[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[Tony Lindgren]

Hi all,

Here are some fixes to get minimal cpuidle support working with omap5
and dra7. Not sure if there are still some unsorted issues on enabling
this on dra7, but at least omap5 has been behving for me for few
weeks with these.

Regards,

Tony


Santosh Shilimkar (1):
  ARM: DRA7: PM: cpuidle MPU CSWR support

Tony Lindgren (4):
  ARM: OMAP5: Fix build for PM code
  ARM: OMAP5: Fix mpuss_early_init
  ARM: OMAP4+: Fix bad fallthrough for cpuidle
  ARM: OMAP5: Enable minimal cpuidle for omap5 retention

 arch/arm/mach-omap2/Makefile              |  2 +-
 arch/arm/mach-omap2/cpuidle44xx.c         | 80 ++++++++++++++++++++++++++++++-
 arch/arm/mach-omap2/io.c                  |  3 +-
 arch/arm/mach-omap2/omap-mpuss-lowpower.c | 33 +++++++++----
 arch/arm/mach-omap2/omap4-sar-layout.h    |  2 +
 arch/arm/mach-omap2/pm44xx.c              |  2 +-
 6 files changed, 108 insertions(+), 14 deletions(-)

-- 
2.9.3

[PATCH 1/5] ARM: OMAP5: Fix build for PM code

[Tony Lindgren]

It's CONFIG_SOC_OMAP5, not CONFIG_ARCH_OMAP5. Looks like make randconfig
builds have not hit this one yet.

Fixes: b3bf289c1c45 ("ARM: OMAP2+: Fix build with CONFIG_SMP and CONFIG_PM
is not set")
Signed-off-by: Tony Lindgren <tony@atomide.com>
---
 arch/arm/mach-omap2/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/mach-omap2/Makefile b/arch/arm/mach-omap2/Makefile
--- a/arch/arm/mach-omap2/Makefile
+++ b/arch/arm/mach-omap2/Makefile
@@ -80,7 +80,7 @@ endif
 # Power Management
 omap-4-5-pm-common			= omap-mpuss-lowpower.o
 obj-$(CONFIG_ARCH_OMAP4)		+= $(omap-4-5-pm-common)
-obj-$(CONFIG_ARCH_OMAP5)		+= $(omap-4-5-pm-common)
+obj-$(CONFIG_SOC_OMAP5)			+= $(omap-4-5-pm-common)
 obj-$(CONFIG_OMAP_PM_NOOP)		+= omap-pm-noop.o
 
 ifeq ($(CONFIG_PM),y)
-- 
2.9.3

[PATCH 2/5] ARM: OMAP5: Fix mpuss_early_init

[Tony Lindgren]

We need to properly initialize mpuss also on omap5 like we do on omap4.
Otherwise we run into similar kexec problems like we had on omap4 when
trying to kexec from a kernel with PM initialized.

Fixes: 0573b957fc21 ("ARM: OMAP4+: Prevent CPU1 related hang with kexec")
Signed-off-by: Tony Lindgren <tony@atomide.com>
---
 arch/arm/mach-omap2/io.c                  |  3 ++-
 arch/arm/mach-omap2/omap-mpuss-lowpower.c | 28 +++++++++++++++++++++-------
 arch/arm/mach-omap2/omap4-sar-layout.h    |  2 ++
 3 files changed, 25 insertions(+), 8 deletions(-)

diff --git a/arch/arm/mach-omap2/io.c b/arch/arm/mach-omap2/io.c
--- a/arch/arm/mach-omap2/io.c
+++ b/arch/arm/mach-omap2/io.c
@@ -717,10 +717,11 @@ void __init omap5_init_early(void)
 			      OMAP2_L4_IO_ADDRESS(OMAP54XX_SCM_BASE));
 	omap2_set_globals_prcm_mpu(OMAP2_L4_IO_ADDRESS(OMAP54XX_PRCM_MPU_BASE));
 	omap2_control_base_init();
-	omap4_pm_init_early();
 	omap2_prcm_base_init();
 	omap5xxx_check_revision();
 	omap4_sar_ram_init();
+	omap4_mpuss_early_init();
+	omap4_pm_init_early();
 	omap54xx_voltagedomains_init();
 	omap54xx_powerdomains_init();
 	omap54xx_clockdomains_init();
diff --git a/arch/arm/mach-omap2/omap-mpuss-lowpower.c b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
--- a/arch/arm/mach-omap2/omap-mpuss-lowpower.c
+++ b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
@@ -371,8 +371,12 @@ int __init omap4_mpuss_init(void)
 	pm_info = &per_cpu(omap4_pm_info, 0x0);
 	if (sar_base) {
 		pm_info->scu_sar_addr = sar_base + SCU_OFFSET0;
-		pm_info->wkup_sar_addr = sar_base +
-					CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
+		if (cpu_is_omap44xx())
+			pm_info->wkup_sar_addr = sar_base +
+				CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
+		else
+			pm_info->wkup_sar_addr = sar_base +
+				OMAP5_CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
 		pm_info->l2x0_sar_addr = sar_base + L2X0_SAVE_OFFSET0;
 	}
 	pm_info->pwrdm = pwrdm_lookup("cpu0_pwrdm");
@@ -391,8 +395,12 @@ int __init omap4_mpuss_init(void)
 	pm_info = &per_cpu(omap4_pm_info, 0x1);
 	if (sar_base) {
 		pm_info->scu_sar_addr = sar_base + SCU_OFFSET1;
-		pm_info->wkup_sar_addr = sar_base +
-					CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
+		if (cpu_is_omap44xx())
+			pm_info->wkup_sar_addr = sar_base +
+				CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
+		else
+			pm_info->wkup_sar_addr = sar_base +
+				OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
 		pm_info->l2x0_sar_addr = sar_base + L2X0_SAVE_OFFSET1;
 	}
 
