Skip to contentSkip to conversation

Convert highbank to use PSCI calls

PATCH New Rob Herring 3 participants 13 messages Feb 27, 2013 – Feb 28, 2013

Subsystems (6)
lore
Share & tools
lore
#0 lore raw

cover Convert highbank to use PSCI calls

From: Rob Herring <[email protected]>

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

Will, I'm using the id field to specify system level events which are highbank specific. I worry that the id field will be abused for things that could be standardized. Having standard PSCI calls for reset and poweroff would avoid this for highbank.

Rob

Rob Herring (4):
ARM: highbank: clean-up some unused includes
dts: calxeda: add ARM PSCI binding
cpuidle: calxeda: add support to use PSCI calls
ARM: highbank: adapt to use ARM PSCI calls

 arch/arm/boot/dts/ecx-common.dtsi |    8 +++++
 arch/arm/mach-highbank/Kconfig    |    2 +-
 arch/arm/mach-highbank/Makefile   |    2 +-
 arch/arm/mach-highbank/core.h     |    3 --
 arch/arm/mach-highbank/highbank.c |   53 +++++------------------------
 arch/arm/mach-highbank/hotplug.c  |   12 +------
 arch/arm/mach-highbank/platsmp.c  |   42 ++++-------------------
 arch/arm/mach-highbank/pm.c       |   25 +++++---------
 arch/arm/mach-highbank/sysregs.h  |   67 ++++++-------------------------------
 arch/arm/mach-highbank/system.c   |   32 ------------------
 drivers/cpuidle/cpuidle-calxeda.c |   54 ++++--------------------------
 drivers/mailbox/pl320-ipc.c       |    2 +-
 12 files changed, 53 insertions(+), 249 deletions(-)
 delete mode 100644 arch/arm/mach-highbank/system.c
Signature · 1 line
1.7.10.4
#1 lore raw

PATCH 1/4 ARM: highbank: clean-up some unused includes

From: Rob Herring <[email protected]>

---
 arch/arm/mach-highbank/highbank.c |    4 ----
 1 file changed, 4 deletions(-)
arch/arm/mach-highbank/highbank.c +0 -4
index a4f9f50..3a5e62c 100644
@@ -17,14 +17,11 @@
 #include <linux/clkdev.h>
 #include <linux/dma-mapping.h>
 #include <linux/io.h>
-#include <linux/irq.h>
 #include <linux/irqchip.h>
-#include <linux/irqdomain.h>
 #include <linux/of.h>
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
 #include <linux/of_address.h>
-#include <linux/smp.h>
 #include <linux/amba/bus.h>
 #include <linux/clk-provider.h>
 
@@ -38,7 +35,6 @@
 #include <asm/hardware/cache-l2x0.h>
 #include <asm/mach/arch.h>
 #include <asm/mach/map.h>
-#include <asm/mach/time.h>
 
 #include "core.h"
 #include "sysregs.h"
Signature · 1 line
1.7.10.4
#2 lore raw

PATCH 2/4 dts: calxeda: add ARM PSCI binding

In reply to Rob Herring

From: Rob Herring <[email protected]>

Add the PSCI binding for Calxeda SOCs.

---
 arch/arm/boot/dts/ecx-common.dtsi |    8 ++++++++
 1 file changed, 8 insertions(+)
arch/arm/boot/dts/ecx-common.dtsi +8 -0
index d61b535..2c7b31e 100644
@@ -19,6 +19,14 @@
 		bootargs = "console=ttyAMA0";
 	};
 
+	psci {
+		compatible	= "arm,psci";
+		method		= "smc";
+		cpu_suspend	= <0x95c10000>;
+		cpu_off		= <0x95c10001>;
+		cpu_on		= <0x95c10002>;
+	};
+
 	soc {
 		#address-cells = <1>;
 		#size-cells = <1>;
Signature · 1 line
1.7.10.4
#3 lore raw

PATCH 3/4 cpuidle: calxeda: add support to use PSCI calls

In reply to Rob Herring

From: Rob Herring <[email protected]>

This updates the Calxeda cpuidle driver to use PSCI calls to powergate cores. This is needed to enable cpuidle for the ECX-2000.

This could possibly become a generic PSCI driver, but there are no other PSCI users in the kernel other than mach-virt.

