Skip to contentSkip to conversation

arm64: Introduce CIX P1 (SKY1) SoC

PATCH Superseded Peter Chen 5 participants 25 messages Jun 9, 2025 – Jul 16, 2025

Subsystems (6)

v9 of 11 Superseded Latest is v11 →

lore
Share & tools
lore

Peter Chen

author
#0 lore raw

v9 cover arm64: Introduce CIX P1 (SKY1) SoC

Cixtech P1 (internal name sky1) is high performance generic Armv9 SoC. Orion O6 is the Arm V9 Motherboard built by Radxa. You could find brief introduction for SoC and related boards at:
https://radxa.com/products/orion/o6#overview

Currently, to run upstream kernel at Orion O6 board, you need to use BIOS released by Radxa, and add "clk_ignore_unused=1" at bootargs.
https://docs.radxa.com/en/orion/o6/bios/install-bios

In this series, we add initial SoC and board support for Kernel building. Since mailbox is used for SCMI clock communication, mailbox driver is added in this series for the minimum SoC support.

Patch 1-2: add dt-binding doc for CIX and its sky1 SoC
Patch 3: add Arm64 build support
Patch 4-5: add CIX mailbox driver which needs to support SCMI clock protocol. Patch 6: add Arm64 defconfig support
Patch 7-8: add initial dts support for SoC and Orion O6 board Patch 9: add MAINTAINERS entry

Below are the review status:
Patch 1-4: received public Reviewed-by or Acked-by tags Patch 6-7: received public Reviewed-by or Acked-by tags Patch 8: received public Tested-by tags

Changes for v9:
- Rebase to v6.16-rc1
- Patch 5: address comments from Jassi Brar
- Patch 9: Add CIX mailbox information at MAINTAINER file

Changes for v8:
- Patch 6: add Krzysztof Kozlowski's Reviewed-by tag
- Patch 7: add Krzysztof Kozlowski's Acked-by tag

Changes for v7:
- Patch 8:
- Refine *_scmi_mem nodes for their properties ordering
- Delete Krzysztof Kozlowski and Fugang Duan's tag due to substantial changes
- Add two Tested-by tags from Enric Balletbo i Serra and Kajetan Puchalski
- Patch 4: Add Krzysztof Kozlowski's Reviewed-by tag
- Squash two patches as one for arm64 defconfig
- Rename clock binding file from sky1-clk.h to cix,sky1.h
- Some of my Sob are missing, add them

Changes for v6:
- Rebase to v6.15-rc2
- Add mailbox driver
- Add device tree description for uart, mailbox and scmi firmware (for clock).

Changes for v5:
- Patch 5: Delete pmu-spe node which need to refine, and add it in future
- Patch 6: Refine MAINTAINERS for all CIX SoC and adding code tree location

Changes for v4:
- Move add MAINTAINERS entry patch to the last, and add two dts files entry in it.
- Add three Krzysztof Kozlowski's Reviewed-by Tags
- For sky1.dtsi, makes below changes:
- Add ppi-partition entry for gic-v3 node, and let pmu-a520 and pmu-a720's interrupt entry get its handle
- Remove gic-v3's #redistributor-regions and redistributor-stride properties
- Change gic-v3's #interrupt-cells as 4, and change all interrupt specifiers accordingly
- Remove "arm,no-tick-in-suspend" for timer due to global counter is at always-on power domain
- Remove timer's clock frequency due to firmware has already set it

Changes for v3:
- Patch 1: Add Krzysztof Kozlowski's Acked-by Tag
- Patch 2: Add Krzysztof Kozlowski's Reviewed-by Tag
- Patch 6: Fix two dts coding sytle issues

Changes for v2:
- Pass dts build check with below commands:
make O=$OUTKNL dt_binding_check DT_SCHEMA_FILES=vendor-prefixes.yaml make O=$OUTKNL dt_binding_check DT_SCHEMA_FILES=arm/cix.yaml make O=$OUTKNL CHECK_DTBS=y W=1 cix/sky1-orion-o6.dtb
- Re-order the patch set, and move vendor-perfixes to the 1st patch.
- Patch 4: Ordered Kconfig config entry by alpha-numerically
- Patch 5: Corrects the Ack tag's name
- Patch 6: see below.
1) Corrects the SoF tag's name
2) Fix several coding sytle issues
3) move linux,cma node to dts file
4) delete memory node, memory size is passed by firmware
5) delete uart2 node which will be added in future patches
6) Improve for pmu and cpu node to stands for more specific cpu model
7) Improve the timer node and add hypervisor virtual timer irq

Fugang Duan (1):
arm64: Kconfig: add ARCH_CIX for cix silicons

Gary Yang (1):
dt-bindings: clock: cix: Add CIX sky1 scmi clock id

Guomin Chen (2):
dt-bindings: mailbox: add cix,sky1-mbox
mailbox: add CIX mailbox driver

Peter Chen (5):
dt-bindings: vendor-prefixes: Add CIX Technology Group Co., Ltd.
dt-bindings: arm: add CIX P1 (SKY1) SoC
arm64: defconfig: Enable CIX SoC
arm64: dts: cix: Add sky1 base dts initial support
MAINTAINERS: Add CIX SoC maintainer entry

 .../devicetree/bindings/arm/cix.yaml          |  26 +
 .../bindings/mailbox/cix,sky1-mbox.yaml       |  71 ++
 .../devicetree/bindings/vendor-prefixes.yaml  |   2 +
 MAINTAINERS                                   |  13 +
 arch/arm64/Kconfig.platforms                  |   6 +
 arch/arm64/boot/dts/Makefile                  |   1 +
 arch/arm64/boot/dts/cix/Makefile              |   2 +
 arch/arm64/boot/dts/cix/sky1-orion-o6.dts     |  39 ++
 arch/arm64/boot/dts/cix/sky1.dtsi             | 331 +++++++++
 arch/arm64/configs/defconfig                  |   2 +
 drivers/mailbox/Kconfig                       |  10 +
 drivers/mailbox/Makefile                      |   2 +
 drivers/mailbox/cix-mailbox.c                 | 635 ++++++++++++++++++
 include/dt-bindings/clock/cix,sky1.h          | 279 ++++++++
 14 files changed, 1419 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/arm/cix.yaml
 create mode 100644 Documentation/devicetree/bindings/mailbox/cix,sky1-mbox.yaml
 create mode 100644 arch/arm64/boot/dts/cix/Makefile
 create mode 100644 arch/arm64/boot/dts/cix/sky1-orion-o6.dts
 create mode 100644 arch/arm64/boot/dts/cix/sky1.dtsi
 create mode 100644 drivers/mailbox/cix-mailbox.c
 create mode 100644 include/dt-bindings/clock/cix,sky1.h
Signature · 1 line
2.25.1
author
#1 lore raw

PATCH v9 1/9 dt-bindings: vendor-prefixes: Add CIX Technology Group Co., Ltd.

CIX Technology Group Co., Ltd. is a high performance Arm SoC design company. Link: https://www.cixtech.com/.

---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)
Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
index 5d2a7a8d3ac6..f258c1f53b3c 100644
@@ -306,6 +306,8 @@ patternProperties:
     description: Cirrus Logic, Inc.
   "^cisco,.*":
     description: Cisco Systems, Inc.
+  "^cix,.*":
+    description: CIX Technology Group Co., Ltd.
   "^clockwork,.*":
     description: Clockwork Tech LLC
   "^cloos,.*":
Signature · 1 line
2.25.1
author
#2 lore raw

PATCH v9 2/9 dt-bindings: arm: add CIX P1 (SKY1) SoC

In reply to Peter Chen

Add device tree bindings for CIX P1 (Internal name sky1) Arm SoC, it consists several SoC models like CP8180, CD8180, etc.

---
 .../devicetree/bindings/arm/cix.yaml          | 26 +++++++++++++++++++
 1 file changed, 26 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/arm/cix.yaml
Documentation/devicetree/bindings/arm/cix.yaml new file +26 -0
new file mode 100644
index 000000000000..114dab4bc4d2
@@ -0,0 +1,26 @@
+# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/arm/cix.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: CIX platforms
+
+maintainers:
+  - Peter Chen <[email protected]>
+  - Fugang Duan <[email protected]>
+
+properties:
+  $nodename:
+    const: '/'
+  compatible:
+    oneOf:
+
+      - description: Radxa Orion O6
+        items:
+          - const: radxa,orion-o6
+          - const: cix,sky1
+
+additionalProperties: true
+
+...
Signature · 1 line
2.25.1
author
#3 lore raw

PATCH v9 3/9 arm64: Kconfig: add ARCH_CIX for cix silicons

In reply to Peter Chen

From: Fugang Duan <[email protected]>

Add ARCH_CIX for CIX SoC series support.

---
 arch/arm64/Kconfig.platforms | 6 ++++++
 1 file changed, 6 insertions(+)
arch/arm64/Kconfig.platforms +6 -0
index a541bb029aa4..5eda66aea359 100644
@@ -106,6 +106,12 @@ config ARCH_BLAIZE
 	help
 	  This enables support for the Blaize SoC family
 
+config ARCH_CIX
+	bool "Cixtech SoC family"
+	help
+	  This enables support for the Cixtech SoC family,
+	  like P1(sky1).
+
 config ARCH_EXYNOS
 	bool "Samsung Exynos SoC family"
 	select COMMON_CLK_SAMSUNG
Signature · 1 line
2.25.1
author
#4 lore raw

PATCH v9 4/9 dt-bindings: mailbox: add cix,sky1-mbox

In reply to Peter Chen

From: Guomin Chen <[email protected]>

Add a dt-binding for the Cixtech Mailbox Controller.

---
Changes for v7:
- Add Krzysztof Kozlowski Reviewed-by tag
- Add my Sob tag

Changes for v3:
- Replace the direction attribute of the mailbox with the strings "rx" and "tx"

 .../bindings/mailbox/cix,sky1-mbox.yaml       | 71 +++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mailbox/cix,sky1-mbox.yaml
Documentation/devicetree/bindings/mailbox/cix,sky1-mbox.yaml new file +71 -0
new file mode 100644
index 000000000000..216186f7cc4d
@@ -0,0 +1,71 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/mailbox/cix,sky1-mbox.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Cixtech mailbox controller
+
+maintainers:
+  - Guomin Chen <[email protected]>
+
+description:
+  The Cixtech mailbox controller, used in the Cixtech Sky1 SoC,
+  is used for message transmission between multiple processors
+  within the SoC, such as the AP, PM, audio DSP, SensorHub MCU,
+  and others
+
+  Each Cixtech mailbox controller is unidirectional, so they are
+  typically used in pairs-one for receiving and one for transmitting.
+
+  Each Cixtech mailbox supports 11 channels with different transmission modes
+    channel 0-7 - Fast channel with 32bit transmit register and IRQ support
+    channel 8   - Doorbell mode,using the mailbox as an interrupt-generating
+                   mechanism.
+    channel 9   - Fifo based channel with 32*32bit depth fifo and IRQ support
+    channel 10  - Reg based channel with 32*32bit transmit register and
+                   Doorbell+transmit acknowledgment IRQ support
+
+properties:
+  compatible:
+    const: cix,sky1-mbox
+
+  reg:
+    maxItems: 1
+
+  interrupts:
+    maxItems: 1
+
+  "#mbox-cells":
+    const: 1
+
+  cix,mbox-dir:
+    $ref: /schemas/types.yaml#/definitions/string
+    description: Direction of the mailbox relative to the AP
+    enum: [tx, rx]
+
+required:
+  - compatible
+  - reg
+  - interrupts
+  - "#mbox-cells"
+  - cix,mbox-dir
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/interrupt-controller/arm-gic.h>
+
+    soc {
+        #address-cells = <2>;
+        #size-cells = <2>;
+
+        mbox_ap2pm: mailbox@30000000 {
+            compatible = "cix,sky1-mbox";
+            reg = <0 0x30000000 0 0x10000>;
+            interrupts = <GIC_SPI 136 IRQ_TYPE_LEVEL_HIGH 0>;
+            #mbox-cells = <1>;
+            cix,mbox-dir = "tx";
+        };
+    };
Signature · 1 line
2.25.1
author
#5 lore raw

PATCH v9 5/9 mailbox: add CIX mailbox driver

In reply to Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

---
Changes for v9:
- Move macro definitions above where they are used
- Remove the brackets around the number
- Merging and sorting local variable definitions
- free the irq in the error path

Changes for v3 (As mailbox patch set):
- Update MODULE_AUTHOR.
- Remove the extra blank lines.

Changes for v2 (As mailbox patch set):
- Update the real name and email address.
- Remove the ACPI header files.
- Update the Copyright from 2024 to 2025.
- Update the Module License from "GPL" to "GPL v2"
- Add an interface for message length to limit potential out-of-bound access

 drivers/mailbox/Kconfig       |  10 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/cix-mailbox.c | 635 ++++++++++++++++++++++++++++++++++
 3 files changed, 647 insertions(+)
 create mode 100644 drivers/mailbox/cix-mailbox.c
drivers/mailbox/Kconfig +10 -0
index 68eeed660a4a..4fef4797b110 100644
@@ -340,4 +340,14 @@ config THEAD_TH1520_MBOX
 	  kernel is running, and E902 core used for power management among other
 	  things.
 
+config CIX_MBOX
+        tristate "CIX Mailbox"
+        depends on ARCH_CIX || COMPILE_TEST
+        depends on OF
+        help
+          Mailbox implementation for CIX IPC system. The controller supports
+          11 mailbox channels with different operating mode and every channel
+          is unidirectional. Say Y here if you want to use the CIX Mailbox
+          support.
+
 endif
drivers/mailbox/Makefile +2 -0
index 13a3448b3271..786a46587ba1 100644
@@ -72,3 +72,5 @@ obj-$(CONFIG_QCOM_CPUCP_MBOX)	+= qcom-cpucp-mbox.o
 obj-$(CONFIG_QCOM_IPCC)		+= qcom-ipcc.o
 
 obj-$(CONFIG_THEAD_TH1520_MBOX)	+= mailbox-th1520.o
+
+obj-$(CONFIG_CIX_MBOX)	+= cix-mailbox.o
drivers/mailbox/cix-mailbox.c new file +635 -0

635 lines not on this page · raw message

Signature · 1 line
2.25.1
author
#6 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 6 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

Hi Jassi,

A gentle ping.

Peter

Quoted 703 lines

---
Changes for v9:
- Move macro definitions above where they are used
- Remove the brackets around the number
- Merging and sorting local variable definitions
- free the irq in the error path

Changes for v3 (As mailbox patch set):
- Update MODULE_AUTHOR.
- Remove the extra blank lines.

Changes for v2 (As mailbox patch set):
- Update the real name and email address.
- Remove the ACPI header files.
- Update the Copyright from 2024 to 2025.
- Update the Module License from "GPL" to "GPL v2"
- Add an interface for message length to limit potential out-of-bound access

 drivers/mailbox/Kconfig       |  10 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/cix-mailbox.c | 635 ++++++++++++++++++++++++++++++++++
 3 files changed, 647 insertions(+)
 create mode 100644 drivers/mailbox/cix-mailbox.c
drivers/mailbox/Kconfig +10 -0
index 68eeed660a4a..4fef4797b110 100644
@@ -340,4 +340,14 @@ config THEAD_TH1520_MBOX
 	  kernel is running, and E902 core used for power management among other
 	  things.
 
