[PATCH v4 32/35] arm: mach-bcm283x: Add ARMV8_MULTIENTRY support

[Patrick Rudolph <patrick.rudolph@9elements.com>]

When ACPI is enabled over FDT the APs cannot be brought out of reset
by the OS using the "FDT spin-table" mechanism, as no FDT is provided
to the OS. The APs must be released out of reset in u-boot and then
brought up in an ACPI compliant fashion.

When ARMV8_MULTIENTRY is specified, the APs are released from reset
and will enter U-Boot after it has been relocated as well.

By default ARMV8_MULTIENTRY is not selected, keeping existing behaviour.

TEST: All APs enter U-Boot when run on qemu-system-aarch64 and on
      real hardware.

Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Cc: Matthias Brugger <mbrugger@suse.com>
Cc: Peter Robinson <pbrobinson@gmail.com>
Cc: Tom Rini <trini@konsulko.com>
---
Changelog v2:
- Drop CPU code from board
- Add CPU driver to release APs from reset
Changelog v4:
- Use the generic armv8 CPU driver
- Probe all CPUs
- Release secondary CPUs in the probe function
- Use of_to_plat() to parse DT
- Drop udelay()
- Select the GICv2 driver
---
 arch/arm/Kconfig              |   1 +
 arch/arm/mach-bcm283x/Kconfig |   7 +-
 arch/arm/mach-bcm283x/init.c  |  10 ++
 drivers/cpu/Makefile          |   2 +-
 drivers/cpu/bcm283x_cpu.c     | 218 ++++++++++++++++++++++++++++++++++
 5 files changed, 236 insertions(+), 2 deletions(-)
 create mode 100644 drivers/cpu/bcm283x_cpu.c

diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index e85a7131d3..72702cd99c 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -651,6 +651,7 @@ config ARCH_ORION5X
 
 config ARCH_BCM283X
 	bool "Broadcom BCM283X family"
+	select CPU
 	select DM
 	select DM_GPIO
 	select DM_SERIAL
diff --git a/arch/arm/mach-bcm283x/Kconfig b/arch/arm/mach-bcm283x/Kconfig
index b3287ce8bc..fad900d965 100644
--- a/arch/arm/mach-bcm283x/Kconfig
+++ b/arch/arm/mach-bcm283x/Kconfig
@@ -12,6 +12,7 @@ config BCM2836
 config BCM2837
 	bool "Broadcom BCM2837 SoC support"
 	depends on ARCH_BCM283X
+	select ARCH_EARLY_INIT_R
 
 config BCM2837_32B
 	bool "Broadcom BCM2837 SoC 32-bit support"
@@ -24,11 +25,14 @@ config BCM2837_64B
 	bool "Broadcom BCM2837 SoC 64-bit support"
 	depends on ARCH_BCM283X
 	select BCM2837
+	select DRIVER_GICV2
 	select ARM64
+	select CPU_ARMV8
 
 config BCM2711
 	bool "Broadcom BCM2711 SoC support"
 	depends on ARCH_BCM283X
+	select ARCH_EARLY_INIT_R
 
 config BCM2711_32B
 	bool "Broadcom BCM2711 SoC 32-bit support"
@@ -42,11 +46,12 @@ config BCM2711_64B
 	bool "Broadcom BCM2711 SoC 64-bit support"
 	depends on ARCH_BCM283X
 	select BCM2711
+	select DRIVER_GICV2
 	select ARM64
+	select CPU_ARMV8
 
 menu "Broadcom BCM283X family"
 	depends on ARCH_BCM283X
-
 choice
 	prompt "Broadcom BCM283X board select"
 	optional
diff --git a/arch/arm/mach-bcm283x/init.c b/arch/arm/mach-bcm283x/init.c
index 7a1de22e0a..c59e557432 100644
--- a/arch/arm/mach-bcm283x/init.c
+++ b/arch/arm/mach-bcm283x/init.c
@@ -9,6 +9,7 @@
 #include <cpu_func.h>
 #include <init.h>
 #include <dm/device.h>
+#include <dm/uclass.h>
 #include <fdt_support.h>
 #include <asm/global_data.h>
 
