[PATCH v4 0/12] Introduce atomic support for SCMI transports

@@ -766,6 +766,10 @@ static int do_xfer(const struct scmi_protocol_handle *ph,
 		return -EINVAL;
 	}
 
+	cinfo = idr_find(&info->tx_idr, pi->proto->id);
+	if (unlikely(!cinfo))
+		return -EINVAL;
+
 	/*
 	 * Initialise protocol id now from protocol handle to avoid it being
 	 * overridden by mistake (or malice) by the protocol code mangling with

@@ -774,10 +778,6 @@ static int do_xfer(const struct scmi_protocol_handle *ph,
 	xfer->hdr.protocol_id = pi->proto->id;
 	reinit_completion(&xfer->done);
 
-	cinfo = idr_find(&info->tx_idr, xfer->hdr.protocol_id);
-	if (unlikely(!cinfo))
-		return -EINVAL;
-
 	trace_scmi_xfer_begin(xfer->transfer_id, xfer->hdr.id,
 			      xfer->hdr.protocol_id, xfer->hdr.seq,
 			      xfer->hdr.poll_completion);

@@ -339,11 +339,19 @@ void scmi_protocol_release(const struct scmi_handle *handle, u8 protocol_id);
  * @dev: Reference to device in the SCMI hierarchy corresponding to this
  *	 channel
  * @handle: Pointer to SCMI entity handle
+ * @no_completion_irq: Flag to indicate that this channel has no completion
+ *		       interrupt mechanism, so it needs SCMI core to poll,
+ *		       using .poll_done(), to determine when a command has
+ *		       completed.
+ *		       This can be dynamically set by transports at run-time
+ *		       inside their provided .chan_setup() when they determine
+ *		       no completion interrupt is available.
  * @transport_info: Transport layer related information
  */
 struct scmi_chan_info {
 	struct device *dev;
 	struct scmi_handle *handle;
+	bool no_completion_irq;
 	void *transport_info;
 };
 

@@ -402,6 +410,8 @@ struct scmi_device *scmi_child_dev_find(struct device *parent,
  *	be pending simultaneously in the system. May be overridden by the
  *	get_max_msg op.
  * @max_msg_size: Maximum size of data per message that can be handled.
+ * @force_polling: Flag to force this whole transport to use SCMI core polling
+ *		   mechanism instead of completion interrupts even if available.
  */
 struct scmi_desc {
 	int (*transport_init)(void);

@@ -410,6 +420,7 @@ struct scmi_desc {
 	int max_rx_timeout_ms;
 	int max_msg;
 	int max_msg_size;
+	bool force_polling;
 };
 
 #ifdef CONFIG_ARM_SCMI_TRANSPORT_MAILBOX

@@ -656,6 +656,15 @@ static void scmi_handle_response(struct scmi_chan_info *cinfo,
 		return;
 	}
 
+	/* Discard unexpected messages when polling is active. */
+	if (xfer->hdr.type != MSG_TYPE_DELAYED_RESP &&
+	    xfer->hdr.poll_completion) {
+		WARN_ON_ONCE(1);
+		dev_dbg(cinfo->dev,
+			"Completion IRQ received but using polling. Ignore.\n");
+		return;
+	}
+
 	/* rx.len could be shrunk in the sync do_xfer, so reset to maxsz */
 	if (xfer->hdr.type == MSG_TYPE_DELAYED_RESP)
 		xfer->rx.len = info->desc->max_msg_size;

@@ -760,16 +769,19 @@ static int do_xfer(const struct scmi_protocol_handle *ph,
 	struct device *dev = info->dev;
 	struct scmi_chan_info *cinfo;
 
-	if (xfer->hdr.poll_completion && !info->desc->ops->poll_done) {
-		dev_warn_once(dev,
-			      "Polling mode is not supported by transport.\n");
-		return -EINVAL;
-	}
-
 	cinfo = idr_find(&info->tx_idr, pi->proto->id);
 	if (unlikely(!cinfo))
 		return -EINVAL;
 
+	if (info->desc->force_polling || cinfo->no_completion_irq) {
+		if (!info->desc->ops->poll_done) {
+			dev_warn_once(dev,
+				      "Polling mode is not supported by transport.\n");
+			return -EINVAL;
+		}
+		xfer->hdr.poll_completion = true;
+	}
+
 	/*
 	 * Initialise protocol id now from protocol handle to avoid it being
 	 * overridden by mistake (or malice) by the protocol code mangling with