+config CIX_MBOX
+        tristate "CIX Mailbox"
+        depends on ARCH_CIX || COMPILE_TEST
+        depends on OF
+        help
+          Mailbox implementation for CIX IPC system. The controller supports
+          11 mailbox channels with different operating mode and every channel
+          is unidirectional. Say Y here if you want to use the CIX Mailbox
+          support.
+
 endif
drivers/mailbox/Makefile +2 -0
index 13a3448b3271..786a46587ba1 100644
@@ -72,3 +72,5 @@ obj-$(CONFIG_QCOM_CPUCP_MBOX)	+= qcom-cpucp-mbox.o
 obj-$(CONFIG_QCOM_IPCC)		+= qcom-ipcc.o
 
 obj-$(CONFIG_THEAD_TH1520_MBOX)	+= mailbox-th1520.o
+
+obj-$(CONFIG_CIX_MBOX)	+= cix-mailbox.o
drivers/mailbox/cix-mailbox.c new file +635 -0
new file mode 100644
index 000000000000..eecb53d59dfe
@@ -0,0 +1,635 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2025 Cix Technology Group Co., Ltd.
+ */
+
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/mailbox_controller.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+
+#include "mailbox.h"
+
+/*
+ * The maximum transmission size is 32 words or 128 bytes.
+ */
+#define CIX_MBOX_MSG_LEN	32	/* Max length = 32 words */
+#define MBOX_MSG_LEN_MASK	0x7fL	/* Max length = 128 bytes */
+
+/* Register define */
+#define REG_MSG(n)	(0x0 + 0x4*(n))			/* 0x0~0x7c */
+#define REG_DB_ACK	REG_MSG(CIX_MBOX_MSG_LEN)	/* 0x80 */
+#define ERR_COMP	(REG_DB_ACK + 0x4)		/* 0x84 */
+#define ERR_COMP_CLR	(REG_DB_ACK + 0x8)		/* 0x88 */
+#define REG_F_INT(IDX)	(ERR_COMP_CLR + 0x4*(IDX+1))	/* 0x8c~0xa8 */
+#define FIFO_WR		(REG_F_INT(MBOX_FAST_IDX+1))	/* 0xac */
+#define FIFO_RD		(FIFO_WR + 0x4)			/* 0xb0 */
+#define FIFO_STAS	(FIFO_WR + 0x8)			/* 0xb4 */
+#define FIFO_WM		(FIFO_WR + 0xc)			/* 0xb8 */
+#define INT_ENABLE	(FIFO_WR + 0x10)		/* 0xbc */
+#define INT_ENABLE_SIDE_B	(FIFO_WR + 0x14)	/* 0xc0 */
+#define INT_CLEAR	(FIFO_WR + 0x18)		/* 0xc4 */
+#define INT_STATUS	(FIFO_WR + 0x1c)		/* 0xc8 */
+#define FIFO_RST	(FIFO_WR + 0x20)		/* 0xcc */
+
+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
+#define MBOX_FAST_IDX		7
+#define MBOX_DB_IDX		8
+#define MBOX_FIFO_IDX		9
+#define MBOX_REG_IDX		10
+#define CIX_MBOX_CHANS		11
+
+#define MBOX_TX		0
+#define MBOX_RX		1
+
+#define DB_INT_BIT	BIT(0)
+#define DB_ACK_INT_BIT	BIT(1)
+
+#define FIFO_WM_DEFAULT		CIX_MBOX_MSG_LEN
+#define FIFO_STAS_WMK		BIT(0)
+#define FIFO_STAS_FULL		BIT(1)
+#define FIFO_STAS_EMPTY		BIT(2)
+#define FIFO_STAS_UFLOW		BIT(3)
+#define FIFO_STAS_OFLOW		BIT(4)
+
+#define FIFO_RST_BIT		BIT(0)
+
+#define DB_INT			BIT(0)
+#define ACK_INT			BIT(1)
+#define FIFO_FULL_INT		BIT(2)
+#define FIFO_EMPTY_INT		BIT(3)
+#define FIFO_WM01_INT		BIT(4)
+#define FIFO_WM10_INT		BIT(5)
+#define FIFO_OFLOW_INT		BIT(6)
+#define FIFO_UFLOW_INT		BIT(7)
+#define FIFO_N_EMPTY_INT	BIT(8)
+#define FAST_CH_INT(IDX)	BIT((IDX)+9)
+
+#define SHMEM_OFFSET 0x80
+
+enum cix_mbox_chan_type {
+	CIX_MBOX_TYPE_DB,
+	CIX_MBOX_TYPE_REG,
+	CIX_MBOX_TYPE_FIFO,
+	CIX_MBOX_TYPE_FAST,
+};
+
+struct cix_mbox_con_priv {
+	enum cix_mbox_chan_type type;
+	struct mbox_chan	*chan;
+	int index;
+};
+
+struct cix_mbox_priv {
+	struct device *dev;
+	int irq;
+	int dir;
+	bool tx_irq_mode;	/* flag of enabling tx's irq mode */
+	void __iomem *base;	/* region for mailbox */
+	unsigned int chan_num;
+	struct cix_mbox_con_priv con_priv[CIX_MBOX_CHANS];
+	struct mbox_chan mbox_chans[CIX_MBOX_CHANS];
+	struct mbox_controller mbox;
+	bool use_shmem;
+};
+
+/*
+ * The CIX mailbox supports four types of transfers:
+ * CIX_MBOX_TYPE_DB, CIX_MBOX_TYPE_FAST, CIX_MBOX_TYPE_REG, and CIX_MBOX_TYPE_FIFO.
+ * For the REG and FIFO types of transfers, the message format is as follows:
+ */
+union cix_mbox_msg_reg_fifo {
+	u32 length;	/* unit is byte */
+	u32 buf[CIX_MBOX_MSG_LEN]; /* buf[0] must be the byte length of this array */
+};
+
+static struct cix_mbox_priv *to_cix_mbox_priv(struct mbox_controller *mbox)
+{
+	return container_of(mbox, struct cix_mbox_priv, mbox);
+}
+
+static void cix_mbox_write(struct cix_mbox_priv *priv, u32 val, u32 offset)
+{
+	if (priv->use_shmem)
+		iowrite32(val, priv->base + offset - SHMEM_OFFSET);
+	else
+		iowrite32(val, priv->base + offset);
+}
+
+static u32 cix_mbox_read(struct cix_mbox_priv *priv, u32 offset)
+{
+	if (priv->use_shmem)
+		return ioread32(priv->base + offset - SHMEM_OFFSET);
+	else
+		return ioread32(priv->base + offset);
+}
+
+static bool mbox_fifo_empty(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+	return ((cix_mbox_read(priv, FIFO_STAS) & FIFO_STAS_EMPTY) ? true : false);
+}
+
+/*
+ *The transmission unit of the CIX mailbox is word.
+ *The byte length should be converted into the word length.
+ */
+static inline u32 mbox_get_msg_size(void *msg)
+{
+	u32 len;
+
+	len = ((u32 *)msg)[0] & MBOX_MSG_LEN_MASK;
+	return DIV_ROUND_UP(len, 4);
+}
+
+static int cix_mbox_send_data_db(struct mbox_chan *chan, void *data)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+	/* trigger doorbell irq */
+	cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+	return 0;
+}
+
+static int cix_mbox_send_data_reg(struct mbox_chan *chan, void *data)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	union cix_mbox_msg_reg_fifo *msg = data;
+	u32 len, i;
+
+	if (!data)
+		return -EINVAL;
+
+	len = mbox_get_msg_size(data);
+	for (i = 0; i < len; i++)
+		cix_mbox_write(priv, msg->buf[i], REG_MSG(i));
+
+	/* trigger doorbell irq */
+	cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+	return 0;
+}
+
+static int cix_mbox_send_data_fifo(struct mbox_chan *chan, void *data)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	union cix_mbox_msg_reg_fifo *msg = data;
+	u32 len, val_32, i;
+
+	if (!data)
+		return -EINVAL;
+
+	len = mbox_get_msg_size(data);
+	cix_mbox_write(priv, len, FIFO_WM);
+	for (i = 0; i < len; i++)
+		cix_mbox_write(priv, msg->buf[i], FIFO_WR);
+
+	/* Enable fifo empty interrupt */
+	val_32 = cix_mbox_read(priv, INT_ENABLE);
+	val_32 |= FIFO_EMPTY_INT;
+	cix_mbox_write(priv, val_32, INT_ENABLE);
+
+	return 0;
+}
+
+static int cix_mbox_send_data_fast(struct mbox_chan *chan, void *data)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+	u32 *arg = (u32 *)data;
+	int index = cp->index;
+
+	if (!data)
+		return -EINVAL;
+
+	if (index < 0 || index > MBOX_FAST_IDX) {
+		dev_err(priv->dev, "Invalid Mbox index %d\n", index);
+		return -EINVAL;
+	}
+
+	cix_mbox_write(priv, arg[0], REG_F_INT(index));
+
+	return 0;
+}
+
+static int cix_mbox_send_data(struct mbox_chan *chan, void *data)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+
+	if (priv->dir != MBOX_TX) {
+		dev_err(priv->dev, "Invalid Mbox dir %d\n", priv->dir);
+		return -EINVAL;
+	}
+
+	switch (cp->type) {
+	case CIX_MBOX_TYPE_DB:
+		cix_mbox_send_data_db(chan, data);
+		break;
+	case CIX_MBOX_TYPE_REG:
+		cix_mbox_send_data_reg(chan, data);
+		break;
+	case CIX_MBOX_TYPE_FIFO:
+		cix_mbox_send_data_fifo(chan, data);
+		break;
+	case CIX_MBOX_TYPE_FAST:
+		cix_mbox_send_data_fast(chan, data);
+		break;
+	default:
+		dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void cix_mbox_isr_db(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	u32 int_status;
+
+	int_status = cix_mbox_read(priv, INT_STATUS);
+
+	if (priv->dir == MBOX_RX) {
+		/* rx interrupt is triggered */
+		if (int_status & DB_INT) {
+			cix_mbox_write(priv, DB_INT, INT_CLEAR);
+			mbox_chan_received_data(chan, NULL);
+			/* trigger ack interrupt */
+			cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+		}
+	} else {
+		/* tx ack interrupt is triggered */
+		if (int_status & ACK_INT) {
+			cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+			mbox_chan_received_data(chan, NULL);
+		}
+	}
+}
+
+static void cix_mbox_isr_reg(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	u32 int_status;
+
+	int_status = cix_mbox_read(priv, INT_STATUS);
+
+	if (priv->dir == MBOX_RX) {
+		/* rx interrupt is triggered */
+		if (int_status & DB_INT) {
+			u32 data[CIX_MBOX_MSG_LEN], len, i;
+
+			cix_mbox_write(priv, DB_INT, INT_CLEAR);
+			data[0] = cix_mbox_read(priv, REG_MSG(0));
+			len = mbox_get_msg_size(data);
+			for (i = 1; i < len; i++)
+				data[i] = cix_mbox_read(priv, REG_MSG(i));
+
+			/* trigger ack interrupt */
+			cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+			mbox_chan_received_data(chan, data);
+		}
+	} else {
+		/* tx ack interrupt is triggered */
+		if (int_status & ACK_INT) {
+			cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+			mbox_chan_txdone(chan, 0);
+		}
+	}
+}
+
+static void cix_mbox_isr_fifo(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	u32 int_status, status;
+
+	int_status = cix_mbox_read(priv, INT_STATUS);
+
+	if (priv->dir == MBOX_RX) {
+		/* FIFO waterMark interrupt is generated */
+		if (int_status & (FIFO_FULL_INT | FIFO_WM01_INT)) {
+			u32 data[CIX_MBOX_MSG_LEN] = { 0 }, i = 0;
+
+			cix_mbox_write(priv, (FIFO_FULL_INT | FIFO_WM01_INT), INT_CLEAR);
+			do {
+				data[i++] = cix_mbox_read(priv, FIFO_RD);
+			} while (!mbox_fifo_empty(chan) && i < CIX_MBOX_MSG_LEN);
+
+			mbox_chan_received_data(chan, data);
+		}
+		/* FIFO underflow is generated */
+		if (int_status & FIFO_UFLOW_INT) {
+			status = cix_mbox_read(priv, FIFO_STAS);
+			dev_err(priv->dev, "fifo underflow: int_stats %d\n", status);
+			cix_mbox_write(priv, FIFO_UFLOW_INT, INT_CLEAR);
+		}
+	} else {
+		/* FIFO empty interrupt is generated */
+		if (int_status & FIFO_EMPTY_INT) {
+			u32 val_32;
+
+			cix_mbox_write(priv, FIFO_EMPTY_INT, INT_CLEAR);
+			/* Disable empty irq*/
+			val_32 = cix_mbox_read(priv, INT_ENABLE);
+			val_32 &= ~FIFO_EMPTY_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE);
+			mbox_chan_txdone(chan, 0);
+		}
+		/* FIFO overflow is generated */
+		if (int_status & FIFO_OFLOW_INT) {
+			status = cix_mbox_read(priv, FIFO_STAS);
+			dev_err(priv->dev, "fifo overlow: int_stats %d\n", status);
+			cix_mbox_write(priv, FIFO_OFLOW_INT, INT_CLEAR);
+		}
+	}
+}
+
+static void cix_mbox_isr_fast(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+	u32 int_status, data;
+
+	/* no irq will be trigger for TX dir mbox */
+	if (priv->dir != MBOX_RX)
+		return;
+
+	int_status = cix_mbox_read(priv, INT_STATUS);
+
+	if (int_status & FAST_CH_INT(cp->index)) {
+		cix_mbox_write(priv, FAST_CH_INT(cp->index), INT_CLEAR);
+		data = cix_mbox_read(priv, REG_F_INT(cp->index));
+		mbox_chan_received_data(chan, &data);
+	}
+}
+
+static irqreturn_t cix_mbox_isr(int irq, void *arg)
+{
+	struct mbox_chan *chan = arg;
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+
+	switch (cp->type) {
+	case CIX_MBOX_TYPE_DB:
+		cix_mbox_isr_db(chan);
+		break;
+	case CIX_MBOX_TYPE_REG:
+		cix_mbox_isr_reg(chan);
+		break;
+	case CIX_MBOX_TYPE_FIFO:
+		cix_mbox_isr_fifo(chan);
+		break;
+	case CIX_MBOX_TYPE_FAST:
+		cix_mbox_isr_fast(chan);
+		break;
+	default:
+		dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+		return IRQ_NONE;
+	}
+
+	return IRQ_HANDLED;
+}
+
+static int cix_mbox_startup(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+	int index = cp->index, ret;
+	u32 val_32;
+
+	ret = request_irq(priv->irq, cix_mbox_isr, 0,
+			  dev_name(priv->dev), chan);
+	if (ret) {
+		dev_err(priv->dev, "Unable to acquire IRQ %d\n", priv->irq);
+		return ret;
+	}
+
+	switch (cp->type) {
+	case CIX_MBOX_TYPE_DB:
+		/* Overwrite txdone_method for DB channel */
+		chan->txdone_method = TXDONE_BY_ACK;
+		fallthrough;
+	case CIX_MBOX_TYPE_REG:
+		if (priv->dir == MBOX_TX) {
+			/* Enable ACK interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE);
+			val_32 |= ACK_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE);
+		} else {
+			/* Enable Doorbell interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 |= DB_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+	case CIX_MBOX_TYPE_FIFO:
+		/* reset fifo */
+		cix_mbox_write(priv, FIFO_RST_BIT, FIFO_RST);
+		/* set default watermark */
+		cix_mbox_write(priv, FIFO_WM_DEFAULT, FIFO_WM);
+		if (priv->dir == MBOX_TX) {
+			/* Enable fifo overflow interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE);
+			val_32 |= FIFO_OFLOW_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE);
+		} else {
+			/* Enable fifo full/underflow interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 |= FIFO_UFLOW_INT|FIFO_WM01_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+	case CIX_MBOX_TYPE_FAST:
+		/* Only RX channel has intterupt */
+		if (priv->dir == MBOX_RX) {
+			if (index < 0 || index > MBOX_FAST_IDX) {
+				dev_err(priv->dev, "Invalid index %d\n", index);
+				ret = -EINVAL;
+				goto failed;
+			}
+			/* enable fast channel interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 |= FAST_CH_INT(index);
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+	default:
+		dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+		ret = -EINVAL;
+		goto failed;
+	}
+	return 0;
+
+failed:
+	free_irq(priv->irq, chan);
+	return ret;
+}
+
+static void cix_mbox_shutdown(struct mbox_chan *chan)
+{
+	struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+	struct cix_mbox_con_priv *cp = chan->con_priv;
+	int index = cp->index;
+	u32 val_32;
+
+	switch (cp->type) {
+	case CIX_MBOX_TYPE_DB:
+	case CIX_MBOX_TYPE_REG:
+		if (priv->dir == MBOX_TX) {
+			/* Disable ACK interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE);
+			val_32 &= ~ACK_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE);
+		} else if (priv->dir == MBOX_RX) {
+			/* Disable Doorbell interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 &= ~DB_INT;
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+	case CIX_MBOX_TYPE_FIFO:
+		if (priv->dir == MBOX_TX) {
+			/* Disable empty/fifo overflow irq*/
+			val_32 = cix_mbox_read(priv, INT_ENABLE);
+			val_32 &= ~(FIFO_EMPTY_INT | FIFO_OFLOW_INT);
+			cix_mbox_write(priv, val_32, INT_ENABLE);
+		} else if (priv->dir == MBOX_RX) {
+			/* Disable fifo WM01/underflow interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 &= ~(FIFO_UFLOW_INT | FIFO_WM01_INT);
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+	case CIX_MBOX_TYPE_FAST:
+		if (priv->dir == MBOX_RX) {
+			if (index < 0 || index > MBOX_FAST_IDX) {
+				dev_err(priv->dev, "Invalid index %d\n", index);
+				break;
+			}
+			/* Disable fast channel interrupt */
+			val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+			val_32 &= ~FAST_CH_INT(index);
+			cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+		}
+		break;
+
+	default:
+		dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+		break;
+	}
+
+	free_irq(priv->irq, chan);
+}
+
+static const struct mbox_chan_ops cix_mbox_chan_ops = {
+	.send_data = cix_mbox_send_data,
+	.startup = cix_mbox_startup,
+	.shutdown = cix_mbox_shutdown,
+};
+
+static void cix_mbox_init(struct cix_mbox_priv *priv)
+{
+	struct cix_mbox_con_priv *cp;
+	int i;
+
+	for (i = 0; i < CIX_MBOX_CHANS; i++) {
+		cp = &priv->con_priv[i];
+		cp->index = i;
+		cp->chan = &priv->mbox_chans[i];
+		priv->mbox_chans[i].con_priv = cp;
+		if (cp->index <= MBOX_FAST_IDX)
+			cp->type = CIX_MBOX_TYPE_FAST;
+		if (cp->index == MBOX_DB_IDX) {
+			cp->type = CIX_MBOX_TYPE_DB;
+			priv->use_shmem = true;
+		}
+		if (cp->index == MBOX_FIFO_IDX)
+			cp->type = CIX_MBOX_TYPE_FIFO;
+		if (cp->index == MBOX_REG_IDX)
+			cp->type = CIX_MBOX_TYPE_REG;
+	}
+}
+
+static int cix_mbox_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct cix_mbox_priv *priv;
+	const char *dir_str;
+	int ret;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->dev = dev;
+	priv->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);
+
+	priv->irq = platform_get_irq(pdev, 0);
+	if (priv->irq < 0)
+		return priv->irq;
+
+	if (device_property_read_string(dev, "cix,mbox-dir", &dir_str)) {
+		dev_err(priv->dev, "cix,mbox_dir property not found\n");
+		return -EINVAL;
+	}
+
+	if (!strcmp(dir_str, "tx"))
+		priv->dir = 0;
+	else if (!strcmp(dir_str, "rx"))
+		priv->dir = 1;
+	else {
+		dev_err(priv->dev, "cix,mbox_dir=%s is not expected\n", dir_str);
+		return -EINVAL;
+	}
+
+	cix_mbox_init(priv);
+
+	priv->mbox.dev = dev;
+	priv->mbox.ops = &cix_mbox_chan_ops;
+	priv->mbox.chans = priv->mbox_chans;
+	priv->mbox.txdone_irq = true;
+	priv->mbox.num_chans = CIX_MBOX_CHANS;
+	priv->mbox.of_xlate = NULL;
+
+	platform_set_drvdata(pdev, priv);
+	ret = devm_mbox_controller_register(dev, &priv->mbox);
+	if (ret)
+		dev_err(dev, "Failed to register mailbox %d\n", ret);
+
+	return ret;
+}
+
+static const struct of_device_id cix_mbox_dt_ids[] = {
+	{ .compatible = "cix,sky1-mbox" },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, cix_mbox_dt_ids);
+
+static struct platform_driver cix_mbox_driver = {
+	.probe = cix_mbox_probe,
+	.driver = {
+		.name = "cix_mbox",
+		.of_match_table = cix_mbox_dt_ids,
+	},
+};
+
+static int __init cix_mailbox_init(void)
+{
+	return platform_driver_register(&cix_mbox_driver);
+}
+arch_initcall(cix_mailbox_init);
+
+MODULE_AUTHOR("Cix Technology Group Co., Ltd.");
+MODULE_DESCRIPTION("CIX mailbox driver");
+MODULE_LICENSE("GPL");
Signature · 1 line
2.25.1
Signature · 2 lines
Best regards,
Peter
maintainer
#7 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