@@ -453,15 +461,21 @@ void __init omap4_mpuss_early_init(void)
 {
 	unsigned long startup_pa;
 
-	if (!cpu_is_omap44xx())
+	if (!(cpu_is_omap44xx() || soc_is_omap54xx()))
 		return;
 
 	sar_base = omap4_get_sar_ram_base();
 
 	if (cpu_is_omap443x())
 		startup_pa = virt_to_phys(omap4_secondary_startup);
-	else
+	else if (cpu_is_omap446x())
 		startup_pa = virt_to_phys(omap4460_secondary_startup);
+	else
+		startup_pa = virt_to_phys(omap5_secondary_startup);
 
-	writel_relaxed(startup_pa, sar_base + CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
+	if (cpu_is_omap44xx())
+		writel_relaxed(startup_pa, sar_base + CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
+	else
+		writel_relaxed(startup_pa, sar_base +
+			       OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
 }
diff --git a/arch/arm/mach-omap2/omap4-sar-layout.h b/arch/arm/mach-omap2/omap4-sar-layout.h
--- a/arch/arm/mach-omap2/omap4-sar-layout.h
+++ b/arch/arm/mach-omap2/omap4-sar-layout.h
@@ -31,6 +31,8 @@
 /* CPUx Wakeup Non-Secure Physical Address offsets in SAR_BANK3 */
 #define CPU0_WAKEUP_NS_PA_ADDR_OFFSET		0xa04
 #define CPU1_WAKEUP_NS_PA_ADDR_OFFSET		0xa08
+#define OMAP5_CPU0_WAKEUP_NS_PA_ADDR_OFFSET	0xe00
+#define OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET	0xe04
 
 #define SAR_BACKUP_STATUS_OFFSET		(SAR_BANK3_OFFSET + 0x500)
 #define SAR_SECURE_RAM_SIZE_OFFSET		(SAR_BANK3_OFFSET + 0x504)
-- 
2.9.3

[PATCH 2/5] ARM: OMAP5: Fix mpuss_early_init

[Tony Lindgren]

* Tony Lindgren <tony@atomide.com> [161026 08:18]:
> We need to properly initialize mpuss also on omap5 like we do on omap4.
> Otherwise we run into similar kexec problems like we had on omap4 when
> trying to kexec from a kernel with PM initialized.
...

> +	else if (cpu_is_omap446x())
>  		startup_pa = virt_to_phys(omap4460_secondary_startup);
> +	else
> +		startup_pa = virt_to_phys(omap5_secondary_startup);

Here we need to check for omap5_secondary_hyp_startup too. And with
omap4 and omap5 being optional, we need to fix up things for make
randconfig builds as reported by the kbuild test robot.

Updated patch below.

Regards,

Tony

8< -----------------------
>From tony Mon Sep 17 00:00:00 2001
From: Tony Lindgren <tony@atomide.com>
Date: Tue, 25 Oct 2016 08:33:32 -0700
Subject: [PATCH] ARM: OMAP5: Fix mpuss_early_init

We need to properly initialize mpuss also on omap5 like we do on omap4.
Otherwise we run into similar kexec problems like we had on omap4 when
trying to kexec from a kernel with PM initialized.

Fixes: 0573b957fc21 ("ARM: OMAP4+: Prevent CPU1 related hang with kexec")
Signed-off-by: Tony Lindgren <tony@atomide.com>
---
 arch/arm/mach-omap2/common.h              | 38 +++++++++++++++++++++++--------
 arch/arm/mach-omap2/io.c                  |  3 ++-
 arch/arm/mach-omap2/omap-mpuss-lowpower.c | 32 ++++++++++++++++++++------
 arch/arm/mach-omap2/omap4-sar-layout.h    |  2 ++
 4 files changed, 58 insertions(+), 17 deletions(-)

diff --git a/arch/arm/mach-omap2/common.h b/arch/arm/mach-omap2/common.h
--- a/arch/arm/mach-omap2/common.h
+++ b/arch/arm/mach-omap2/common.h
@@ -262,8 +262,6 @@ extern void __iomem *omap4_get_sar_ram_base(void);
 extern void omap4_mpuss_early_init(void);
 extern void omap_do_wfi(void);
 