@@ -1499,6 +1511,11 @@ static int scmi_chan_setup(struct scmi_info *info, struct device *dev,
 	if (ret)
 		return ret;
 
+	if (tx && (cinfo->no_completion_irq || info->desc->force_polling))
+		dev_info(dev,
+			 "Enabled polling mode for TX channel - prot_id:%d\n",
+			 prot_id);
+
 idr_alloc:
 	ret = idr_alloc(idr, cinfo, prot_id, prot_id + 1, GFP_KERNEL);
 	if (ret != prot_id) {

@@ -412,6 +412,8 @@ struct scmi_device *scmi_child_dev_find(struct device *parent,
  * @max_msg_size: Maximum size of data per message that can be handled.
  * @force_polling: Flag to force this whole transport to use SCMI core polling
  *		   mechanism instead of completion interrupts even if available.
+ * @atomic_capable: Flag to indicate that this transport is assured not to sleep
+ *		    on the TX path.
  */
 struct scmi_desc {
 	int (*transport_init)(void);

@@ -421,6 +423,7 @@ struct scmi_desc {
 	int max_msg;
 	int max_msg_size;
 	bool force_polling;
+	bool atomic_capable;
 };
 
 #ifdef CONFIG_ARM_SCMI_TRANSPORT_MAILBOX

@@ -681,6 +681,10 @@ static void scmi_handle_response(struct scmi_chan_info *cinfo,
 		scmi_clear_channel(info, cinfo);
 		complete(xfer->async_done);
 	} else {
+		/*
+		 * This same xfer->done completion is used in atomic mode as a
+		 * flag for polling.
+		 */
 		complete(&xfer->done);
 	}
 

@@ -733,8 +737,6 @@ static void xfer_put(const struct scmi_protocol_handle *ph,
 	__scmi_xfer_put(&info->tx_minfo, xfer);
 }
 
-#define SCMI_MAX_POLL_TO_NS	(100 * NSEC_PER_USEC)
-
 static bool scmi_xfer_done_no_timeout(struct scmi_chan_info *cinfo,
 				      struct scmi_xfer *xfer, ktime_t stop)
 {

@@ -749,6 +751,90 @@ static bool scmi_xfer_done_no_timeout(struct scmi_chan_info *cinfo,
 	       ktime_after(ktime_get(), stop);
 }
 
+static bool xfer_complete_or_timeout(struct completion *done, ktime_t stop)
+{
+	return try_wait_for_completion(done) || ktime_after(ktime_get(), stop);
+}
+
+static int spin_for_completion_timeout(struct completion *done, int timeout_ms)
+{
+	ktime_t stop = ktime_add_ms(ktime_get(), timeout_ms);
+
+	spin_until_cond(xfer_complete_or_timeout(done, stop));
+	if (ktime_after(ktime_get(), stop))
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+/**
+ * scmi_wait_for_message_response  - An helper to group all the possible ways of
+ * waiting for a synchronous message response.
+ *
+ * @cinfo: SCMI channel info
+ * @xfer: Reference to the transfer being waited for.
+ *
+ * Chooses waiting strategy (sleep-waiting vs busy-waiting) depending on flags
+ * configuration like xfer->hdr.poll_completion and scmi_desc.atomic.capable.
+ *
+ * Return: 0 on Success, error otherwise.
+ */
+static int scmi_wait_for_message_response(struct scmi_chan_info *cinfo,
+					  struct scmi_xfer *xfer)
+{
+	struct scmi_info *info = handle_to_scmi_info(cinfo->handle);
+	struct device *dev = info->dev;
+	int ret = 0, timeout_ms = info->desc->max_rx_timeout_ms;
+
+	if (!xfer->hdr.poll_completion) {
+		if (!info->desc->atomic_capable) {
+			if (!wait_for_completion_timeout(&xfer->done,
+							 msecs_to_jiffies(timeout_ms))) {
+				dev_err(dev, "timed out in resp(caller: %pS)\n",
+					(void *)_RET_IP_);
+				ret = -ETIMEDOUT;
+			}
+		} else {
+			/* Poll on xfer->done waiting for completion by interrupt */
+			ret = spin_for_completion_timeout(&xfer->done,
+							  timeout_ms);
+			if (ret)
+				dev_err(dev,
+					"timed out in resp(caller: %pS) - atomic\n",
+					(void *)_RET_IP_);
+		}
+	} else {
+		/*
+		 * Poll on xfer using transport provided .poll_done();
+		 * assumes no completion interrupt was available.
+		 */
+		ktime_t stop = ktime_add_ms(ktime_get(), timeout_ms);
+
+		spin_until_cond(scmi_xfer_done_no_timeout(cinfo, xfer, stop));
+		if (ktime_before(ktime_get(), stop)) {
+			unsigned long flags;
+
+			/*
+			 * Do not fetch_response if an out-of-order
+			 * delayed response is being processed.
+			 */
+			spin_lock_irqsave(&xfer->lock, flags);
+			if (xfer->state == SCMI_XFER_SENT_OK) {
+				info->desc->ops->fetch_response(cinfo, xfer);
+				xfer->state = SCMI_XFER_RESP_OK;
+			}
+			spin_unlock_irqrestore(&xfer->lock, flags);
+		} else {
+			dev_err(dev,
+				"timed out in resp(caller: %pS) - polling\n",
+				(void *)_RET_IP_);
+			ret = -ETIMEDOUT;
+		}
+	}
+
+	return ret;
+}
+
 /**
  * do_xfer() - Do one transfer
  *