In reply to Peter Chen
Quoted 7 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

Quoted 5 lines

This is the only driver holding up the merge of the CIX platform, so I had a closer look myself. The driver looks well written overall, and I see a lot of details that have come up in previous versions are addressed.

The one thing that stuck out to me is the design of having multiple types of mailbox in one driver, which feels out of scope for a simple mailbox.

drivers/mailbox/cix-mailbox.c line 175

Quoted 12 lines
drivers/mailbox/cix-mailbox.c new file +12 -0
+static int cix_mbox_send_data_reg(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       union cix_mbox_msg_reg_fifo *msg = data;
+       u32 len, i;
+
+       if (!data)
+               return -EINVAL;
+
+       len = mbox_get_msg_size(data);
+       for (i = 0; i < len; i++)
+               cix_mbox_write(priv, msg->buf[i], REG_MSG(i));

In particular, this bit seems to do more than just what I think of as a simple mailbox that should have fixed-length messages, it feels more like a generic message passing interface between device drivers and firmware or some microcontroller.

What is the purpose here?

Quoted 2 lines
(quoted diff) +2 -0
+static int cix_mbox_send_data(struct mbox_chan *chan, void *data)
+{

...

drivers/mailbox/cix-mailbox.c line 247

Quoted 13 lines
drivers/mailbox/cix-mailbox.c new file +13 -0
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               cix_mbox_send_data_db(chan, data);
+               break;
+       case CIX_MBOX_TYPE_REG:
+               cix_mbox_send_data_reg(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               cix_mbox_send_data_fifo(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               cix_mbox_send_data_fast(chan, data);
+               break;

Similarly, this also exceeds the complexity I would expect in a simple controller, it feels like there should either be four separate drivers that implement one type of interface, or a much higher-level abstraction.

Is there a document that details how the messages are structured and what the users are? How does a user of the mailbox know the size of a message to pass down?

Arnd

maintainer
#8 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 9 lines · Arnd Bergmann
Quoted 7 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

Quoted 9 lines
Quoted 5 lines

This is the only driver holding up the merge of the CIX platform, so I had a closer look myself.

Sorry I wasn't made aware of this. Also I normally let the drivers roast until second half
hoping other platform folks find issues - I have reduced imposing my opinions on platform
specific code because it is usually met with some sledge-hammer requirement.

Quoted 5 lines

The one thing that stuck out to me is the design of having multiple types of mailbox in one driver, which feels out of scope for a simple mailbox.

Yes, if not all modes are used currently, maybe drop unused ones.

I will review the rest separately.

Thanks.

#9 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 30 lines · Jassi Brar

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL

Quoted 9 lines · Arnd Bergmann
Quoted 7 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

Quoted 9 lines
Quoted 5 lines

This is the only driver holding up the merge of the CIX platform, so I had a closer look myself.

Sorry I wasn't made aware of this. Also I normally let the drivers roast until second half
hoping other platform folks find issues - I have reduced imposing my opinions on platform
specific code because it is usually met with some sledge-hammer requirement.

Quoted 1 line
Quoted 6 lines
Quoted 4 lines

The one thing that stuck out to me is the design of having multiple types of mailbox in one driver, which feels out of scope for a simple mailbox.

Yes, if not all modes are used currently, maybe drop unused ones.

Hi Jassi Brar & Arnd Bergmann
Many thanks to both of you for your responses.

In the CIX platform, each mailbox controller is equipped with 11 channels, categorized into four communication types: Doorbell, REG, FIFO, and FAST.

The Doorbell/REG transfer modes are used for the SCMI framework,
while the FIFO transfer mode is used for the remoteproc/rpmsg framework. Currently, there are no clients using the FAST mode, but there may be in the future.

In the CIX Sky1 SoC, there are 4 pairs of mailbox controllers (since each mailbox controller is unidirectional, 2 mailbox controllers are required to establish a single bidirectional communication link, one for receiving and one for transmitting):
AP <--> PM (using Doorbell transfer mode),
AP <--> SE (using REG transfer mode),
AP <--> DSP (using FIFO transfer mode),
AP <--> SensorHub (using FIFO transfer mode).

Since this patchset is only a minimal system that includes only the necessary subsystems for booting, it contains only the SCMI clock client using Doorbell mode. Other clients using different transfer modes are not included in this patchset, but they will be added later.

Supporting multiple types of transfer modes is a feature of this driver that leverages the hardware capabilities. We are not the only ones with a relatively complex mailbox design; for example, the NXP mailbox driver also supports multiple types of transfer modes.

Thanks
Guomin Chen

Quoted 2 lines

Thanks.

maintainer
#10 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

In reply to Peter Chen
Quoted 82 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

---
Changes for v9:
- Move macro definitions above where they are used
- Remove the brackets around the number
- Merging and sorting local variable definitions
- free the irq in the error path

Changes for v3 (As mailbox patch set):
- Update MODULE_AUTHOR.
- Remove the extra blank lines.

Changes for v2 (As mailbox patch set):
- Update the real name and email address.
- Remove the ACPI header files.
- Update the Copyright from 2024 to 2025.
- Update the Module License from "GPL" to "GPL v2"
- Add an interface for message length to limit potential out-of-bound access

 drivers/mailbox/Kconfig       |  10 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/cix-mailbox.c | 635 ++++++++++++++++++++++++++++++++++
 3 files changed, 647 insertions(+)
 create mode 100644 drivers/mailbox/cix-mailbox.c
drivers/mailbox/Kconfig +10 -0
index 68eeed660a4a..4fef4797b110 100644
@@ -340,4 +340,14 @@ config THEAD_TH1520_MBOX
          kernel is running, and E902 core used for power management among other
          things.
 
+config CIX_MBOX
+        tristate "CIX Mailbox"
+        depends on ARCH_CIX || COMPILE_TEST
+        depends on OF
+        help
+          Mailbox implementation for CIX IPC system. The controller supports
+          11 mailbox channels with different operating mode and every channel
+          is unidirectional. Say Y here if you want to use the CIX Mailbox
+          support.
+
 endif
drivers/mailbox/Makefile +2 -0
index 13a3448b3271..786a46587ba1 100644
@@ -72,3 +72,5 @@ obj-$(CONFIG_QCOM_CPUCP_MBOX) += qcom-cpucp-mbox.o
 obj-$(CONFIG_QCOM_IPCC)                += qcom-ipcc.o
 
 obj-$(CONFIG_THEAD_TH1520_MBOX)        += mailbox-th1520.o
+
+obj-$(CONFIG_CIX_MBOX) += cix-mailbox.o
drivers/mailbox/cix-mailbox.c new file +10 -0
new file mode 100644
index 000000000000..eecb53d59dfe
@@ -0,0 +1,635 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2025 Cix Technology Group Co., Ltd.
+ */
+
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>

drivers/mailbox/cix-mailbox.c line 20

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+#include <linux/mailbox_controller.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+
+#include "mailbox.h"
+
+/*
+ * The maximum transmission size is 32 words or 128 bytes.
+ */
+#define CIX_MBOX_MSG_LEN       32      /* Max length = 32 words */

maybe call it CIX_MBOX_MAX_MSG_WORDS

Quoted 19 lines
drivers/mailbox/cix-mailbox.c new file +19 -0
+#define MBOX_MSG_LEN_MASK      0x7fL   /* Max length = 128 bytes */
+
+/* Register define */
+#define REG_MSG(n)     (0x0 + 0x4*(n))                 /* 0x0~0x7c */
+#define REG_DB_ACK     REG_MSG(CIX_MBOX_MSG_LEN)       /* 0x80 */
+#define ERR_COMP       (REG_DB_ACK + 0x4)              /* 0x84 */
+#define ERR_COMP_CLR   (REG_DB_ACK + 0x8)              /* 0x88 */
+#define REG_F_INT(IDX) (ERR_COMP_CLR + 0x4*(IDX+1))    /* 0x8c~0xa8 */
+#define FIFO_WR                (REG_F_INT(MBOX_FAST_IDX+1))    /* 0xac */
+#define FIFO_RD                (FIFO_WR + 0x4)                 /* 0xb0 */
+#define FIFO_STAS      (FIFO_WR + 0x8)                 /* 0xb4 */
+#define FIFO_WM                (FIFO_WR + 0xc)                 /* 0xb8 */
+#define INT_ENABLE     (FIFO_WR + 0x10)                /* 0xbc */
+#define INT_ENABLE_SIDE_B      (FIFO_WR + 0x14)        /* 0xc0 */
+#define INT_CLEAR      (FIFO_WR + 0x18)                /* 0xc4 */
+#define INT_STATUS     (FIFO_WR + 0x1c)                /* 0xc8 */
+#define FIFO_RST       (FIFO_WR + 0x20)                /* 0xcc */
+
+/* [0~7] Fast channel

drivers/mailbox/cix-mailbox.c line 49

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

And s/MBOX/CIX/ to keep everything CIX specific. Here and elsewhere.

Quoted 38 lines
drivers/mailbox/cix-mailbox.c new file +38 -0
+#define MBOX_TX                0
+#define MBOX_RX                1
+
+#define DB_INT_BIT     BIT(0)
+#define DB_ACK_INT_BIT BIT(1)
+
+#define FIFO_WM_DEFAULT                CIX_MBOX_MSG_LEN
+#define FIFO_STAS_WMK          BIT(0)
+#define FIFO_STAS_FULL         BIT(1)
+#define FIFO_STAS_EMPTY                BIT(2)
+#define FIFO_STAS_UFLOW                BIT(3)
+#define FIFO_STAS_OFLOW                BIT(4)
+
+#define FIFO_RST_BIT           BIT(0)
+
+#define DB_INT                 BIT(0)
+#define ACK_INT                        BIT(1)
+#define FIFO_FULL_INT          BIT(2)
+#define FIFO_EMPTY_INT         BIT(3)
+#define FIFO_WM01_INT          BIT(4)
+#define FIFO_WM10_INT          BIT(5)
+#define FIFO_OFLOW_INT         BIT(6)
+#define FIFO_UFLOW_INT         BIT(7)
+#define FIFO_N_EMPTY_INT       BIT(8)
+#define FAST_CH_INT(IDX)       BIT((IDX)+9)
+
+#define SHMEM_OFFSET 0x80
+
+enum cix_mbox_chan_type {
+       CIX_MBOX_TYPE_DB,
+       CIX_MBOX_TYPE_REG,
+       CIX_MBOX_TYPE_FIFO,
+       CIX_MBOX_TYPE_FAST,
+};
+
+struct cix_mbox_con_priv {
+       enum cix_mbox_chan_type type;
+       struct mbox_chan        *chan;

drivers/mailbox/cix-mailbox.c line 97

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+       int index;
+};
+
+struct cix_mbox_priv {
+       struct device *dev;
+       int irq;
+       int dir;
+       bool tx_irq_mode;       /* flag of enabling tx's irq mode */
+       void __iomem *base;     /* region for mailbox */
+       unsigned int chan_num;

tx_irq_mode and chan_num are unused

Quoted 27 lines
drivers/mailbox/cix-mailbox.c new file +27 -0
+       struct cix_mbox_con_priv con_priv[CIX_MBOX_CHANS];
+       struct mbox_chan mbox_chans[CIX_MBOX_CHANS];
+       struct mbox_controller mbox;
+       bool use_shmem;
+};
+
+/*
+ * The CIX mailbox supports four types of transfers:
+ * CIX_MBOX_TYPE_DB, CIX_MBOX_TYPE_FAST, CIX_MBOX_TYPE_REG, and CIX_MBOX_TYPE_FIFO.
+ * For the REG and FIFO types of transfers, the message format is as follows:
+ */
+union cix_mbox_msg_reg_fifo {
+       u32 length;     /* unit is byte */
+       u32 buf[CIX_MBOX_MSG_LEN]; /* buf[0] must be the byte length of this array */
+};
+
+static struct cix_mbox_priv *to_cix_mbox_priv(struct mbox_controller *mbox)
+{
+       return container_of(mbox, struct cix_mbox_priv, mbox);
+}
+
+static void cix_mbox_write(struct cix_mbox_priv *priv, u32 val, u32 offset)
+{
+       if (priv->use_shmem)
+               iowrite32(val, priv->base + offset - SHMEM_OFFSET);
+       else
+               iowrite32(val, priv->base + offset);

drivers/mailbox/cix-mailbox.c line 134

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+}
+
+static u32 cix_mbox_read(struct cix_mbox_priv *priv, u32 offset)
+{
+       if (priv->use_shmem)
+               return ioread32(priv->base + offset - SHMEM_OFFSET);
+       else
+               return ioread32(priv->base + offset);
+}
+

use_shmem is set for only CIX_MBOX_TYPE_DB, but it affects every read/write. Maybe instead adjust the base for TYPE_DB?

