[PATCH 0/7] memory: renesas-rpc-if: Optimize transfer overhead and implement runtime PM autosuspend

[PATCH 0/7] memory: renesas-rpc-if: Optimize transfer overhead and implement runtime PM autosuspend

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Hi all,

This patch series optimizes the Renesas RPC-IF driver by significantly
reducing software overhead in both the direct memory-mapped (dirmap) and
manual transfer paths, maximizing transaction throughput, and introducing
runtime PM autosuspend to prevent aggressive power gating during
consecutive flash operations.

---
Performance Results:

These changes were benchmarked using flash_speed on the RZ/T2H EVK with 
xSPI0/1 interfaces operating in 1S-1S-1S mode for write operations:

xSPI0(MX25LW51245G):
                 Without        With Fixes
  Write Speed:  806 KiB/s       942 KiB/s
  Read Speed:   5333 KiB/s      5361 KiB/s

xSPI1(AT25SF128A):
  Write Speed:  467 KiB/s       501 KiB/s
  Read Speed:   17964 KiB/s     18285 KiB/s


Note:
- Patches have been rebased on top of next-20260714.
- Ive kept patches 6 and 7 separate as they belong
  to different subsystems (SPI and memory) and may
  be applied independently.

Cheers,
Prabhakar

Lad Prabhakar (7):
  memory: renesas-rpc-if: Mark XSPI_CDD1BUF0 as volatile
  memory: renesas-rpc-if: Move static bridge configurations out of
    dirmap hot paths
  memory: renesas-rpc-if: Consolidate XSPI_CDTBUF0 updates in write path
  memory: renesas-rpc-if: Consolidate XSPI_CDTBUF0 updates in read path
  memory: renesas-rpc-if: Consolidate command setup register programming
  spi: spi-rpc-if: Implement 200ms runtime PM autosuspend delay
  memory: renesas-rpc-if: Use runtime PM autosuspend after transfers

 drivers/memory/renesas-rpc-if.c | 87 ++++++++++++++-------------------
 drivers/spi/spi-rpc-if.c        |  4 ++
 2 files changed, 40 insertions(+), 51 deletions(-)

-- 
2.54.0

[PATCH 1/7] memory: renesas-rpc-if: Mark XSPI_CDD1BUF0 as volatile

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Add XSPI_CDD1BUF0 to the xSPI regmap volatile register table.

XSPI_CDD1BUF0 is a data buffer register used during manual transfers and,
like XSPI_CDD0BUF0, its contents are transient and should not be cached
by regmap. Mark it as volatile so reads always access the hardware
register.

Fixes: 687cac9559d8 ("memory: renesas-rpc-if: Add RZ/G3E xSPI support")
Cc: stable@vger.kernel.org
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index 3755956ae906..104c28532c20 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -34,6 +34,7 @@ static const struct regmap_access_table rpcif_volatile_table = {
 
 static const struct regmap_range xspi_volatile_ranges[] = {
 	regmap_reg_range(XSPI_CDD0BUF0, XSPI_CDD0BUF0),
+	regmap_reg_range(XSPI_CDD1BUF0, XSPI_CDD1BUF0),
 };
 
 static const struct regmap_access_table xspi_volatile_table = {
-- 
2.54.0

[PATCH 2/7] memory: renesas-rpc-if: Move static bridge configurations out of dirmap hot paths

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Move the initialization of the xSPI Bridge Control Register 0
(XSPI_BMCTL0) and Bridge Configuration Register (XSPI_BMCFG) from the
dirmap read and write implementations into the hardware initialization
function.

The bridge configuration values remain unchanged across sequential dirmap
transfers. Keeping these regmap_update_bits() calls inside the hot paths
(xspi_dirmap_read_impl and xspi_dirmap_write) introduces significant
software overhead and redundant register locking cycles.

Statically setting CS0ACC to 0x3 during xspi_hw_init_impl() permanently
enables the memory area for both read and write operations.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 26 ++++++++------------------
 1 file changed, 8 insertions(+), 18 deletions(-)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index 104c28532c20..50294e5d2bf5 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -312,6 +312,14 @@ static int xspi_hw_init_impl(struct rpcif_priv *xspi, bool hyperflash)
 	regmap_update_bits(xspi->regmap, XSPI_INTE, XSPI_INTE_CMDCMPE,
 			   XSPI_INTE_CMDCMPE);
 