---
 drivers/cpuidle/cpuidle-calxeda.c |   54 +++++--------------------------------
 1 file changed, 6 insertions(+), 48 deletions(-)
drivers/cpuidle/cpuidle-calxeda.c +6 -48
index e1aab38..eee8074 100644
@@ -20,67 +20,25 @@
 
 #include <linux/cpuidle.h>
 #include <linux/init.h>
-#include <linux/io.h>
 #include <linux/of.h>
-#include <linux/time.h>
-#include <linux/delay.h>
-#include <linux/suspend.h>
 #include <asm/cpuidle.h>
-#include <asm/proc-fns.h>
-#include <asm/smp_scu.h>
 #include <asm/suspend.h>
-#include <asm/cacheflush.h>
-#include <asm/cp15.h>
-
-extern void highbank_set_cpu_jump(int cpu, void *jump_addr);
-extern void *scu_base_addr;
+#include <asm/psci.h>
 
 static struct cpuidle_device __percpu *calxeda_idle_cpuidle_devices;
 
-static inline unsigned int get_auxcr(void)
-{
-	unsigned int val;
-	asm("mrc p15, 0, %0, c1, c0, 1	@ get AUXCR" : "=r" (val) : : "cc");
-	return val;
-}
-
-static inline void set_auxcr(unsigned int val)
-{
-	asm volatile("mcr p15, 0, %0, c1, c0, 1	@ set AUXCR"
-	  : : "r" (val) : "cc");
-	isb();
-}
-
-static noinline void calxeda_idle_restore(void)
-{
-	set_cr(get_cr() | CR_C);
-	set_auxcr(get_auxcr() | 0x40);
-	scu_power_mode(scu_base_addr, SCU_PM_NORMAL);
-}
-
 static int calxeda_idle_finish(unsigned long val)
 {
-	/* Already flushed cache, but do it again as the outer cache functions
-	 * dirty the cache with spinlocks */
-	flush_cache_all();
-
-	set_auxcr(get_auxcr() & ~0x40);
-	set_cr(get_cr() & ~CR_C);
-
-	scu_power_mode(scu_base_addr, SCU_PM_DORMANT);
-
-	cpu_do_idle();
-
-	/* Restore things if we didn't enter power-gating */
-	calxeda_idle_restore();
-	return 1;
+	const struct psci_power_state ps = {
+		.type = PSCI_POWER_STATE_TYPE_POWER_DOWN,
+	};
+	return psci_ops.cpu_suspend(ps, __pa(cpu_resume));
 }
 
 static int calxeda_pwrdown_idle(struct cpuidle_device *dev,
 				struct cpuidle_driver *drv,
 				int index)
 {
-	highbank_set_cpu_jump(smp_processor_id(), cpu_resume);
 	cpu_suspend(0, calxeda_idle_finish);
 	return index;
 }
@@ -123,7 +81,7 @@ static int __init calxeda_cpuidle_init(void)
 	struct cpuidle_device *dev;
 	struct cpuidle_driver *drv = &calxeda_idle_driver;
 
-	if (!of_machine_is_compatible("calxeda,highbank"))
+	if (!of_machine_is_compatible("calxeda,highbank") || !psci_ops.cpu_suspend)
 		return -ENODEV;
 
 	ret = cpuidle_register_driver(drv);
Signature · 1 line
1.7.10.4
#4 lore raw

PATCH 4/4 ARM: highbank: adapt to use ARM PSCI calls

In reply to Rob Herring

From: Rob Herring <[email protected]>

This adapts highbank to use the ARM PSCI calls for cpu power control. Secondary core boot, cpu hotplug, suspend/resume, reset and poweroff are all converted to use PSCI calls. Doing this removes direct access to the SCU and factors out the A9 vs. A15 differences.

---
 arch/arm/mach-highbank/Kconfig    |    2 +-
 arch/arm/mach-highbank/Makefile   |    2 +-
 arch/arm/mach-highbank/core.h     |    3 --
 arch/arm/mach-highbank/highbank.c |   49 +++++----------------------
 arch/arm/mach-highbank/hotplug.c  |   12 +------
 arch/arm/mach-highbank/platsmp.c  |   42 ++++-------------------
 arch/arm/mach-highbank/pm.c       |   25 +++++---------
 arch/arm/mach-highbank/sysregs.h  |   67 ++++++-------------------------------
 arch/arm/mach-highbank/system.c   |   32 ------------------
 drivers/mailbox/pl320-ipc.c       |    2 +-
 10 files changed, 39 insertions(+), 197 deletions(-)
 delete mode 100644 arch/arm/mach-highbank/system.c
arch/arm/mach-highbank/Kconfig +1 -1
index 44b12f9..9dc0c9e 100644
@@ -5,13 +5,13 @@ config ARCH_HIGHBANK
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select ARM_AMBA
 	select ARM_GIC
+	select ARM_PSCI
 	select ARM_TIMER_SP804
 	select CACHE_L2X0
 	select CLKDEV_LOOKUP
 	select COMMON_CLK
 	select CPU_V7
 	select GENERIC_CLOCKEVENTS
-	select HAVE_ARM_SCU
 	select HAVE_SMP
 	select MAILBOX
 	select PL320_MBOX
arch/arm/mach-highbank/Makefile +1 -1
index 8a1ef57..c1dde52 100644
@@ -1,4 +1,4 @@
-obj-y					:= highbank.o system.o smc.o
+obj-y					:= highbank.o smc.o
 
