[PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Daniel Golle]

Add compatible strings for MediaTek SoCs where the hardware random number
generator is accessed via a vendor-defined Secure Monitor Call (SMC)
rather than direct MMIO register access:

  - mediatek,mt7981-rng
  - mediatek,mt7987-rng
  - mediatek,mt7988-rng

These variants require no reg, clocks, or clock-names properties since
the RNG hardware is managed by ARM Trusted Firmware-A.

Relax the $nodename pattern to also allow 'rng' in addition to the
existing 'rng@...' pattern.

Add a second example showing the minimal SMC variant binding.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
---
v3:
 * drop not: in compatible conditional
 * add reg/clocks/clock-names: false for mt7981-rng
 * add else: requiring reg/clocks/clock-names for others

v2: express compatibilities with fallback

 .../devicetree/bindings/rng/mtk-rng.yaml      | 32 ++++++++++++++++---
 1 file changed, 28 insertions(+), 4 deletions(-)

diff --git a/Documentation/devicetree/bindings/rng/mtk-rng.yaml b/Documentation/devicetree/bindings/rng/mtk-rng.yaml
index 7e8dc62e5d3a6..8fe6c209ab1e5 100644
--- a/Documentation/devicetree/bindings/rng/mtk-rng.yaml
+++ b/Documentation/devicetree/bindings/rng/mtk-rng.yaml
@@ -11,12 +11,13 @@ maintainers:
 
 properties:
   $nodename:
-    pattern: "^rng@[0-9a-f]+$"
+    pattern: "^rng(@[0-9a-f]+)?$"
 
   compatible:
     oneOf:
       - enum:
           - mediatek,mt7623-rng
+          - mediatek,mt7981-rng
       - items:
           - enum:
               - mediatek,mt7622-rng
@@ -25,6 +26,11 @@ properties:
               - mediatek,mt8365-rng
               - mediatek,mt8516-rng
           - const: mediatek,mt7623-rng
+      - items:
+          - enum:
+              - mediatek,mt7987-rng
+              - mediatek,mt7988-rng
+          - const: mediatek,mt7981-rng
 
   reg:
     maxItems: 1
@@ -38,9 +44,23 @@ properties:
 
 required:
   - compatible
-  - reg
-  - clocks
-  - clock-names
+
+allOf:
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: mediatek,mt7981-rng
+    then:
+      properties:
+        reg: false
+        clocks: false
+        clock-names: false
+    else:
+      required:
+        - reg
+        - clocks
+        - clock-names
 
 additionalProperties: false
 
@@ -53,3 +73,7 @@ examples:
             clocks = <&infracfg CLK_INFRA_TRNG>;
             clock-names = "rng";
     };
+  - |
+    rng {
+            compatible = "mediatek,mt7981-rng";
+    };
-- 
2.53.0

[PATCH v3 2/2] hwrng: mtk - add support for hw access via SMCC

[Daniel Golle]

Newer versions of ARM TrustedFirmware-A on MediaTek's ARMv8 SoCs no longer
allow accessing the TRNG from outside of the trusted firmware.
Instead, a vendor-defined custom Secure Monitor Call can be used to
acquire random bytes.

Add support for newer SoCs (MT7981, MT7987, MT7988).

As TF-A for the MT7986 may either follow the old or the new
convention, the best bet is to test if firmware blocks direct access
to the hwrng and if so, expect the SMCC interface to be usable.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
---
v3: unchanged
v2: unchanged

 drivers/char/hw_random/mtk-rng.c | 127 ++++++++++++++++++++++++++-----
 1 file changed, 106 insertions(+), 21 deletions(-)

diff --git a/drivers/char/hw_random/mtk-rng.c b/drivers/char/hw_random/mtk-rng.c
index 5808d09d12c45..8f5856b59ad66 100644
--- a/drivers/char/hw_random/mtk-rng.c
+++ b/drivers/char/hw_random/mtk-rng.c
@@ -3,6 +3,7 @@
  * Driver for Mediatek Hardware Random Number Generator
  *
  * Copyright (C) 2017 Sean Wang <sean.wang@mediatek.com>
+ * Copyright (C) 2026 Daniel Golle <daniel@makrotopia.org>
  */
 #define MTK_RNG_DEV KBUILD_MODNAME
 