+	regmap_update_bits(xspi->regmap, XSPI_BMCTL0,
+			   XSPI_BMCTL0_CS0ACC(0xff), XSPI_BMCTL0_CS0ACC(0x03));
+
+	regmap_update_bits(xspi->regmap, XSPI_BMCFG,
+			   XSPI_BMCFG_WRMD | XSPI_BMCFG_MWRCOMB |
+			   XSPI_BMCFG_MWRSIZE(0xff) | XSPI_BMCFG_PREEN,
+			   0 | XSPI_BMCFG_MWRCOMB | XSPI_BMCFG_MWRSIZE(0x0f) |
+			   XSPI_BMCFG_PREEN);
 	return 0;
 }
 
@@ -869,15 +877,6 @@ static size_t xspi_dirmap_read_impl(struct rpcif_priv *xspi, u64 offs,
 			   XSPI_CMCFG1_RDCMD_UPPER_BYTE(xspi->command) |
 			   XSPI_CMCFG1_RDLATE(xspi->dummy));
 
-	regmap_update_bits(xspi->regmap, XSPI_BMCTL0, XSPI_BMCTL0_CS0ACC(0xff),
-			   XSPI_BMCTL0_CS0ACC(0x01));
-
-	regmap_update_bits(xspi->regmap, XSPI_BMCFG,
-			   XSPI_BMCFG_WRMD | XSPI_BMCFG_MWRCOMB |
-			   XSPI_BMCFG_MWRSIZE(0xff) | XSPI_BMCFG_PREEN,
-			   0 | XSPI_BMCFG_MWRCOMB | XSPI_BMCFG_MWRSIZE(0x0f) |
-			   XSPI_BMCFG_PREEN);
-
 	regmap_update_bits(xspi->regmap, XSPI_LIOCFGCS0, XSPI_LIOCFG_PRTMD(0x3ff),
 			   XSPI_LIOCFG_PRTMD(xspi->proto));
 
@@ -945,15 +944,6 @@ ssize_t xspi_dirmap_write(struct device *dev, u64 offs, size_t len, const void *
 			   XSPI_CMCFG2_WRCMD_UPPER(xspi->command) |
 			   XSPI_CMCFG2_WRLATE(xspi->dummy));
 
-	regmap_update_bits(xspi->regmap, XSPI_BMCTL0,
-			   XSPI_BMCTL0_CS0ACC(0xff), XSPI_BMCTL0_CS0ACC(0x03));
-
-	regmap_update_bits(xspi->regmap, XSPI_BMCFG,
-			   XSPI_BMCFG_WRMD | XSPI_BMCFG_MWRCOMB |
-			   XSPI_BMCFG_MWRSIZE(0xff) | XSPI_BMCFG_PREEN,
-			   0 | XSPI_BMCFG_MWRCOMB | XSPI_BMCFG_MWRSIZE(0x0f) |
-			   XSPI_BMCFG_PREEN);
-
 	regmap_update_bits(xspi->regmap, XSPI_LIOCFGCS0, XSPI_LIOCFG_PRTMD(0x3ff),
 			   XSPI_LIOCFG_PRTMD(xspi->proto));
 
-- 
2.54.0

[PATCH 3/7] memory: renesas-rpc-if: Consolidate XSPI_CDTBUF0 updates in write path

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Program the XSPI_CDTBUF0 TRTYPE, DATASIZE and ADDSIZE fields with a
single regmap_update_bits() call in the RPCIF_DATA_OUT transfer path.

The existing code updates these fields through three separate
read-modify-write operations even though all values are known before the
transfer request is issued. Combine the updates into a single register
access to reduce overhead and simplify the configuration sequence.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index 50294e5d2bf5..e406afc42616 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -660,17 +660,12 @@ static int xspi_manual_xfer_impl(struct rpcif_priv *xspi)
 			u32 bytes_left = xspi->xferlen - pos;
 			u32 nbytes, data[2], *p = data;
 
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_TRTYPE, XSPI_CDTBUF_TRTYPE);
-
 			nbytes = bytes_left >= max ? max : bytes_left;
 
 			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_DATASIZE(0xf),
-					   XSPI_CDTBUF_DATASIZE(nbytes));
-
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
+					   XSPI_CDTBUF_TRTYPE | XSPI_CDTBUF_DATASIZE(0xf) |
 					   XSPI_CDTBUF_ADDSIZE(0x7),
+					   XSPI_CDTBUF_TRTYPE | XSPI_CDTBUF_DATASIZE(nbytes) |
 					   XSPI_CDTBUF_ADDSIZE(xspi->addr_nbytes));
 
 			memcpy(data, xspi->buffer + pos, nbytes);
-- 
2.54.0

[PATCH 4/7] memory: renesas-rpc-if: Consolidate XSPI_CDTBUF0 updates in read path

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Program the XSPI_CDTBUF0 fields used by the RPCIF_DATA_IN transfer path
through a single regmap_update_bits() call.