@@ -763,7 +849,6 @@ static int do_xfer(const struct scmi_protocol_handle *ph,
 		   struct scmi_xfer *xfer)
 {
 	int ret;
-	int timeout;
 	const struct scmi_protocol_instance *pi = ph_to_pi(ph);
 	struct scmi_info *info = handle_to_scmi_info(pi->handle);
 	struct device *dev = info->dev;

@@ -810,36 +895,7 @@ static int do_xfer(const struct scmi_protocol_handle *ph,
 		return ret;
 	}
 
-	if (xfer->hdr.poll_completion) {
-		ktime_t stop = ktime_add_ns(ktime_get(), SCMI_MAX_POLL_TO_NS);
-
-		spin_until_cond(scmi_xfer_done_no_timeout(cinfo, xfer, stop));
-		if (ktime_before(ktime_get(), stop)) {
-			unsigned long flags;
-
-			/*
-			 * Do not fetch_response if an out-of-order delayed
-			 * response is being processed.
-			 */
-			spin_lock_irqsave(&xfer->lock, flags);
-			if (xfer->state == SCMI_XFER_SENT_OK) {
-				info->desc->ops->fetch_response(cinfo, xfer);
-				xfer->state = SCMI_XFER_RESP_OK;
-			}
-			spin_unlock_irqrestore(&xfer->lock, flags);
-		} else {
-			ret = -ETIMEDOUT;
-		}
-	} else {
-		/* And we wait for the response. */
-		timeout = msecs_to_jiffies(info->desc->max_rx_timeout_ms);
-		if (!wait_for_completion_timeout(&xfer->done, timeout)) {
-			dev_err(dev, "timed out in resp(caller: %pS)\n",
-				(void *)_RET_IP_);
-			ret = -ETIMEDOUT;
-		}
-	}
-
+	ret = scmi_wait_for_message_response(cinfo, xfer);
 	if (!ret && xfer->hdr.status)
 		ret = scmi_to_linux_errno(xfer->hdr.status);
 

@@ -861,7 +917,7 @@ static void reset_rx_to_maxsz(const struct scmi_protocol_handle *ph,
 	xfer->rx.len = info->desc->max_msg_size;
 }
 
-#define SCMI_MAX_RESPONSE_TIMEOUT	(2 * MSEC_PER_SEC)
+#define SCMI_DRESP_TIMEOUT	(2 * MSEC_PER_SEC)
 
