diff mbox series

[v4,2/2] firmware: arm_scmi: Support 'reg-io-width' property for shared memory

Message ID 20240827182450.3608307-3-florian.fainelli@broadcom.com (mailing list archive)
State New, archived
Headers show
Series Support for I/O width within ARM SCMI SHMEM | expand

Commit Message

Florian Fainelli Aug. 27, 2024, 6:24 p.m. UTC
Some shared memory areas might only support a certain access width,
such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
on ARM64 by making both 8-bit and 64-bit accesses to such memory.

Update the shmem layer to support reading from and writing to such
shared memory area using the specified I/O width in the Device Tree. The
various transport layers making use of the shmem.c code are updated
accordingly to pass the I/O accessors that they store.

Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
---
 drivers/firmware/arm_scmi/common.h            | 32 +++++++-
 .../arm_scmi/scmi_transport_mailbox.c         | 13 +++-
 .../firmware/arm_scmi/scmi_transport_optee.c  | 11 ++-
 .../firmware/arm_scmi/scmi_transport_smc.c    | 11 ++-
 drivers/firmware/arm_scmi/shmem.c             | 78 +++++++++++++++++--
 5 files changed, 124 insertions(+), 21 deletions(-)

Comments

Sudeep Holla Sept. 3, 2024, 3:40 p.m. UTC | #1
On Tue, Aug 27, 2024 at 11:24:50AM -0700, Florian Fainelli wrote:
> Some shared memory areas might only support a certain access width,
> such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
> on ARM64 by making both 8-bit and 64-bit accesses to such memory.
>
> Update the shmem layer to support reading from and writing to such
> shared memory area using the specified I/O width in the Device Tree. The
> various transport layers making use of the shmem.c code are updated
> accordingly to pass the I/O accessors that they store.
>

This looks good to me now, much simpler. I will push this to -next soon,
but it won't be for v6.12. I have already sent PR for that. I want this
to be in -next for longer just to see if anyone has any comments and
doesn't break any platform(which it shouldn't anyways).

Just hoping if anyone looks at it and have feedback once it is in -next.
I will apply formally at v6.12-rc1 and report back if no one complains
until then.

--
Regards,
Sudeep
Sudeep Holla Oct. 18, 2024, 12:57 p.m. UTC | #2
On Tue, Sep 03, 2024 at 04:40:00PM +0100, Sudeep Holla wrote:
> On Tue, Aug 27, 2024 at 11:24:50AM -0700, Florian Fainelli wrote:
> > Some shared memory areas might only support a certain access width,
> > such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
> > on ARM64 by making both 8-bit and 64-bit accesses to such memory.
> >
> > Update the shmem layer to support reading from and writing to such
> > shared memory area using the specified I/O width in the Device Tree. The
> > various transport layers making use of the shmem.c code are updated
> > accordingly to pass the I/O accessors that they store.
> >
> 
> This looks good to me now, much simpler. I will push this to -next soon,
> but it won't be for v6.12. I have already sent PR for that. I want this
> to be in -next for longer just to see if anyone has any comments and
> doesn't break any platform(which it shouldn't anyways).
> 
> Just hoping if anyone looks at it and have feedback once it is in -next.
> I will apply formally at v6.12-rc1 and report back if no one complains
> until then.
> 

Hi Florian,

Just thought I will check with you if the content is -next are fine as I now
recall I did the rebase as this patch was original posted before the rework
of transport as modules were merged. Please confirm if you are happy with the
rebase as you see in -next. I also had to rebase it on recent fixes that
Justin added as there were trivial conflicts.

Another thing I wanted to check is if [1] series has any impact on this.
IIUC no, but it would be good to give a go in terms of testing just in case
that as well lands in -next.

--
Regards,
Sudeep

[1] https://lore.kernel.org/all/20241010123627.695191-1-jvetter@kalrayinc.com
Sudeep Holla Oct. 24, 2024, 11:05 a.m. UTC | #3
Gentle ping! Not sure if my earlier email got into spam or didn't land
in lore/ML. Just thought of checking again.