 plus_sec := $(call as-instr,.arch_extension sec,+sec)
 AFLAGS_smc.o				:=-Wa,-march=armv7-a$(plus_sec)
arch/arm/mach-highbank/core.h +0 -3
index 3f65206..b8bcde7 100644
@@ -1,9 +1,6 @@
 #ifndef __HIGHBANK_CORE_H
 #define __HIGHBANK_CORE_H
 
-extern void highbank_set_cpu_jump(int cpu, void *jump_addr);
-extern void highbank_restart(char, const char *);
-extern void __iomem *scu_base_addr;
 
 #ifdef CONFIG_PM_SLEEP
 extern void highbank_pm_init(void);
arch/arm/mach-highbank/highbank.c +8 -41

77 lines not on this page · raw message

arch/arm/mach-highbank/hotplug.c +1 -11
index f30c528..dc65b0f 100644
@@ -15,25 +15,15 @@
  */
 #include <linux/kernel.h>
 
-#include <asm/cacheflush.h>
-
-#include "core.h"
 #include "sysregs.h"
 
-extern void secondary_startup(void);
-
 /*
  * platform-specific code to shutdown a CPU
  *
  */
 void __ref highbank_cpu_die(unsigned int cpu)
 {
-	flush_cache_all();
-
-	highbank_set_cpu_jump(cpu, phys_to_virt(0));
-	highbank_set_core_pwr();
-
-	cpu_do_idle();
+	highbank_cpu_off(0, 0);
 
 	/* We should never return from idle */
 	panic("highbank: cpu %d unexpectedly exit from shutdown\n", cpu);
arch/arm/mach-highbank/platsmp.c +7 -35

58 lines not on this page · raw message

arch/arm/mach-highbank/pm.c +9 -16
index 04eddb4..d1f511b 100644
@@ -16,27 +16,21 @@
 
 #include <linux/cpu_pm.h>
 #include <linux/init.h>
-#include <linux/io.h>
 #include <linux/suspend.h>
 
-#include <asm/cacheflush.h>
-#include <asm/proc-fns.h>
 #include <asm/suspend.h>
 
-#include "core.h"
 #include "sysregs.h"
 
 static int highbank_suspend_finish(unsigned long val)
 {
-	outer_flush_all();
-	outer_disable();
+	const struct psci_power_state ps = {
+		.id = HB_PWR_SUSPEND,
+		.type = PSCI_POWER_STATE_TYPE_POWER_DOWN,
+		.affinity_level = 1,
+	};
 
-	highbank_set_pwr_suspend();
-
-	cpu_do_idle();
-
-	highbank_clear_pwr_request();
-	return 0;
+	return psci_ops.cpu_suspend(ps, __pa(cpu_resume));
 }
 
 static int highbank_pm_enter(suspend_state_t state)
@@ -44,15 +38,11 @@ static int highbank_pm_enter(suspend_state_t state)
 	cpu_pm_enter();
 	cpu_cluster_pm_enter();
 
-	highbank_set_cpu_jump(0, cpu_resume);
 	cpu_suspend(0, highbank_suspend_finish);
 
 	cpu_cluster_pm_exit();
 	cpu_pm_exit();
 
-	highbank_smc1(0x102, 0x1);
-	if (scu_base_addr)
-		scu_enable(scu_base_addr);
 	return 0;
 }
 
@@ -63,5 +53,8 @@ static const struct platform_suspend_ops highbank_pm_ops = {
 
 void __init highbank_pm_init(void)
 {
+	if (!psci_ops.cpu_suspend)
+		return;
+
 	suspend_set_ops(&highbank_pm_ops);
 }
arch/arm/mach-highbank/sysregs.h +11 -56