 /**
  * do_xfer_with_response() - Do one transfer and wait until the delayed

@@ -870,22 +926,57 @@ static void reset_rx_to_maxsz(const struct scmi_protocol_handle *ph,
  * @ph: Pointer to SCMI protocol handle
  * @xfer: Transfer to initiate and wait for response
  *
+ * Avois sleeping in favour of busy-waiting if the underlying transport was
+ * declared as .atomic_capable.
+ *
+ * Note that using asynchronous commands when running on top of atomic
+ * transports should be avoided since it could cause long busy-waiting here,
+ * but, once a transport is declared atomic, upper layers using the SCMI stack
+ * can freely make assumptions about the 'non-sleeping' nature of the stack
+ * (e.g. Clock framework) and it cannot be excluded that asynchronous commands
+ * could be exposed by the platform and so used.
+ *
+ * The only other option would have been to refrain from using any asynchronous
+ * command even if made available, when an atomic transport is detected, and
+ * instead forcibly use the synchronous version (thing that can be easily
+ * attained at the protocol layer), but this would also have led to longer
+ * stalls of the channel for synchronous commands and possibly timeouts.
+ * (in other words there is usually a good reason if a platform provides an
+ *  asynchronous version of a command and we should prefer to use it)
+ *
  * Return: -ETIMEDOUT in case of no delayed response, if transmit error,
  *	return corresponding error, else if all goes well, return 0.
  */
 static int do_xfer_with_response(const struct scmi_protocol_handle *ph,
 				 struct scmi_xfer *xfer)
 {
-	int ret, timeout = msecs_to_jiffies(SCMI_MAX_RESPONSE_TIMEOUT);
+	int ret, timeout = msecs_to_jiffies(SCMI_DRESP_TIMEOUT);
+	const struct scmi_protocol_instance *pi = ph_to_pi(ph);
+	struct scmi_info *info = handle_to_scmi_info(pi->handle);
 	DECLARE_COMPLETION_ONSTACK(async_response);
 
 	xfer->async_done = &async_response;
 
 	ret = do_xfer(ph, xfer);
 	if (!ret) {
-		if (!wait_for_completion_timeout(xfer->async_done, timeout))
-			ret = -ETIMEDOUT;
-		else if (xfer->hdr.status)
+		if (!info->desc->atomic_capable) {
+			if (!wait_for_completion_timeout(xfer->async_done,
+							 timeout)) {
+				dev_err(ph->dev,
+					"timed out in delayed resp(caller: %pS)\n",
+					(void *)_RET_IP_);
+				ret = -ETIMEDOUT;
+			}
+		} else {
+			ret = spin_for_completion_timeout(xfer->async_done,
+							  SCMI_DRESP_TIMEOUT);
+			if (ret)
+				dev_err(ph->dev,
+					"timed out in delayed resp(caller: %pS) - atomic\n",
+					(void *)_RET_IP_);
+		}
+
+		if (!ret && xfer->hdr.status)
 			ret = scmi_to_linux_errno(xfer->hdr.status);
 	}
 

@@ -33,6 +33,34 @@ TRACE_EVENT(scmi_xfer_begin,
 		__entry->seq, __entry->poll)
 );
 
+TRACE_EVENT(scmi_xfer_response_wait,
+	TP_PROTO(int transfer_id, u8 msg_id, u8 protocol_id, u16 seq,
+		 bool poll, bool atomic),
+	TP_ARGS(transfer_id, msg_id, protocol_id, seq, poll, atomic),
+
+	TP_STRUCT__entry(
+		__field(int, transfer_id)
+		__field(u8, msg_id)
+		__field(u8, protocol_id)
+		__field(u16, seq)
+		__field(bool, poll)
+		__field(bool, atomic)
+	),
+
+	TP_fast_assign(
+		__entry->transfer_id = transfer_id;
+		__entry->msg_id = msg_id;
+		__entry->protocol_id = protocol_id;
+		__entry->seq = seq;
+		__entry->poll = poll;
+		__entry->atomic = atomic;
+	),
+
+	TP_printk("transfer_id=%d msg_id=%u protocol_id=%u seq=%u poll=%u atomic=%u",
+		__entry->transfer_id, __entry->msg_id, __entry->protocol_id,
+		__entry->seq, __entry->poll, __entry->atomic)
+);
+
 TRACE_EVENT(scmi_xfer_end,
 	TP_PROTO(int transfer_id, u8 msg_id, u8 protocol_id, u16 seq,
 		 int status),

@@ -786,6 +786,11 @@ static int scmi_wait_for_message_response(struct scmi_chan_info *cinfo,
 	struct device *dev = info->dev;
 	int ret = 0, timeout_ms = info->desc->max_rx_timeout_ms;
 