On Fri, Oct 18, 2024 at 01:57:09PM +0100, Sudeep Holla wrote:
> On Tue, Sep 03, 2024 at 04:40:00PM +0100, Sudeep Holla wrote:
> > On Tue, Aug 27, 2024 at 11:24:50AM -0700, Florian Fainelli wrote:
> > > Some shared memory areas might only support a certain access width,
> > > such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
> > > on ARM64 by making both 8-bit and 64-bit accesses to such memory.
> > >
> > > Update the shmem layer to support reading from and writing to such
> > > shared memory area using the specified I/O width in the Device Tree. The
> > > various transport layers making use of the shmem.c code are updated
> > > accordingly to pass the I/O accessors that they store.
> > >
> > 
> > This looks good to me now, much simpler. I will push this to -next soon,
> > but it won't be for v6.12. I have already sent PR for that. I want this
> > to be in -next for longer just to see if anyone has any comments and
> > doesn't break any platform(which it shouldn't anyways).
> > 
> > Just hoping if anyone looks at it and have feedback once it is in -next.
> > I will apply formally at v6.12-rc1 and report back if no one complains
> > until then.
> > 
> 
> Hi Florian,
> 
> Just thought I will check with you if the content is -next are fine as I now
> recall I did the rebase as this patch was original posted before the rework
> of transport as modules were merged. Please confirm if you are happy with the
> rebase as you see in -next. I also had to rebase it on recent fixes that
> Justin added as there were trivial conflicts.
> 
> Another thing I wanted to check is if [1] series has any impact on this.
> IIUC no, but it would be good to give a go in terms of testing just in case
> that as well lands in -next.
> 
> --
> Regards,
> Sudeep
> 
> [1] https://lore.kernel.org/all/20241010123627.695191-1-jvetter@kalrayinc.com
Florian Fainelli Oct. 24, 2024, 4:45 p.m. UTC | #4
On 10/24/24 04:05, Sudeep Holla wrote:
> Gentle ping! Not sure if my earlier email got into spam or didn't land
> in lore/ML. Just thought of checking again.

You did not land in spam, just being quite busy.

> 
> On Fri, Oct 18, 2024 at 01:57:09PM +0100, Sudeep Holla wrote:
>> On Tue, Sep 03, 2024 at 04:40:00PM +0100, Sudeep Holla wrote:
>>> On Tue, Aug 27, 2024 at 11:24:50AM -0700, Florian Fainelli wrote:
>>>> Some shared memory areas might only support a certain access width,
>>>> such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
>>>> on ARM64 by making both 8-bit and 64-bit accesses to such memory.
>>>>
>>>> Update the shmem layer to support reading from and writing to such
>>>> shared memory area using the specified I/O width in the Device Tree. The
>>>> various transport layers making use of the shmem.c code are updated
>>>> accordingly to pass the I/O accessors that they store.
>>>>
>>>
>>> This looks good to me now, much simpler. I will push this to -next soon,
>>> but it won't be for v6.12. I have already sent PR for that. I want this
>>> to be in -next for longer just to see if anyone has any comments and
>>> doesn't break any platform(which it shouldn't anyways).
>>>
>>> Just hoping if anyone looks at it and have feedback once it is in -next.
>>> I will apply formally at v6.12-rc1 and report back if no one complains
>>> until then.
>>>
>>
>> Hi Florian,
>>
>> Just thought I will check with you if the content is -next are fine as I now
>> recall I did the rebase as this patch was original posted before the rework
>> of transport as modules were merged. Please confirm if you are happy with the
>> rebase as you see in -next. I also had to rebase it on recent fixes that
>> Justin added as there were trivial conflicts.
>>
>> Another thing I wanted to check is if [1] series has any impact on this.
>> IIUC no, but it would be good to give a go in terms of testing just in case
>> that as well lands in -next.

linux-next as of today (2024-10-24) still works good on the affected 
platform, thanks for asking!
Sudeep Holla Oct. 25, 2024, 8:57 a.m. UTC | #5
On Thu, Oct 24, 2024 at 09:45:25AM -0700, Florian Fainelli wrote:
> On 10/24/24 04:05, Sudeep Holla wrote:
> > Gentle ping! Not sure if my earlier email got into spam or didn't land
> > in lore/ML. Just thought of checking again.
>
> You did not land in spam, just being quite busy.
>

Ah good, at times our email server acts up, so there is always some
doubt about it 
diff mbox series

Patch

diff --git a/drivers/firmware/arm_scmi/common.h b/drivers/firmware/arm_scmi/common.h
index 69928bbd01c2..a13f79b37c99 100644
--- a/drivers/firmware/arm_scmi/common.h
+++ b/drivers/firmware/arm_scmi/common.h
@@ -316,6 +316,26 @@  enum scmi_bad_msg {
 	MSG_MBOX_SPURIOUS = -5,
 };
 
