Add support for a fake, para-virtualised machine

Quoted 9 lines · Will Deacon

Documentation/devicetree/bindings/arm/psci.txt new file +9 -0

+
+ - function-base : The base ID from which the functions are offset.
+
+Main node optional properties:
+
+ - cpu_suspend   : Offset of CPU_SUSPEND ID from function-base
+
+ - cpu_off       : Offset of CPU_OFF ID from function-base
+

Quoted 3 lines

+ - cpu_on : Offset of CPU_ON ID from function-base
+
+ - migrate : Offset of MIGRATE ID from function-base

What is the benefit of the "function-base" property over just having 32 bit IDs for each function. For all I can tell, the interface does not rely on the numbers to be consecutive, so removing the function-base attribute would make the binding simpler as well as more flexible.

Quoted 9 lines · Will Deacon

Documentation/devicetree/bindings/arm/psci.txt new file +9 -0

+
+ - function-base : The base ID from which the functions are offset.
+
+Main node optional properties:
+
+ - cpu_suspend   : Offset of CPU_SUSPEND ID from function-base
+
+ - cpu_off       : Offset of CPU_OFF ID from function-base
+

Quoted 3 lines

+ - cpu_on : Offset of CPU_ON ID from function-base
+
+ - migrate : Offset of MIGRATE ID from function-base

What is the benefit of the "function-base" property over just having 32 bit IDs for each function. For all I can tell, the interface does not rely on the numbers to be consecutive, so removing the function-base attribute would make the binding simpler as well as more flexible.

This patch adds support for the Power State Coordination Interface defined by ARM, allowing Linux to request CPU-centric power-management operations from firmware implementing the PSCI protocol.

Signed-off-by Will Deacon <[email protected]>

+/*
+ * The following two functions are invoked via the invoke_psci_fn pointer
+ * and will not be inlined, allowing us to piggyback on the AAPCS.
+ */

arch/arm/kernel/psci.c new file +27 -0

+static int __invoke_psci_fn_hvc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__HVC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int __invoke_psci_fn_smc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__SMC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}

arch/arm/kernel/psci.c new file +40 -0

+	return psci_to_linux_errno(err);
+}
+
+static int psci_cpu_off(struct psci_power_state state)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_OFF];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, 0, 0);
+	return psci_to_linux_errno(err);
+}
+
+static int psci_cpu_on(unsigned long cpuid, unsigned long entry_point)
+{
+	int err;
+	u32 fn;
+
+	fn = psci_function_id[PSCI_FN_CPU_ON];
+	err = invoke_psci_fn(fn, cpuid, entry_point, 0);
+	return psci_to_linux_errno(err);
+}
+
+static int psci_migrate(unsigned long cpuid)
+{
+	int err;
+	u32 fn;
+
+	fn = psci_function_id[PSCI_FN_MIGRATE];
+	err = invoke_psci_fn(fn, cpuid, 0, 0);
+	return psci_to_linux_errno(err);
+}
+
+static const struct of_device_id psci_of_match[] __initconst = {
+	{ .compatible = "arm,psci",	},
+	{},
+};
+
+static int __init psci_init(void)

Quoted 5 lines · Will Deacon

This patch adds support for the Power State Coordination Interface defined by ARM, allowing Linux to request CPU-centric power-management operations from firmware implementing the PSCI protocol.

Signed-off-by Will Deacon <[email protected]>

[...]

Quoted 4 lines

+/*
+ * The following two functions are invoked via the invoke_psci_fn pointer
+ * and will not be inlined, allowing us to piggyback on the AAPCS.
+ */

To make sure the code is always in sync with the intent, you could mark those with noinline as well.

Quoted 31 lines

arch/arm/kernel/psci.c new file +31 -0

+static int __invoke_psci_fn_hvc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__HVC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int __invoke_psci_fn_smc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__SMC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{

arch/arm/kernel/psci.c line 119

Quoted 6 lines

arch/arm/kernel/psci.c new file +6 -0

+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

arch/arm/kernel/psci.c line 169

Quoted 11 lines

arch/arm/kernel/psci.c new file +11 -0

+static int __init psci_init(void)
+{
+	struct device_node *np;
+	const char *method;
+	u32 base, id;
+
+	np = of_find_matching_node(NULL, psci_of_match);
+	if (!np)
+		return 0;
+
+	pr_info("probing function IDs from device-tree\n");

Having "probing function IDs from device-tree" in the middle of a kernel log isn't very informative. Better make this more useful or remove it.

arch/arm/kernel/psci.c line 172

Quoted 3 lines

arch/arm/kernel/psci.c new file +3 -0

+
+	if (of_property_read_u32(np, "function-base", &base)) {
+		pr_warning("missing \"function-base\" property\n");

Same thing here: this lacks context in a kernel log. And so on for the other occurrences.

Quoted 12 lines · Nicolas Pitre

arch/arm/kernel/psci.c line 119

Quoted 9 lines · Will Deacon

arch/arm/kernel/psci.c new file +9 -0

+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

That's a hangover from when entry_point was a void *. I'll fix that, thanks.

Quoted 15 lines

arch/arm/kernel/psci.c line 169

Quoted 11 lines

arch/arm/kernel/psci.c new file +11 -0

+static int __init psci_init(void)
+{
+	struct device_node *np;
+	const char *method;
+	u32 base, id;
+
+	np = of_find_matching_node(NULL, psci_of_match);
+	if (!np)
+		return 0;
+
+	pr_info("probing function IDs from device-tree\n");

Having "probing function IDs from device-tree" in the middle of a kernel log isn't very informative. Better make this more useful or remove it.

Quoted 6 lines

arch/arm/kernel/psci.c line 172

Quoted 3 lines

arch/arm/kernel/psci.c new file +3 -0

+
+	if (of_property_read_u32(np, "function-base", &base)) {
+		pr_warning("missing \"function-base\" property\n");

Same thing here: this lacks context in a kernel log. And so on for the other occurrences.