@@ -17,6 +18,8 @@
 #include <linux/of.h>
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
+#include <linux/arm-smccc.h>
+#include <linux/soc/mediatek/mtk_sip_svc.h>
 
 /* Runtime PM autosuspend timeout: */
 #define RNG_AUTOSUSPEND_TIMEOUT		100
@@ -30,6 +33,11 @@
 
 #define RNG_DATA			0x08
 
+/* Driver feature flags */
+#define MTK_RNG_SMC			BIT(0)
+
+#define MTK_SIP_KERNEL_GET_RND		MTK_SIP_SMC_CMD(0x550)
+
 #define to_mtk_rng(p)	container_of(p, struct mtk_rng, rng)
 
 struct mtk_rng {
@@ -37,6 +45,7 @@ struct mtk_rng {
 	struct clk *clk;
 	struct hwrng rng;
 	struct device *dev;
+	unsigned long flags;
 };
 
 static int mtk_rng_init(struct hwrng *rng)
@@ -103,6 +112,56 @@ static int mtk_rng_read(struct hwrng *rng, void *buf, size_t max, bool wait)
 	return retval || !wait ? retval : -EIO;
 }
 
+static int mtk_rng_read_smc(struct hwrng *rng, void *buf, size_t max,
+			    bool wait)
+{
+	struct arm_smccc_res res;
+	int retval = 0;
+
+	while (max >= sizeof(u32)) {
+		arm_smccc_smc(MTK_SIP_KERNEL_GET_RND, 0, 0, 0, 0, 0, 0, 0,
+			      &res);
+		if (res.a0)
+			break;
+
+		*(u32 *)buf = res.a1;
+		retval += sizeof(u32);
+		buf += sizeof(u32);
+		max -= sizeof(u32);
+	}
+
+	return retval || !wait ? retval : -EIO;
+}
+
+static bool mtk_rng_hw_accessible(struct mtk_rng *priv)
+{
+	u32 val;
+	int err;
+
+	err = clk_prepare_enable(priv->clk);
+	if (err)
+		return false;
+
+	val = readl(priv->base + RNG_CTRL);
+	val |= RNG_EN;
+	writel(val, priv->base + RNG_CTRL);
+
+	val = readl(priv->base + RNG_CTRL);
+
+	if (val & RNG_EN) {
+		/* HW is accessible, clean up: disable RNG and clock */
+		writel(val & ~RNG_EN, priv->base + RNG_CTRL);
+		clk_disable_unprepare(priv->clk);
+		return true;
+	}
+
+	/*
+	 * If TF-A blocks direct access, the register reads back as 0.
+	 * Leave the clock enabled as TF-A needs it.
+	 */
+	return false;
+}
+
 static int mtk_rng_probe(struct platform_device *pdev)
 {
 	int ret;
@@ -114,23 +173,42 @@ static int mtk_rng_probe(struct platform_device *pdev)
 
 	priv->dev = &pdev->dev;
 	priv->rng.name = pdev->name;
-#ifndef CONFIG_PM
-	priv->rng.init = mtk_rng_init;
-	priv->rng.cleanup = mtk_rng_cleanup;
-#endif
-	priv->rng.read = mtk_rng_read;
 	priv->rng.quality = 900;
-
-	priv->clk = devm_clk_get(&pdev->dev, "rng");
-	if (IS_ERR(priv->clk)) {
-		ret = PTR_ERR(priv->clk);
-		dev_err(&pdev->dev, "no clock for device: %d\n", ret);
-		return ret;
+	priv->flags = (unsigned long)device_get_match_data(&pdev->dev);
+
+	if (!(priv->flags & MTK_RNG_SMC)) {
+		priv->clk = devm_clk_get(&pdev->dev, "rng");
+		if (IS_ERR(priv->clk)) {
+			ret = PTR_ERR(priv->clk);
+			dev_err(&pdev->dev, "no clock for device: %d\n", ret);
+			return ret;
+		}
+
+		priv->base = devm_platform_ioremap_resource(pdev, 0);
+		if (IS_ERR(priv->base))
+			return PTR_ERR(priv->base);
+
+		if (IS_ENABLED(CONFIG_HAVE_ARM_SMCCC) &&
+		    of_device_is_compatible(pdev->dev.of_node,
+					    "mediatek,mt7986-rng") &&
+		    !mtk_rng_hw_accessible(priv)) {
+			priv->flags |= MTK_RNG_SMC;
+			dev_info(&pdev->dev,
+				 "HW RNG not MMIO accessible, using SMC\n");
+		}
 	}
 