+/* Used for compactness and signature validation of the function pointers being
+ * passed.
+ */
+typedef void (*shmem_copy_toio_t)(void __iomem *to, const void *from,
+				  size_t count);
+typedef void (*shmem_copy_fromio_t)(void *to, const void __iomem *from,
+				    size_t count);
+
+/**
+ * struct scmi_shmem_io_ops  - I/O operations to read from/write to
+ * Shared Memory
+ *
+ * @toio: Copy data to the shared memory area
+ * @fromio: Copy data from the shared memory area
+ */
+struct scmi_shmem_io_ops {
+	shmem_copy_fromio_t fromio;
+	shmem_copy_toio_t toio;
+};
+
 /* shmem related declarations */
 struct scmi_shared_mem;
 
@@ -336,13 +356,16 @@  struct scmi_shared_mem;
 struct scmi_shared_mem_operations {
 	void (*tx_prepare)(struct scmi_shared_mem __iomem *shmem,
 			   struct scmi_xfer *xfer,
-			   struct scmi_chan_info *cinfo);
+			   struct scmi_chan_info *cinfo,
+			   shmem_copy_toio_t toio);
 	u32 (*read_header)(struct scmi_shared_mem __iomem *shmem);
 
 	void (*fetch_response)(struct scmi_shared_mem __iomem *shmem,
-			       struct scmi_xfer *xfer);
+			       struct scmi_xfer *xfer,
+			       shmem_copy_fromio_t fromio);
 	void (*fetch_notification)(struct scmi_shared_mem __iomem *shmem,
-				   size_t max_len, struct scmi_xfer *xfer);
+				   size_t max_len, struct scmi_xfer *xfer,
+				   shmem_copy_fromio_t fromio);
 	void (*clear_channel)(struct scmi_shared_mem __iomem *shmem);
 	bool (*poll_done)(struct scmi_shared_mem __iomem *shmem,
 			  struct scmi_xfer *xfer);
@@ -350,7 +373,8 @@  struct scmi_shared_mem_operations {
 	bool (*channel_intr_enabled)(struct scmi_shared_mem __iomem *shmem);
 	void __iomem *(*setup_iomap)(struct scmi_chan_info *cinfo,
 				     struct device *dev,
-				     bool tx, struct resource *res);
+				     bool tx, struct resource *res,
+				     struct scmi_shmem_io_ops **ops);
 };
 
 const struct scmi_shared_mem_operations *scmi_shared_mem_operations_get(void);
diff --git a/drivers/firmware/arm_scmi/scmi_transport_mailbox.c b/drivers/firmware/arm_scmi/scmi_transport_mailbox.c
index dc5ca894d5eb..1a2e90e5c765 100644
--- a/drivers/firmware/arm_scmi/scmi_transport_mailbox.c
+++ b/drivers/firmware/arm_scmi/scmi_transport_mailbox.c
@@ -25,6 +25,7 @@ 
  * @chan_platform_receiver: Optional Platform Receiver mailbox unidirectional channel
  * @cinfo: SCMI channel info
  * @shmem: Transmit/Receive shared memory area
+ * @io_ops: Transport specific I/O operations
  */
 struct scmi_mailbox {
 	struct mbox_client cl;
@@ -33,6 +34,7 @@  struct scmi_mailbox {
 	struct mbox_chan *chan_platform_receiver;
 	struct scmi_chan_info *cinfo;
 	struct scmi_shared_mem __iomem *shmem;
+	struct scmi_shmem_io_ops *io_ops;
 };
 
 #define client_to_scmi_mailbox(c) container_of(c, struct scmi_mailbox, cl)
@@ -43,7 +45,8 @@  static void tx_prepare(struct mbox_client *cl, void *m)
 {
 	struct scmi_mailbox *smbox = client_to_scmi_mailbox(cl);
 
-	core->shmem->tx_prepare(smbox->shmem, m, smbox->cinfo);
+	core->shmem->tx_prepare(smbox->shmem, m, smbox->cinfo,
+				smbox->io_ops->toio);
 }
 
 static void rx_callback(struct mbox_client *cl, void *m)
@@ -197,7 +200,8 @@  static int mailbox_chan_setup(struct scmi_chan_info *cinfo, struct device *dev,
 	if (!smbox)
 		return -ENOMEM;
 