+	trace_scmi_xfer_response_wait(xfer->transfer_id, xfer->hdr.id,
+				      xfer->hdr.protocol_id, xfer->hdr.seq,
+				      xfer->hdr.poll_completion,
+				      info->desc->atomic_capable);
+
 	if (!xfer->hdr.poll_completion) {
 		if (!info->desc->atomic_capable) {
 			if (!wait_for_completion_timeout(&xfer->done,

@@ -1416,6 +1416,21 @@ static void scmi_devm_protocol_put(struct scmi_device *sdev, u8 protocol_id)
 	WARN_ON(ret);
 }
 
+/**
+ * scmi_is_transport_atomic  - Method to check if underlying transport for an
+ * SCMI instance is configured as atomic.
+ *
+ * @handle: A reference to the SCMI platform instance.
+ *
+ * Return: True if transport is configured as atomic
+ */
+static bool scmi_is_transport_atomic(const struct scmi_handle *handle)
+{
+	struct scmi_info *info = handle_to_scmi_info(handle);
+
+	return info->desc->atomic_capable;
+}
+
 static inline
 struct scmi_handle *scmi_handle_get_from_info_unlocked(struct scmi_info *info)
 {

@@ -1948,6 +1963,7 @@ static int scmi_probe(struct platform_device *pdev)
 	handle->version = &info->version;
 	handle->devm_protocol_get = scmi_devm_protocol_get;
 	handle->devm_protocol_put = scmi_devm_protocol_put;
+	handle->is_transport_atomic = scmi_is_transport_atomic;
 
 	if (desc->ops->link_supplier) {
 		ret = desc->ops->link_supplier(dev);

@@ -612,6 +612,13 @@ struct scmi_notify_ops {
  * @devm_protocol_get: devres managed method to acquire a protocol and get specific
  *		       operations and a dedicated protocol handler
  * @devm_protocol_put: devres managed method to release a protocol
+ * @is_transport_atomic: method to check if the underlying transport for this
+ *			 instance handle is configured to support atomic
+ *			 transactions for commands.
+ *			 Some users of the SCMI stack in the upper layers could
+ *			 be interested to know if they can assume SCMI
+ *			 command transactions associated to this handle will
+ *			 never sleep and act accordingly.
  * @notify_ops: pointer to set of notifications related operations
  */
 struct scmi_handle {

@@ -622,6 +629,7 @@ struct scmi_handle {
 		(*devm_protocol_get)(struct scmi_device *sdev, u8 proto,
 				     struct scmi_protocol_handle **ph);
 	void (*devm_protocol_put)(struct scmi_device *sdev, u8 proto);
+	bool (*is_transport_atomic)(const struct scmi_handle *handle);
 
 	const struct scmi_notify_ops *notify_ops;
 };

@@ -88,21 +88,42 @@ static void scmi_clk_disable(struct clk_hw *hw)
 	scmi_proto_clk_ops->disable(clk->ph, clk->id);
 }
 
+/*
+ * We can provide enable/disable atomic callbacks only if the underlying SCMI
+ * transport for this SCMI instance is configured to handle synchronous commands
+ * in an atomic manner.
+ *
+ * When no SCMI atomic transport support is available we instead provide the
+ * prepare/unprepare API, as allowed by the clock framework where atomic calls
+ * are not available.
+ *
+ * Note that the provided clk_ops implementations are the same in both cases,
+ * using indeed the same underlying SCMI clock_protocol operations, it's just
+ * that they are assured to act in an atomic manner or not depending on the
+ * actual underlying SCMI transport configuration.
+ *
+ * Two distinct sets of clk_ops are provided since we could have multiple SCMI
+ * instances with different underlying transport quality, so they cannot be
+ * shared.
+ */
 static const struct clk_ops scmi_clk_ops = {
 	.recalc_rate = scmi_clk_recalc_rate,
 	.round_rate = scmi_clk_round_rate,
 	.set_rate = scmi_clk_set_rate,
-	/*
-	 * We can't provide enable/disable callback as we can't perform the same
-	 * in atomic context. Since the clock framework provides standard API
-	 * clk_prepare_enable that helps cases using clk_enable in non-atomic
-	 * context, it should be fine providing prepare/unprepare.
-	 */
 	.prepare = scmi_clk_enable,
 	.unprepare = scmi_clk_disable,
 };
 