The existing code updates TRTYPE, DATASIZE, ADDSIZE and LATE using
multiple read-modify-write operations on the same register, even though
all field values are known before the transfer starts.

Build the register mask and value first, then perform a single register
update. This reduces register accesses and simplifies the transfer setup
without changing functionality.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 31 +++++++++++++++++--------------
 1 file changed, 17 insertions(+), 14 deletions(-)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index e406afc42616..c768e9c18556 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -697,30 +697,33 @@ static int xspi_manual_xfer_impl(struct rpcif_priv *xspi)
 		while (pos < xspi->xferlen) {
 			u32 bytes_left = xspi->xferlen - pos;
 			u32 nbytes, data[2], *p = data;
-
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_TRTYPE,
-					   ~(u32)XSPI_CDTBUF_TRTYPE);
+			u32 cdtbuf0_mask, cdtbuf0_val;
 
 			/* nbytes can be up to 8 bytes */
 			nbytes = bytes_left >= max ? max : bytes_left;
 
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_DATASIZE(0xf),
-					   XSPI_CDTBUF_DATASIZE(nbytes));
+			/* clear TRTYPE */
+			cdtbuf0_mask = XSPI_CDTBUF_TRTYPE;
+			cdtbuf0_val = 0;
 
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_ADDSIZE(0x7),
-					   XSPI_CDTBUF_ADDSIZE(xspi->addr_nbytes));
+			/* program DATASIZE */
+			cdtbuf0_mask |= XSPI_CDTBUF_DATASIZE(0xf);
+			cdtbuf0_val |= XSPI_CDTBUF_DATASIZE(nbytes);
+
+			/* program ADDSIZE */
+			cdtbuf0_mask |= XSPI_CDTBUF_ADDSIZE(0x7);
+			cdtbuf0_val |= XSPI_CDTBUF_ADDSIZE(xspi->addr_nbytes);
+
+			/* program LATE */
+			cdtbuf0_mask |= XSPI_CDTBUF_LATE(0x1f);
+			cdtbuf0_val |= XSPI_CDTBUF_LATE(xspi->dummy);
+
+			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0, cdtbuf0_mask, cdtbuf0_val);
 
 			if (xspi->addr_nbytes)
 				regmap_write(xspi->regmap, XSPI_CDABUF0,
 					     xspi->smadr + pos);
 
-			regmap_update_bits(xspi->regmap, XSPI_CDTBUF0,
-					   XSPI_CDTBUF_LATE(0x1f),
-					   XSPI_CDTBUF_LATE(xspi->dummy));
-
 			regmap_update_bits(xspi->regmap, XSPI_CDCTL0,
 					   XSPI_CDCTL0_TRREQ, XSPI_CDCTL0_TRREQ);
 
-- 
2.54.0

[PATCH 5/7] memory: renesas-rpc-if: Consolidate command setup register programming

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Simplify the command setup sequence in xspi_manual_xfer_impl() by
combining updates to XSPI_CDCTL0 and XSPI_CDTBUF0.

Clear the TRNUM and TRREQ fields in XSPI_CDCTL0 with a single
regmap_update_bits() call. Likewise, program the command, command size
and address size fields in XSPI_CDTBUF0 with a single regmap_write(),
eliminating the subsequent ADDSIZE update. The intermediate write of
zero to XSPI_CDABUF0 is also redundant, as the register is immediately
programmed with the transfer address.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index c768e9c18556..d98991266887 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -636,18 +636,14 @@ static int xspi_manual_xfer_impl(struct rpcif_priv *xspi)
 	u32 pos = 0, max = 8;
 	int ret = 0;
 
-	regmap_update_bits(xspi->regmap, XSPI_CDCTL0, XSPI_CDCTL0_TRNUM(0x3),
+	/* Clear transaction number and request */
+	regmap_update_bits(xspi->regmap, XSPI_CDCTL0,
+			   XSPI_CDCTL0_TRNUM(0x3) | XSPI_CDCTL0_TRREQ,
 			   XSPI_CDCTL0_TRNUM(0));
 
-	regmap_update_bits(xspi->regmap, XSPI_CDCTL0, XSPI_CDCTL0_TRREQ, 0);
-
 	regmap_write(xspi->regmap, XSPI_CDTBUF0,
-		     XSPI_CDTBUF_CMDSIZE(0x1) | XSPI_CDTBUF_CMD_FIELD(xspi->command));
-
-	regmap_write(xspi->regmap, XSPI_CDABUF0, 0);
-
-	regmap_update_bits(xspi->regmap, XSPI_CDTBUF0, XSPI_CDTBUF_ADDSIZE(0x7),
-			   XSPI_CDTBUF_ADDSIZE(xspi->addr_nbytes));
+		     XSPI_CDTBUF_CMDSIZE(0x1) | XSPI_CDTBUF_CMD_FIELD(xspi->command) |
+		     XSPI_CDTBUF_ADDSIZE(xspi->addr_nbytes));
 
 	regmap_write(xspi->regmap, XSPI_CDABUF0, xspi->smadr);
 
-- 
2.54.0

[PATCH 6/7] spi: spi-rpc-if: Implement 200ms runtime PM autosuspend delay

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Enable and configure the runtime PM autosuspend infrastructure during
the controller probe sequence to safeguard chunked flash operations.