-	smbox->shmem = core->shmem->setup_iomap(cinfo, dev, tx, NULL);
+	smbox->shmem = core->shmem->setup_iomap(cinfo, dev, tx, NULL,
+						&smbox->io_ops);
 	if (IS_ERR(smbox->shmem))
 		return PTR_ERR(smbox->shmem);
 
@@ -298,7 +302,7 @@  static void mailbox_fetch_response(struct scmi_chan_info *cinfo,
 {
 	struct scmi_mailbox *smbox = cinfo->transport_info;
 
-	core->shmem->fetch_response(smbox->shmem, xfer);
+	core->shmem->fetch_response(smbox->shmem, xfer, smbox->io_ops->fromio);
 }
 
 static void mailbox_fetch_notification(struct scmi_chan_info *cinfo,
@@ -306,7 +310,8 @@  static void mailbox_fetch_notification(struct scmi_chan_info *cinfo,
 {
 	struct scmi_mailbox *smbox = cinfo->transport_info;
 
-	core->shmem->fetch_notification(smbox->shmem, max_len, xfer);
+	core->shmem->fetch_notification(smbox->shmem, max_len, xfer,
+					smbox->io_ops->fromio);
 }
 
 static void mailbox_clear_channel(struct scmi_chan_info *cinfo)
diff --git a/drivers/firmware/arm_scmi/scmi_transport_optee.c b/drivers/firmware/arm_scmi/scmi_transport_optee.c
index 08911f40d1ff..2be4124c6826 100644
--- a/drivers/firmware/arm_scmi/scmi_transport_optee.c
+++ b/drivers/firmware/arm_scmi/scmi_transport_optee.c
@@ -114,6 +114,7 @@  enum scmi_optee_pta_cmd {
  * @req.shmem: Virtual base address of the shared memory
  * @req.msg: Shared memory protocol handle for SCMI request and
  *   synchronous response
+ * @io_ops: Transport specific I/O operations
  * @tee_shm: TEE shared memory handle @req or NULL if using IOMEM shmem
  * @link: Reference in agent's channel list
  */
@@ -128,6 +129,7 @@  struct scmi_optee_channel {
 		struct scmi_shared_mem __iomem *shmem;
 		struct scmi_msg_payld *msg;
 	} req;
+	struct scmi_shmem_io_ops *io_ops;
 	struct tee_shm *tee_shm;
 	struct list_head link;
 };
@@ -350,7 +352,8 @@  static int setup_dynamic_shmem(struct device *dev, struct scmi_optee_channel *ch
 static int setup_static_shmem(struct device *dev, struct scmi_chan_info *cinfo,
 			      struct scmi_optee_channel *channel)
 {
-	channel->req.shmem = core->shmem->setup_iomap(cinfo, dev, true, NULL);
+	channel->req.shmem = core->shmem->setup_iomap(cinfo, dev, true, NULL,
+						      &channel->io_ops);
 	if (IS_ERR(channel->req.shmem))
 		return PTR_ERR(channel->req.shmem);
 
@@ -465,7 +468,8 @@  static int scmi_optee_send_message(struct scmi_chan_info *cinfo,
 		ret = invoke_process_msg_channel(channel,
 						 core->msg->command_size(xfer));
 	} else {
-		core->shmem->tx_prepare(channel->req.shmem, xfer, cinfo);
+		core->shmem->tx_prepare(channel->req.shmem, xfer, cinfo,
+					channel->io_ops->toio);
 		ret = invoke_process_smt_channel(channel);
 	}
 
@@ -484,7 +488,8 @@  static void scmi_optee_fetch_response(struct scmi_chan_info *cinfo,
 		core->msg->fetch_response(channel->req.msg,
 					  channel->rx_len, xfer);
 	else
-		core->shmem->fetch_response(channel->req.shmem, xfer);
+		core->shmem->fetch_response(channel->req.shmem, xfer,
+					    channel->io_ops->fromio);
 }
 