Actually, these are all prefixed with "psci: " thanks to the pr_fmt definition at the top of the file.

Quoted 14 lines · Will Deacon

Quoted 12 lines · Nicolas Pitre

arch/arm/kernel/psci.c line 119

Quoted 9 lines · Will Deacon

arch/arm/kernel/psci.c new file +9 -0

+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

Quoted 1 line

That's a hangover from when entry_point was a void *. I'll fix that, thanks.

Hopefully you didn't pass virtual pointers to the PSCI call, did you? :-)

Quoted 14 lines · Will Deacon

Quoted 12 lines · Nicolas Pitre

arch/arm/kernel/psci.c line 119

Quoted 9 lines · Will Deacon

arch/arm/kernel/psci.c new file +9 -0

+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

Quoted 1 line

That's a hangover from when entry_point was a void *. I'll fix that, thanks.

Hopefully you didn't pass virtual pointers to the PSCI call, did you? :-)

Quoted 12 lines · Nicolas Pitre

arch/arm/kernel/psci.c line 119

Quoted 9 lines · Will Deacon

arch/arm/kernel/psci.c new file +9 -0

+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

That's a hangover from when entry_point was a void *. I'll fix that, thanks.

Quoted 15 lines

arch/arm/kernel/psci.c line 169

Quoted 11 lines

arch/arm/kernel/psci.c new file +11 -0

+static int __init psci_init(void)
+{
+	struct device_node *np;
+	const char *method;
+	u32 base, id;
+
+	np = of_find_matching_node(NULL, psci_of_match);
+	if (!np)
+		return 0;
+
+	pr_info("probing function IDs from device-tree\n");

Having "probing function IDs from device-tree" in the middle of a kernel log isn't very informative. Better make this more useful or remove it.

Quoted 6 lines

arch/arm/kernel/psci.c line 172

Quoted 3 lines

arch/arm/kernel/psci.c new file +3 -0

+
+	if (of_property_read_u32(np, "function-base", &base)) {
+		pr_warning("missing \"function-base\" property\n");

Same thing here: this lacks context in a kernel log. And so on for the other occurrences.

Actually, these are all prefixed with "psci: " thanks to the pr_fmt definition at the top of the file.

Quoted 5 lines · Will Deacon

This patch adds support for the Power State Coordination Interface defined by ARM, allowing Linux to request CPU-centric power-management operations from firmware implementing the PSCI protocol.

Signed-off-by Will Deacon <[email protected]>

[...]

Quoted 4 lines

+/*
+ * The following two functions are invoked via the invoke_psci_fn pointer
+ * and will not be inlined, allowing us to piggyback on the AAPCS.
+ */

To make sure the code is always in sync with the intent, you could mark those with noinline as well.

Quoted 31 lines

arch/arm/kernel/psci.c new file +31 -0

+static int __invoke_psci_fn_hvc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__HVC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int __invoke_psci_fn_smc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__SMC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int psci_cpu_suspend(struct psci_power_state state,
+			    unsigned long entry_point)
+{

arch/arm/kernel/psci.c line 119

Quoted 6 lines

arch/arm/kernel/psci.c new file +6 -0

+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_SUSPEND];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, (u32)entry_point, 0);

Why do you need the u32 cast here?

arch/arm/kernel/psci.c line 169

Quoted 11 lines

arch/arm/kernel/psci.c new file +11 -0

+static int __init psci_init(void)
+{
+	struct device_node *np;
+	const char *method;
+	u32 base, id;
+
+	np = of_find_matching_node(NULL, psci_of_match);
+	if (!np)
+		return 0;
+
+	pr_info("probing function IDs from device-tree\n");

Having "probing function IDs from device-tree" in the middle of a kernel log isn't very informative. Better make this more useful or remove it.

arch/arm/kernel/psci.c line 172

Quoted 3 lines

arch/arm/kernel/psci.c new file +3 -0

+
+	if (of_property_read_u32(np, "function-base", &base)) {
+		pr_warning("missing \"function-base\" property\n");

Same thing here: this lacks context in a kernel log. And so on for the other occurrences.

This patch adds support for the Power State Coordination Interface defined by ARM, allowing Linux to request CPU-centric power-management operations from firmware implementing the PSCI protocol.

Signed-off-by Will Deacon <[email protected]>

+/*
+ * The following two functions are invoked via the invoke_psci_fn pointer
+ * and will not be inlined, allowing us to piggyback on the AAPCS.
+ */

arch/arm/kernel/psci.c new file +27 -0

+static int __invoke_psci_fn_hvc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__HVC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}
+
+static int __invoke_psci_fn_smc(u32 function_id, u32 arg0, u32 arg1, u32 arg2)
+{
+	asm volatile(
+			__asmeq("%0", "r0")
+			__asmeq("%1", "r1")
+			__asmeq("%2", "r2")
+			__asmeq("%3", "r3")
+			__SMC(0)
+		: "+r" (function_id)
+		: "r" (arg0), "r" (arg1), "r" (arg2));
+
+	return function_id;
+}

arch/arm/kernel/psci.c new file +40 -0

+	return psci_to_linux_errno(err);
+}
+
+static int psci_cpu_off(struct psci_power_state state)
+{
+	int err;
+	u32 fn, power_state;
+
+	fn = psci_function_id[PSCI_FN_CPU_OFF];
+	power_state = psci_power_state_pack(state);
+	err = invoke_psci_fn(fn, power_state, 0, 0);
+	return psci_to_linux_errno(err);
+}
+
+static int psci_cpu_on(unsigned long cpuid, unsigned long entry_point)
+{
+	int err;
+	u32 fn;
+
+	fn = psci_function_id[PSCI_FN_CPU_ON];
+	err = invoke_psci_fn(fn, cpuid, entry_point, 0);
+	return psci_to_linux_errno(err);
+}
+
+static int psci_migrate(unsigned long cpuid)
+{
+	int err;
+	u32 fn;
+
+	fn = psci_function_id[PSCI_FN_MIGRATE];
+	err = invoke_psci_fn(fn, cpuid, 0, 0);
+	return psci_to_linux_errno(err);
+}
+
+static const struct of_device_id psci_of_match[] __initconst = {
+	{ .compatible = "arm,psci",	},
+	{},
+};
+
+static int __init psci_init(void)

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

It only mandates a GIC and architected timers. Fits nicely with a multiplatform zImage. Uses very little silicon area.