The underlying hardware controller handles memory-mapped page programming
by dividing data transfers into an automated sequence of consecutive
64-byte chunks. To prevent the power management framework from
aggressively gating the interface between these individual chunk frames
or during immediate out-of-band status checks, an explicit 200ms delay
window is required.

Configure this temporal cushion using pm_runtime_set_autosuspend_delay()
and pm_runtime_use_autosuspend() at probe time, ensuring proper cleanup
via pm_runtime_dont_use_autosuspend() in error and driver removal
pathways. This guarantees interface continuity across the full lifecycle
of a multi-chunk write operation.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/spi/spi-rpc-if.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/spi/spi-rpc-if.c b/drivers/spi/spi-rpc-if.c
index b63c7856e758..9a740342d0f7 100644
--- a/drivers/spi/spi-rpc-if.c
+++ b/drivers/spi/spi-rpc-if.c
@@ -161,6 +161,8 @@ static int rpcif_spi_probe(struct platform_device *pdev)
 
 	ctlr->dev.of_node = parent->of_node;
 
+	pm_runtime_set_autosuspend_delay(rpc->dev, 200);
+	pm_runtime_use_autosuspend(rpc->dev);
 	pm_runtime_enable(rpc->dev);
 
 	ctlr->num_chipselect = 1;
@@ -183,6 +185,7 @@ static int rpcif_spi_probe(struct platform_device *pdev)
 	return 0;
 
 out_disable_rpm:
+	pm_runtime_dont_use_autosuspend(rpc->dev);
 	pm_runtime_disable(rpc->dev);
 	return error;
 }
@@ -193,6 +196,7 @@ static void rpcif_spi_remove(struct platform_device *pdev)
 	struct rpcif *rpc = spi_controller_get_devdata(ctlr);
 
 	spi_unregister_controller(ctlr);
+	pm_runtime_dont_use_autosuspend(rpc->dev);
 	pm_runtime_disable(rpc->dev);
 }
 
-- 
2.54.0

[PATCH 7/7] memory: renesas-rpc-if: Use runtime PM autosuspend after transfers

[Prabhakar]

From: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>

Replace pm_runtime_put() with pm_runtime_put_autosuspend() after manual
transfers and direct memory-mapped read/write operations.

Flash page programming may be performed as a sequence of closely spaced
transfer operations. Calling pm_runtime_put() after each operation allows
the runtime PM core to suspend the controller between successive
transfers, introducing unnecessary suspend/resume cycles and potentially
interrupting multi-step transfer sequences.

Use pm_runtime_put_autosuspend() instead so the controller remains
active while transfers continue, allowing it to suspend only after an
idle period.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
---
 drivers/memory/renesas-rpc-if.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/memory/renesas-rpc-if.c b/drivers/memory/renesas-rpc-if.c
index d98991266887..ec901a5a99aa 100644
--- a/drivers/memory/renesas-rpc-if.c
+++ b/drivers/memory/renesas-rpc-if.c
@@ -776,7 +776,7 @@ int rpcif_manual_xfer(struct device *dev)
 
 	ret = rpc->info->impl->manual_xfer(rpc);
 
-	pm_runtime_put(dev);
+	pm_runtime_put_autosuspend(dev);
 
 	return ret;
 }
@@ -891,7 +891,7 @@ ssize_t rpcif_dirmap_read(struct device *dev, u64 offs, size_t len, void *buf)
 
 	read = rpc->info->impl->dirmap_read(rpc, offs, len, buf);
 
-	pm_runtime_put(dev);
+	pm_runtime_put_autosuspend(dev);
 
 	return read;
 }
@@ -948,7 +948,7 @@ ssize_t xspi_dirmap_write(struct device *dev, u64 offs, size_t len, const void *
 		regmap_update_bits(xspi->regmap, XSPI_BMCTL1,
 				   XSPI_BMCTL1_MWRPUSH, XSPI_BMCTL1_MWRPUSH);
 
-	pm_runtime_put(dev);
+	pm_runtime_put_autosuspend(dev);
 
 	return writebytes;
 }
-- 
2.54.0