 static void scmi_optee_mark_txdone(struct scmi_chan_info *cinfo, int ret,
diff --git a/drivers/firmware/arm_scmi/scmi_transport_smc.c b/drivers/firmware/arm_scmi/scmi_transport_smc.c
index c6c69a17a9cc..04e715ec1570 100644
--- a/drivers/firmware/arm_scmi/scmi_transport_smc.c
+++ b/drivers/firmware/arm_scmi/scmi_transport_smc.c
@@ -45,6 +45,7 @@ 
  * @irq: An optional IRQ for completion
  * @cinfo: SCMI channel info
  * @shmem: Transmit/Receive shared memory area
+ * @io_ops: Transport specific I/O operations
  * @shmem_lock: Lock to protect access to Tx/Rx shared memory area.
  *		Used when NOT operating in atomic mode.
  * @inflight: Atomic flag to protect access to Tx/Rx shared memory area.
@@ -60,6 +61,7 @@  struct scmi_smc {
 	int irq;
 	struct scmi_chan_info *cinfo;
 	struct scmi_shared_mem __iomem *shmem;
+	struct scmi_shmem_io_ops *io_ops;
 	/* Protect access to shmem area */
 	struct mutex shmem_lock;
 #define INFLIGHT_NONE	MSG_TOKEN_MAX
@@ -144,7 +146,8 @@  static int smc_chan_setup(struct scmi_chan_info *cinfo, struct device *dev,
 	if (!scmi_info)
 		return -ENOMEM;
 
-	scmi_info->shmem = core->shmem->setup_iomap(cinfo, dev, tx, &res);
+	scmi_info->shmem = core->shmem->setup_iomap(cinfo, dev, tx, &res,
+						    &scmi_info->io_ops);
 	if (IS_ERR(scmi_info->shmem))
 		return PTR_ERR(scmi_info->shmem);
 
@@ -229,7 +232,8 @@  static int smc_send_message(struct scmi_chan_info *cinfo,
 	 */
 	smc_channel_lock_acquire(scmi_info, xfer);
 
-	core->shmem->tx_prepare(scmi_info->shmem, xfer, cinfo);
+	core->shmem->tx_prepare(scmi_info->shmem, xfer, cinfo,
+				scmi_info->io_ops->toio);
 
 	if (scmi_info->cap_id != ULONG_MAX)
 		arm_smccc_1_1_invoke(scmi_info->func_id, scmi_info->cap_id, 0,
@@ -253,7 +257,8 @@  static void smc_fetch_response(struct scmi_chan_info *cinfo,
 {
 	struct scmi_smc *scmi_info = cinfo->transport_info;
 
-	core->shmem->fetch_response(scmi_info->shmem, xfer);
+	core->shmem->fetch_response(scmi_info->shmem, xfer,
+				    scmi_info->io_ops->fromio);
 }
 
 static void smc_mark_txdone(struct scmi_chan_info *cinfo, int ret,
diff --git a/drivers/firmware/arm_scmi/shmem.c b/drivers/firmware/arm_scmi/shmem.c
index 01d8a9398fe8..e9f30ab671a8 100644
--- a/drivers/firmware/arm_scmi/shmem.c
+++ b/drivers/firmware/arm_scmi/shmem.c
@@ -34,9 +34,59 @@  struct scmi_shared_mem {
 	u8 msg_payload[];
 };
 
+static inline void shmem_memcpy_fromio32(void *to,
+					 const void __iomem *from,
+					 size_t count)
+{
+	WARN_ON(!IS_ALIGNED((unsigned long)from, 4) ||
+		!IS_ALIGNED((unsigned long)to, 4) ||
+		count % 4);
+
+	__ioread32_copy(to, from, count / 4);
+}
+
+static inline void shmem_memcpy_toio32(void __iomem *to,
+				       const void *from,
+				       size_t count)
+{
+	WARN_ON(!IS_ALIGNED((unsigned long)to, 4) ||
+		!IS_ALIGNED((unsigned long)from, 4) ||
+		count % 4);
+
+	__iowrite32_copy(to, from, count / 4);
+}
+
+static struct scmi_shmem_io_ops shmem_io_ops32 = {
+	.fromio	= shmem_memcpy_fromio32,
+	.toio	= shmem_memcpy_toio32,
+};
+
+/* Wrappers are needed for proper memcpy_{from,to}_io expansion by the
+ * pre-processor.
+ */
+static inline void shmem_memcpy_fromio(void *to,
+				       const void __iomem *from,
+				       size_t count)
+{
+	memcpy_fromio(to, from, count);
+}
+
+static inline void shmem_memcpy_toio(void __iomem *to,
+				     const void *from,
+				     size_t count)
+{
+	memcpy_toio(to, from, count);
+}
+
+static struct scmi_shmem_io_ops shmem_io_ops_default = {
+	.fromio = shmem_memcpy_fromio,
+	.toio	= shmem_memcpy_toio,
+};
+
 static void shmem_tx_prepare(struct scmi_shared_mem __iomem *shmem,
 			     struct scmi_xfer *xfer,
-			     struct scmi_chan_info *cinfo)
+			     struct scmi_chan_info *cinfo,
+			     shmem_copy_toio_t copy_toio)
 {
 	ktime_t stop;
 