@@ -199,6 +200,15 @@ int mach_cpu_init(void)
 					       "brcm,bcm2712-pm");
 	if (offset > soc)
 		rpi_wdog_base = fdt_get_base_address(gd->fdt_blob, offset);
+	return 0;
+}
+
+int arch_early_init_r(void)
+{
+	/* The older RPis do not have a CPU driver yet,
+	 * thus do not check for errors here...
+	 */
+	uclass_probe_all(UCLASS_CPU);
 
 	return 0;
 }
diff --git a/drivers/cpu/Makefile b/drivers/cpu/Makefile
index 773395693a..eaf494706e 100644
--- a/drivers/cpu/Makefile
+++ b/drivers/cpu/Makefile
@@ -6,7 +6,7 @@
 
 obj-$(CONFIG_CPU) += cpu-uclass.o
 
-
+obj-$(CONFIG_ARCH_BCM283X) += bcm283x_cpu.o
 obj-$(CONFIG_ARCH_BMIPS) += bmips_cpu.o
 obj-$(CONFIG_ARCH_IMX8) += imx8_cpu.o
 obj-$(CONFIG_ARCH_AT91) += at91_cpu.o
diff --git a/drivers/cpu/bcm283x_cpu.c b/drivers/cpu/bcm283x_cpu.c
new file mode 100644
index 0000000000..5ed75b3754
--- /dev/null
+++ b/drivers/cpu/bcm283x_cpu.c
@@ -0,0 +1,218 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2024 9elements GmbH
+ */
+
+#include <acpi/acpigen.h>
+#include <asm/cache.h>
+#include <asm/io.h>
+#include <asm/global_data.h>
+#include <asm/system.h>
+#include <asm/armv8/cpu.h>
+#include <asm-generic/sections.h>
+#include <cpu.h>
+#include <cpu_func.h>
+#include <dm.h>
+#include <fdt_support.h>
+#include <linux/bitops.h>
+#include <linux/clk-provider.h>
+#include <linux/delay.h>
+#include "armv8_cpu.h"
+
+DECLARE_GLOBAL_DATA_PTR;
+
+struct bcm_plat {
+	u64 release_addr;
+};
+
+static int cpu_bcm_get_desc(const struct udevice *dev, char *buf, int size)
+{
+	struct cpu_plat *plat = dev_get_parent_plat(dev);
+	const char *name;
+	int ret;
+
+	if (size < 32)
+		return -ENOSPC;
+
+	if (device_is_compatible(dev, "arm,cortex-a53"))
+		name = "A53";
+	else if (device_is_compatible(dev, "arm,cortex-a72"))
+		name = "A72";
+	else
+		name = "?";
+
+	ret = snprintf(buf, size, "Broadcom Cortex-%s at %u MHz\n",
+		       name, plat->timebase_freq);
+
+	return 0;
+}
+
+static int cpu_bcm_get_info(const struct udevice *dev, struct cpu_info *info)
+{
+	struct cpu_plat *plat = dev_get_parent_plat(dev);
+
+	info->cpu_freq = plat->timebase_freq * 1000;
+	info->features = BIT(CPU_FEAT_L1_CACHE) | BIT(CPU_FEAT_MMU);
+
+	return 0;
+}
+
+static int cpu_bcm_get_count(const struct udevice *dev)
+{
+	return uclass_id_count(UCLASS_CPU);
+}
+
+static int cpu_bcm_get_vendor(const struct udevice *dev,  char *buf, int size)
+{
+	snprintf(buf, size, "Broadcom");
+
+	return 0;
+}
+
+static int cpu_bcm_is_current(struct udevice *dev)
+{
+#ifdef CONFIG_ARM64
+	struct cpu_plat *plat = dev_get_parent_plat(dev);
+
+	if (plat->cpu_id == (read_mpidr() & 0xffff))
+		return 1;
+#endif
+
+	return 0;
+}
+
+/**
+ * bcm_cpu_on - Releases the secondary CPU from it's spintable
+ *
+ * Write the CPU's spintable mailbox and let the CPU enter U-Boot.
+ *
+ * @dev: Device to start
+ * @return: zero on success or error code on failure.
+ */
+static int bcm_cpu_on(struct udevice *dev)
+{
+	struct bcm_plat *plat = dev_get_plat(dev);