-extern void omap4_secondary_startup(void);
-extern void omap4460_secondary_startup(void);
 
 #ifdef CONFIG_SMP
 /* Needed for secondary core boot */
@@ -275,16 +273,11 @@ extern void omap4_cpu_die(unsigned int cpu);
 extern int omap4_cpu_kill(unsigned int cpu);
 
 extern const struct smp_operations omap4_smp_ops;
-
-extern void omap5_secondary_startup(void);
-extern void omap5_secondary_hyp_startup(void);
 #endif
 
 #if defined(CONFIG_SMP) && defined(CONFIG_PM)
 extern int omap4_mpuss_init(void);
 extern int omap4_enter_lowpower(unsigned int cpu, unsigned int power_state);
-extern int omap4_finish_suspend(unsigned long cpu_state);
-extern void omap4_cpu_resume(void);
 extern int omap4_hotplug_cpu(unsigned int cpu, unsigned int power_state);
 #else
 static inline int omap4_enter_lowpower(unsigned int cpu,
@@ -305,14 +298,41 @@ static inline int omap4_mpuss_init(void)
 	return 0;
 }
 
+#endif
+
+#ifdef CONFIG_ARCH_OMAP4
+void omap4_secondary_startup(void);
+void omap4460_secondary_startup(void);
+int omap4_finish_suspend(unsigned long cpu_state);
+void omap4_cpu_resume(void);
+#else
+static inline void omap4_secondary_startup(void)
+{
+}
+
+static inline void omap4460_secondary_startup(void)
+{
+}
 static inline int omap4_finish_suspend(unsigned long cpu_state)
 {
 	return 0;
 }
-
 static inline void omap4_cpu_resume(void)
-{}
+{
+}
+#endif
 
+#if defined(CONFIG_SOC_OMAP5) || defined(CONFIG_SOC_DRA7XX)
+void omap5_secondary_startup(void);
+void omap5_secondary_hyp_startup(void);
+#else
+static inline void omap5_secondary_startup(void)
+{
+}
+
+static inline void omap5_secondary_hyp_startup(void)
+{
+}
 #endif
 
 void pdata_quirks_init(const struct of_device_id *);
diff --git a/arch/arm/mach-omap2/io.c b/arch/arm/mach-omap2/io.c
--- a/arch/arm/mach-omap2/io.c
+++ b/arch/arm/mach-omap2/io.c
@@ -717,10 +717,11 @@ void __init omap5_init_early(void)
 			      OMAP2_L4_IO_ADDRESS(OMAP54XX_SCM_BASE));
 	omap2_set_globals_prcm_mpu(OMAP2_L4_IO_ADDRESS(OMAP54XX_PRCM_MPU_BASE));
 	omap2_control_base_init();
-	omap4_pm_init_early();
 	omap2_prcm_base_init();
 	omap5xxx_check_revision();
 	omap4_sar_ram_init();
+	omap4_mpuss_early_init();
+	omap4_pm_init_early();
 	omap54xx_voltagedomains_init();
 	omap54xx_powerdomains_init();
 	omap54xx_clockdomains_init();
diff --git a/arch/arm/mach-omap2/omap-mpuss-lowpower.c b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
--- a/arch/arm/mach-omap2/omap-mpuss-lowpower.c
+++ b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
@@ -48,6 +48,7 @@
 #include <asm/smp_scu.h>
 #include <asm/pgalloc.h>
 #include <asm/suspend.h>
+#include <asm/virt.h>
 #include <asm/hardware/cache-l2x0.h>
 
 #include "soc.h"
@@ -371,8 +372,12 @@ int __init omap4_mpuss_init(void)
 	pm_info = &per_cpu(omap4_pm_info, 0x0);
 	if (sar_base) {
 		pm_info->scu_sar_addr = sar_base + SCU_OFFSET0;
-		pm_info->wkup_sar_addr = sar_base +
-					CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
+		if (cpu_is_omap44xx())
+			pm_info->wkup_sar_addr = sar_base +
+				CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
+		else
+			pm_info->wkup_sar_addr = sar_base +
+				OMAP5_CPU0_WAKEUP_NS_PA_ADDR_OFFSET;
 		pm_info->l2x0_sar_addr = sar_base + L2X0_SAVE_OFFSET0;
 	}
 	pm_info->pwrdm = pwrdm_lookup("cpu0_pwrdm");
@@ -391,8 +396,12 @@ int __init omap4_mpuss_init(void)
 	pm_info = &per_cpu(omap4_pm_info, 0x1);
 	if (sar_base) {
 		pm_info->scu_sar_addr = sar_base + SCU_OFFSET1;
-		pm_info->wkup_sar_addr = sar_base +
-					CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
+		if (cpu_is_omap44xx())
+			pm_info->wkup_sar_addr = sar_base +
+				CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
+		else
+			pm_info->wkup_sar_addr = sar_base +
+				OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET;
 		pm_info->l2x0_sar_addr = sar_base + L2X0_SAVE_OFFSET1;
 	}
 
