>> + init_completion(&msg->done);
>> + scpi_chan->t = msg;
>> +
>> + ret = mbox_send_message(scpi_chan->chan, &msg->cmd);
>
> If slot is initialized to cmd, then you can pass msg itself above.
> Then you can evaluate how much this function deviates from
> scpi_send_message and try to re-use.

The function deviates quite a lot since the queing is not used.

>
>> + if (ret < 0)
>> + goto out;
>> +
>> + if (!wait_for_completion_timeout(&msg->done, MAX_RX_TIMEOUT))
>> + ret = -ETIMEDOUT;
>> + else
>> + /* first status word */
>> + ret = msg->status;
>> +out:
>> + mutex_unlock(&scpi_chan->xfers_lock);
>> +
>> + put_scpi_xfer(msg, scpi_chan);
>> + /* SCPI error codes > 0, translate them to Linux scale*/
>> + return ret > 0 ? scpi_to_linux_errno(ret) : ret;
>> +}
>> +
>> static int __scpi_send_message(u8 cmd, void *tx_buf, unsigned int tx_len,
>> void *rx_buf, unsigned int rx_len, bool extn)
>> {
>>
>
> [Nit]: Not sure if we need this as a separate patch. It might just
> generate warnings, anyways we can merge into one later.

I'll prefer to have functionnaly separate patches for now to clarify the changes.
I'll eventually merge them for the final apply if needed.

Thanks,
Neil