Quoted 266 lines
drivers/mailbox/cix-mailbox.c new file +266 -0
+static bool mbox_fifo_empty(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+       return ((cix_mbox_read(priv, FIFO_STAS) & FIFO_STAS_EMPTY) ? true : false);
+}
+
+/*
+ *The transmission unit of the CIX mailbox is word.
+ *The byte length should be converted into the word length.
+ */
+static inline u32 mbox_get_msg_size(void *msg)
+{
+       u32 len;
+
+       len = ((u32 *)msg)[0] & MBOX_MSG_LEN_MASK;
+       return DIV_ROUND_UP(len, 4);
+}
+
+static int cix_mbox_send_data_db(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+       /* trigger doorbell irq */
+       cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_reg(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       union cix_mbox_msg_reg_fifo *msg = data;
+       u32 len, i;
+
+       if (!data)
+               return -EINVAL;
+
+       len = mbox_get_msg_size(data);
+       for (i = 0; i < len; i++)
+               cix_mbox_write(priv, msg->buf[i], REG_MSG(i));
+
+       /* trigger doorbell irq */
+       cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_fifo(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       union cix_mbox_msg_reg_fifo *msg = data;
+       u32 len, val_32, i;
+
+       if (!data)
+               return -EINVAL;
+
+       len = mbox_get_msg_size(data);
+       cix_mbox_write(priv, len, FIFO_WM);
+       for (i = 0; i < len; i++)
+               cix_mbox_write(priv, msg->buf[i], FIFO_WR);
+
+       /* Enable fifo empty interrupt */
+       val_32 = cix_mbox_read(priv, INT_ENABLE);
+       val_32 |= FIFO_EMPTY_INT;
+       cix_mbox_write(priv, val_32, INT_ENABLE);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_fast(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       u32 *arg = (u32 *)data;
+       int index = cp->index;
+
+       if (!data)
+               return -EINVAL;
+
+       if (index < 0 || index > MBOX_FAST_IDX) {
+               dev_err(priv->dev, "Invalid Mbox index %d\n", index);
+               return -EINVAL;
+       }
+
+       cix_mbox_write(priv, arg[0], REG_F_INT(index));
+
+       return 0;
+}
+
+static int cix_mbox_send_data(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+
+       if (priv->dir != MBOX_TX) {
+               dev_err(priv->dev, "Invalid Mbox dir %d\n", priv->dir);
+               return -EINVAL;
+       }
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               cix_mbox_send_data_db(chan, data);
+               break;
+       case CIX_MBOX_TYPE_REG:
+               cix_mbox_send_data_reg(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               cix_mbox_send_data_fifo(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               cix_mbox_send_data_fast(chan, data);
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               return -EINVAL;
+       }
+       return 0;
+}
+
+static void cix_mbox_isr_db(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* rx interrupt is triggered */
+               if (int_status & DB_INT) {
+                       cix_mbox_write(priv, DB_INT, INT_CLEAR);
+                       mbox_chan_received_data(chan, NULL);
+                       /* trigger ack interrupt */
+                       cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+               }
+       } else {
+               /* tx ack interrupt is triggered */
+               if (int_status & ACK_INT) {
+                       cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+                       mbox_chan_received_data(chan, NULL);
+               }
+       }
+}
+
+static void cix_mbox_isr_reg(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* rx interrupt is triggered */
+               if (int_status & DB_INT) {
+                       u32 data[CIX_MBOX_MSG_LEN], len, i;
+
+                       cix_mbox_write(priv, DB_INT, INT_CLEAR);
+                       data[0] = cix_mbox_read(priv, REG_MSG(0));
+                       len = mbox_get_msg_size(data);
+                       for (i = 1; i < len; i++)
+                               data[i] = cix_mbox_read(priv, REG_MSG(i));
+
+                       /* trigger ack interrupt */
+                       cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+                       mbox_chan_received_data(chan, data);
+               }
+       } else {
+               /* tx ack interrupt is triggered */
+               if (int_status & ACK_INT) {
+                       cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+                       mbox_chan_txdone(chan, 0);
+               }
+       }
+}
+
+static void cix_mbox_isr_fifo(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status, status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* FIFO waterMark interrupt is generated */
+               if (int_status & (FIFO_FULL_INT | FIFO_WM01_INT)) {
+                       u32 data[CIX_MBOX_MSG_LEN] = { 0 }, i = 0;
+
+                       cix_mbox_write(priv, (FIFO_FULL_INT | FIFO_WM01_INT), INT_CLEAR);
+                       do {
+                               data[i++] = cix_mbox_read(priv, FIFO_RD);
+                       } while (!mbox_fifo_empty(chan) && i < CIX_MBOX_MSG_LEN);
+
+                       mbox_chan_received_data(chan, data);
+               }
+               /* FIFO underflow is generated */
+               if (int_status & FIFO_UFLOW_INT) {
+                       status = cix_mbox_read(priv, FIFO_STAS);
+                       dev_err(priv->dev, "fifo underflow: int_stats %d\n", status);
+                       cix_mbox_write(priv, FIFO_UFLOW_INT, INT_CLEAR);
+               }
+       } else {
+               /* FIFO empty interrupt is generated */
+               if (int_status & FIFO_EMPTY_INT) {
+                       u32 val_32;
+
+                       cix_mbox_write(priv, FIFO_EMPTY_INT, INT_CLEAR);
+                       /* Disable empty irq*/
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~FIFO_EMPTY_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+                       mbox_chan_txdone(chan, 0);
+               }
+               /* FIFO overflow is generated */
+               if (int_status & FIFO_OFLOW_INT) {
+                       status = cix_mbox_read(priv, FIFO_STAS);
+                       dev_err(priv->dev, "fifo overlow: int_stats %d\n", status);
+                       cix_mbox_write(priv, FIFO_OFLOW_INT, INT_CLEAR);
+               }
+       }
+}
+
+static void cix_mbox_isr_fast(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       u32 int_status, data;
+
+       /* no irq will be trigger for TX dir mbox */
+       if (priv->dir != MBOX_RX)
+               return;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (int_status & FAST_CH_INT(cp->index)) {
+               cix_mbox_write(priv, FAST_CH_INT(cp->index), INT_CLEAR);
+               data = cix_mbox_read(priv, REG_F_INT(cp->index));
+               mbox_chan_received_data(chan, &data);
+       }
+}
+
+static irqreturn_t cix_mbox_isr(int irq, void *arg)
+{
+       struct mbox_chan *chan = arg;
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               cix_mbox_isr_db(chan);
+               break;
+       case CIX_MBOX_TYPE_REG:
+               cix_mbox_isr_reg(chan);
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               cix_mbox_isr_fifo(chan);
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               cix_mbox_isr_fast(chan);
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               return IRQ_NONE;
+       }
+
+       return IRQ_HANDLED;
+}

drivers/mailbox/cix-mailbox.c line 410

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+
+static int cix_mbox_startup(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       int index = cp->index, ret;
+       u32 val_32;
+
+       ret = request_irq(priv->irq, cix_mbox_isr, 0,
+                         dev_name(priv->dev), chan);

The same irq is requested for each channel. How do you expect it to work? Maybe request it just once in probe and pass the 'priv' instead of 'chan' , and in the cix_mbox_isr handle according to INT_STATUS

Quoted 62 lines
drivers/mailbox/cix-mailbox.c new file +62 -0
+       if (ret) {
+               dev_err(priv->dev, "Unable to acquire IRQ %d\n", priv->irq);
+               return ret;
+       }
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               /* Overwrite txdone_method for DB channel */
+               chan->txdone_method = TXDONE_BY_ACK;
+               fallthrough;
+       case CIX_MBOX_TYPE_REG:
+               if (priv->dir == MBOX_TX) {
+                       /* Enable ACK interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 |= ACK_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else {
+                       /* Enable Doorbell interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= DB_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               /* reset fifo */
+               cix_mbox_write(priv, FIFO_RST_BIT, FIFO_RST);
+               /* set default watermark */
+               cix_mbox_write(priv, FIFO_WM_DEFAULT, FIFO_WM);
+               if (priv->dir == MBOX_TX) {
+                       /* Enable fifo overflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 |= FIFO_OFLOW_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else {
+                       /* Enable fifo full/underflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= FIFO_UFLOW_INT|FIFO_WM01_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               /* Only RX channel has intterupt */
+               if (priv->dir == MBOX_RX) {
+                       if (index < 0 || index > MBOX_FAST_IDX) {
+                               dev_err(priv->dev, "Invalid index %d\n", index);
+                               ret = -EINVAL;
+                               goto failed;
+                       }
+                       /* enable fast channel interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= FAST_CH_INT(index);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               ret = -EINVAL;
+               goto failed;
+       }
+       return 0;
+
+failed:

drivers/mailbox/cix-mailbox.c line 482

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+       free_irq(priv->irq, chan);
+       return ret;
+}
+
+static void cix_mbox_shutdown(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       int index = cp->index;
+       u32 val_32;

Never saw this style before, may simply use val

Quoted 143 lines
drivers/mailbox/cix-mailbox.c new file +143 -0
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+       case CIX_MBOX_TYPE_REG:
+               if (priv->dir == MBOX_TX) {
+                       /* Disable ACK interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~ACK_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else if (priv->dir == MBOX_RX) {
+                       /* Disable Doorbell interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~DB_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               if (priv->dir == MBOX_TX) {
+                       /* Disable empty/fifo overflow irq*/
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~(FIFO_EMPTY_INT | FIFO_OFLOW_INT);
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else if (priv->dir == MBOX_RX) {
+                       /* Disable fifo WM01/underflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~(FIFO_UFLOW_INT | FIFO_WM01_INT);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               if (priv->dir == MBOX_RX) {
+                       if (index < 0 || index > MBOX_FAST_IDX) {
+                               dev_err(priv->dev, "Invalid index %d\n", index);
+                               break;
+                       }
+                       /* Disable fast channel interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~FAST_CH_INT(index);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               break;
+       }
+
+       free_irq(priv->irq, chan);
+}
+
+static const struct mbox_chan_ops cix_mbox_chan_ops = {
+       .send_data = cix_mbox_send_data,
+       .startup = cix_mbox_startup,
+       .shutdown = cix_mbox_shutdown,
+};
+
+static void cix_mbox_init(struct cix_mbox_priv *priv)
+{
+       struct cix_mbox_con_priv *cp;
+       int i;
+
+       for (i = 0; i < CIX_MBOX_CHANS; i++) {
+               cp = &priv->con_priv[i];
+               cp->index = i;
+               cp->chan = &priv->mbox_chans[i];
+               priv->mbox_chans[i].con_priv = cp;
+               if (cp->index <= MBOX_FAST_IDX)
+                       cp->type = CIX_MBOX_TYPE_FAST;
+               if (cp->index == MBOX_DB_IDX) {
+                       cp->type = CIX_MBOX_TYPE_DB;
+                       priv->use_shmem = true;
+               }
+               if (cp->index == MBOX_FIFO_IDX)
+                       cp->type = CIX_MBOX_TYPE_FIFO;
+               if (cp->index == MBOX_REG_IDX)
+                       cp->type = CIX_MBOX_TYPE_REG;
+       }
+}
+
+static int cix_mbox_probe(struct platform_device *pdev)
+{
+       struct device *dev = &pdev->dev;
+       struct cix_mbox_priv *priv;
+       const char *dir_str;
+       int ret;
+
+       priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+       if (!priv)
+               return -ENOMEM;
+
+       priv->dev = dev;
+       priv->base = devm_platform_ioremap_resource(pdev, 0);
+       if (IS_ERR(priv->base))
+               return PTR_ERR(priv->base);
+
+       priv->irq = platform_get_irq(pdev, 0);
+       if (priv->irq < 0)
+               return priv->irq;
+
+       if (device_property_read_string(dev, "cix,mbox-dir", &dir_str)) {
+               dev_err(priv->dev, "cix,mbox_dir property not found\n");
+               return -EINVAL;
+       }
+
+       if (!strcmp(dir_str, "tx"))
+               priv->dir = 0;
+       else if (!strcmp(dir_str, "rx"))
+               priv->dir = 1;
+       else {
+               dev_err(priv->dev, "cix,mbox_dir=%s is not expected\n", dir_str);
+               return -EINVAL;
+       }
+
+       cix_mbox_init(priv);
+
+       priv->mbox.dev = dev;
+       priv->mbox.ops = &cix_mbox_chan_ops;
+       priv->mbox.chans = priv->mbox_chans;
+       priv->mbox.txdone_irq = true;
+       priv->mbox.num_chans = CIX_MBOX_CHANS;
+       priv->mbox.of_xlate = NULL;
+
+       platform_set_drvdata(pdev, priv);
+       ret = devm_mbox_controller_register(dev, &priv->mbox);
+       if (ret)
+               dev_err(dev, "Failed to register mailbox %d\n", ret);
+
+       return ret;
+}
+
+static const struct of_device_id cix_mbox_dt_ids[] = {
+       { .compatible = "cix,sky1-mbox" },
+       { },
+};
+MODULE_DEVICE_TABLE(of, cix_mbox_dt_ids);
+
+static struct platform_driver cix_mbox_driver = {
+       .probe = cix_mbox_probe,
+       .driver = {
+               .name = "cix_mbox",
+               .of_match_table = cix_mbox_dt_ids,
+       },
+};

drivers/mailbox/cix-mailbox.c line 635

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+
+static int __init cix_mailbox_init(void)
+{
+       return platform_driver_register(&cix_mbox_driver);
+}
+arch_initcall(cix_mailbox_init);
+
+MODULE_AUTHOR("Cix Technology Group Co., Ltd.");
+MODULE_DESCRIPTION("CIX mailbox driver");
+MODULE_LICENSE("GPL");

GPL v2 ? according to the SPDX-License-Identifier

And please make sure you run scripts/checkpatch.pl

thanks

#11 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 90 lines · Jassi Brar

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL

Quoted 85 lines · Peter Chen

From: Guomin Chen <[email protected]>

The CIX mailbox controller, used in the Cix SoCs, like sky1. facilitates message transmission between multiple processors within the SoC, such as the AP, PM, audio DSP, SensorHub MCU, and others.

---
Changes for v9:
- Move macro definitions above where they are used
- Remove the brackets around the number
- Merging and sorting local variable definitions
- free the irq in the error path

Changes for v3 (As mailbox patch set):
- Update MODULE_AUTHOR.
- Remove the extra blank lines.

Changes for v2 (As mailbox patch set):
- Update the real name and email address.
- Remove the ACPI header files.
- Update the Copyright from 2024 to 2025.
- Update the Module License from "GPL" to "GPL v2"
- Add an interface for message length to limit potential out-of-bound access

 drivers/mailbox/Kconfig       |  10 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/cix-mailbox.c | 635 ++++++++++++++++++++++++++++++++++
 3 files changed, 647 insertions(+)
 create mode 100644 drivers/mailbox/cix-mailbox.c
drivers/mailbox/Kconfig +10 -0
index 68eeed660a4a..4fef4797b110 100644
@@ -340,4 +340,14 @@ config THEAD_TH1520_MBOX
          kernel is running, and E902 core used for power management among other
          things.
 
+config CIX_MBOX
+        tristate "CIX Mailbox"
+        depends on ARCH_CIX || COMPILE_TEST
+        depends on OF
+        help
+          Mailbox implementation for CIX IPC system. The controller supports
+          11 mailbox channels with different operating mode and every channel
+          is unidirectional. Say Y here if you want to use the CIX Mailbox
+          support.
+
 endif
drivers/mailbox/Makefile +2 -0
index 13a3448b3271..786a46587ba1 100644
@@ -72,3 +72,5 @@ obj-$(CONFIG_QCOM_CPUCP_MBOX) += qcom-cpucp-mbox.o
 obj-$(CONFIG_QCOM_IPCC)                += qcom-ipcc.o
 
 obj-$(CONFIG_THEAD_TH1520_MBOX)        += mailbox-th1520.o
+
+obj-$(CONFIG_CIX_MBOX) += cix-mailbox.o
drivers/mailbox/cix-mailbox.c new file +13 -0
new file mode 100644
index 000000000000..eecb53d59dfe
@@ -0,0 +1,635 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2025 Cix Technology Group Co., Ltd.
+ */
+
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/mailbox_controller.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
Quoted 8 lines

drivers/mailbox/cix-mailbox.c line 20

Quoted 7 lines
drivers/mailbox/cix-mailbox.c new file +7 -0
+
+#include "mailbox.h"
+
+/*
+ * The maximum transmission size is 32 words or 128 bytes.
+ */
+#define CIX_MBOX_MSG_LEN       32      /* Max length = 32 words */

maybe call it CIX_MBOX_MAX_MSG_WORDS

Yes, CIX_MBOX_MAX_MSG_WORDS is descriptive.

Quoted 23 lines
Quoted 23 lines
drivers/mailbox/cix-mailbox.c new file +23 -0
+#define MBOX_MSG_LEN_MASK      0x7fL   /* Max length = 128 bytes */
+
+/* Register define */
+#define REG_MSG(n)     (0x0 + 0x4*(n))                 /* 0x0~0x7c */
+#define REG_DB_ACK     REG_MSG(CIX_MBOX_MSG_LEN)       /* 0x80 */
+#define ERR_COMP       (REG_DB_ACK + 0x4)              /* 0x84 */
+#define ERR_COMP_CLR   (REG_DB_ACK + 0x8)              /* 0x88 */
+#define REG_F_INT(IDX) (ERR_COMP_CLR + 0x4*(IDX+1))    /* 0x8c~0xa8 */
+#define FIFO_WR                (REG_F_INT(MBOX_FAST_IDX+1))    /* 0xac */
+#define FIFO_RD                (FIFO_WR + 0x4)                 /* 0xb0 */
+#define FIFO_STAS      (FIFO_WR + 0x8)                 /* 0xb4 */
+#define FIFO_WM                (FIFO_WR + 0xc)                 /* 0xb8 */
+#define INT_ENABLE     (FIFO_WR + 0x10)                /* 0xbc */
+#define INT_ENABLE_SIDE_B      (FIFO_WR + 0x14)        /* 0xc0 */
+#define INT_CLEAR      (FIFO_WR + 0x18)                /* 0xc4 */
+#define INT_STATUS     (FIFO_WR + 0x1c)                /* 0xc8 */
+#define FIFO_RST       (FIFO_WR + 0x20)                /* 0xcc */
+
+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 6 lines
drivers/mailbox/cix-mailbox.c new file +6 -0
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

As mentioned in the previous email, a single controller can support multiple different channels.