@@ -453,15 +462,24 @@ void __init omap4_mpuss_early_init(void)
 {
 	unsigned long startup_pa;
 
-	if (!cpu_is_omap44xx())
+	if (!(cpu_is_omap44xx() || soc_is_omap54xx()))
 		return;
 
 	sar_base = omap4_get_sar_ram_base();
 
 	if (cpu_is_omap443x())
 		startup_pa = virt_to_phys(omap4_secondary_startup);
-	else
+	else if (cpu_is_omap446x())
 		startup_pa = virt_to_phys(omap4460_secondary_startup);
+	else if ((__boot_cpu_mode & MODE_MASK) == HYP_MODE)
+		startup_pa = virt_to_phys(omap5_secondary_hyp_startup);
+	else
+		startup_pa = virt_to_phys(omap5_secondary_startup);
 
-	writel_relaxed(startup_pa, sar_base + CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
+	if (cpu_is_omap44xx())
+		writel_relaxed(startup_pa, sar_base +
+			       CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
+	else
+		writel_relaxed(startup_pa, sar_base +
+			       OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET);
 }
diff --git a/arch/arm/mach-omap2/omap4-sar-layout.h b/arch/arm/mach-omap2/omap4-sar-layout.h
--- a/arch/arm/mach-omap2/omap4-sar-layout.h
+++ b/arch/arm/mach-omap2/omap4-sar-layout.h
@@ -31,6 +31,8 @@
 /* CPUx Wakeup Non-Secure Physical Address offsets in SAR_BANK3 */
 #define CPU0_WAKEUP_NS_PA_ADDR_OFFSET		0xa04
 #define CPU1_WAKEUP_NS_PA_ADDR_OFFSET		0xa08
+#define OMAP5_CPU0_WAKEUP_NS_PA_ADDR_OFFSET	0xe00
+#define OMAP5_CPU1_WAKEUP_NS_PA_ADDR_OFFSET	0xe04
 
 #define SAR_BACKUP_STATUS_OFFSET		(SAR_BANK3_OFFSET + 0x500)
 #define SAR_SECURE_RAM_SIZE_OFFSET		(SAR_BANK3_OFFSET + 0x504)
-- 
2.9.3

[PATCH 3/5] ARM: OMAP4+: Fix bad fallthrough for cpuidle

[Tony Lindgren]

We don't want to fall through to a bunch of errors for retention
if PM_OMAP4_CPU_OSWR_DISABLE is not configured for a SoC.

Fixes: 6099dd37c669 ("ARM: OMAP5 / DRA7: Enable CPU RET on suspend")
Signed-off-by: Tony Lindgren <tony@atomide.com>
---
 arch/arm/mach-omap2/omap-mpuss-lowpower.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-omap2/omap-mpuss-lowpower.c b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
--- a/arch/arm/mach-omap2/omap-mpuss-lowpower.c
+++ b/arch/arm/mach-omap2/omap-mpuss-lowpower.c
@@ -244,10 +244,9 @@ int omap4_enter_lowpower(unsigned int cpu, unsigned int power_state)
 		save_state = 1;
 		break;
 	case PWRDM_POWER_RET:
-		if (IS_PM44XX_ERRATUM(PM_OMAP4_CPU_OSWR_DISABLE)) {
+		if (IS_PM44XX_ERRATUM(PM_OMAP4_CPU_OSWR_DISABLE))
 			save_state = 0;
-			break;
-		}
+		break;
 	default:
 		/*
 		 * CPUx CSWR is invalid hardware state. Also CPUx OSWR
-- 
2.9.3

[PATCH 4/5] ARM: DRA7: PM: cpuidle MPU CSWR support

[Tony Lindgren]

From: Santosh Shilimkar <santosh.shilimkar@ti.com>

Add DRA74/72 CPUIDLE support.

This patch adds MPUSS low power states in cpuidle.

        C1 - CPU0 WFI + CPU1 WFI + MPU ON
        C2 - CPU0 RET + CPU1 RET + MPU CSWR

Tested on DRA74/72-EVM for C1 and C2 states.

NOTE: DRA7 does not do voltage scaling as part of retention transition
and has Mercury which speeds up transition paths - Latency numbers are
based on measurements done by toggling GPIOs.

Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[ j-keerthy at ti.com rework on 3.14]
Signed-off-by: Keerthy <j-keerthy@ti.com>
[nm at ti.com: updates based on profiling]
[tony at atomide.com: dropped CPUIDLE_FLAG_TIME_VALID no longer used]
Signed-off-by: Nishanth Menon <nm@ti.com>
---
 arch/arm/mach-omap2/cpuidle44xx.c | 80 ++++++++++++++++++++++++++++++++++++++-
 arch/arm/mach-omap2/pm44xx.c      |  2 +-
 2 files changed, 80 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-omap2/cpuidle44xx.c b/arch/arm/mach-omap2/cpuidle44xx.c
--- a/arch/arm/mach-omap2/cpuidle44xx.c
+++ b/arch/arm/mach-omap2/cpuidle44xx.c
@@ -21,6 +21,7 @@
 #include "common.h"
 #include "pm.h"
 #include "prm.h"
+#include "soc.h"
 #include "clockdomain.h"
 
 #define MAX_CPUS	2
@@ -30,6 +31,7 @@ struct idle_statedata {
 	u32 cpu_state;
 	u32 mpu_logic_state;
 	u32 mpu_state;
+	u32 mpu_state_vote;
 };
 
 static struct idle_statedata omap4_idle_data[] = {
@@ -50,12 +52,26 @@ static struct idle_statedata omap4_idle_data[] = {
 	},
 };
 
+static struct idle_statedata dra7_idle_data[] = {
+	{
+		.cpu_state = PWRDM_POWER_ON,
+		.mpu_state = PWRDM_POWER_ON,
+		.mpu_logic_state = PWRDM_POWER_ON,
+	},
+	{
+		.cpu_state = PWRDM_POWER_RET,
+		.mpu_state = PWRDM_POWER_RET,
+		.mpu_logic_state = PWRDM_POWER_RET,
+	},
+};
+
 static struct powerdomain *mpu_pd, *cpu_pd[MAX_CPUS];
 static struct clockdomain *cpu_clkdm[MAX_CPUS];
 
 static atomic_t abort_barrier;
 static bool cpu_done[MAX_CPUS];
 static struct idle_statedata *state_ptr = &omap4_idle_data[0];
+static DEFINE_RAW_SPINLOCK(mpu_lock);
 
 /* Private functions */
 
@@ -77,6 +93,32 @@ static int omap_enter_idle_simple(struct cpuidle_device *dev,
 	return index;
 }
 
+static int omap_enter_idle_smp(struct cpuidle_device *dev,
+			       struct cpuidle_driver *drv,
+			       int index)
+{
+	struct idle_statedata *cx = state_ptr + index;
+	unsigned long flag;
+
+	raw_spin_lock_irqsave(&mpu_lock, flag);
+	cx->mpu_state_vote++;
+	if (cx->mpu_state_vote == num_online_cpus()) {
+		pwrdm_set_logic_retst(mpu_pd, cx->mpu_logic_state);
+		omap_set_pwrdm_state(mpu_pd, cx->mpu_state);
+	}
+	raw_spin_unlock_irqrestore(&mpu_lock, flag);
+
+	omap4_enter_lowpower(dev->cpu, cx->cpu_state);
+
+	raw_spin_lock_irqsave(&mpu_lock, flag);
+	if (cx->mpu_state_vote == num_online_cpus())
+		omap_set_pwrdm_state(mpu_pd, PWRDM_POWER_ON);
+	cx->mpu_state_vote--;
+	raw_spin_unlock_irqrestore(&mpu_lock, flag);
+
+	return index;
+}
+
 static int omap_enter_idle_coupled(struct cpuidle_device *dev,
 			struct cpuidle_driver *drv,
 			int index)
@@ -220,6 +262,32 @@ static struct cpuidle_driver omap4_idle_driver = {
 	.safe_state_index = 0,
 };
 
+static struct cpuidle_driver dra7_idle_driver = {
+	.name				= "dra7_idle",
+	.owner				= THIS_MODULE,
+	.states = {
+		{
+			/* C1 - CPU0 ON + CPU1 ON + MPU ON */
+			.exit_latency = 2 + 2,
+			.target_residency = 5,
+			.enter = omap_enter_idle_simple,
+			.name = "C1",
+			.desc = "CPUx WFI, MPUSS ON"
+		},
+		{
+			/* C2 - CPU0 RET + CPU1 RET + MPU CSWR */
+			.exit_latency = 48 + 60,
+			.target_residency = 100,
+			.flags = CPUIDLE_FLAG_TIMER_STOP,
+			.enter = omap_enter_idle_smp,
+			.name = "C2",
+			.desc = "CPUx CSWR, MPUSS CSWR",
+		},
+	},
+	.state_count = ARRAY_SIZE(dra7_idle_data),
+	.safe_state_index = 0,
+};
+
 /* Public functions */
 
 /**
@@ -230,6 +298,16 @@ static struct cpuidle_driver omap4_idle_driver = {
  */
 int __init omap4_idle_init(void)
 {
+	struct cpuidle_driver *idle_driver;
+
+	if (soc_is_dra7xx()) {
+		state_ptr = &dra7_idle_data[0];
+		idle_driver = &dra7_idle_driver;
+	} else {
+		state_ptr = &omap4_idle_data[0];
+		idle_driver = &omap4_idle_driver;
+	}
+
 	mpu_pd = pwrdm_lookup("mpu_pwrdm");
 	cpu_pd[0] = pwrdm_lookup("cpu0_pwrdm");
 	cpu_pd[1] = pwrdm_lookup("cpu1_pwrdm");
@@ -244,5 +322,5 @@ int __init omap4_idle_init(void)
 	/* Configure the broadcast timer on each cpu */
 	on_each_cpu(omap_setup_broadcast_timer, NULL, 1);
 