-	priv->base = devm_platform_ioremap_resource(pdev, 0);
-	if (IS_ERR(priv->base))
-		return PTR_ERR(priv->base);
+	if (priv->flags & MTK_RNG_SMC) {
+		if (!IS_ENABLED(CONFIG_HAVE_ARM_SMCCC))
+			return -ENODEV;
+		priv->rng.read = mtk_rng_read_smc;
+	} else {
+#ifndef CONFIG_PM
+		priv->rng.init = mtk_rng_init;
+		priv->rng.cleanup = mtk_rng_cleanup;
+#endif
+		priv->rng.read = mtk_rng_read;
+	}
 
 	ret = devm_hwrng_register(&pdev->dev, &priv->rng);
 	if (ret) {
@@ -139,12 +217,15 @@ static int mtk_rng_probe(struct platform_device *pdev)
 		return ret;
 	}
 
-	dev_set_drvdata(&pdev->dev, priv);
-	pm_runtime_set_autosuspend_delay(&pdev->dev, RNG_AUTOSUSPEND_TIMEOUT);
-	pm_runtime_use_autosuspend(&pdev->dev);
-	ret = devm_pm_runtime_enable(&pdev->dev);
-	if (ret)
-		return ret;
+	if (!(priv->flags & MTK_RNG_SMC)) {
+		dev_set_drvdata(&pdev->dev, priv);
+		pm_runtime_set_autosuspend_delay(&pdev->dev,
+						 RNG_AUTOSUSPEND_TIMEOUT);
+		pm_runtime_use_autosuspend(&pdev->dev);
+		ret = devm_pm_runtime_enable(&pdev->dev);
+		if (ret)
+			return ret;
+	}
 
 	dev_info(&pdev->dev, "registered RNG driver\n");
 
@@ -181,8 +262,11 @@ static const struct dev_pm_ops mtk_rng_pm_ops = {
 #endif	/* CONFIG_PM */
 
 static const struct of_device_id mtk_rng_match[] = {
-	{ .compatible = "mediatek,mt7986-rng" },
 	{ .compatible = "mediatek,mt7623-rng" },
+	{ .compatible = "mediatek,mt7981-rng", .data = (void *)MTK_RNG_SMC },
+	{ .compatible = "mediatek,mt7986-rng" },
+	{ .compatible = "mediatek,mt7987-rng", .data = (void *)MTK_RNG_SMC },
+	{ .compatible = "mediatek,mt7988-rng", .data = (void *)MTK_RNG_SMC },
 	{},
 };
 MODULE_DEVICE_TABLE(of, mtk_rng_match);
@@ -200,4 +284,5 @@ module_platform_driver(mtk_rng_driver);
 
 MODULE_DESCRIPTION("Mediatek Random Number Generator Driver");
 MODULE_AUTHOR("Sean Wang <sean.wang@mediatek.com>");
+MODULE_AUTHOR("Daniel Golle <daniel@makrotopia.org>");
 MODULE_LICENSE("GPL");
-- 
2.53.0

Re: [PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Krzysztof Kozlowski]

On Sun, Apr 19, 2026 at 01:05:01PM +0100, Daniel Golle wrote:
> +    rng {
> +            compatible = "mediatek,mt7981-rng";

I asked at v1. Reminded at v2. Nothing serious, but repeating myself is
pointless and kind of waste of time.

Best regards,
Krzysztof

Re: [PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Daniel Golle]

On Mon, Apr 20, 2026 at 04:07:33PM +0200, Krzysztof Kozlowski wrote:
> On Sun, Apr 19, 2026 at 01:05:01PM +0100, Daniel Golle wrote:
> > +    rng {
> > +            compatible = "mediatek,mt7981-rng";
> 
> I asked at v1. Reminded at v2. Nothing serious, but repeating myself is
> pointless and kind of waste of time.

Replying *once* telling what you would actually want, or replying to
me asking back would have helped enormously:
https://patchwork.kernel.org/comment/26880354/

All I can see is that you concluded "no improvements" without telling
*what it is you would like to see improved*.