Quoted 1 line

And s/MBOX/CIX/ to keep everything CIX specific. Here and elsewhere.

Okay, I will change it in the next version.

Quoted 41 lines
Quoted 41 lines
drivers/mailbox/cix-mailbox.c new file +41 -0
+#define MBOX_TX                0
+#define MBOX_RX                1
+
+#define DB_INT_BIT     BIT(0)
+#define DB_ACK_INT_BIT BIT(1)
+
+#define FIFO_WM_DEFAULT                CIX_MBOX_MSG_LEN
+#define FIFO_STAS_WMK          BIT(0)
+#define FIFO_STAS_FULL         BIT(1)
+#define FIFO_STAS_EMPTY                BIT(2)
+#define FIFO_STAS_UFLOW                BIT(3)
+#define FIFO_STAS_OFLOW                BIT(4)
+
+#define FIFO_RST_BIT           BIT(0)
+
+#define DB_INT                 BIT(0)
+#define ACK_INT                        BIT(1)
+#define FIFO_FULL_INT          BIT(2)
+#define FIFO_EMPTY_INT         BIT(3)
+#define FIFO_WM01_INT          BIT(4)
+#define FIFO_WM10_INT          BIT(5)
+#define FIFO_OFLOW_INT         BIT(6)
+#define FIFO_UFLOW_INT         BIT(7)
+#define FIFO_N_EMPTY_INT       BIT(8)
+#define FAST_CH_INT(IDX)       BIT((IDX)+9)
+
+#define SHMEM_OFFSET 0x80
+
+enum cix_mbox_chan_type {
+       CIX_MBOX_TYPE_DB,
+       CIX_MBOX_TYPE_REG,
+       CIX_MBOX_TYPE_FIFO,
+       CIX_MBOX_TYPE_FAST,
+};
+
+struct cix_mbox_con_priv {
+       enum cix_mbox_chan_type type;
+       struct mbox_chan        *chan;
+       int index;
+};
+
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 97

Quoted 7 lines
drivers/mailbox/cix-mailbox.c new file +7 -0
+struct cix_mbox_priv {
+       struct device *dev;
+       int irq;
+       int dir;
+       bool tx_irq_mode;       /* flag of enabling tx's irq mode */
+       void __iomem *base;     /* region for mailbox */
+       unsigned int chan_num;

tx_irq_mode and chan_num are unused

Okay,I will remove it in the next version.

Quoted 31 lines
Quoted 31 lines
drivers/mailbox/cix-mailbox.c new file +31 -0
+       struct cix_mbox_con_priv con_priv[CIX_MBOX_CHANS];
+       struct mbox_chan mbox_chans[CIX_MBOX_CHANS];
+       struct mbox_controller mbox;
+       bool use_shmem;
+};
+
+/*
+ * The CIX mailbox supports four types of transfers:
+ * CIX_MBOX_TYPE_DB, CIX_MBOX_TYPE_FAST, CIX_MBOX_TYPE_REG, and CIX_MBOX_TYPE_FIFO.
+ * For the REG and FIFO types of transfers, the message format is as follows:
+ */
+union cix_mbox_msg_reg_fifo {
+       u32 length;     /* unit is byte */
+       u32 buf[CIX_MBOX_MSG_LEN]; /* buf[0] must be the byte length of this array */
+};
+
+static struct cix_mbox_priv *to_cix_mbox_priv(struct mbox_controller *mbox)
+{
+       return container_of(mbox, struct cix_mbox_priv, mbox);
+}
+
+static void cix_mbox_write(struct cix_mbox_priv *priv, u32 val, u32 offset)
+{
+       if (priv->use_shmem)
+               iowrite32(val, priv->base + offset - SHMEM_OFFSET);
+       else
+               iowrite32(val, priv->base + offset);
+}
+
+static u32 cix_mbox_read(struct cix_mbox_priv *priv, u32 offset)
+{
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 134

Quoted 6 lines
drivers/mailbox/cix-mailbox.c new file +6 -0
+       if (priv->use_shmem)
+               return ioread32(priv->base + offset - SHMEM_OFFSET);
+       else
+               return ioread32(priv->base + offset);
+}
+

use_shmem is set for only CIX_MBOX_TYPE_DB, but it affects every read/write. Maybe instead adjust the base for TYPE_DB?

The reason we introduced use_shmem here is that we had to adjust the base address of TYPE_DB to resolve the reg conflict in the DTS. This change has virtually no impact on performance.

Quoted 271 lines
Quoted 271 lines
drivers/mailbox/cix-mailbox.c new file +271 -0
+static bool mbox_fifo_empty(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+       return ((cix_mbox_read(priv, FIFO_STAS) & FIFO_STAS_EMPTY) ? true : false);
+}
+
+/*
+ *The transmission unit of the CIX mailbox is word.
+ *The byte length should be converted into the word length.
+ */
+static inline u32 mbox_get_msg_size(void *msg)
+{
+       u32 len;
+
+       len = ((u32 *)msg)[0] & MBOX_MSG_LEN_MASK;
+       return DIV_ROUND_UP(len, 4);
+}
+
+static int cix_mbox_send_data_db(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+
+       /* trigger doorbell irq */
+       cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_reg(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       union cix_mbox_msg_reg_fifo *msg = data;
+       u32 len, i;
+
+       if (!data)
+               return -EINVAL;
+
+       len = mbox_get_msg_size(data);
+       for (i = 0; i < len; i++)
+               cix_mbox_write(priv, msg->buf[i], REG_MSG(i));
+
+       /* trigger doorbell irq */
+       cix_mbox_write(priv, DB_INT_BIT, REG_DB_ACK);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_fifo(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       union cix_mbox_msg_reg_fifo *msg = data;
+       u32 len, val_32, i;
+
+       if (!data)
+               return -EINVAL;
+
+       len = mbox_get_msg_size(data);
+       cix_mbox_write(priv, len, FIFO_WM);
+       for (i = 0; i < len; i++)
+               cix_mbox_write(priv, msg->buf[i], FIFO_WR);
+
+       /* Enable fifo empty interrupt */
+       val_32 = cix_mbox_read(priv, INT_ENABLE);
+       val_32 |= FIFO_EMPTY_INT;
+       cix_mbox_write(priv, val_32, INT_ENABLE);
+
+       return 0;
+}
+
+static int cix_mbox_send_data_fast(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       u32 *arg = (u32 *)data;
+       int index = cp->index;
+
+       if (!data)
+               return -EINVAL;
+
+       if (index < 0 || index > MBOX_FAST_IDX) {
+               dev_err(priv->dev, "Invalid Mbox index %d\n", index);
+               return -EINVAL;
+       }
+
+       cix_mbox_write(priv, arg[0], REG_F_INT(index));
+
+       return 0;
+}
+
+static int cix_mbox_send_data(struct mbox_chan *chan, void *data)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+
+       if (priv->dir != MBOX_TX) {
+               dev_err(priv->dev, "Invalid Mbox dir %d\n", priv->dir);
+               return -EINVAL;
+       }
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               cix_mbox_send_data_db(chan, data);
+               break;
+       case CIX_MBOX_TYPE_REG:
+               cix_mbox_send_data_reg(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               cix_mbox_send_data_fifo(chan, data);
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               cix_mbox_send_data_fast(chan, data);
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               return -EINVAL;
+       }
+       return 0;
+}
+
+static void cix_mbox_isr_db(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* rx interrupt is triggered */
+               if (int_status & DB_INT) {
+                       cix_mbox_write(priv, DB_INT, INT_CLEAR);
+                       mbox_chan_received_data(chan, NULL);
+                       /* trigger ack interrupt */
+                       cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+               }
+       } else {
+               /* tx ack interrupt is triggered */
+               if (int_status & ACK_INT) {
+                       cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+                       mbox_chan_received_data(chan, NULL);
+               }
+       }
+}
+
+static void cix_mbox_isr_reg(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* rx interrupt is triggered */
+               if (int_status & DB_INT) {
+                       u32 data[CIX_MBOX_MSG_LEN], len, i;
+
+                       cix_mbox_write(priv, DB_INT, INT_CLEAR);
+                       data[0] = cix_mbox_read(priv, REG_MSG(0));
+                       len = mbox_get_msg_size(data);
+                       for (i = 1; i < len; i++)
+                               data[i] = cix_mbox_read(priv, REG_MSG(i));
+
+                       /* trigger ack interrupt */
+                       cix_mbox_write(priv, DB_ACK_INT_BIT, REG_DB_ACK);
+                       mbox_chan_received_data(chan, data);
+               }
+       } else {
+               /* tx ack interrupt is triggered */
+               if (int_status & ACK_INT) {
+                       cix_mbox_write(priv, ACK_INT, INT_CLEAR);
+                       mbox_chan_txdone(chan, 0);
+               }
+       }
+}
+
+static void cix_mbox_isr_fifo(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       u32 int_status, status;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (priv->dir == MBOX_RX) {
+               /* FIFO waterMark interrupt is generated */
+               if (int_status & (FIFO_FULL_INT | FIFO_WM01_INT)) {
+                       u32 data[CIX_MBOX_MSG_LEN] = { 0 }, i = 0;
+
+                       cix_mbox_write(priv, (FIFO_FULL_INT | FIFO_WM01_INT), INT_CLEAR);
+                       do {
+                               data[i++] = cix_mbox_read(priv, FIFO_RD);
+                       } while (!mbox_fifo_empty(chan) && i < CIX_MBOX_MSG_LEN);
+
+                       mbox_chan_received_data(chan, data);
+               }
+               /* FIFO underflow is generated */
+               if (int_status & FIFO_UFLOW_INT) {
+                       status = cix_mbox_read(priv, FIFO_STAS);
+                       dev_err(priv->dev, "fifo underflow: int_stats %d\n", status);
+                       cix_mbox_write(priv, FIFO_UFLOW_INT, INT_CLEAR);
+               }
+       } else {
+               /* FIFO empty interrupt is generated */
+               if (int_status & FIFO_EMPTY_INT) {
+                       u32 val_32;
+
+                       cix_mbox_write(priv, FIFO_EMPTY_INT, INT_CLEAR);
+                       /* Disable empty irq*/
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~FIFO_EMPTY_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+                       mbox_chan_txdone(chan, 0);
+               }
+               /* FIFO overflow is generated */
+               if (int_status & FIFO_OFLOW_INT) {
+                       status = cix_mbox_read(priv, FIFO_STAS);
+                       dev_err(priv->dev, "fifo overlow: int_stats %d\n", status);
+                       cix_mbox_write(priv, FIFO_OFLOW_INT, INT_CLEAR);
+               }
+       }
+}
+
+static void cix_mbox_isr_fast(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       u32 int_status, data;
+
+       /* no irq will be trigger for TX dir mbox */
+       if (priv->dir != MBOX_RX)
+               return;
+
+       int_status = cix_mbox_read(priv, INT_STATUS);
+
+       if (int_status & FAST_CH_INT(cp->index)) {
+               cix_mbox_write(priv, FAST_CH_INT(cp->index), INT_CLEAR);
+               data = cix_mbox_read(priv, REG_F_INT(cp->index));
+               mbox_chan_received_data(chan, &data);
+       }
+}
+
+static irqreturn_t cix_mbox_isr(int irq, void *arg)
+{
+       struct mbox_chan *chan = arg;
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               cix_mbox_isr_db(chan);
+               break;
+       case CIX_MBOX_TYPE_REG:
+               cix_mbox_isr_reg(chan);
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               cix_mbox_isr_fifo(chan);
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               cix_mbox_isr_fast(chan);
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               return IRQ_NONE;
+       }
+
+       return IRQ_HANDLED;
+}
+
+static int cix_mbox_startup(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 410

Quoted 5 lines
drivers/mailbox/cix-mailbox.c new file +5 -0
+       int index = cp->index, ret;
+       u32 val_32;
+
+       ret = request_irq(priv->irq, cix_mbox_isr, 0,
+                         dev_name(priv->dev), chan);

The same irq is requested for each channel. How do you expect it to work? Maybe request it just once in probe and pass the 'priv' instead of 'chan' , and in the cix_mbox_isr handle according to INT_STATUS

For the same mailbox controller, there won't be multiple channels simultaneously requesting the same IRQ, so there won't be an issue here. As you mentioned, if we need to handle multiple channels working concurrently in the future, we would need to modify cix_mbox_isr. However, that is not required at the moment.

Quoted 66 lines
Quoted 66 lines
drivers/mailbox/cix-mailbox.c new file +66 -0
+       if (ret) {
+               dev_err(priv->dev, "Unable to acquire IRQ %d\n", priv->irq);
+               return ret;
+       }
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+               /* Overwrite txdone_method for DB channel */
+               chan->txdone_method = TXDONE_BY_ACK;
+               fallthrough;
+       case CIX_MBOX_TYPE_REG:
+               if (priv->dir == MBOX_TX) {
+                       /* Enable ACK interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 |= ACK_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else {
+                       /* Enable Doorbell interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= DB_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               /* reset fifo */
+               cix_mbox_write(priv, FIFO_RST_BIT, FIFO_RST);
+               /* set default watermark */
+               cix_mbox_write(priv, FIFO_WM_DEFAULT, FIFO_WM);
+               if (priv->dir == MBOX_TX) {
+                       /* Enable fifo overflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 |= FIFO_OFLOW_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else {
+                       /* Enable fifo full/underflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= FIFO_UFLOW_INT|FIFO_WM01_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               /* Only RX channel has intterupt */
+               if (priv->dir == MBOX_RX) {
+                       if (index < 0 || index > MBOX_FAST_IDX) {
+                               dev_err(priv->dev, "Invalid index %d\n", index);
+                               ret = -EINVAL;
+                               goto failed;
+                       }
+                       /* enable fast channel interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 |= FAST_CH_INT(index);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               ret = -EINVAL;
+               goto failed;
+       }
+       return 0;
+
+failed:
+       free_irq(priv->irq, chan);
+       return ret;
+}
+
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 482

Quoted 6 lines
drivers/mailbox/cix-mailbox.c new file +6 -0
+static void cix_mbox_shutdown(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       int index = cp->index;
+       u32 val_32;

Never saw this style before, may simply use val

Yes,I will change it in the next version.