+	ulong *start_address;
+
+	if (!plat || plat->release_addr == ~0ULL)
+		return -ENODATA;
+
+	start_address = map_physmem(plat->release_addr, sizeof(uintptr_t), MAP_NOCACHE);
+
+	/* Point secondary CPU to U-Boot entry */
+	*start_address = (uintptr_t)_start;
+
+	/* Make sure the other CPUs see the written start address */
+	if (!CONFIG_IS_ENABLED(SYS_DCACHE_OFF))
+		flush_dcache_all();
+
+	/* Send an event to wake up the secondary CPU. */
+	asm("dsb	ishst\n"
+	    "sev");
+
+	unmap_physmem(start_address, MAP_NOCACHE);
+
+	return 0;
+}
+
+static const struct cpu_ops cpu_bcm_ops = {
+	.get_desc	= cpu_bcm_get_desc,
+	.get_info	= cpu_bcm_get_info,
+	.get_count	= cpu_bcm_get_count,
+	.get_vendor	= cpu_bcm_get_vendor,
+	.is_current	= cpu_bcm_is_current,
+};
+
+static const struct udevice_id cpu_bcm_ids[] = {
+	{ .compatible = "arm,cortex-a53" },	/* RPi 3 */
+	{ .compatible = "arm,cortex-a72" },	/* RPi 4 */
+	{ }
+};
+
+static int bcm_cpu_bind(struct udevice *dev)
+{
+	struct cpu_plat *plat = dev_get_parent_plat(dev);
+
+	plat->cpu_id = dev_read_addr(dev);
+
+	return 0;
+}
+
+/**
+ * bcm_cpu_of_to_plat - Gather spin-table release address
+ *
+ * Read the spin-table release address to allow all seconary CPUs to enter
+ * U-Boot when necessary.
+ *
+ * @dev: Device to start
+ */
+static int bcm_cpu_of_to_plat(struct udevice *dev)
+{
+	struct bcm_plat *plat = dev_get_plat(dev);
+	const char *prop;
+
+	if (CONFIG_IS_ENABLED(ARMV8_MULTIENTRY)) {
+		plat->release_addr = ~0ULL;
+
+		prop = dev_read_string(dev, "enable-method");
+		if (!prop || strcmp(prop, "spin-table"))
+			return -ENODEV;
+
+		plat->release_addr = dev_read_u64_default(dev, "cpu-release-addr", ~0ULL);
+
+		if (plat->release_addr == ~0ULL)
+			return -ENODEV;
+	}
+
+	return 0;
+}
+
+static int bcm_cpu_probe(struct udevice *dev)
+{
+	struct cpu_plat *plat = dev_get_parent_plat(dev);
+	struct clk clk;
+	int ret;
+
+	/* Get a clock if it exists */
+	ret = clk_get_by_index(dev, 0, &clk);
+	if (!ret) {
+		ret = clk_enable(&clk);
+		if (ret && (ret != -ENOSYS || ret != -EOPNOTSUPP))
+			return ret;
+		ret = clk_get_rate(&clk);
+		if (IS_ERR_VALUE(ret))
+			return ret;
+		plat->timebase_freq = ret;
+	}
+
+	/*
+	 * The armstub holds the secondary CPUs in a spinloop. When
+	 * ARMV8_MULTIENTRY is enabled release the secondary CPUs and
+	 * let them enter U-Boot as well.
+	 */
+	if (CONFIG_IS_ENABLED(ARMV8_MULTIENTRY)) {
+		ret = bcm_cpu_on(dev);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+struct acpi_ops bcm283x_cpu_acpi_ops = {
+	.fill_ssdt	= armv8_cpu_fill_ssdt,
+	.fill_madt	= armv8_cpu_fill_madt,
+};
+
+U_BOOT_DRIVER(cpu_bcm_drv) = {
+	.name		= "bcm283x_cpu",
+	.id		= UCLASS_CPU,
+	.of_match	= cpu_bcm_ids,
+	.ops		= &cpu_bcm_ops,
+	.probe		= bcm_cpu_probe,
+	.bind		= bcm_cpu_bind,
+	.of_to_plat	= bcm_cpu_of_to_plat,
+	.plat_auto	= sizeof(struct bcm_plat),
+	.flags		= 0,
+	ACPI_OPS_PTR(&bcm283x_cpu_acpi_ops)
+};
-- 
2.46.0