-	return cpuidle_register(&omap4_idle_driver, cpu_online_mask);
+	return cpuidle_register(idle_driver, cpu_online_mask);
 }
diff --git a/arch/arm/mach-omap2/pm44xx.c b/arch/arm/mach-omap2/pm44xx.c
--- a/arch/arm/mach-omap2/pm44xx.c
+++ b/arch/arm/mach-omap2/pm44xx.c
@@ -287,7 +287,7 @@ int __init omap4_pm_init(void)
 	/* Overwrite the default cpu_do_idle() */
 	arm_pm_idle = omap_default_idle;
 
-	if (cpu_is_omap44xx())
+	if (cpu_is_omap44xx() || soc_is_dra7xx())
 		omap4_idle_init();
 
 err2:
-- 
2.9.3

[PATCH 4/5] ARM: DRA7: PM: cpuidle MPU CSWR support

[Tony Lindgren]

* Tony Lindgren <tony@atomide.com> [161026 08:18]:
> From: Santosh Shilimkar <santosh.shilimkar@ti.com>
> 
> Add DRA74/72 CPUIDLE support.

And as Nishnath noted, that hardware does not support deeper
idle states properly. So below is patches 4 and 5 updated and
squashed together into a single patche to enable cpuidle only
for omap5.

Regards,

Tony

8< -------------------------
>From tony Mon Sep 17 00:00:00 2001
From: Santosh Shilimkar <santosh.shilimkar@ti.com>
Date: Tue, 25 Oct 2016 08:33:32 -0700
Subject: [PATCH] ARM: OMAP5: Add basic cpuidle MPU CSWR support

Add OMAP5 CPUIDLE support.

This patch adds MPUSS low power states in cpuidle.

        C1 - CPU0 WFI + CPU1 WFI + MPU ON
        C2 - CPU0 RET + CPU1 RET + MPU CSWR

Modified from TI kernel tree commit 605967fd2205 ("ARM: DRA7: PM:
cpuidle MPU CSWR support") except enable cpuidle for omap5 instead
of dra7.

According to Nishanth Menon <nm@ti.com>, cpuidle on dra7 is not
supported properly in the hardware so we don't want to enable it.
However, for omap5 this adds some nice power savings. Note that
the TI 3.8 based tree has other cpuidle states that we may be able
to enable later on.

On omap5-uevm, the power consumption eventually settles down to about
920mW with ehci-omap and ohci-omap3 unloaded compared to about 1.7W
without these patches. Note that it seems to take few minutes after
booting for the idle power to go down to 920mW from 1.3W, no idea so
far what might be causing that.

Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[ j-keerthy at ti.com rework on 3.14]
Signed-off-by: Keerthy <j-keerthy@ti.com>
[nm at ti.com: updates based on profiling]
[tony at atomide.com: dropped CPUIDLE_FLAG_TIME_VALID no longer used,
changed for omap5 only as requested by Nishanth, updated comments]
Signed-off-by: Nishanth Menon <nm@ti.com>
---
 arch/arm/mach-omap2/cpuidle44xx.c | 80 ++++++++++++++++++++++++++++++++++++++-
 arch/arm/mach-omap2/pm44xx.c      |  2 +-
 2 files changed, 80 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-omap2/cpuidle44xx.c b/arch/arm/mach-omap2/cpuidle44xx.c
--- a/arch/arm/mach-omap2/cpuidle44xx.c
+++ b/arch/arm/mach-omap2/cpuidle44xx.c
@@ -21,6 +21,7 @@
 #include "common.h"
 #include "pm.h"
 #include "prm.h"
+#include "soc.h"
 #include "clockdomain.h"
 
 #define MAX_CPUS	2
@@ -30,6 +31,7 @@ struct idle_statedata {
 	u32 cpu_state;
 	u32 mpu_logic_state;
 	u32 mpu_state;
+	u32 mpu_state_vote;
 };
 
 static struct idle_statedata omap4_idle_data[] = {
@@ -50,12 +52,26 @@ static struct idle_statedata omap4_idle_data[] = {
 	},
 };
 
+static struct idle_statedata omap5_idle_data[] = {
+	{
+		.cpu_state = PWRDM_POWER_ON,
+		.mpu_state = PWRDM_POWER_ON,
+		.mpu_logic_state = PWRDM_POWER_ON,
+	},
+	{
+		.cpu_state = PWRDM_POWER_RET,
+		.mpu_state = PWRDM_POWER_RET,
+		.mpu_logic_state = PWRDM_POWER_RET,
+	},
+};
+
 static struct powerdomain *mpu_pd, *cpu_pd[MAX_CPUS];
 static struct clockdomain *cpu_clkdm[MAX_CPUS];
 
 static atomic_t abort_barrier;
 static bool cpu_done[MAX_CPUS];
 static struct idle_statedata *state_ptr = &omap4_idle_data[0];
+static DEFINE_RAW_SPINLOCK(mpu_lock);
 