Signed-off-by Marc Zyngier <[email protected]>

Signed-off-by Will Deacon <[email protected]>

---
 arch/arm/Kconfig            |  2 ++
 arch/arm/Makefile           |  1 +
 arch/arm/mach-virt/Kconfig  |  9 +++++++
 arch/arm/mach-virt/Makefile |  5 ++++
 arch/arm/mach-virt/virt.c   | 65 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 82 insertions(+)
 create mode 100644 arch/arm/mach-virt/Kconfig
 create mode 100644 arch/arm/mach-virt/Makefile
 create mode 100644 arch/arm/mach-virt/virt.c

arch/arm/Kconfig +2 -0

index 2a04375..552d72e0 100644
@@ -1127,6 +1127,8 @@ source "arch/arm/mach-versatile/Kconfig"
 source "arch/arm/mach-vexpress/Kconfig"
 source "arch/arm/plat-versatile/Kconfig"
 
+source "arch/arm/mach-virt/Kconfig"
+
 source "arch/arm/mach-w90x900/Kconfig"
 
 # Definitions to make life easier

arch/arm/Makefile +1 -0

index 5f914fc..e8232ad 100644
@@ -192,6 +192,7 @@ machine-$(CONFIG_ARCH_SOCFPGA)		+= socfpga
 machine-$(CONFIG_ARCH_SPEAR13XX)	+= spear13xx
 machine-$(CONFIG_ARCH_SPEAR3XX)		+= spear3xx
 machine-$(CONFIG_MACH_SPEAR600)		+= spear6xx
+machine-$(CONFIG_ARCH_VIRT)		+= virt
 machine-$(CONFIG_ARCH_ZYNQ)		+= zynq
 
 # Platform directory name.  This list is sorted alphanumerically

arch/arm/mach-virt/Kconfig new file +9 -0

new file mode 100644
index 0000000..a568a2a
@@ -0,0 +1,9 @@
+config ARCH_VIRT
+	bool "Dummy Virtual Machine" if ARCH_MULTI_V7
+	select ARCH_WANT_OPTIONAL_GPIOLIB
+	select ARM_GIC
+	select ARM_ARCH_TIMER
+	select HAVE_SMP
+	select CPU_V7
+	select SPARSE_IRQ
+	select USE_OF

arch/arm/mach-virt/Makefile new file +5 -0

new file mode 100644
index 0000000..7ddbfa6
@@ -0,0 +1,5 @@
+#
+# Makefile for the linux kernel.
+#
+
+obj-y					:= virt.o

arch/arm/mach-virt/virt.c new file +65 -0

new file mode 100644
index 0000000..174b9da
@@ -0,0 +1,65 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Authors: Will Deacon <[email protected]>,
+ *          Marc Zyngier <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/of_irq.h>
+#include <linux/of_platform.h>
+
+#include <asm/arch_timer.h>
+#include <asm/hardware/gic.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/time.h>
+
+const static struct of_device_id irq_match[] = {
+	{ .compatible = "arm,cortex-a15-gic", .data = gic_of_init, },
+	{}
+};
+
+static void __init gic_init_irq(void)
+{
+	of_irq_init(irq_match);
+}
+
+static void __init virt_init(void)
+{
+	of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);
+}
+
+static void __init virt_timer_init(void)
+{
+	WARN_ON(arch_timer_of_register() != 0);
+	WARN_ON(arch_timer_sched_clock_init() != 0);
+}
+
+static const char *virt_dt_match[] = {
+	"linux,dummy-virt",
+	NULL
+};
+
+static struct sys_timer virt_timer = {
+	.init = virt_timer_init,
+};
+
+DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
+	.init_irq	= gic_init_irq,
+	.handle_irq     = gic_handle_irq,
+	.timer		= &virt_timer,
+	.init_machine	= virt_init,
+	.dt_compat	= virt_dt_match,
+MACHINE_END

Signature · 7 lines

1.8.0


_______________________________________________
Xen-devel mailing list
Xen-devel at lists.xen.org
http://lists.xen.org/xen-devel

Quoted 10 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

It only mandates a GIC and architected timers. Fits nicely with a multiplatform zImage. Uses very little silicon area.

Signed-off-by Marc Zyngier <[email protected]>

Signed-off-by Will Deacon <[email protected]>

Quoted 10 lines

---
 arch/arm/Kconfig            |  2 ++
 arch/arm/Makefile           |  1 +
 arch/arm/mach-virt/Kconfig  |  9 +++++++
 arch/arm/mach-virt/Makefile |  5 ++++
 arch/arm/mach-virt/virt.c   | 65 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 82 insertions(+)
 create mode 100644 arch/arm/mach-virt/Kconfig
 create mode 100644 arch/arm/mach-virt/Makefile
 create mode 100644 arch/arm/mach-virt/virt.c

Should it come along with a DTS?

Hi Stefano,

Quoted 22 lines · Stefano Stabellini

Quoted 10 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

It only mandates a GIC and architected timers. Fits nicely with a multiplatform zImage. Uses very little silicon area.