@@ -73,7 +123,7 @@  static void shmem_tx_prepare(struct scmi_shared_mem __iomem *shmem,
 	iowrite32(sizeof(shmem->msg_header) + xfer->tx.len, &shmem->length);
 	iowrite32(pack_scmi_header(&xfer->hdr), &shmem->msg_header);
 	if (xfer->tx.buf)
-		memcpy_toio(shmem->msg_payload, xfer->tx.buf, xfer->tx.len);
+		copy_toio(shmem->msg_payload, xfer->tx.buf, xfer->tx.len);
 }
 
 static u32 shmem_read_header(struct scmi_shared_mem __iomem *shmem)
@@ -82,7 +132,8 @@  static u32 shmem_read_header(struct scmi_shared_mem __iomem *shmem)
 }
 
 static void shmem_fetch_response(struct scmi_shared_mem __iomem *shmem,
-				 struct scmi_xfer *xfer)
+				 struct scmi_xfer *xfer,
+				 shmem_copy_fromio_t copy_fromio)
 {
 	size_t len = ioread32(&shmem->length);
 
@@ -91,11 +142,12 @@  static void shmem_fetch_response(struct scmi_shared_mem __iomem *shmem,
 	xfer->rx.len = min_t(size_t, xfer->rx.len, len > 8 ? len - 8 : 0);
 
 	/* Take a copy to the rx buffer.. */
-	memcpy_fromio(xfer->rx.buf, shmem->msg_payload + 4, xfer->rx.len);
+	copy_fromio(xfer->rx.buf, shmem->msg_payload + 4, xfer->rx.len);
 }
 
 static void shmem_fetch_notification(struct scmi_shared_mem __iomem *shmem,
-				     size_t max_len, struct scmi_xfer *xfer)
+				     size_t max_len, struct scmi_xfer *xfer,
+				     shmem_copy_fromio_t copy_fromio)
 {
 	size_t len = ioread32(&shmem->length);
 
@@ -103,7 +155,7 @@  static void shmem_fetch_notification(struct scmi_shared_mem __iomem *shmem,
 	xfer->rx.len = min_t(size_t, max_len, len > 4 ? len - 4 : 0);
 
 	/* Take a copy to the rx buffer.. */
-	memcpy_fromio(xfer->rx.buf, shmem->msg_payload, xfer->rx.len);
+	copy_fromio(xfer->rx.buf, shmem->msg_payload, xfer->rx.len);
 }
 
 static void shmem_clear_channel(struct scmi_shared_mem __iomem *shmem)
@@ -139,7 +191,8 @@  static bool shmem_channel_intr_enabled(struct scmi_shared_mem __iomem *shmem)
 
 static void __iomem *shmem_setup_iomap(struct scmi_chan_info *cinfo,
 				       struct device *dev, bool tx,
-				       struct resource *res)
+				       struct resource *res,
+				       struct scmi_shmem_io_ops **ops)
 {
 	struct device_node *shmem __free(device_node);
 	const char *desc = tx ? "Tx" : "Rx";
@@ -148,6 +201,7 @@  static void __iomem *shmem_setup_iomap(struct scmi_chan_info *cinfo,
 	struct resource lres = {};
 	resource_size_t size;
 	void __iomem *addr;
+	u32 reg_io_width;
 
 	shmem = of_parse_phandle(cdev->of_node, "shmem", idx);
 	if (!shmem)
@@ -173,6 +227,16 @@  static void __iomem *shmem_setup_iomap(struct scmi_chan_info *cinfo,
 		return IOMEM_ERR_PTR(-EADDRNOTAVAIL);
 	}
 
+	of_property_read_u32(shmem, "reg-io-width", &reg_io_width);
+	switch (reg_io_width) {
+	case 4:
+		*ops = &shmem_io_ops32;
+		break;
+	default:
+		*ops = &shmem_io_ops_default;
+		break;
+	}
+
 	return addr;
 }