-static int scmi_clk_ops_init(struct device *dev, struct scmi_clk *sclk)
+static const struct clk_ops scmi_atomic_clk_ops = {
+	.recalc_rate = scmi_clk_recalc_rate,
+	.round_rate = scmi_clk_round_rate,
+	.set_rate = scmi_clk_set_rate,
+	.enable = scmi_clk_enable,
+	.disable = scmi_clk_disable,
+};
+
+static int scmi_clk_ops_init(struct device *dev, struct scmi_clk *sclk,
+			     const struct clk_ops *scmi_ops)
 {
 	int ret;
 	unsigned long min_rate, max_rate;

@@ -110,7 +131,7 @@ static int scmi_clk_ops_init(struct device *dev, struct scmi_clk *sclk)
 	struct clk_init_data init = {
 		.flags = CLK_GET_RATE_NOCACHE,
 		.num_parents = 0,
-		.ops = &scmi_clk_ops,
+		.ops = scmi_ops,
 		.name = sclk->info->name,
 	};
 

@@ -145,6 +166,7 @@ static int scmi_clocks_probe(struct scmi_device *sdev)
 	struct device_node *np = dev->of_node;
 	const struct scmi_handle *handle = sdev->handle;
 	struct scmi_protocol_handle *ph;
+	bool is_atomic;
 
 	if (!handle)
 		return -ENODEV;

@@ -168,6 +190,8 @@ static int scmi_clocks_probe(struct scmi_device *sdev)
 	clk_data->num = count;
 	hws = clk_data->hws;
 
+	is_atomic = handle->is_transport_atomic(handle);
+
 	for (idx = 0; idx < count; idx++) {
 		struct scmi_clk *sclk;
 

@@ -184,7 +208,9 @@ static int scmi_clocks_probe(struct scmi_device *sdev)
 		sclk->id = idx;
 		sclk->ph = ph;
 
-		err = scmi_clk_ops_init(dev, sclk);
+		err = scmi_clk_ops_init(dev, sclk,
+					is_atomic ? &scmi_atomic_clk_ops :
+					&scmi_clk_ops);
 		if (err) {
 			dev_err(dev, "failed to register clock %d\n", idx);
 			devm_kfree(dev, sclk);

@@ -488,4 +488,5 @@ const struct scmi_desc scmi_virtio_desc = {
 	.max_rx_timeout_ms = 60000, /* for non-realtime virtio devices */
 	.max_msg = 0, /* overridden by virtio_get_max_msg() */
 	.max_msg_size = VIRTIO_SCMI_MAX_MSG_SIZE,
+	.atomic_capable = true,
 };

@@ -25,8 +25,6 @@
  * @shmem: Transmit/Receive shared memory area
  * @shmem_lock: Lock to protect access to Tx/Rx shared memory area
  * @func_id: smc/hvc call function id
- * @irq: Optional; employed when platforms indicates msg completion by intr.
- * @tx_complete: Optional, employed only when irq is valid.
  */
 
 struct scmi_smc {

@@ -34,15 +32,14 @@ struct scmi_smc {
 	struct scmi_shared_mem __iomem *shmem;
 	struct mutex shmem_lock;
 	u32 func_id;
-	int irq;
-	struct completion tx_complete;
 };
 
 static irqreturn_t smc_msg_done_isr(int irq, void *data)
 {
 	struct scmi_smc *scmi_info = data;
 
-	complete(&scmi_info->tx_complete);
+	scmi_rx_callback(scmi_info->cinfo,
+			 shmem_read_header(scmi_info->shmem), NULL);
 
 	return IRQ_HANDLED;
 }

@@ -111,8 +108,8 @@ static int smc_chan_setup(struct scmi_chan_info *cinfo, struct device *dev,
 			dev_err(dev, "failed to setup SCMI smc irq\n");
 			return ret;
 		}
-		init_completion(&scmi_info->tx_complete);
-		scmi_info->irq = irq;
+	} else {
+		cinfo->no_completion_irq = true;
 	}
 
 	scmi_info->func_id = func_id;