82 lines not on this page · raw message

arch/arm/mach-highbank/system.c deleted +0 -32
deleted file mode 100644
index 37d8384..0000000
@@ -1,32 +0,0 @@
-/*
- * Copyright 2011 Calxeda, Inc.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope 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/io.h>
-#include <asm/proc-fns.h>
-
-#include "core.h"
-#include "sysregs.h"
-
-void highbank_restart(char mode, const char *cmd)
-{
-	if (mode == 'h')
-		highbank_set_pwr_hard_reset();
-	else
-		highbank_set_pwr_soft_reset();
-
-	while (1)
-		cpu_do_idle();
-}
-
drivers/mailbox/pl320-ipc.c +1 -1
index c45b3ae..4127e2e 100644
@@ -196,4 +196,4 @@ static int __init ipc_init(void)
 {
 	return amba_driver_register(&pl320_driver);
 }
-module_init(ipc_init);
+arch_initcall(ipc_init);
Signature · 1 line
1.7.10.4
#5 lore raw

cover Convert highbank to use PSCI calls

In reply to Rob Herring
Quoted 6 lines · Rob Herring

From: Rob Herring <[email protected]>

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

Hi Rob,

This looks pretty cool. Can you or Will comment on what is still left for the SMP ops that is actually required as platform specific code?

Specifically, I'm wondering if we could have a set of default smp_ops based on this that we could use with the default machine descriptor [1] and any platform that has SMP enabled but a NULL smp_ops pointer.

Arnd

[1] https://patchwork.kernel.org/patch/2074871/

#6 lore raw

cover Convert highbank to use PSCI calls

Quoted 8 lines · Arnd Bergmann
Quoted 6 lines · Rob Herring

From: Rob Herring <[email protected]>

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

Quoted 8 lines

Hi Rob,

This looks pretty cool. Can you or Will comment on what is still left for the SMP ops that is actually required as platform specific code?

Specifically, I'm wondering if we could have a set of default smp_ops based on this that we could use with the default machine descriptor [1] and any platform that has SMP enabled but a NULL smp_ops pointer.

My smp code is exactly the same as mach-virt with this, but adds cpu_die for hotplug. smp_secondary_init will go away with Catalin's series to use notifiers for the gic secondary init. This leaves smp_boot_secondary.

We could move these remaining smp_ops to psci.c and create psci_smp_ops. That probably needs to be after the gic notifier changes.

Your patch, psci smp ops, and converting the arch timers to use OF_CLKSRC_DECLARE will eliminate mach-virt unless Will has more things planned for it.

Rob

#7 lore raw

cover Convert highbank to use PSCI calls

Quoted 17 lines · Rob Herring
Quoted 11 lines · Arnd Bergmann
Quoted 4 lines

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

This looks pretty cool. Can you or Will comment on what is still left for the SMP ops that is actually required as platform specific code?

Specifically, I'm wondering if we could have a set of default smp_ops based on this that we could use with the default machine descriptor [1] and any platform that has SMP enabled but a NULL smp_ops pointer.

My smp code is exactly the same as mach-virt with this, but adds cpu_die for hotplug. smp_secondary_init will go away with Catalin's series to use notifiers for the gic secondary init. This leaves smp_boot_secondary.

Quoted 6 lines

We could move these remaining smp_ops to psci.c and create psci_smp_ops. That probably needs to be after the gic notifier changes.

Your patch, psci smp ops, and converting the arch timers to use OF_CLKSRC_DECLARE will eliminate mach-virt unless Will has more things planned for it.

As long as we remain compatible with "linux,dummy-virt" then I'm not fussed what happens to the support code. I don't envisage adding anything else to it that can't live under drivers/ (e.g. new virtio devices).

Will

#10 lore raw

cover Convert highbank to use PSCI calls

In reply to Rob Herring

Hi Rob,

Quoted 6 lines · Rob Herring

From: Rob Herring <[email protected]>

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

This is great, nice work!

Quoted 4 lines

Will, I'm using the id field to specify system level events which are highbank specific. I worry that the id field will be abused for things that could be standardized. Having standard PSCI calls for reset and poweroff would avoid this for highbank.

I don't like using the ID field for this because it places some of the number space in the device tree binding and other parts in the platform code. It might be better to use the bottom 16 bits of the power_state argument to CPU_OFF (which are designated as platform specific anyway) to indicate that you actually want a system-wide operation to take place.

Since this document is currently undergoing a fairly major update (which will hopefully be released soon), I'll go and check with the authors to see if they have already considered this.

Cheers,

Will

#11 lore raw

cover Convert highbank to use PSCI calls

Quoted 17 lines · Will Deacon

Hi Rob,

Quoted 6 lines · Rob Herring

From: Rob Herring <[email protected]>

This series converts highbank to use PSCI calls for secondary core boot, hotplug, cpuidle, suspend/resume, reset, and poweroff. Doing this removes about 200 lines of platform code (1/3 of it) and abstracts out the A9 and A15 differences (some of which are not yet in place).

