[PATCH v4 0/2] mtd: rawnand: renesas: Runtime PM use

[PATCH v4 0/2] mtd: rawnand: renesas: Runtime PM use

[Miquel Raynal]

There was a small mistake when first introducing this controller driver:
the power-domain property was missing in the device tree because there
was only one controller supported at this time (UART) and this is a
Synopsis IP which did not support power domains. The idea is to always
use these power domains when available, so let's add it to the bindings,
the DT and use it from the driver through the runtimpe PM API instead of
doing raw clk API calls.

Changes in v4:
* Use pm_runtime_resume_and_get() instead of pm_runtime_get_sync().

Changes in v3:
* Moved the pm_runtime calls earlier in the probe to avoid possible
  register accesses while the clocks have not yet been enabled.
 
Changes in v2:
* Dropped the DT patch merged by Geert.
* Used devm_pm_runtime_enable() instead of pm_runtime_enable() and
  dropped the pm_runtime_disable() calls.
* Used pm_runtime_resume_and_get() instead of pm_runtime_get_sync().
* Collected the tags on the bindings.

Miquel Raynal (2):
  dt-bindings: mtd: renesas: Fix the NAND controller description
  mtd: rawnand: renesas: Use runtime PM instead of the raw clock API

 .../bindings/mtd/renesas-nandc.yaml           |  5 ++
 .../mtd/nand/raw/renesas-nand-controller.c    | 51 +++++++++----------
 2 files changed, 28 insertions(+), 28 deletions(-)

-- 
2.27.0


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH v4 1/2] dt-bindings: mtd: renesas: Fix the NAND controller description

[Miquel Raynal]

Add the missing power-domain property which is needed on all the
RZ/N1 SoC IPs.

Suggested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Rob Herring <robh@kernel.org>
---
 Documentation/devicetree/bindings/mtd/renesas-nandc.yaml | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/Documentation/devicetree/bindings/mtd/renesas-nandc.yaml b/Documentation/devicetree/bindings/mtd/renesas-nandc.yaml
index 2870d36361c4..7b18bc5cc8b3 100644
--- a/Documentation/devicetree/bindings/mtd/renesas-nandc.yaml
+++ b/Documentation/devicetree/bindings/mtd/renesas-nandc.yaml
@@ -36,11 +36,15 @@ properties:
       - const: hclk
       - const: eclk
 
+  power-domains:
+    maxItems: 1
+
 required:
   - compatible
   - reg
   - clocks
   - clock-names
+  - power-domains
   - interrupts
 
 unevaluatedProperties: false
@@ -56,6 +60,7 @@ examples:
         interrupts = <GIC_SPI 58 IRQ_TYPE_LEVEL_HIGH>;
         clocks = <&sysctrl R9A06G032_HCLK_NAND>, <&sysctrl R9A06G032_CLK_NAND>;
         clock-names = "hclk", "eclk";
+        power-domains = <&sysctrl>;
         #address-cells = <1>;
         #size-cells = <0>;
     };
-- 
2.27.0


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH v4 2/2] mtd: rawnand: renesas: Use runtime PM instead of the raw clock API

[Miquel Raynal]

This NAND controller is part of a well defined power domain handled by
the runtime PM core. Let's keep the harmony with the other RZ/N1 drivers
and exclusively use the runtime PM API to enable/disable the clocks.

We still need to retrieve the external clock rate in order to derive the
NAND timings, but that is not a big deal, we can still do that in the
probe and just save this value to reuse it later.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 .../mtd/nand/raw/renesas-nand-controller.c    | 51 +++++++++----------
 1 file changed, 23 insertions(+), 28 deletions(-)

diff --git a/drivers/mtd/nand/raw/renesas-nand-controller.c b/drivers/mtd/nand/raw/renesas-nand-controller.c
index 6db063b230a9..1620e25a1147 100644
--- a/drivers/mtd/nand/raw/renesas-nand-controller.c
+++ b/drivers/mtd/nand/raw/renesas-nand-controller.c
@@ -16,6 +16,7 @@
 #include <linux/mtd/rawnand.h>
 #include <linux/of.h>
 #include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
 #include <linux/slab.h>
 
 #define COMMAND_REG 0x00
@@ -216,8 +217,7 @@ struct rnandc {
 	struct nand_controller controller;
 	struct device *dev;
 	void __iomem *regs;
-	struct clk *hclk;
-	struct clk *eclk;
+	unsigned long ext_clk_rate;
 	unsigned long assigned_cs;
 	struct list_head chips;
 	struct nand_chip *selected_chip;
@@ -891,7 +891,7 @@ static int rnandc_setup_interface(struct nand_chip *chip, int chipnr,
 {
 	struct rnand_chip *rnand = to_rnand(chip);
 	struct rnandc *rnandc = to_rnandc(chip->controller);
-	unsigned int period_ns = 1000000000 / clk_get_rate(rnandc->eclk);
+	unsigned int period_ns = 1000000000 / rnandc->ext_clk_rate;
 	const struct nand_sdr_timings *sdr;
 	unsigned int cyc, cle, ale, bef_dly, ca_to_data;
 
@@ -1319,6 +1319,7 @@ static int rnandc_chips_init(struct rnandc *rnandc)
 static int rnandc_probe(struct platform_device *pdev)
 {
 	struct rnandc *rnandc;
+	struct clk *eclk;
 	int irq, ret;
 
 	rnandc = devm_kzalloc(&pdev->dev, sizeof(*rnandc), GFP_KERNEL);
@@ -1335,29 +1336,26 @@ static int rnandc_probe(struct platform_device *pdev)
 	if (IS_ERR(rnandc->regs))
 		return PTR_ERR(rnandc->regs);
 