 /* Private functions */
 
@@ -77,6 +93,32 @@ static int omap_enter_idle_simple(struct cpuidle_device *dev,
 	return index;
 }
 
+static int omap_enter_idle_smp(struct cpuidle_device *dev,
+			       struct cpuidle_driver *drv,
+			       int index)
+{
+	struct idle_statedata *cx = state_ptr + index;
+	unsigned long flag;
+
+	raw_spin_lock_irqsave(&mpu_lock, flag);
+	cx->mpu_state_vote++;
+	if (cx->mpu_state_vote == num_online_cpus()) {
+		pwrdm_set_logic_retst(mpu_pd, cx->mpu_logic_state);
+		omap_set_pwrdm_state(mpu_pd, cx->mpu_state);
+	}
+	raw_spin_unlock_irqrestore(&mpu_lock, flag);
+
+	omap4_enter_lowpower(dev->cpu, cx->cpu_state);
+
+	raw_spin_lock_irqsave(&mpu_lock, flag);
+	if (cx->mpu_state_vote == num_online_cpus())
+		omap_set_pwrdm_state(mpu_pd, PWRDM_POWER_ON);
+	cx->mpu_state_vote--;
+	raw_spin_unlock_irqrestore(&mpu_lock, flag);
+
+	return index;
+}
+
 static int omap_enter_idle_coupled(struct cpuidle_device *dev,
 			struct cpuidle_driver *drv,
 			int index)
@@ -220,6 +262,32 @@ static struct cpuidle_driver omap4_idle_driver = {
 	.safe_state_index = 0,
 };
 
+static struct cpuidle_driver omap5_idle_driver = {
+	.name				= "omap5_idle",
+	.owner				= THIS_MODULE,
+	.states = {
+		{
+			/* C1 - CPU0 ON + CPU1 ON + MPU ON */
+			.exit_latency = 2 + 2,
+			.target_residency = 5,
+			.enter = omap_enter_idle_simple,
+			.name = "C1",
+			.desc = "CPUx WFI, MPUSS ON"
+		},
+		{
+			/* C2 - CPU0 RET + CPU1 RET + MPU CSWR */
+			.exit_latency = 48 + 60,
+			.target_residency = 100,
+			.flags = CPUIDLE_FLAG_TIMER_STOP,
+			.enter = omap_enter_idle_smp,
+			.name = "C2",
+			.desc = "CPUx CSWR, MPUSS CSWR",
+		},
+	},
+	.state_count = ARRAY_SIZE(omap5_idle_data),
+	.safe_state_index = 0,
+};
+
 /* Public functions */
 
 /**
@@ -230,6 +298,16 @@ static struct cpuidle_driver omap4_idle_driver = {
  */
 int __init omap4_idle_init(void)
 {
+	struct cpuidle_driver *idle_driver;
+
+	if (soc_is_omap54xx()) {
+		state_ptr = &omap5_idle_data[0];
+		idle_driver = &omap5_idle_driver;
+	} else {
+		state_ptr = &omap4_idle_data[0];
+		idle_driver = &omap4_idle_driver;
+	}
+
 	mpu_pd = pwrdm_lookup("mpu_pwrdm");
 	cpu_pd[0] = pwrdm_lookup("cpu0_pwrdm");
 	cpu_pd[1] = pwrdm_lookup("cpu1_pwrdm");
@@ -244,5 +322,5 @@ int __init omap4_idle_init(void)
 	/* Configure the broadcast timer on each cpu */
 	on_each_cpu(omap_setup_broadcast_timer, NULL, 1);
 
-	return cpuidle_register(&omap4_idle_driver, cpu_online_mask);
+	return cpuidle_register(idle_driver, cpu_online_mask);
 }
diff --git a/arch/arm/mach-omap2/pm44xx.c b/arch/arm/mach-omap2/pm44xx.c
--- a/arch/arm/mach-omap2/pm44xx.c
+++ b/arch/arm/mach-omap2/pm44xx.c
@@ -287,7 +287,7 @@ int __init omap4_pm_init(void)
 	/* Overwrite the default cpu_do_idle() */
 	arm_pm_idle = omap_default_idle;
 
-	if (cpu_is_omap44xx())
+	if (cpu_is_omap44xx() || soc_is_omap54xx())
 		omap4_idle_init();
 
 err2:
-- 
2.9.3

[PATCH 5/5] ARM: OMAP5: Enable minimal cpuidle for omap5 retention

[Tony Lindgren]

This allows the CPUs to enter retention during idle and takes the idle
power consumption down quite a bit. On omap5-uevm, the power consumption
eventually settles down to about 920mW with ehci-omap and ohci-omap3
unloaded compared to about 1.7W without these patches. Note that it
seems to take few minutes after booting for the idle power to go down
to 920mW from 1.3W, no idea so far what might be causing that.