This is great, nice work!

Quoted 4 lines

Will, I'm using the id field to specify system level events which are highbank specific. I worry that the id field will be abused for things that could be standardized. Having standard PSCI calls for reset and poweroff would avoid this for highbank.

Quoted 5 lines

I don't like using the ID field for this because it places some of the number space in the device tree binding and other parts in the platform code. It might be better to use the bottom 16 bits of the power_state argument to CPU_OFF (which are designated as platform specific anyway) to indicate that you actually want a system-wide operation to take place.

The platform specific ID in the power state field of CPU_OFF is what I'm using. But there is no reason for this to be platform specific when possibly all platforms need this.

Quoted 3 lines

Since this document is currently undergoing a fairly major update (which will hopefully be released soon), I'll go and check with the authors to see if they have already considered this.

It would be nice to have some visibility to this as this is an ABI and will break if things change.

Rob

#8 lore raw

cover Convert highbank to use PSCI calls

In reply to Will Deacon
Quoted 3 lines · Will Deacon

As long as we remain compatible with "linux,dummy-virt" then I'm not fussed what happens to the support code. I don't envisage adding anything else to it that can't live under drivers/ (e.g. new virtio devices).

The idea with my patch is that you don't actually have to check anything against "linux,dummy-virt" or another top-level "compatible value.

Basically any device tree would boot as long as all device you want to use (including timer, irqchip, smp CPU setup method, ...) can be brought up using the DT information. The SMP stuff is the only bit missing right now, any other field of the machine_desc structure is already optional with my patch, and if all fields are NULL, you can leave out the entire structure. The best part of this is that an empty machine_desc is exactly what we have on arm64, so any machine that is supported on arm64 could also be supported on arch/arm without any machine specific code.

Arnd

#9 lore raw

cover Convert highbank to use PSCI calls

Quoted 7 lines · Arnd Bergmann
Quoted 3 lines · Will Deacon

As long as we remain compatible with "linux,dummy-virt" then I'm not fussed what happens to the support code. I don't envisage adding anything else to it that can't live under drivers/ (e.g. new virtio devices).

The idea with my patch is that you don't actually have to check anything against "linux,dummy-virt" or another top-level "compatible value.

Great, I've got no problem killing of the short-lived mach-virt then!

Quoted 9 lines

Basically any device tree would boot as long as all device you want to use (including timer, irqchip, smp CPU setup method, ...) can be brought up using the DT information. The SMP stuff is the only bit missing right now, any other field of the machine_desc structure is already optional with my patch, and if all fields are NULL, you can leave out the entire structure. The best part of this is that an empty machine_desc is exactly what we have on arm64, so any machine that is supported on arm64 could also be supported on arch/arm without any machine specific code.

Keeping on the same topic: using a PSCI implementation as the default SMP ops would extend the arm/arm64 compatibility to SMP booting, as both are PSCI-capable.

Will

#12 lore raw

cover Convert highbank to use PSCI calls

In reply to Rob Herring
Quoted 13 lines · Rob Herring
Quoted 11 lines · Will Deacon
Quoted 4 lines · Rob Herring

Will, I'm using the id field to specify system level events which are highbank specific. I worry that the id field will be abused for things that could be standardized. Having standard PSCI calls for reset and poweroff would avoid this for highbank.

I don't like using the ID field for this because it places some of the number space in the device tree binding and other parts in the platform code. It might be better to use the bottom 16 bits of the power_state argument to CPU_OFF (which are designated as platform specific anyway) to indicate that you actually want a system-wide operation to take place.

Quoted 3 lines

The platform specific ID in the power state field of CPU_OFF is what I'm using. But there is no reason for this to be platform specific when possibly all platforms need this.

Yes, sorry, I misread your description above and assumed you were referring to the *function* ID.

Quoted 6 lines
Quoted 3 lines

Since this document is currently undergoing a fairly major update (which will hopefully be released soon), I'll go and check with the authors to see if they have already considered this.

It would be nice to have some visibility to this as this is an ABI and will break if things change.

As long as things are versioned sensibly, we just don't pretend to be compatible with versions of the ABI that we don't support; otherwise PSCI would be unable to evolve as a specification. The main thing is to avoid implementing features as platform-specific parts when an upcoming version of the specification actually describes them in a different way.

We should talk more about this at Connect, where hopefully some other platforms can be converted too.

Will