-	/* APB clock */
-	rnandc->hclk = devm_clk_get(&pdev->dev, "hclk");
-	if (IS_ERR(rnandc->hclk))
-		return PTR_ERR(rnandc->hclk);
-
-	/* External NAND bus clock */
-	rnandc->eclk = devm_clk_get(&pdev->dev, "eclk");
-	if (IS_ERR(rnandc->eclk))
-		return PTR_ERR(rnandc->eclk);
-
-	ret = clk_prepare_enable(rnandc->hclk);
-	if (ret)
+	devm_pm_runtime_enable(&pdev->dev);
+	ret = pm_runtime_resume_and_get(&pdev->dev);
+	if (ret < 0)
 		return ret;
 
-	ret = clk_prepare_enable(rnandc->eclk);
-	if (ret)
-		goto disable_hclk;
+	/* The external NAND bus clock rate is needed for computing timings */
+	eclk = clk_get(&pdev->dev, "eclk");
+	if (IS_ERR(eclk)) {
+		ret = PTR_ERR(eclk);
+		goto dis_runtime_pm;
+	}
+
+	rnandc->ext_clk_rate = clk_get_rate(eclk);
+	clk_put(eclk);
 
 	rnandc_dis_interrupts(rnandc);
 	irq = platform_get_irq_optional(pdev, 0);
 	if (irq == -EPROBE_DEFER) {
 		ret = irq;
-		goto disable_eclk;
+		goto dis_runtime_pm;
 	} else if (irq < 0) {
 		dev_info(&pdev->dev, "No IRQ found, fallback to polling\n");
 		rnandc->use_polling = true;
@@ -1365,12 +1363,12 @@ static int rnandc_probe(struct platform_device *pdev)
 		ret = devm_request_irq(&pdev->dev, irq, rnandc_irq_handler, 0,
 				       "renesas-nand-controller", rnandc);
 		if (ret < 0)
-			goto disable_eclk;
+			goto dis_runtime_pm;
 	}
 
 	ret = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
 	if (ret)
-		goto disable_eclk;
+		goto dis_runtime_pm;
 
 	rnandc_clear_fifo(rnandc);
 
@@ -1378,14 +1376,12 @@ static int rnandc_probe(struct platform_device *pdev)
 
 	ret = rnandc_chips_init(rnandc);
 	if (ret)
-		goto disable_eclk;
+		goto dis_runtime_pm;
 
 	return 0;
 
-disable_eclk:
-	clk_disable_unprepare(rnandc->eclk);
-disable_hclk:
-	clk_disable_unprepare(rnandc->hclk);
+dis_runtime_pm:
+	pm_runtime_put(&pdev->dev);
 
 	return ret;
 }
@@ -1396,8 +1392,7 @@ static int rnandc_remove(struct platform_device *pdev)
 
 	rnandc_chips_cleanup(rnandc);
 
-	clk_disable_unprepare(rnandc->eclk);
-	clk_disable_unprepare(rnandc->hclk);
+	pm_runtime_put(&pdev->dev);
 
 	return 0;
 }
-- 
2.27.0


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH v4 2/2] mtd: rawnand: renesas: Use runtime PM instead of the raw clock API

[Geert Uytterhoeven]

On Fri, May 13, 2022 at 12:50 PM Miquel Raynal
<miquel.raynal@bootlin.com> wrote:
> This NAND controller is part of a well defined power domain handled by
> the runtime PM core. Let's keep the harmony with the other RZ/N1 drivers
> and exclusively use the runtime PM API to enable/disable the clocks.
>
> We still need to retrieve the external clock rate in order to derive the
> NAND timings, but that is not a big deal, we can still do that in the
> probe and just save this value to reuse it later.
>
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH v4 0/2] mtd: rawnand: renesas: Runtime PM use

[Miquel Raynal]

Hi,

miquel.raynal@bootlin.com wrote on Fri, 13 May 2022 12:49:55 +0200:

> There was a small mistake when first introducing this controller driver:
> the power-domain property was missing in the device tree because there
> was only one controller supported at this time (UART) and this is a
> Synopsis IP which did not support power domains. The idea is to always
> use these power domains when available, so let's add it to the bindings,
> the DT and use it from the driver through the runtimpe PM API instead of
> doing raw clk API calls.
> 
> Changes in v4:
> * Use pm_runtime_resume_and_get() instead of pm_runtime_get_sync().
> 
> Changes in v3:
> * Moved the pm_runtime calls earlier in the probe to avoid possible
>   register accesses while the clocks have not yet been enabled.
>  
> Changes in v2:
> * Dropped the DT patch merged by Geert.
> * Used devm_pm_runtime_enable() instead of pm_runtime_enable() and
>   dropped the pm_runtime_disable() calls.
> * Used pm_runtime_resume_and_get() instead of pm_runtime_get_sync().
> * Collected the tags on the bindings.
> 
> Miquel Raynal (2):
>   dt-bindings: mtd: renesas: Fix the NAND controller description
>   mtd: rawnand: renesas: Use runtime PM instead of the raw clock API
> 
>  .../bindings/mtd/renesas-nandc.yaml           |  5 ++
>  .../mtd/nand/raw/renesas-nand-controller.c    | 51 +++++++++----------
>  2 files changed, 28 insertions(+), 28 deletions(-)
> 

Series applied on nand/next.

Thanks,
Miquèl

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/