Quoted 156 lines
Quoted 143 lines
drivers/mailbox/cix-mailbox.c new file +143 -0
+
+       switch (cp->type) {
+       case CIX_MBOX_TYPE_DB:
+       case CIX_MBOX_TYPE_REG:
+               if (priv->dir == MBOX_TX) {
+                       /* Disable ACK interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~ACK_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else if (priv->dir == MBOX_RX) {
+                       /* Disable Doorbell interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~DB_INT;
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FIFO:
+               if (priv->dir == MBOX_TX) {
+                       /* Disable empty/fifo overflow irq*/
+                       val_32 = cix_mbox_read(priv, INT_ENABLE);
+                       val_32 &= ~(FIFO_EMPTY_INT | FIFO_OFLOW_INT);
+                       cix_mbox_write(priv, val_32, INT_ENABLE);
+               } else if (priv->dir == MBOX_RX) {
+                       /* Disable fifo WM01/underflow interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~(FIFO_UFLOW_INT | FIFO_WM01_INT);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+       case CIX_MBOX_TYPE_FAST:
+               if (priv->dir == MBOX_RX) {
+                       if (index < 0 || index > MBOX_FAST_IDX) {
+                               dev_err(priv->dev, "Invalid index %d\n", index);
+                               break;
+                       }
+                       /* Disable fast channel interrupt */
+                       val_32 = cix_mbox_read(priv, INT_ENABLE_SIDE_B);
+                       val_32 &= ~FAST_CH_INT(index);
+                       cix_mbox_write(priv, val_32, INT_ENABLE_SIDE_B);
+               }
+               break;
+
+       default:
+               dev_err(priv->dev, "Invalid channel type: %d\n", cp->type);
+               break;
+       }
+
+       free_irq(priv->irq, chan);
+}
+
+static const struct mbox_chan_ops cix_mbox_chan_ops = {
+       .send_data = cix_mbox_send_data,
+       .startup = cix_mbox_startup,
+       .shutdown = cix_mbox_shutdown,
+};
+
+static void cix_mbox_init(struct cix_mbox_priv *priv)
+{
+       struct cix_mbox_con_priv *cp;
+       int i;
+
+       for (i = 0; i < CIX_MBOX_CHANS; i++) {
+               cp = &priv->con_priv[i];
+               cp->index = i;
+               cp->chan = &priv->mbox_chans[i];
+               priv->mbox_chans[i].con_priv = cp;
+               if (cp->index <= MBOX_FAST_IDX)
+                       cp->type = CIX_MBOX_TYPE_FAST;
+               if (cp->index == MBOX_DB_IDX) {
+                       cp->type = CIX_MBOX_TYPE_DB;
+                       priv->use_shmem = true;
+               }
+               if (cp->index == MBOX_FIFO_IDX)
+                       cp->type = CIX_MBOX_TYPE_FIFO;
+               if (cp->index == MBOX_REG_IDX)
+                       cp->type = CIX_MBOX_TYPE_REG;
+       }
+}
+
+static int cix_mbox_probe(struct platform_device *pdev)
+{
+       struct device *dev = &pdev->dev;
+       struct cix_mbox_priv *priv;
+       const char *dir_str;
+       int ret;
+
+       priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+       if (!priv)
+               return -ENOMEM;
+
+       priv->dev = dev;
+       priv->base = devm_platform_ioremap_resource(pdev, 0);
+       if (IS_ERR(priv->base))
+               return PTR_ERR(priv->base);
+
+       priv->irq = platform_get_irq(pdev, 0);
+       if (priv->irq < 0)
+               return priv->irq;
+
+       if (device_property_read_string(dev, "cix,mbox-dir", &dir_str)) {
+               dev_err(priv->dev, "cix,mbox_dir property not found\n");
+               return -EINVAL;
+       }
+
+       if (!strcmp(dir_str, "tx"))
+               priv->dir = 0;
+       else if (!strcmp(dir_str, "rx"))
+               priv->dir = 1;
+       else {
+               dev_err(priv->dev, "cix,mbox_dir=%s is not expected\n", dir_str);
+               return -EINVAL;
+       }
+
+       cix_mbox_init(priv);
+
+       priv->mbox.dev = dev;
+       priv->mbox.ops = &cix_mbox_chan_ops;
+       priv->mbox.chans = priv->mbox_chans;
+       priv->mbox.txdone_irq = true;
+       priv->mbox.num_chans = CIX_MBOX_CHANS;
+       priv->mbox.of_xlate = NULL;
+
+       platform_set_drvdata(pdev, priv);
+       ret = devm_mbox_controller_register(dev, &priv->mbox);
+       if (ret)
+               dev_err(dev, "Failed to register mailbox %d\n", ret);
+
+       return ret;
+}
+
+static const struct of_device_id cix_mbox_dt_ids[] = {
+       { .compatible = "cix,sky1-mbox" },
+       { },
+};
+MODULE_DEVICE_TABLE(of, cix_mbox_dt_ids);
+
+static struct platform_driver cix_mbox_driver = {
+       .probe = cix_mbox_probe,
+       .driver = {
+               .name = "cix_mbox",
+               .of_match_table = cix_mbox_dt_ids,
+       },
+};

drivers/mailbox/cix-mailbox.c line 635

Quoted 10 lines
drivers/mailbox/cix-mailbox.c new file +10 -0
+
+static int __init cix_mailbox_init(void)
+{
+       return platform_driver_register(&cix_mbox_driver);
+}
+arch_initcall(cix_mailbox_init);
+
+MODULE_AUTHOR("Cix Technology Group Co., Ltd.");
+MODULE_DESCRIPTION("CIX mailbox driver");
+MODULE_LICENSE("GPL");

GPL v2 ? according to the SPDX-License-Identifier

Quoted 1 line

And please make sure you run scripts/checkpatch.pl

Yes, I'm also confused here. I was previously using GPL v2, but when I ran checkpatch.pl, it triggered a warning due to commit bf7fbeeae6db ("module: Cure the MODULE_LICENSE 'GPL' vs. 'GPL v2' bogosity"). So I changed it to GPL.

Thanks
Guomin Chen

maintainer
#12 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

On Tue, Jul 8, 2025 at 8:54 AM Guomin chen <[email protected]> wrote:
....

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */

Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 6 lines
drivers/mailbox/cix-mailbox.c new file +6 -0
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

Quoted 3 lines

As mentioned in the previous email, a single controller can support multiple different channels.

OK. I am not too worried about having all variants in one driver esp when it is manageable and share the code.
Unless I am overlooking something. Arnd?

Quoted 11 lines
Quoted 11 lines

drivers/mailbox/cix-mailbox.c line 134

Quoted 8 lines
drivers/mailbox/cix-mailbox.c new file +8 -0
+static u32 cix_mbox_read(struct cix_mbox_priv *priv, u32 offset)
+{
+       if (priv->use_shmem)
+               return ioread32(priv->base + offset - SHMEM_OFFSET);
+       else
+               return ioread32(priv->base + offset);
+}
+

use_shmem is set for only CIX_MBOX_TYPE_DB, but it affects every read/write. Maybe instead adjust the base for TYPE_DB?

Quoted 4 lines

The reason we introduced use_shmem here is that we had to adjust the base address of TYPE_DB to resolve the reg conflict in the DTS. This change has virtually no impact on performance.

Yes, I can see it should have no impact on performance and I think adjusting the base once
during init is cleaner than checking the flag every read/write. But wait... use_shmem is a controller wide flag, and isn't priv->use_shmem always set to true in cix_mbox_init() ? Is the driver even tested?
....

Quoted 13 lines
Quoted 4 lines
Quoted 4 lines
drivers/mailbox/cix-mailbox.c new file +4 -0
+static int cix_mbox_startup(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 410

Quoted 5 lines
drivers/mailbox/cix-mailbox.c new file +5 -0
+       int index = cp->index, ret;
+       u32 val_32;
+
+       ret = request_irq(priv->irq, cix_mbox_isr, 0,
+                         dev_name(priv->dev), chan);

The same irq is requested for each channel. How do you expect it to work? Maybe request it just once in probe and pass the 'priv' instead of 'chan' , and in the cix_mbox_isr handle according to INT_STATUS

Quoted 6 lines

For the same mailbox controller, there won't be multiple channels simultaneously requesting the same IRQ, so there won't be an issue here. As you mentioned, if we need to handle multiple channels working concurrently in the future, we would need to modify cix_mbox_isr. However, that is not required at the moment.

Is it too hard to do it right already?

....

Quoted 6 lines
Quoted 6 lines

drivers/mailbox/cix-mailbox.c line 635

Quoted 3 lines
drivers/mailbox/cix-mailbox.c new file +3 -0
+MODULE_AUTHOR("Cix Technology Group Co., Ltd.");
+MODULE_DESCRIPTION("CIX mailbox driver");
+MODULE_LICENSE("GPL");

GPL v2 ? according to the SPDX-License-Identifier

Quoted 7 lines
Quoted 1 line

And please make sure you run scripts/checkpatch.pl

Yes, I'm also confused here. I was previously using GPL v2, but when I ran checkpatch.pl, it triggered a warning due to commit bf7fbeeae6db ("module: Cure the MODULE_LICENSE 'GPL' vs. 'GPL v2' bogosity"). So I changed it to GPL.

Sorry, I was wrong. It should simply be GPL

Thanks.

#13 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 43 lines · Jassi Brar

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL

On Tue, Jul 8, 2025 at 8:54 AM Guomin chen <[email protected]> wrote:
....

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */

Quoted 9 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 6 lines
drivers/mailbox/cix-mailbox.c new file +6 -0
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

Quoted 3 lines

As mentioned in the previous email, a single controller can support multiple different channels.

OK. I am not too worried about having all variants in one driver esp when it is manageable and share the code.
Unless I am overlooking something. Arnd?

Quoted 11 lines
Quoted 11 lines

drivers/mailbox/cix-mailbox.c line 134

Quoted 8 lines
drivers/mailbox/cix-mailbox.c new file +8 -0
+static u32 cix_mbox_read(struct cix_mbox_priv *priv, u32 offset)
+{
+       if (priv->use_shmem)
+               return ioread32(priv->base + offset - SHMEM_OFFSET);
+       else
+               return ioread32(priv->base + offset);
+}
+

use_shmem is set for only CIX_MBOX_TYPE_DB, but it affects every read/write. Maybe instead adjust the base for TYPE_DB?

Quoted 4 lines

The reason we introduced use_shmem here is that we had to adjust the base address of TYPE_DB to resolve the reg conflict in the DTS. This change has virtually no impact on performance.

Quoted 7 lines

Yes, I can see it should have no impact on performance and I think adjusting the base once
during init is cleaner than checking the flag every read/write. But wait... use_shmem is a controller wide flag, and isn't priv->use_shmem always set to true in cix_mbox_init() ? Is the driver even tested?
....

Yes, we did perform testing before sending out the patch. The test here shows no issues because there aren’t more clients. There indeed exists a problem with priv->use_shmem always being set to true in cix_mbox_init().
So I will add a restriction in the probe function in the next version.

Quoted 13 lines
Quoted 13 lines
Quoted 13 lines

drivers/mailbox/cix-mailbox.c line 410

Quoted 9 lines
drivers/mailbox/cix-mailbox.c new file +9 -0
+static int cix_mbox_startup(struct mbox_chan *chan)
+{
+       struct cix_mbox_priv *priv = to_cix_mbox_priv(chan->mbox);
+       struct cix_mbox_con_priv *cp = chan->con_priv;
+       int index = cp->index, ret;
+       u32 val_32;
+
+       ret = request_irq(priv->irq, cix_mbox_isr, 0,
+                         dev_name(priv->dev), chan);

The same irq is requested for each channel. How do you expect it to work? Maybe request it just once in probe and pass the 'priv' instead of 'chan' , and in the cix_mbox_isr handle according to INT_STATUS

Quoted 8 lines
Quoted 6 lines

For the same mailbox controller, there won't be multiple channels simultaneously requesting the same IRQ, so there won't be an issue here. As you mentioned, if we need to handle multiple channels working concurrently in the future, we would need to modify cix_mbox_isr. However, that is not required at the moment.

Is it too hard to do it right already?

We haven't set the IRQF_SHARED flag here because there is no scenario where a single mailbox controller supports multiple channel clients simultaneously. And, the ISR still requires the channel as an argument.
While this approach does not support multiple clients in parallel, it does allow for better sequential support of multiple clients.

So we have placed it in the client's startup/shutdown lifecycle rather than in the probe function.

Thanks
Guomin Chen

maintainer
#14 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

In reply to Jassi Brar
Quoted 16 lines · Jassi Brar

On Tue, Jul 8, 2025 at 8:54 AM Guomin chen <[email protected]> wrote:
....

Quoted 14 lines
Quoted 14 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 11 lines
drivers/mailbox/cix-mailbox.c new file +11 -0
+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

Quoted 6 lines
Quoted 3 lines

As mentioned in the previous email, a single controller can support multiple different channels.

OK. I am not too worried about having all variants in one driver esp when it is manageable and share the code.
Unless I am overlooking something. Arnd?

My main worry here is that the types are all quite different: while the doorbell and fast mailboxes are what a lot of other drivers have, the FIFO mode does not seem to be a good fit for the mailbox subsystem but instead looks like a more generic firmware interface with variable length messages.

For those, I think a higher-level driver with fixed data structures passed through the hardware interface seems more appropriate.

Are there any other mailbox drivers that just use the mailbox to tunnel variable-length messages?

Arnd

maintainer
#15 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 25 lines · Arnd Bergmann
Quoted 16 lines · Jassi Brar

On Tue, Jul 8, 2025 at 8:54 AM Guomin chen <[email protected]> wrote:
....

Quoted 14 lines
Quoted 14 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 11 lines
drivers/mailbox/cix-mailbox.c new file +11 -0
+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

Quoted 6 lines
Quoted 3 lines

As mentioned in the previous email, a single controller can support multiple different channels.

OK. I am not too worried about having all variants in one driver esp when it is manageable and share the code.
Unless I am overlooking something. Arnd?

Quoted 9 lines

My main worry here is that the types are all quite different: while the doorbell and fast mailboxes are what a lot of other drivers have, the FIFO mode does not seem to be a good fit for the mailbox subsystem but instead looks like a more generic firmware interface with variable length messages.

For those, I think a higher-level driver with fixed data structures passed through the hardware interface seems more appropriate.

Yes. But sometimes when the data structures of a protocol are not bigger than FIFO depth, the platform may choose to use the FIFO mode. I see it as platform dependent.

Quoted 3 lines

Are there any other mailbox drivers that just use the mailbox to tunnel variable-length messages?

From a quick look, Armada 37xx and Hi6220 have fifo though they fill them up fully for each transfer.

Thanks

#16 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 43 lines · Jassi Brar

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL

Quoted 25 lines · Arnd Bergmann
Quoted 16 lines · Jassi Brar

On Tue, Jul 8, 2025 at 8:54 AM Guomin chen <[email protected]> wrote:
....

Quoted 14 lines
Quoted 14 lines

drivers/mailbox/cix-mailbox.c line 49

Quoted 11 lines
drivers/mailbox/cix-mailbox.c new file +11 -0
+/* [0~7] Fast channel
+ * [8] doorbell base channel
+ * [9]fifo base channel
+ * [10] register base channel
+ */
+#define MBOX_FAST_IDX          7
+#define MBOX_DB_IDX            8
+#define MBOX_FIFO_IDX          9
+#define MBOX_REG_IDX           10
+#define CIX_MBOX_CHANS         11
+

if it is not really a single controller owning different channels, maybe implement only what you currently use.

Quoted 6 lines
Quoted 3 lines

As mentioned in the previous email, a single controller can support multiple different channels.

OK. I am not too worried about having all variants in one driver esp when it is manageable and share the code.
Unless I am overlooking something. Arnd?

Quoted 9 lines

My main worry here is that the types are all quite different: while the doorbell and fast mailboxes are what a lot of other drivers have, the FIFO mode does not seem to be a good fit for the mailbox subsystem but instead looks like a more generic firmware interface with variable length messages.

For those, I think a higher-level driver with fixed data structures passed through the hardware interface seems more appropriate.

Yes. But sometimes when the data structures of a protocol are not bigger than FIFO depth, the platform may choose to use the FIFO mode. I see it as platform dependent.

Quoted 6 lines
Quoted 3 lines

Are there any other mailbox drivers that just use the mailbox to tunnel variable-length messages?

From a quick look, Armada 37xx and Hi6220 have fifo though they fill them up fully for each transfer.

Yes, both Armada 37xx and Hi6220 support FIFO functionality, and they fill the FIFO with each transfer.

Since the cix mailbox hardware supports messages with a maximum length of 128 bytes, different clients transmit messages of varying lengths, such as the cix DSP using 8 bytes, the cix sensorhub using 12 bytes, etc.

Therefore, the cix mailbox driver has been modified to support variable- length messages of up to 128 bytes. This allows for more compact and flexible support of various clients.

Thanks
Guomin Chen

maintainer
#17 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 24 lines · Guomin chen
Quoted 14 lines · Jassi Brar
Quoted 9 lines · Arnd Bergmann

My main worry here is that the types are all quite different: while the doorbell and fast mailboxes are what a lot of other drivers have, the FIFO mode does not seem to be a good fit for the mailbox subsystem but instead looks like a more generic firmware interface with variable length messages.

For those, I think a higher-level driver with fixed data structures passed through the hardware interface seems more appropriate.

Yes. But sometimes when the data structures of a protocol are not bigger than FIFO depth, the platform may choose to use the FIFO mode. I see it as platform dependent.

Quoted 6 lines
Quoted 3 lines

Are there any other mailbox drivers that just use the mailbox to tunnel variable-length messages?

From a quick look, Armada 37xx and Hi6220 have fifo though they fill them up fully for each transfer.

Yes, both Armada 37xx and Hi6220 support FIFO functionality, and they fill the FIFO with each transfer.

Quoted 7 lines

Since the cix mailbox hardware supports messages with a maximum length of 128 bytes, different clients transmit messages of varying lengths, such as the cix DSP using 8 bytes, the cix sensorhub using 12 bytes, etc.

Therefore, the cix mailbox driver has been modified to support variable- length messages of up to 128 bytes. This allows for more compact and flexible support of various clients.

Thanks, this makes sense to me, and I have no other objections if this is an established way to use the subsystem. I wonder if there is a way to abstract it further though, since it would appear that the same thing should be possible on any device that has a FIFO to buffer more than a single fixed-length message.

Jassi, are there any remaining issues on your side that need to be fixed before merging the initial driver? It would be nice if I could merge all nine patches through the soc tree for 6.17 if the current version, or a feature-reduced variant of the mailbox driver is ok.

Arnd

maintainer
#18 lore raw

Re: [PATCH v9 5/9] mailbox: add CIX mailbox driver

Quoted 40 lines · Arnd Bergmann
Quoted 24 lines · Guomin chen
Quoted 14 lines · Jassi Brar
Quoted 9 lines · Arnd Bergmann

My main worry here is that the types are all quite different: while the doorbell and fast mailboxes are what a lot of other drivers have, the FIFO mode does not seem to be a good fit for the mailbox subsystem but instead looks like a more generic firmware interface with variable length messages.

For those, I think a higher-level driver with fixed data structures passed through the hardware interface seems more appropriate.

Yes. But sometimes when the data structures of a protocol are not bigger than FIFO depth, the platform may choose to use the FIFO mode. I see it as platform dependent.