Signed-off-by Marc Zyngier <[email protected]>

Signed-off-by Will Deacon <[email protected]>

Quoted 10 lines

---
 arch/arm/Kconfig            |  2 ++
 arch/arm/Makefile           |  1 +
 arch/arm/mach-virt/Kconfig  |  9 +++++++
 arch/arm/mach-virt/Makefile |  5 ++++
 arch/arm/mach-virt/virt.c   | 65 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 82 insertions(+)
 create mode 100644 arch/arm/mach-virt/Kconfig
 create mode 100644 arch/arm/mach-virt/Makefile
 create mode 100644 arch/arm/mach-virt/virt.c

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU. Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

Hi Stefano,

Quoted 22 lines · Stefano Stabellini

Quoted 10 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

It only mandates a GIC and architected timers. Fits nicely with a multiplatform zImage. Uses very little silicon area.

Signed-off-by Marc Zyngier <[email protected]>

Signed-off-by Will Deacon <[email protected]>

Quoted 10 lines

---
 arch/arm/Kconfig            |  2 ++
 arch/arm/Makefile           |  1 +
 arch/arm/mach-virt/Kconfig  |  9 +++++++
 arch/arm/mach-virt/Makefile |  5 ++++
 arch/arm/mach-virt/virt.c   | 65 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 82 insertions(+)
 create mode 100644 arch/arm/mach-virt/Kconfig
 create mode 100644 arch/arm/mach-virt/Makefile
 create mode 100644 arch/arm/mach-virt/virt.c

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU. Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

Hi Stefano,

Quoted 5 lines · Stefano Stabellini

Quoted 4 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU.

Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

Hi Will,

Quoted 8 lines · Will Deacon

Hi Stefano,

Quoted 5 lines · Stefano Stabellini

Quoted 4 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

[...]

Quoted 3 lines

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU.

I assume multiple virtio-mmio peripherals are hiding behind what you seem to advertising here as plain old memory?

Quoted 3 lines

Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

I disagree. I think an example DTS would be fairly useful, if only for the full list of peripherals you're using on the platform.

Quoted 2 lines

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

If you'd much prefer to post the command line, tools version, etc. that you're using to generate the DTB, rather than the DTS, that'd be better than nothing.

It seems like Rob Herring's earlier question about whether the dummy platform is really justified never got answered. I think sending a sample DTS out with the patchset would help "highlight where we need to do more work on DT driving the initialization."

Lastly, I'm somewhat curious, why virtio-mmio console rather than DCC?

Hi Christopher,

Quoted 5 lines · Christopher Covington

Hi Will,

Quoted 2 lines · Will Deacon

Hi Stefano,

Quoted 5 lines

Quoted 2 lines

The only things the platform needs are GIC, timers, memory and a CPU.

I assume multiple virtio-mmio peripherals are hiding behind what you seem to be advertising here as plain old memory?

No. Memory is memory. Virtio peripherals are created outside of the memory range. They end up having rings and descriptor in memory, but that's not any different from what you have with an fairly complicated DMA capable hardware device.

Here's what kvmtool has been seen to generate, with the parameters I used a few minutes ago:

/dts-v1/;

/memreserve/ 0x000000008fff0000 0x0000000000001000;

/ {
	interrupt-parent = <0x1>;
	compatible = "linux,dummy-virt";

Does it help?

What would be the point of using DCC?

We would have to trap on each access...

...and then we'd have to invent yet another mechanism to channel the console to userspace.

Not to mention that I like to be able to actually input something on a console, not just read from it.

Quoted 1 line

What would be the point of using DCC?

It seemed like ARM-Ltd.-architected peripherals were picked for the timer and interrupt controller, so I wondered why not for the console as well. As best I'm aware, unless one ventures into the PrimeCell line with the PL011, DCC is the closest match for an officially architected console mechanism.

Quoted 6 lines · Christopher Covington

Quoted 1 line

What would be the point of using DCC?

It seemed like ARM-Ltd.-architected peripherals were picked for the timer and interrupt controller, so I wondered why not for the console as well. As best I'm aware, unless one ventures into the PrimeCell line with the PL011, DCC is the closest match for an officially architected console mechanism.

I certainly wouldn't want to mandate the presence of any devices that require emulation in the DT for a virtual machine, including DCC, as simple as it might be.

Quoted 6 lines · Christopher Covington

Quoted 1 line

What would be the point of using DCC?

It seemed like ARM-Ltd.-architected peripherals were picked for the timer and interrupt controller, so I wondered why not for the console as well. As best I'm aware, unless one ventures into the PrimeCell line with the PL011, DCC is the closest match for an officially architected console mechanism.

I certainly wouldn't want to mandate the presence of any devices that require emulation in the DT for a virtual machine, including DCC, as simple as it might be.

Quoted 1 line

What would be the point of using DCC?

It seemed like ARM-Ltd.-architected peripherals were picked for the timer and interrupt controller, so I wondered why not for the console as well. As best I'm aware, unless one ventures into the PrimeCell line with the PL011, DCC is the closest match for an officially architected console mechanism.

Hi Christopher,

Quoted 5 lines · Christopher Covington

Hi Will,

Quoted 2 lines · Will Deacon

Hi Stefano,

Quoted 5 lines

Quoted 2 lines

The only things the platform needs are GIC, timers, memory and a CPU.

I assume multiple virtio-mmio peripherals are hiding behind what you seem to be advertising here as plain old memory?

No. Memory is memory. Virtio peripherals are created outside of the memory range. They end up having rings and descriptor in memory, but that's not any different from what you have with an fairly complicated DMA capable hardware device.

Here's what kvmtool has been seen to generate, with the parameters I used a few minutes ago:

/dts-v1/;

/memreserve/ 0x000000008fff0000 0x0000000000001000;

/ {
	interrupt-parent = <0x1>;
	compatible = "linux,dummy-virt";

Does it help?