@@ -142,26 +139,21 @@ static int smc_send_message(struct scmi_chan_info *cinfo,
 	struct scmi_smc *scmi_info = cinfo->transport_info;
 	struct arm_smccc_res res;
 
+	/*
+	 * Channel lock will be released only once response has been
+	 * surely fully retrieved, so after .mark_txdone()
+	 */
 	mutex_lock(&scmi_info->shmem_lock);
-
 	shmem_tx_prepare(scmi_info->shmem, xfer);
 
-	if (scmi_info->irq)
-		reinit_completion(&scmi_info->tx_complete);
-
 	arm_smccc_1_1_invoke(scmi_info->func_id, 0, 0, 0, 0, 0, 0, 0, &res);
 
-	if (scmi_info->irq)
-		wait_for_completion(&scmi_info->tx_complete);
-
-	scmi_rx_callback(scmi_info->cinfo,
-			 shmem_read_header(scmi_info->shmem), NULL);
-
-	mutex_unlock(&scmi_info->shmem_lock);
-
 	/* Only SMCCC_RET_NOT_SUPPORTED is valid error code */
-	if (res.a0)
+	if (res.a0) {
+		mutex_unlock(&scmi_info->shmem_lock);
 		return -EOPNOTSUPP;
+	}
+
 	return 0;
 }
 

@@ -173,6 +165,13 @@ static void smc_fetch_response(struct scmi_chan_info *cinfo,
 	shmem_fetch_response(scmi_info->shmem, xfer);
 }
 
+static void smc_mark_txdone(struct scmi_chan_info *cinfo, int ret)
+{
+	struct scmi_smc *scmi_info = cinfo->transport_info;
+
+	mutex_unlock(&scmi_info->shmem_lock);
+}
+
 static bool
 smc_poll_done(struct scmi_chan_info *cinfo, struct scmi_xfer *xfer)
 {

@@ -186,6 +185,7 @@ static const struct scmi_transport_ops scmi_smc_ops = {
 	.chan_setup = smc_chan_setup,
 	.chan_free = smc_chan_free,
 	.send_message = smc_send_message,
+	.mark_txdone = smc_mark_txdone,
 	.fetch_response = smc_fetch_response,
 	.poll_done = smc_poll_done,
 };

@@ -7,6 +7,7 @@
  */
 
 #include <linux/arm-smccc.h>
+#include <linux/atomic.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/interrupt.h>

@@ -14,6 +15,7 @@
 #include <linux/of.h>
 #include <linux/of_address.h>
 #include <linux/of_irq.h>
+#include <linux/processor.h>
 #include <linux/slab.h>
 
 #include "common.h"

@@ -23,14 +25,15 @@
  *
  * @cinfo: SCMI channel info
  * @shmem: Transmit/Receive shared memory area
- * @shmem_lock: Lock to protect access to Tx/Rx shared memory area
+ * @inflight: Atomic flag to protect access to Tx/Rx shared memory area
  * @func_id: smc/hvc call function id
  */
 
 struct scmi_smc {
 	struct scmi_chan_info *cinfo;
 	struct scmi_shared_mem __iomem *shmem;
-	struct mutex shmem_lock;
+#define INFLIGHT_NONE	MSG_TOKEN_MAX
+	atomic_t inflight;
 	u32 func_id;
 };
 

@@ -114,7 +117,7 @@ static int smc_chan_setup(struct scmi_chan_info *cinfo, struct device *dev,
 
 	scmi_info->func_id = func_id;
 	scmi_info->cinfo = cinfo;
-	mutex_init(&scmi_info->shmem_lock);
+	atomic_set(&scmi_info->inflight, INFLIGHT_NONE);
 	cinfo->transport_info = scmi_info;
 
 	return 0;

@@ -133,6 +136,15 @@ static int smc_chan_free(int id, void *p, void *data)
 	return 0;
 }
 
+static bool smc_xfer_inflight(struct scmi_xfer *xfer, atomic_t *inflight)
+{
+	int ret;
+
+	ret = atomic_cmpxchg(inflight, INFLIGHT_NONE, xfer->hdr.seq);
+
+	return ret == INFLIGHT_NONE;
+}
+
 static int smc_send_message(struct scmi_chan_info *cinfo,
 			    struct scmi_xfer *xfer)
 {

@@ -140,17 +152,18 @@ static int smc_send_message(struct scmi_chan_info *cinfo,
 	struct arm_smccc_res res;
 