Quoted 6 lines
Quoted 3 lines

Are there any other mailbox drivers that just use the mailbox to tunnel variable-length messages?

From a quick look, Armada 37xx and Hi6220 have fifo though they fill them up fully for each transfer.

Yes, both Armada 37xx and Hi6220 support FIFO functionality, and they fill the FIFO with each transfer.

Quoted 7 lines

Since the cix mailbox hardware supports messages with a maximum length of 128 bytes, different clients transmit messages of varying lengths, such as the cix DSP using 8 bytes, the cix sensorhub using 12 bytes, etc.

Therefore, the cix mailbox driver has been modified to support variable- length messages of up to 128 bytes. This allows for more compact and flexible support of various clients.

Thanks, this makes sense to me, and I have no other objections if this is an established way to use the subsystem. I wonder if there is a way to abstract it further though, since it would appear that the same thing should be possible on any device that has a FIFO to buffer more than a single fixed-length message.

Quoted 6 lines

Jassi, are there any remaining issues on your side that need to be fixed before merging the initial driver? It would be nice if I could merge all nine patches through the soc tree for 6.17 if the current version, or a feature-reduced variant of the mailbox driver is ok.

I have pointed out some, and Guomin said there is a revision coming. That should be good.

-j

author
#19 lore raw

PATCH v9 6/9 arm64: defconfig: Enable CIX SoC

In reply to Peter Chen

- Enable CIX SoC support at ARM64 defconfig
- Enable CIX mailbox
At CIX SoC platforms, the clock handling uses Arm SCMI protocol, the physical clock access is at sub processor, so it needs to enable mailbox by default.

---
Changes for v8:
- Add Krzysztof Kozlowski's Reviewed-by tag Changes for v7:
- Add both CIX SoC and mailbox configuration at one patch

 arch/arm64/configs/defconfig | 2 ++
 1 file changed, 2 insertions(+)
arch/arm64/configs/defconfig +2 -0
index 897fc686e6a9..181c5f91b032 100644
@@ -45,6 +45,7 @@ CONFIG_ARCH_BCMBCA=y
 CONFIG_ARCH_BRCMSTB=y
 CONFIG_ARCH_BERLIN=y
 CONFIG_ARCH_BLAIZE=y
+CONFIG_ARCH_CIX=y
 CONFIG_ARCH_EXYNOS=y
 CONFIG_ARCH_SPARX5=y
 CONFIG_ARCH_K3=y
@@ -1445,6 +1446,7 @@ CONFIG_BCM2835_MBOX=y
 CONFIG_QCOM_APCS_IPC=y
 CONFIG_MTK_ADSP_MBOX=m
 CONFIG_QCOM_IPCC=y
+CONFIG_CIX_MBOX=y
 CONFIG_ROCKCHIP_IOMMU=y
 CONFIG_TEGRA_IOMMU_SMMU=y
 CONFIG_ARM_SMMU=y
Signature · 1 line
2.25.1
author
#20 lore raw

PATCH v9 7/9 dt-bindings: clock: cix: Add CIX sky1 scmi clock id

In reply to Peter Chen

From: Gary Yang <[email protected]>

Add device tree bindings for the scmi clock id on
Cix sky1 platform.

---
Changes for v8:
- Add Krzysztof Kozlowski's Acked-by tag

Changes for v7:
- Rename clock binding file from sky1-clk.h to cix,sky1.h
- Add my Sob

 include/dt-bindings/clock/cix,sky1.h | 279 +++++++++++++++++++++++++++
 1 file changed, 279 insertions(+)
 create mode 100644 include/dt-bindings/clock/cix,sky1.h
include/dt-bindings/clock/cix,sky1.h new file +279 -0
new file mode 100644
index 000000000000..9245ebd1e80a
@@ -0,0 +1,279 @@
+/* SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) */
+/*
+ * Copyright 2024-2025 Cix Technology Group Co., Ltd.
+ */
+
+#ifndef _DT_BINDINGS_CLK_CIX_SKY1_H
+#define _DT_BINDINGS_CLK_CIX_SKY1_H
+
+#define CLK_TREE_CPU_GICxCLK			0
+#define CLK_TREE_CPU_PPUCLK			1
+#define CLK_TREE_CPU_PERIPHCLK			2
+#define CLK_TREE_DSU_CLK			3
+#define CLK_TREE_DSU_PCLK			4
+#define CLK_TREE_CPU_CLK_BC0			5
+#define CLK_TREE_CPU_CLK_BC1			6
+#define CLK_TREE_CPU_CLK_BC2			7
+#define CLK_TREE_CPU_CLK_BC3			8
+#define CLK_TREE_CPU_CLK_MC0			9
+#define CLK_TREE_CPU_CLK_MC1			10
+#define CLK_TREE_CPU_CLK_MC2			11
+#define CLK_TREE_CPU_CLK_MC3			12
+#define CLK_TREE_CPU_CLK_LC0			13
+#define CLK_TREE_CPU_CLK_LC1			14
+#define CLK_TREE_CPU_CLK_LC2			15
+#define CLK_TREE_CPU_CLK_LC3			16
+#define CLK_TREE_CSI_CTRL0_PCLK			17
+#define CLK_TREE_CSI_CTRL1_PCLK			18
+#define CLK_TREE_CSI_CTRL2_PCLK			19
+#define CLK_TREE_CSI_CTRL3_PCLK			20
+#define CLK_TREE_CSI_DMA0_PCLK			21
+#define CLK_TREE_CSI_DMA1_PCLK			22
+#define CLK_TREE_CSI_DMA2_PCLK			23
+#define CLK_TREE_CSI_DMA3_PCLK			24
+#define CLK_TREE_CSI_PHY0_PSM			25
+#define CLK_TREE_CSI_PHY1_PSM			26
+#define CLK_TREE_CSI_PHY0_APBCLK		27
+#define CLK_TREE_CSI_PHY1_APBCLK		28
+#define CLK_TREE_FCH_APB_CLK			29
+#define CLK_TREE_GPU_CLK_400M			30
+#define CLK_TREE_GPU_CLK_CORE			31
+#define CLK_TREE_GPU_CLK_STACKS			32
+#define CLK_TREE_DP0_PIXEL0			33
+#define CLK_TREE_DP0_PIXEL1			34
+#define CLK_TREE_DP1_PIXEL0			35
+#define CLK_TREE_DP1_PIXEL1			36
+#define CLK_TREE_DP2_PIXEL0			37
+#define CLK_TREE_DP2_PIXEL1			38
+#define CLK_TREE_DP3_PIXEL0			39
+#define CLK_TREE_DP3_PIXEL1			40
+#define CLK_TREE_DP4_PIXEL0			41
+#define CLK_TREE_DP4_PIXEL1			42
+#define CLK_TREE_DPU_CLK			43
+#define CLK_TREE_DPU0_ACLK			44
+#define CLK_TREE_DPU1_ACLK			45
+#define CLK_TREE_DPU2_ACLK			46
+#define CLK_TREE_DPU3_ACLK			47
+#define CLK_TREE_DPU4_ACLK			48
+#define CLK_TREE_DPC0_VIDCLK0			49
+#define CLK_TREE_DPC0_VIDCLK1			50
+#define CLK_TREE_DPC1_VIDCLK0			51
+#define CLK_TREE_DPC1_VIDCLK1			52
+#define CLK_TREE_DPC2_VIDCLK0			53
+#define CLK_TREE_DPC2_VIDCLK1			54
+#define CLK_TREE_DPC3_VIDCLK0			55
+#define CLK_TREE_DPC3_VIDCLK1			56
+#define CLK_TREE_DPC4_VIDCLK0			57
+#define CLK_TREE_DPC4_VIDCLK1			58
+#define CLK_TREE_DPC0_APBCLK			59
+#define CLK_TREE_DPC1_APBCLK			60
+#define CLK_TREE_DPC2_APBCLK			61
+#define CLK_TREE_DPC3_APBCLK			62
+#define CLK_TREE_DPC4_APBCLK			63
+#define CLK_TREE_NPU_MEMCLK			64
+#define CLK_TREE_NPU_SYSCLK			65
+#define CLK_TREE_NPU_DBGCLK			66
+#define CLK_TREE_VPU_APBCLK			67
+#define CLK_TREE_ISP_ACLK			68
+#define CLK_TREE_ISP_SCLK			69
+#define CLK_TREE_AUDIO_CLK4			70
+#define CLK_TREE_AUDIO_CLK5			71
+#define CLK_TREE_CAMERA_MCLK0			72
+#define CLK_TREE_CAMERA_MCLK1			73
+#define CLK_TREE_CAMERA_MCLK2			74
+#define CLK_TREE_CAMERA_MCLK3			75
+#define CLK_TREE_AUDIO_CLK0			76
+#define CLK_TREE_AUDIO_CLK1			77
+#define CLK_TREE_AUDIO_CLK2			78
+#define CLK_TREE_AUDIO_CLK3			79
+#define CLK_TREE_MM_NI700_CLK			80
+#define CLK_TREE_SYS_NI700_CLK			81
+#define CLK_TREE_GMAC0_ACLK			82
+#define CLK_TREE_GMAC1_ACLK			83
+#define CLK_TREE_GMAC0_DIV_ACLK			84
+#define CLK_TREE_GMAC0_DIV_TXCLK		85
+#define CLK_TREE_GMAC0_RGMII0_TXCLK		86
+#define CLK_TREE_GMAC1_DIV_ACLK			87
+#define CLK_TREE_GMAC1_DIV_TXCLK		88
+#define CLK_TREE_GMAC1_RGMII0_TXCLK		89
+#define CLK_TREE_GMAC0_PCLK			90
+#define CLK_TREE_GMAC1_PCLK			91
+#define CLK_TREE_USB2_0_AXI_GATE		92
+#define CLK_TREE_USB2_0_APB_GATE		93
+#define CLK_TREE_USB2_1_AXI_GATE		94
+#define CLK_TREE_USB2_1_APB_GATE		95
+#define CLK_TREE_USB2_2_AXI_GATE		96
+#define CLK_TREE_USB2_2_APB_GATE		97
+#define CLK_TREE_USB2_3_AXI_GATE		98
+#define CLK_TREE_USB2_3_APB_GATE		99
+#define CLK_TREE_USB2_0_PHY_GATE		100
+#define CLK_TREE_USB2_1_PHY_GATE		101
+#define CLK_TREE_USB2_2_PHY_GATE		102
+#define CLK_TREE_USB2_3_PHY_GATE		103
+#define CLK_TREE_USB3C_DRD_AXI_GATE		104
+#define CLK_TREE_USB3C_DRD_APB_GATE		105
+#define CLK_TREE_USB3C_DRD_PHY2_GATE		106
+#define CLK_TREE_USB3C_DRD_PHY3_GATE		107
+#define CLK_TREE_USB3C_0_AXI_GATE		108
+#define CLK_TREE_USB3C_0_APB_GATE		109
+#define CLK_TREE_USB3C_0_PHY2_GATE		110
+#define CLK_TREE_USB3C_0_PHY3_GATE		111
+#define CLK_TREE_USB3C_1_AXI_GATE		112
+#define CLK_TREE_USB3C_1_APB_GATE		113
+#define CLK_TREE_USB3C_1_PHY2_GATE		114
+#define CLK_TREE_USB3C_1_PHY3_GATE		115
+#define CLK_TREE_USB3C_2_AXI_GATE		116
+#define CLK_TREE_USB3C_2_APB_GATE		117
+#define CLK_TREE_USB3C_2_PHY2_GATE		118
+#define CLK_TREE_USB3C_2_PHY3_GATE		119
+#define CLK_TREE_USB3A_0_AXI_GATE		120
+#define CLK_TREE_USB3A_0_APB_GATE		121
+#define CLK_TREE_USB3A_0_PHY2_GATE		122
+#define CLK_TREE_USB3A_1_AXI_GATE		123
+#define CLK_TREE_USB3A_1_APB_GATE		124
+#define CLK_TREE_USB3A_1_PHY2_GATE		125
+#define CLK_TREE_USB3A_PHY3_GATE		126
+#define CLK_TREE_USB2_0_CLK_SOF			127
+#define CLK_TREE_USB2_1_CLK_SOF			128
+#define CLK_TREE_USB2_2_CLK_SOF			129
+#define CLK_TREE_USB2_3_CLK_SOF			130
+#define CLK_TREE_USB3C_DRD_CLK_SOF		131
+#define CLK_TREE_USB3C_H0_CLK_SOF		132
+#define CLK_TREE_USB3C_H1_CLK_SOF		133
+#define CLK_TREE_USB3C_H2_CLK_SOF		134
+#define CLK_TREE_USB3A_H0_CLK_SOF		135
+#define CLK_TREE_USB3A_H1_CLK_SOF		136
+#define CLK_TREE_USB2_0_CLK_LPM			137
+#define CLK_TREE_USB2_1_CLK_LPM			138
+#define CLK_TREE_USB2_2_CLK_LPM			139
+#define CLK_TREE_USB2_3_CLK_LPM			140
+#define CLK_TREE_USB3C_DRD_CLK_LPM		141
+#define CLK_TREE_USB3C_H0_CLK_LPM		142
+#define CLK_TREE_USB3C_H1_CLK_LPM		143
+#define CLK_TREE_USB3C_H2_CLK_LPM		144
+#define CLK_TREE_USB3A_H0_CLK_LPM		145
+#define CLK_TREE_USB3A_H1_CLK_LPM		146
+#define CLK_TREE_USB2_0_PHY_REF			147
+#define CLK_TREE_USB2_1_PHY_REF			148
+#define CLK_TREE_USB2_2_PHY_REF			149
+#define CLK_TREE_USB2_3_PHY_REF			150
+#define CLK_TREE_USB3C_DRD_PHY_REF		151
+#define CLK_TREE_USB3C_H0_PHY_REF		152
+#define CLK_TREE_USB3C_H1_PHY_REF		153
+#define CLK_TREE_USB3C_H2_PHY_REF		154
+#define CLK_TREE_USB3A_H0_PHY_REF		155
+#define CLK_TREE_USB3A_H1_PHY_REF		156
+#define CLK_TREE_USB3C_DRD_PHY_x4_REF		157
+#define CLK_TREE_USB3C_H0_PHY_x4_REF		158
+#define CLK_TREE_USB3C_H1_PHY_x4_REF		159
+#define CLK_TREE_USB3C_H2_PHY_x4_REF		160
+#define CLK_TREE_USB3A_PHY_x2_REF		161
+#define CLK_TREE_PCIE_X8CTRL_APB		162
+#define CLK_TREE_PCIE_X4CTRL_APB		163
+#define CLK_TREE_PCIE_X2CTRL_APB		164
+#define CLK_TREE_PCIE_X1_0CTRL_APB		165
+#define CLK_TREE_PCIE_X1_1CTRL_APB		166
+#define CLK_TREE_PCIE_X8_PHY_APB		167
+#define CLK_TREE_PCIE_X4_PHY_APB		168
+#define CLK_TREE_PCIE_X211_PHY_APB		169
+#define CLK_TREE_PCIE_NI700_CLK			170
+#define CLK_TREE_PCIE_CTRL0_CLK			171
+#define CLK_TREE_PCIE_CTRL1_CLK			172
+#define CLK_TREE_PCIE_CTRL2_CLK			173
+#define CLK_TREE_PCIE_CTRL3_CLK			174
+#define CLK_TREE_PCIE_CTRL4_CLK			175
+#define CLK_TREE_CSI_CTRL0_SYSCLK		176
+#define CLK_TREE_CSI_CTRL1_SYSCLK		177
+#define CLK_TREE_CSI_CTRL2_SYSCLK		178
+#define CLK_TREE_CSI_CTRL3_SYSCLK		179
+#define CLK_TREE_CSI_CTRL0_PIXEL0_CLK		180
+#define CLK_TREE_CSI_CTRL0_PIXEL1_CLK		181
+#define CLK_TREE_CSI_CTRL0_PIXEL2_CLK		182
+#define CLK_TREE_CSI_CTRL0_PIXEL3_CLK		183
+#define CLK_TREE_CSI_CTRL1_PIXEL0_CLK		184
+#define CLK_TREE_CSI_CTRL2_PIXEL0_CLK		185
+#define CLK_TREE_CSI_CTRL2_PIXEL1_CLK		186
+#define CLK_TREE_CSI_CTRL2_PIXEL2_CLK		187
+#define CLK_TREE_CSI_CTRL2_PIXEL3_CLK		188
+#define CLK_TREE_CSI_CTRL3_PIXEL0_CLK		189
+#define CLK_TREE_CI700_GCLK0			190
+#define CLK_TREE_DDRC0_ACLK_CLK			191
+#define CLK_TREE_DDRC1_ACLK_CLK			192
+#define CLK_TREE_DDRC2_ACLK_CLK			193
+#define CLK_TREE_DDRC3_ACLK_CLK			194
+#define CLK_TREE_DDRC0_DFICLK_CLK		195
+#define CLK_TREE_DDRC1_DFICLK_CLK		196
+#define CLK_TREE_DDRC2_DFICLK_CLK		197
+#define CLK_TREE_DDRC3_DFICLK_CLK		198
+#define CLK_TREE_PHY0_SYNC_CLK			199
+#define CLK_TREE_PHY1_SYNC_CLK			200
+#define CLK_TREE_PHY2_SYNC_CLK			201
+#define CLK_TREE_PHY3_SYNC_CLK			202
+#define CLK_TREE_PHY0_BYPASS_CLK		203
+#define CLK_TREE_PHY1_BYPASS_CLK		204
+#define CLK_TREE_PHY2_BYPASS_CLK		205
+#define CLK_TREE_PHY3_BYPASS_CLK		206
+#define CLK_TREE_DDRC_0_APB			207
+#define CLK_TREE_DDRC_1_APB			208
+#define CLK_TREE_DDRC_2_APB			209
+#define CLK_TREE_DDRC_3_APB			210
+#define CLK_TREE_TZC400_0_APB			211
+#define CLK_TREE_TZC400_1_APB			212
+#define CLK_TREE_TZC400_2_APB			213
+#define CLK_TREE_TZC400_3_APB			214
+#define CLK_TREE_S5_SENSOR_HUB_25M		215
+#define CLK_TREE_S5_SENSOR_HUB_400M		216
+#define CLK_TREE_S5_CSS600_100M			217
+#define CLK_TREE_S5_DFD_800M			218
+#define CLK_TREE_S5_CSU_SE_800M			219
+#define CLK_TREE_S5_CSU_PM_800M			220
+#define CLK_TREE_PCIE_REF_B0			221
+#define CLK_TREE_PCIE_REF_B1			222
+#define CLK_TREE_PCIE_REF_B2			223
+#define CLK_TREE_PCIE_REF_B3			224
+#define CLK_TREE_PCIE_REF_B4			225
+#define CLK_TREE_PCIE_REF_PHY_X8		226
+#define CLK_TREE_PCIE_REF_PHY_X4		227
+#define CLK_TREE_PCIE_REF_PHY_X211		228
+#define CLK_TREE_GMAC_REC_CLK			229
+#define CLK_TREE_GPUTOP_PLL			230
+#define CLK_TREE_GPUCORE_PLL			231
+#define CLK_TREE_CPU_PLL_LIT			232
+#define CLK_TREE_CPU_PLL0			233
+#define CLK_TREE_CPU_PLL1			234
+#define CLK_TREE_CPU_PLL2			235
+#define CLK_TREE_CPU_PLL3			236
+#define CLK_TREE_FCH_I3C0_FUNC			237
+#define CLK_TREE_FCH_I3C1_FUNC			238
+#define CLK_TREE_FCH_DMA_ACLK			239
+#define CLK_TREE_FCH_XSPI_FUNC			240
+#define CLK_TREE_FCH_XSPI_MACLK			241
+#define CLK_TREE_FCH_TIMER_FUN			242
+#define CLK_TREE_FCH_APB_IO_S0			243
+#define CLK_TREE_FCH_I3C0_APB			244
+#define CLK_TREE_FCH_I3C1_APB			245
+#define CLK_TREE_FCH_UART0_APB			246
+#define CLK_TREE_FCH_UART1_APB			247
+#define CLK_TREE_FCH_UART2_APB			248
+#define CLK_TREE_FCH_UART3_APB			249
+#define CLK_TREE_FCH_SPI0_APB			250
+#define CLK_TREE_FCH_SPI1_APB			251
+#define CLK_TREE_FCH_XSPI_APB			252
+#define CLK_TREE_FCH_I2C0_APB			253
+#define CLK_TREE_FCH_I2C1_APB			254
+#define CLK_TREE_FCH_I2C2_APB			255
+#define CLK_TREE_FCH_I2C3_APB			256
+#define CLK_TREE_FCH_I2C4_APB			257
+#define CLK_TREE_FCH_I2C5_APB			258
+#define CLK_TREE_FCH_I2C6_APB			259
+#define CLK_TREE_FCH_I2C7_APB			260
+#define CLK_TREE_FCH_TIMER_APB			261
+#define CLK_TREE_FCH_GPIO_APB			262
+#define CLK_TREE_FCH_UART0_FUNC			263
+#define CLK_TREE_FCH_UART1_FUNC			264
+#define CLK_TREE_FCH_UART2_FUNC			265
+#define CLK_TREE_FCH_UART3_FUNC			266
+/* 267~271 not used by AP, skip */
+#define CLK_TREE_GPU_CLK_200M			272
+
+#endif
Signature · 1 line
2.25.1
author
#21 lore raw