What would be the point of using DCC?

We would have to trap on each access...

...and then we'd have to invent yet another mechanism to channel the console to userspace.

Not to mention that I like to be able to actually input something on a console, not just read from it.

Hi Will,

Quoted 8 lines · Will Deacon

Hi Stefano,

Quoted 5 lines · Stefano Stabellini

Quoted 4 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

[...]

Quoted 3 lines

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU.

I assume multiple virtio-mmio peripherals are hiding behind what you seem to advertising here as plain old memory?

Quoted 3 lines

Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

I disagree. I think an example DTS would be fairly useful, if only for the full list of peripherals you're using on the platform.

Quoted 2 lines

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

If you'd much prefer to post the command line, tools version, etc. that you're using to generate the DTB, rather than the DTS, that'd be better than nothing.

It seems like Rob Herring's earlier question about whether the dummy platform is really justified never got answered. I think sending a sample DTS out with the patchset would help "highlight where we need to do more work on DT driving the initialization."

Lastly, I'm somewhat curious, why virtio-mmio console rather than DCC?

Hi Stefano,

Quoted 5 lines · Stefano Stabellini

Quoted 4 lines · Will Deacon

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

Quoted 1 line

Should it come along with a DTS?

The only things the platform needs are GIC, timers, memory and a CPU.

Furthermore, the location, size, frequency etc properties of these aren't fixed, so a dts would be fairly useless because it will probably not match the particular mach-virt instance you're targetting.

For kvmtool, I've been generating the device-tree at runtime based on how kvmtool is invoked and it's been working pretty well so far.

From: Marc Zyngier <[email protected]>

Add support for the smallest, dumbest possible platform, to be used as a guest for KVM or other hypervisors.

It only mandates a GIC and architected timers. Fits nicely with a multiplatform zImage. Uses very little silicon area.

Signed-off-by Marc Zyngier <[email protected]>

Signed-off-by Will Deacon <[email protected]>

---
 arch/arm/Kconfig            |  2 ++
 arch/arm/Makefile           |  1 +
 arch/arm/mach-virt/Kconfig  |  9 +++++++
 arch/arm/mach-virt/Makefile |  5 ++++
 arch/arm/mach-virt/virt.c   | 65 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 82 insertions(+)
 create mode 100644 arch/arm/mach-virt/Kconfig
 create mode 100644 arch/arm/mach-virt/Makefile
 create mode 100644 arch/arm/mach-virt/virt.c

arch/arm/Kconfig +2 -0

index 2a04375..552d72e0 100644
@@ -1127,6 +1127,8 @@ source "arch/arm/mach-versatile/Kconfig"
 source "arch/arm/mach-vexpress/Kconfig"
 source "arch/arm/plat-versatile/Kconfig"
 
+source "arch/arm/mach-virt/Kconfig"
+
 source "arch/arm/mach-w90x900/Kconfig"
 
 # Definitions to make life easier

arch/arm/Makefile +1 -0

index 5f914fc..e8232ad 100644
@@ -192,6 +192,7 @@ machine-$(CONFIG_ARCH_SOCFPGA)		+= socfpga
 machine-$(CONFIG_ARCH_SPEAR13XX)	+= spear13xx
 machine-$(CONFIG_ARCH_SPEAR3XX)		+= spear3xx
 machine-$(CONFIG_MACH_SPEAR600)		+= spear6xx
+machine-$(CONFIG_ARCH_VIRT)		+= virt
 machine-$(CONFIG_ARCH_ZYNQ)		+= zynq
 
 # Platform directory name.  This list is sorted alphanumerically

arch/arm/mach-virt/Kconfig new file +9 -0

new file mode 100644
index 0000000..a568a2a
@@ -0,0 +1,9 @@
+config ARCH_VIRT
+	bool "Dummy Virtual Machine" if ARCH_MULTI_V7
+	select ARCH_WANT_OPTIONAL_GPIOLIB
+	select ARM_GIC
+	select ARM_ARCH_TIMER
+	select HAVE_SMP
+	select CPU_V7
+	select SPARSE_IRQ
+	select USE_OF

arch/arm/mach-virt/Makefile new file +5 -0

new file mode 100644
index 0000000..7ddbfa6
@@ -0,0 +1,5 @@
+#
+# Makefile for the linux kernel.
+#
+
+obj-y					:= virt.o

arch/arm/mach-virt/virt.c new file +65 -0

new file mode 100644
index 0000000..174b9da
@@ -0,0 +1,65 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Authors: Will Deacon <[email protected]>,
+ *          Marc Zyngier <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/of_irq.h>
+#include <linux/of_platform.h>
+
+#include <asm/arch_timer.h>
+#include <asm/hardware/gic.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/time.h>
+
+const static struct of_device_id irq_match[] = {
+	{ .compatible = "arm,cortex-a15-gic", .data = gic_of_init, },
+	{}
+};
+
+static void __init gic_init_irq(void)
+{
+	of_irq_init(irq_match);
+}
+
+static void __init virt_init(void)
+{
+	of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);
+}
+
+static void __init virt_timer_init(void)
+{
+	WARN_ON(arch_timer_of_register() != 0);
+	WARN_ON(arch_timer_sched_clock_init() != 0);
+}
+
+static const char *virt_dt_match[] = {
+	"linux,dummy-virt",
+	NULL
+};
+
+static struct sys_timer virt_timer = {
+	.init = virt_timer_init,
+};
+
+DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
+	.init_irq	= gic_init_irq,
+	.handle_irq     = gic_handle_irq,
+	.timer		= &virt_timer,
+	.init_machine	= virt_init,
+	.dt_compat	= virt_dt_match,
+MACHINE_END