You want me to drop the whole example? Drop the compatible?
I did drop (and replace) the negative list with a positive list, and
thought that was what you have asked me for
https://patchwork.kernel.org/comment/26817847/

This binding is dead simple: It's a compatible, describing the identical
hardware now hidden behind and SMC call.

Re: [PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Krzysztof Kozlowski]

On 20/04/2026 16:24, Daniel Golle wrote:
> On Mon, Apr 20, 2026 at 04:07:33PM +0200, Krzysztof Kozlowski wrote:
>> On Sun, Apr 19, 2026 at 01:05:01PM +0100, Daniel Golle wrote:
>>> +    rng {
>>> +            compatible = "mediatek,mt7981-rng";
>>
>> I asked at v1. Reminded at v2. Nothing serious, but repeating myself is
>> pointless and kind of waste of time.
> 
> Replying *once* telling what you would actually want, or replying to
> me asking back would have helped enormously:
> https://patchwork.kernel.org/comment/26880354/
> 
> All I can see is that you concluded "no improvements" without telling
> *what it is you would like to see improved*.
> 

Yes, and then you should go to v1 and read the review. There was only
single comment in this spot, so trivial to find.

AGAIN:

Use four spaces for indentation.

Best regards,
Krzysztof

Re: [PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Daniel Golle]

On Mon, Apr 20, 2026 at 04:43:00PM +0200, Krzysztof Kozlowski wrote:
> On 20/04/2026 16:24, Daniel Golle wrote:
> > On Mon, Apr 20, 2026 at 04:07:33PM +0200, Krzysztof Kozlowski wrote:
> >> On Sun, Apr 19, 2026 at 01:05:01PM +0100, Daniel Golle wrote:
> >>> +    rng {
> >>> +            compatible = "mediatek,mt7981-rng";
> >>
> >> I asked at v1. Reminded at v2. Nothing serious, but repeating myself is
> >> pointless and kind of waste of time.
> > 
> > Replying *once* telling what you would actually want, or replying to
> > me asking back would have helped enormously:
> > https://patchwork.kernel.org/comment/26880354/
> > 
> > All I can see is that you concluded "no improvements" without telling
> > *what it is you would like to see improved*.
> > 
> 
> Yes, and then you should go to v1 and read the review. There was only
> single comment in this spot, so trivial to find.
> 
> AGAIN:
> 
> Use four spaces for indentation.

Thank you, that IS helpful.

I've read the "no improvements" statement as an overall conclusion and
not even considered it to be specific to any *place* without further
context.

Is that the only remaining problem you see in the binding right now?

Re: [PATCH v3 1/2] dt-bindings: rng: mtk-rng: add SMC-based TRNG variants

[Krzysztof Kozlowski]

On 20/04/2026 16:48, Daniel Golle wrote:
> On Mon, Apr 20, 2026 at 04:43:00PM +0200, Krzysztof Kozlowski wrote:
>> On 20/04/2026 16:24, Daniel Golle wrote:
>>> On Mon, Apr 20, 2026 at 04:07:33PM +0200, Krzysztof Kozlowski wrote:
>>>> On Sun, Apr 19, 2026 at 01:05:01PM +0100, Daniel Golle wrote:
>>>>> +    rng {
>>>>> +            compatible = "mediatek,mt7981-rng";
>>>>
>>>> I asked at v1. Reminded at v2. Nothing serious, but repeating myself is
>>>> pointless and kind of waste of time.
>>>
>>> Replying *once* telling what you would actually want, or replying to
>>> me asking back would have helped enormously:
>>> https://patchwork.kernel.org/comment/26880354/
>>>
>>> All I can see is that you concluded "no improvements" without telling
>>> *what it is you would like to see improved*.
>>>
>>
>> Yes, and then you should go to v1 and read the review. There was only
>> single comment in this spot, so trivial to find.
>>
>> AGAIN:
>>
>> Use four spaces for indentation.
> 
> Thank you, that IS helpful.
> 
> I've read the "no improvements" statement as an overall conclusion and
> not even considered it to be specific to any *place* without further
> context.
> 
> Is that the only remaining problem you see in the binding right now?

Yes at least quickly looking.

Best regards,
Krzysztof