PATCH v9 8/9 arm64: dts: cix: Add sky1 base dts initial support

In reply to Peter Chen

CIX SKY1 SoC is high performance Armv9 SoC designed by Cixtech, and Orion O6 is the motherboard launched by Radxa. See below for detail:
https://docs.radxa.com/en/orion/o6/getting-started/introduction

In this commit, it only adds limited components for running initramfs at Orion O6.

---
Changes for v7:
- Refine *_scmi_mem nodes for their properties ordering
- Delete Krzysztof Kozlowski and Fugang Duan's tag due to substantial changes
- Add two Tested-by tags from Enric Balletbo i Serra and Kajetan Puchalski

Changes for v6:
- Add mailbox, scmi and uart support

Changes for v5:
- Delete pmu-spe node which need to refine, and add it in future

Changes for v4:
- Add ppi-partition entry for gic-v3 node, and let pmu-a520 and pmu-a720's interrupt entry get its handle
- Remove gic-v3's #redistributor-regions and redistributor-stride properties
- Change gic-v3's #interrupt-cells as 4, and change all interrupt specifiers accordingly
- Remove "arm,no-tick-in-suspend" for timer due to global counter is at always-on power domain
- Remove timer's clock frequency due to firmware has already set it
- Add Krzysztof Kozlowski's reviewed-by

Changes for v3:
- Fix two dts coding sytle issues

Changes for v2:
- Corrects the SoF tag's name
- Fix several coding sytle issues
- move linux,cma node to dts file
- delete memory node, memory size is passed by firmware
- delete uart2 node which will be added in future patches
- Improve for pmu and cpu node to stands for more specific cpu model
- Improve the timer node and add hypervisor virtual timer irq
- Pass "make O=$OUTKNL CHECK_DTBS=y W=1 cix/sky1-orion-o6.dtb"

 arch/arm64/boot/dts/Makefile              |   1 +
 arch/arm64/boot/dts/cix/Makefile          |   2 +
 arch/arm64/boot/dts/cix/sky1-orion-o6.dts |  39 +++
 arch/arm64/boot/dts/cix/sky1.dtsi         | 331 ++++++++++++++++++++++
 4 files changed, 373 insertions(+)
 create mode 100644 arch/arm64/boot/dts/cix/Makefile
 create mode 100644 arch/arm64/boot/dts/cix/sky1-orion-o6.dts
 create mode 100644 arch/arm64/boot/dts/cix/sky1.dtsi
arch/arm64/boot/dts/Makefile +1 -0
index 79b73a21ddc2..8e7ccd0027bd 100644
@@ -13,6 +13,7 @@ subdir-y += bitmain
 subdir-y += blaize
 subdir-y += broadcom
 subdir-y += cavium
+subdir-y += cix
 subdir-y += exynos
 subdir-y += freescale
 subdir-y += hisilicon
arch/arm64/boot/dts/cix/Makefile new file +2 -0
new file mode 100644
index 000000000000..ed3713982012
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+dtb-$(CONFIG_ARCH_CIX) += sky1-orion-o6.dtb
arch/arm64/boot/dts/cix/sky1-orion-o6.dts new file +39 -0
new file mode 100644
index 000000000000..d74964d53c3b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: BSD-3-Clause
+/*
+ * Copyright 2025 Cix Technology Group Co., Ltd.
+ *
+ */
+
+/dts-v1/;
+
+#include "sky1.dtsi"
+/ {
+	model = "Radxa Orion O6";
+	compatible = "radxa,orion-o6", "cix,sky1";
+
+	aliases {
+		serial2 = &uart2;
+	};
+
+	chosen {
+		stdout-path = &uart2;
+	};
+
+	reserved-memory {
+		#address-cells = <2>;
+		#size-cells = <2>;
+		ranges;
+
+		linux,cma {
+			compatible = "shared-dma-pool";
+			reusable;
+			size = <0x0 0x28000000>;
+			linux,cma-default;
+		};
+	};
+
+};
+
+&uart2 {
+	status = "okay";
+};
arch/arm64/boot/dts/cix/sky1.dtsi new file +331 -0

331 lines not on this page · raw message

Signature · 1 line
2.25.1
#22 lore raw

Re: [PATCH v9 8/9] arm64: dts: cix: Add sky1 base dts initial support

arch/arm64/boot/dts/cix/sky1.dtsi line 162

Quoted 7 lines · Peter Chen
arch/arm64/boot/dts/cix/sky1.dtsi new file +7 -0
+
+	firmware {
+		ap_to_pm_scmi: scmi {
+			compatible = "arm,scmi";
+			mbox-names = "tx", "rx";
+			mboxes = <&mbox_ap2pm 8>, <&mbox_pm2ap 8>;
+			shmem = <&ap2pm_scmi_mem &pm2ap_scmi_mem>;

These are two entries, so two <>.

arch/arm64/boot/dts/cix/sky1.dtsi line 170

Quoted 8 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +8 -0
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			scmi_clk: protocol@14 {
+				reg = <0x14>;
+				#clock-cells = <1>;
+			};
+

Drop blank line

Quoted 13 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +13 -0
+		};
+	};
+
+	pmu-a520 {
+		compatible = "arm,cortex-a520-pmu";
+		interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_LOW &ppi_partition0>;
+	};
+
+	pmu-a720 {
+		compatible = "arm,cortex-a720-pmu";
+		interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_LOW &ppi_partition1>;
+	};
+

...

Quoted 8 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +8 -0
+
+		mbox_ap2se: mailbox@5060000 {
+			compatible = "cix,sky1-mbox";
+			reg = <0x0 0x05060000 0x0 0x10000>;
+			interrupts = <GIC_SPI 378 IRQ_TYPE_LEVEL_HIGH 0>;
+			#mbox-cells = <1>;
+			cix,mbox-dir = "tx";
+		};

arch/arm64/boot/dts/cix/sky1.dtsi line 248

Quoted 10 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +10 -0
+
+		mbox_se2ap: mailbox@5070000 {
+			compatible = "cix,sky1-mbox";
+			reg = <0x0 0x05070000 0x0 0x10000>;
+			interrupts = <GIC_SPI 379 IRQ_TYPE_LEVEL_HIGH 0>;
+			#mbox-cells = <1>;
+			cix,mbox-dir = "rx";
+		};
+
+		ap2pm_scmi_mem: ap2pm-shmem@6590000 {

This should be just shmem@

Quoted 4 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +4 -0
+			compatible = "arm,scmi-shmem";
+			reg = <0x0 0x06590000 0x0 0x80>;
+			reg-io-width = <4>;
+		};

arch/arm64/boot/dts/cix/sky1.dtsi line 262

Quoted 10 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +10 -0
+
+		mbox_ap2pm: mailbox@6590080 {
+			compatible = "cix,sky1-mbox";
+			reg = <0x0 0x06590080 0x0 0xff80>;
+			interrupts = <GIC_SPI 363 IRQ_TYPE_LEVEL_HIGH 0>;
+			#mbox-cells = <1>;
+			cix,mbox-dir = "tx";
+		};
+
+		pm2ap_scmi_mem: pm2ap-shmem@65a0000 {

Same here

arch/arm64/boot/dts/cix/sky1.dtsi line 266

Quoted 4 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +4 -0
+			compatible = "arm,scmi-shmem";
+			reg = <0x0 0x065a0000 0x0 0x80>;
+			reg-io-width = <4>;
+		};

Best regards,
Krzysztof

author
#23 lore raw

Re: [PATCH v9 8/9] arm64: dts: cix: Add sky1 base dts initial support

On 25-07-05 10:20:34, Krzysztof Kozlowski wrote:

Thanks, Krzysztof.

All below commands will be addressed at next revision patch set.

Peter

Quoted 84 lines · Peter Chen

arch/arm64/boot/dts/cix/sky1.dtsi line 162

Quoted 7 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +7 -0
+
+     firmware {
+             ap_to_pm_scmi: scmi {
+                     compatible = "arm,scmi";
+                     mbox-names = "tx", "rx";
+                     mboxes = <&mbox_ap2pm 8>, <&mbox_pm2ap 8>;
+                     shmem = <&ap2pm_scmi_mem &pm2ap_scmi_mem>;

These are two entries, so two <>.

arch/arm64/boot/dts/cix/sky1.dtsi line 170

Quoted 8 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +8 -0
+                     #address-cells = <1>;
+                     #size-cells = <0>;
+
+                     scmi_clk: protocol@14 {
+                             reg = <0x14>;
+                             #clock-cells = <1>;
+                     };
+

Drop blank line

Quoted 13 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +13 -0
+             };
+     };
+
+     pmu-a520 {
+             compatible = "arm,cortex-a520-pmu";
+             interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_LOW &ppi_partition0>;
+     };
+
+     pmu-a720 {
+             compatible = "arm,cortex-a720-pmu";
+             interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_LOW &ppi_partition1>;
+     };
+

...

Quoted 8 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +8 -0
+
+             mbox_ap2se: mailbox@5060000 {
+                     compatible = "cix,sky1-mbox";
+                     reg = <0x0 0x05060000 0x0 0x10000>;
+                     interrupts = <GIC_SPI 378 IRQ_TYPE_LEVEL_HIGH 0>;
+                     #mbox-cells = <1>;
+                     cix,mbox-dir = "tx";
+             };

arch/arm64/boot/dts/cix/sky1.dtsi line 248

Quoted 10 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +10 -0
+
+             mbox_se2ap: mailbox@5070000 {
+                     compatible = "cix,sky1-mbox";
+                     reg = <0x0 0x05070000 0x0 0x10000>;
+                     interrupts = <GIC_SPI 379 IRQ_TYPE_LEVEL_HIGH 0>;
+                     #mbox-cells = <1>;
+                     cix,mbox-dir = "rx";
+             };
+
+             ap2pm_scmi_mem: ap2pm-shmem@6590000 {

This should be just shmem@

Quoted 4 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +4 -0
+                     compatible = "arm,scmi-shmem";
+                     reg = <0x0 0x06590000 0x0 0x80>;
+                     reg-io-width = <4>;
+             };

arch/arm64/boot/dts/cix/sky1.dtsi line 262

Quoted 10 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +10 -0
+
+             mbox_ap2pm: mailbox@6590080 {
+                     compatible = "cix,sky1-mbox";
+                     reg = <0x0 0x06590080 0x0 0xff80>;
+                     interrupts = <GIC_SPI 363 IRQ_TYPE_LEVEL_HIGH 0>;
+                     #mbox-cells = <1>;
+                     cix,mbox-dir = "tx";
+             };
+
+             pm2ap_scmi_mem: pm2ap-shmem@65a0000 {

Same here

arch/arm64/boot/dts/cix/sky1.dtsi line 266

Quoted 4 lines
arch/arm64/boot/dts/cix/sky1.dtsi new file +4 -0
+                     compatible = "arm,scmi-shmem";
+                     reg = <0x0 0x065a0000 0x0 0x80>;
+                     reg-io-width = <4>;
+             };

Best regards,
Krzysztof

Signature · 2 lines
Best regards,
Peter
author
#24 lore raw

PATCH v9 9/9 MAINTAINERS: Add CIX SoC maintainer entry

In reply to Peter Chen

Using this entry as the maintainers information for CIX SoCs.

---
Changes for v9:
- Add mailbox driver information

 MAINTAINERS | 13 +++++++++++++
 1 file changed, 13 insertions(+)
MAINTAINERS +13 -0
index a92290fffa16..7f8bee29bb8f 100644
@@ -2473,6 +2473,19 @@ F:	arch/arm/boot/compressed/misc-ep93xx.h
 F:	arch/arm/mach-ep93xx/
 F:	drivers/iio/adc/ep93xx_adc.c
 
+ARM/CIX SOC SUPPORT
+M:	Peter Chen <[email protected]>
+M:	Fugang Duan <[email protected]>
+R:	CIX Linux Kernel Upstream Group <[email protected]>
+L:	[email protected] (moderated for non-subscribers)
+S:	Maintained
+T:	git git://git.kernel.org/pub/scm/linux/kernel/git/peter.chen/cix.git
+F:	Documentation/devicetree/bindings/arm/cix.yaml
+F:	Documentation/devicetree/bindings/mailbox/cix,sky1-mbox.yaml
+F:	arch/arm64/boot/dts/cix/
+F:	drivers/mailbox/cix-mailbox.c
+K:	\bcix\b
+
 ARM/CLKDEV SUPPORT
 M:	Russell King <[email protected]>
 L:	[email protected] (moderated for non-subscribers)
Signature · 1 line
2.25.1