Signature · 7 lines

1.8.0


_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

---
 arch/arm/mach-virt/Kconfig   |  1 +
 arch/arm/mach-virt/Makefile  |  1 +
 arch/arm/mach-virt/platsmp.c | 76 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mach-virt/virt.c    |  6 ++++
 4 files changed, 84 insertions(+)
 create mode 100644 arch/arm/mach-virt/platsmp.c

arch/arm/mach-virt/Kconfig +1 -0

index a568a2a..8958f0d 100644
@@ -3,6 +3,7 @@ config ARCH_VIRT
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select ARM_GIC
 	select ARM_ARCH_TIMER
+	select ARM_PSCI
 	select HAVE_SMP
 	select CPU_V7
 	select SPARSE_IRQ

arch/arm/mach-virt/Makefile +1 -0

index 7ddbfa6..042afc1 100644
@@ -3,3 +3,4 @@
 #
 
 obj-y					:= virt.o
+obj-$(CONFIG_SMP)			+= platsmp.o

arch/arm/mach-virt/platsmp.c new file +38 -0

new file mode 100644
index 0000000..930362b
@@ -0,0 +1,76 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Author: Will Deacon <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/of.h>
+
+#include <asm/psci.h>
+#include <asm/smp_plat.h>
+#include <asm/hardware/gic.h>
+
+extern void secondary_startup(void);
+
+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {

arch/arm/mach-virt/platsmp.c new file +27 -0

+	set_smp_cross_call(gic_raise_softirq);
+}
+
+static void __init virt_smp_prepare_cpus(unsigned int max_cpus)
+{
+}
+
+static int __cpuinit virt_boot_secondary(unsigned int cpu,
+					 struct task_struct *idle)
+{
+	if (psci_ops.cpu_on)
+		return psci_ops.cpu_on(cpu_logical_map(cpu),
+				       __pa(secondary_startup));
+	return -ENODEV;
+}
+
+static void __cpuinit virt_secondary_init(unsigned int cpu)
+{
+	gic_secondary_init(0);
+}
+
+struct smp_operations __initdata virt_smp_ops = {
+	.smp_init_cpus		= virt_smp_init_cpus,
+	.smp_prepare_cpus	= virt_smp_prepare_cpus,
+	.smp_secondary_init	= virt_secondary_init,
+	.smp_boot_secondary	= virt_boot_secondary,
+};

arch/arm/mach-virt/virt.c +1 -0

index 174b9da..d764835 100644
@@ -20,6 +20,7 @@
 
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/smp.h>
 
 #include <asm/arch_timer.h>
 #include <asm/hardware/gic.h>

arch/arm/mach-virt/virt.c +2 -0

+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

Quoted 4 lines · Will Deacon

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

Quoted 13 lines

arch/arm/mach-virt/platsmp.c new file +13 -0

+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {
+		if (cpu < NR_CPUS)
+			set_cpu_possible(cpu, true);
+		cpu++;
+	}

arch/arm/mach-virt/platsmp.c line 48

Quoted 6 lines

arch/arm/mach-virt/platsmp.c new file +6 -0

+
+	/* sanity check */
+	if (cpu > NR_CPUS)
+		pr_warning("no. of cores (%d) greater than configured maximum "
+			   "of %d - clipping\n",
+			   cpu, NR_CPUS);

Since commit 5587164eea you shouldn't need any of the above.

arch/arm/mach-virt/virt.c static struct sys_timer virt_timer = { · line 62

Quoted 3 lines

arch/arm/mach-virt/virt.c +3 -0

+#ifdef CONFIG_SMP
+extern struct smp_operations virt_smp_ops;
+#endif

You don't need to surround prototype declaration here, unless your goal was to define a dummy virt_smp_ops when CONFIG_SMP is not selected? Otherwise the reference below would break compilation.

Quoted 12 lines

arch/arm/mach-virt/virt.c +2 -0

+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

Quoted 4 lines · Will Deacon

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

Quoted 13 lines

arch/arm/mach-virt/platsmp.c new file +13 -0

+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {
+		if (cpu < NR_CPUS)
+			set_cpu_possible(cpu, true);
+		cpu++;
+	}

arch/arm/mach-virt/platsmp.c line 48

Quoted 6 lines

arch/arm/mach-virt/platsmp.c new file +6 -0

+
+	/* sanity check */
+	if (cpu > NR_CPUS)
+		pr_warning("no. of cores (%d) greater than configured maximum "
+			   "of %d - clipping\n",
+			   cpu, NR_CPUS);

Since commit 5587164eea you shouldn't need any of the above.

arch/arm/mach-virt/virt.c static struct sys_timer virt_timer = { · line 62

Quoted 3 lines

arch/arm/mach-virt/virt.c +3 -0

+#ifdef CONFIG_SMP
+extern struct smp_operations virt_smp_ops;
+#endif

You don't need to surround prototype declaration here, unless your goal was to define a dummy virt_smp_ops when CONFIG_SMP is not selected? Otherwise the reference below would break compilation.

Quoted 12 lines

arch/arm/mach-virt/virt.c +2 -0

+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

---
 arch/arm/mach-virt/Kconfig   |  1 +
 arch/arm/mach-virt/Makefile  |  1 +
 arch/arm/mach-virt/platsmp.c | 76 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mach-virt/virt.c    |  6 ++++
 4 files changed, 84 insertions(+)
 create mode 100644 arch/arm/mach-virt/platsmp.c

arch/arm/mach-virt/Kconfig +1 -0

index a568a2a..8958f0d 100644
@@ -3,6 +3,7 @@ config ARCH_VIRT
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select ARM_GIC
 	select ARM_ARCH_TIMER
+	select ARM_PSCI
 	select HAVE_SMP
 	select CPU_V7
 	select SPARSE_IRQ