 	/*
-	 * Channel lock will be released only once response has been
+	 * Channel will be released only once response has been
 	 * surely fully retrieved, so after .mark_txdone()
 	 */
-	mutex_lock(&scmi_info->shmem_lock);
+	spin_until_cond(smc_xfer_inflight(xfer, &scmi_info->inflight));
+
 	shmem_tx_prepare(scmi_info->shmem, xfer);
 
 	arm_smccc_1_1_invoke(scmi_info->func_id, 0, 0, 0, 0, 0, 0, 0, &res);
 
 	/* Only SMCCC_RET_NOT_SUPPORTED is valid error code */
 	if (res.a0) {
-		mutex_unlock(&scmi_info->shmem_lock);
+		atomic_set(&scmi_info->inflight, INFLIGHT_NONE);
 		return -EOPNOTSUPP;
 	}
 

@@ -169,7 +182,7 @@ static void smc_mark_txdone(struct scmi_chan_info *cinfo, int ret)
 {
 	struct scmi_smc *scmi_info = cinfo->transport_info;
 
-	mutex_unlock(&scmi_info->shmem_lock);
+	atomic_set(&scmi_info->inflight, INFLIGHT_NONE);
 }
 
 static bool

@@ -195,4 +208,5 @@ const struct scmi_desc scmi_smc_desc = {
 	.max_rx_timeout_ms = 30,
 	.max_msg = 20,
 	.max_msg_size = 128,
+	.atomic_capable = true,
 };

@@ -412,6 +412,13 @@ struct scmi_device *scmi_child_dev_find(struct device *parent,
  * @max_msg_size: Maximum size of data per message that can be handled.
  * @force_polling: Flag to force this whole transport to use SCMI core polling
  *		   mechanism instead of completion interrupts even if available.
+ * @sync_cmds_atomic_replies: Flag to indicate that the transport assures
+ *			      synchronous-command messages are atomically
+ *			      completed on .send_message: no need to poll
+ *			      actively waiting for a response.
+ *			      Used by core internally only when polling is
+ *			      selected as a waiting for reply method: i.e.
+ *			      if a completion irq was found use that anyway.
  * @atomic_capable: Flag to indicate that this transport is assured not to sleep
  *		    on the TX path.
  */

@@ -423,6 +430,7 @@ struct scmi_desc {
 	int max_msg;
 	int max_msg_size;
 	bool force_polling;
+	bool sync_cmds_atomic_replies;
 	bool atomic_capable;
 };
 

@@ -809,14 +809,24 @@ static int scmi_wait_for_message_response(struct scmi_chan_info *cinfo,
 					(void *)_RET_IP_);
 		}
 	} else {
-		/*
-		 * Poll on xfer using transport provided .poll_done();
-		 * assumes no completion interrupt was available.
-		 */
-		ktime_t stop = ktime_add_ms(ktime_get(), timeout_ms);
+		if (!info->desc->sync_cmds_atomic_replies) {
+			/*
+			 * Poll on xfer using transport provided .poll_done();
+			 * assumes no completion interrupt was available.
+			 */
+			ktime_t stop = ktime_add_ms(ktime_get(), timeout_ms);
+
+			spin_until_cond(scmi_xfer_done_no_timeout(cinfo,
+								  xfer, stop));
+			if (ktime_after(ktime_get(), stop)) {
+				dev_err(dev,
+					"timed out in resp(caller: %pS) - polling\n",
+					(void *)_RET_IP_);
+				ret = -ETIMEDOUT;
+			}
+		}
 
-		spin_until_cond(scmi_xfer_done_no_timeout(cinfo, xfer, stop));
-		if (ktime_before(ktime_get(), stop)) {
+		if (!ret) {
 			unsigned long flags;
 
 			/*

@@ -829,11 +839,6 @@ static int scmi_wait_for_message_response(struct scmi_chan_info *cinfo,
 				xfer->state = SCMI_XFER_RESP_OK;
 			}
 			spin_unlock_irqrestore(&xfer->lock, flags);
-		} else {
-			dev_err(dev,
-				"timed out in resp(caller: %pS) - polling\n",
-				(void *)_RET_IP_);
-			ret = -ETIMEDOUT;
 		}
 	}