arch/arm/mach-virt/Makefile +1 -0

index 7ddbfa6..042afc1 100644
@@ -3,3 +3,4 @@
 #
 
 obj-y					:= virt.o
+obj-$(CONFIG_SMP)			+= platsmp.o

arch/arm/mach-virt/platsmp.c new file +38 -0

new file mode 100644
index 0000000..930362b
@@ -0,0 +1,76 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Author: Will Deacon <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/of.h>
+
+#include <asm/psci.h>
+#include <asm/smp_plat.h>
+#include <asm/hardware/gic.h>
+
+extern void secondary_startup(void);
+
+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {

arch/arm/mach-virt/platsmp.c new file +27 -0

+	set_smp_cross_call(gic_raise_softirq);
+}
+
+static void __init virt_smp_prepare_cpus(unsigned int max_cpus)
+{
+}
+
+static int __cpuinit virt_boot_secondary(unsigned int cpu,
+					 struct task_struct *idle)
+{
+	if (psci_ops.cpu_on)
+		return psci_ops.cpu_on(cpu_logical_map(cpu),
+				       __pa(secondary_startup));
+	return -ENODEV;
+}
+
+static void __cpuinit virt_secondary_init(unsigned int cpu)
+{
+	gic_secondary_init(0);
+}
+
+struct smp_operations __initdata virt_smp_ops = {
+	.smp_init_cpus		= virt_smp_init_cpus,
+	.smp_prepare_cpus	= virt_smp_prepare_cpus,
+	.smp_secondary_init	= virt_secondary_init,
+	.smp_boot_secondary	= virt_boot_secondary,
+};

arch/arm/mach-virt/virt.c +1 -0

index 174b9da..d764835 100644
@@ -20,6 +20,7 @@
 
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/smp.h>
 
 #include <asm/arch_timer.h>
 #include <asm/hardware/gic.h>

arch/arm/mach-virt/virt.c +2 -0

+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

 arch/arm/mach-virt/Kconfig   |  1 +
 arch/arm/mach-virt/Makefile  |  1 +
 arch/arm/mach-virt/platsmp.c | 76 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mach-virt/virt.c    |  6 ++++
 4 files changed, 84 insertions(+)
 create mode 100644 arch/arm/mach-virt/platsmp.c

arch/arm/mach-virt/Kconfig +0 -0

index a568a2a..8958f0d 100644

arch/arm/mach-virt/Makefile +1 -0

index 7ddbfa6..042afc1 100644
@@ -3,3 +3,4 @@
 #
 
 obj-y					:= virt.o
+obj-$(CONFIG_SMP)			+= platsmp.o

arch/arm/mach-virt/platsmp.c new file +54 -0

new file mode 100644
index 0000000..930362b
@@ -0,0 +1,76 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Author: Will Deacon <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/of.h>
+
+#include <asm/psci.h>
+#include <asm/smp_plat.h>
+#include <asm/hardware/gic.h>
+
+extern void secondary_startup(void);
+
+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {
+		if (cpu < NR_CPUS)
+			set_cpu_possible(cpu, true);
+		cpu++;
+	}
+
+	/* sanity check */
+	if (cpu > NR_CPUS)
+		pr_warning("no. of cores (%d) greater than configured maximum "
+			   "of %d - clipping\n",
+			   cpu, NR_CPUS);
+
+	set_smp_cross_call(gic_raise_softirq);
+}
+
+static void __init virt_smp_prepare_cpus(unsigned int max_cpus)
+{

arch/arm/mach-virt/platsmp.c new file +11 -0

+static void __cpuinit virt_secondary_init(unsigned int cpu)
+{
+	gic_secondary_init(0);
+}
+
+struct smp_operations __initdata virt_smp_ops = {
+	.smp_init_cpus		= virt_smp_init_cpus,
+	.smp_prepare_cpus	= virt_smp_prepare_cpus,
+	.smp_secondary_init	= virt_secondary_init,
+	.smp_boot_secondary	= virt_boot_secondary,
+};

arch/arm/mach-virt/virt.c +6 -0

index 174b9da..d764835 100644
@@ -20,6 +20,7 @@
 
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/smp.h>
 
 #include <asm/arch_timer.h>
 #include <asm/hardware/gic.h>

@@ -56,10 +57,15 @@ static struct sys_timer virt_timer = {
 	.init = virt_timer_init,
 };
 
+#ifdef CONFIG_SMP
+extern struct smp_operations virt_smp_ops;
+#endif
+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

Quoted 12 lines · Will Deacon

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

 arch/arm/mach-virt/Kconfig   |  1 +
 arch/arm/mach-virt/Makefile  |  1 +
 arch/arm/mach-virt/platsmp.c | 76 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mach-virt/virt.c    |  6 ++++
 4 files changed, 84 insertions(+)
 create mode 100644 arch/arm/mach-virt/platsmp.c

arch/arm/mach-virt/Kconfig config ARCH_VIRT · line 9

Quoted 12 lines

arch/arm/mach-virt/Kconfig +1 -0

index a568a2a..8958f0d 100644
@@ -3,6 +3,7 @@ config ARCH_VIRT
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select ARM_GIC
 	select ARM_ARCH_TIMER
+	select ARM_PSCI
 	select HAVE_SMP
 	select CPU_V7
 	select SPARSE_IRQ

Considering that PSCI is actually needed only to boot secondary cpus, maybe we want to select it if CONFIG_SMP is enabled?

Quoted 74 lines

arch/arm/mach-virt/Makefile +1 -0

index 7ddbfa6..042afc1 100644
@@ -3,3 +3,4 @@
 #
 
 obj-y					:= virt.o
+obj-$(CONFIG_SMP)			+= platsmp.o

arch/arm/mach-virt/platsmp.c new file +59 -0

new file mode 100644
index 0000000..930362b
@@ -0,0 +1,76 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Author: Will Deacon <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/of.h>
+
+#include <asm/psci.h>
+#include <asm/smp_plat.h>
+#include <asm/hardware/gic.h>
+
+extern void secondary_startup(void);
+
+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {
+		if (cpu < NR_CPUS)
+			set_cpu_possible(cpu, true);
+		cpu++;
+	}
+
+	/* sanity check */
+	if (cpu > NR_CPUS)
+		pr_warning("no. of cores (%d) greater than configured maximum "
+			   "of %d - clipping\n",
+			   cpu, NR_CPUS);
+
+	set_smp_cross_call(gic_raise_softirq);
+}
+
+static void __init virt_smp_prepare_cpus(unsigned int max_cpus)
+{
+}
+
+static int __cpuinit virt_boot_secondary(unsigned int cpu,
+					 struct task_struct *idle)
+{

arch/arm/mach-virt/platsmp.c line 64

Quoted 5 lines

arch/arm/mach-virt/platsmp.c new file +5 -0

+	if (psci_ops.cpu_on)
+		return psci_ops.cpu_on(cpu_logical_map(cpu),
+				       __pa(secondary_startup));
+	return -ENODEV;
+}

Isn't there a better way to check whether PSCI is actually "enabled", as in present in the device tree and initialized correctly?

Maybe we need a pcsi_enabled() static inline of some sort?

Quoted 41 lines

arch/arm/mach-virt/platsmp.c new file +11 -0

+static void __cpuinit virt_secondary_init(unsigned int cpu)
+{
+	gic_secondary_init(0);
+}
+
+struct smp_operations __initdata virt_smp_ops = {
+	.smp_init_cpus		= virt_smp_init_cpus,
+	.smp_prepare_cpus	= virt_smp_prepare_cpus,
+	.smp_secondary_init	= virt_secondary_init,
+	.smp_boot_secondary	= virt_boot_secondary,
+};

arch/arm/mach-virt/virt.c +6 -0

index 174b9da..d764835 100644
@@ -20,6 +20,7 @@
 
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/smp.h>
 
 #include <asm/arch_timer.h>
 #include <asm/hardware/gic.h>

@@ -56,10 +57,15 @@ static struct sys_timer virt_timer = {
 	.init = virt_timer_init,
 };
 
+#ifdef CONFIG_SMP
+extern struct smp_operations virt_smp_ops;
+#endif
+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0

This patch adds support for SMP to mach-virt using the PSCI infrastructure.

Signed-off-by Will Deacon <[email protected]>

 arch/arm/mach-virt/Kconfig   |  1 +
 arch/arm/mach-virt/Makefile  |  1 +
 arch/arm/mach-virt/platsmp.c | 76 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mach-virt/virt.c    |  6 ++++
 4 files changed, 84 insertions(+)
 create mode 100644 arch/arm/mach-virt/platsmp.c

arch/arm/mach-virt/Kconfig +0 -0

index a568a2a..8958f0d 100644

arch/arm/mach-virt/Makefile +1 -0

index 7ddbfa6..042afc1 100644
@@ -3,3 +3,4 @@
 #
 
 obj-y					:= virt.o
+obj-$(CONFIG_SMP)			+= platsmp.o

arch/arm/mach-virt/platsmp.c new file +54 -0

new file mode 100644
index 0000000..930362b
@@ -0,0 +1,76 @@
+/*
+ * Dummy Virtual Machine - does what it says on the tin.
+ *
+ * Copyright (C) 2012 ARM Ltd
+ * Author: Will Deacon <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/of.h>
+
+#include <asm/psci.h>
+#include <asm/smp_plat.h>
+#include <asm/hardware/gic.h>
+
+extern void secondary_startup(void);
+
+/*
+ * Enumerate the possible CPU set from the device tree.
+ */
+static void __init virt_smp_init_cpus(void)
+{
+	struct device_node *dn = NULL;
+	int cpu = 0;
+
+	while ((dn = of_find_node_by_type(dn, "cpu"))) {
+		if (cpu < NR_CPUS)
+			set_cpu_possible(cpu, true);
+		cpu++;
+	}
+
+	/* sanity check */
+	if (cpu > NR_CPUS)
+		pr_warning("no. of cores (%d) greater than configured maximum "
+			   "of %d - clipping\n",
+			   cpu, NR_CPUS);
+
+	set_smp_cross_call(gic_raise_softirq);
+}
+
+static void __init virt_smp_prepare_cpus(unsigned int max_cpus)
+{

arch/arm/mach-virt/platsmp.c new file +11 -0

+static void __cpuinit virt_secondary_init(unsigned int cpu)
+{
+	gic_secondary_init(0);
+}
+
+struct smp_operations __initdata virt_smp_ops = {
+	.smp_init_cpus		= virt_smp_init_cpus,
+	.smp_prepare_cpus	= virt_smp_prepare_cpus,
+	.smp_secondary_init	= virt_secondary_init,
+	.smp_boot_secondary	= virt_boot_secondary,
+};

arch/arm/mach-virt/virt.c +6 -0

index 174b9da..d764835 100644
@@ -20,6 +20,7 @@
 
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/smp.h>
 
 #include <asm/arch_timer.h>
 #include <asm/hardware/gic.h>

@@ -56,10 +57,15 @@ static struct sys_timer virt_timer = {
 	.init = virt_timer_init,
 };
 
+#ifdef CONFIG_SMP
+extern struct smp_operations virt_smp_ops;
+#endif
+
 DT_MACHINE_START(VIRT, "Dummy Virtual Machine")
 	.init_irq	= gic_init_irq,
 	.handle_irq     = gic_handle_irq,
 	.timer		= &virt_timer,
 	.init_machine	= virt_init,
+	.smp		= smp_ops(virt_smp_ops),
 	.dt_compat	= virt_dt_match,
 MACHINE_END

Signature · 1 line

1.8.0