Note that the TI kernel supports more modes in the v3.8 kernel, but
the omap5 specific states can be added later.

Signed-off-by: Tony Lindgren <tony@atomide.com>
---
 arch/arm/mach-omap2/cpuidle44xx.c | 2 +-
 arch/arm/mach-omap2/pm44xx.c      | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-omap2/cpuidle44xx.c b/arch/arm/mach-omap2/cpuidle44xx.c
--- a/arch/arm/mach-omap2/cpuidle44xx.c
+++ b/arch/arm/mach-omap2/cpuidle44xx.c
@@ -300,7 +300,7 @@ int __init omap4_idle_init(void)
 {
 	struct cpuidle_driver *idle_driver;
 
-	if (soc_is_dra7xx()) {
+	if (soc_is_dra7xx() || soc_is_omap54xx()) {
 		state_ptr = &dra7_idle_data[0];
 		idle_driver = &dra7_idle_driver;
 	} else {
diff --git a/arch/arm/mach-omap2/pm44xx.c b/arch/arm/mach-omap2/pm44xx.c
--- a/arch/arm/mach-omap2/pm44xx.c
+++ b/arch/arm/mach-omap2/pm44xx.c
@@ -287,7 +287,7 @@ int __init omap4_pm_init(void)
 	/* Overwrite the default cpu_do_idle() */
 	arm_pm_idle = omap_default_idle;
 
-	if (cpu_is_omap44xx() || soc_is_dra7xx())
+	if (cpu_is_omap44xx() || soc_is_omap54xx() || soc_is_dra7xx())
 		omap4_idle_init();
 
 err2:
-- 
2.9.3

[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[Santosh Shilimkar]

On 10/26/2016 8:16 AM, Tony Lindgren wrote:
> Hi all,
>
> Here are some fixes to get minimal cpuidle support working with omap5
> and dra7. Not sure if there are still some unsorted issues on enabling
> this on dra7, but at least omap5 has been behving for me for few
> weeks with these.
>
> Regards,
>
> Tony
>
>
> Santosh Shilimkar (1):
>   ARM: DRA7: PM: cpuidle MPU CSWR support
>
> Tony Lindgren (4):
>   ARM: OMAP5: Fix build for PM code
>   ARM: OMAP5: Fix mpuss_early_init
>   ARM: OMAP4+: Fix bad fallthrough for cpuidle
>   ARM: OMAP5: Enable minimal cpuidle for omap5 retention
>
Feel free to add my ack if you need one.

Acked-by: Santosh Shilimkar <ssantosh@kernel.org>

[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[Nishanth Menon]

Tony,
On 10/26/2016 10:16 AM, Tony Lindgren wrote:
> Hi all,
>
> Here are some fixes to get minimal cpuidle support working with omap5
> and dra7. Not sure if there are still some unsorted issues on enabling
> this on dra7, but at least omap5 has been behving for me for few
> weeks with these.

I am OK with cpuidle for OMAP5, but NOT dra7. we have enabled this 
series of patches close to a couple of years back and hit quiet a few 
issues that do not have silicon resolution. CPU states lower than WFI 
will not work in cpuidle for DRA7 and has already been "descoped" from 
TI investigation charter. details of the issue remain TI internal, 
given that the descope was already done.

Again, OMAP5 has only had limited exposure from verification point of 
view - and all of it TI internal (unlike OMAP4 which has had 
production devices go ahead) -> at the very least, I'd link it with 
experimental.

-- 
Regards,
Nishanth Menon

[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[Tony Lindgren]

* Nishanth Menon <nm@ti.com> [161026 13:57]:
> Tony,
> On 10/26/2016 10:16 AM, Tony Lindgren wrote:
> > Hi all,
> > 
> > Here are some fixes to get minimal cpuidle support working with omap5
> > and dra7. Not sure if there are still some unsorted issues on enabling
> > this on dra7, but at least omap5 has been behving for me for few
> > weeks with these.
> 
> I am OK with cpuidle for OMAP5, but NOT dra7. we have enabled this series of
> patches close to a couple of years back and hit quiet a few issues that do
> not have silicon resolution. CPU states lower than WFI will not work in
> cpuidle for DRA7 and has already been "descoped" from TI investigation
> charter. details of the issue remain TI internal, given that the descope was
> already done.

Hmm OK, no point adding it if not supported by hardware. I'll modify the
series for omap5 only.

> Again, OMAP5 has only had limited exposure from verification point of view -
> and all of it TI internal (unlike OMAP4 which has had production devices go
> ahead) -> at the very least, I'd link it with experimental.

Sure, this is a subset already to let's see how that works before adding
any other modes.

Regards,

Tony

[PATCH 0/5] Minimal cpuidle fixes for omap5 and dra7

[H. Nikolaus Schaller]

> Am 26.10.2016 um 17:16 schrieb Tony Lindgren <tony@atomide.com>:
> 
> Hi all,
> 
> Here are some fixes to get minimal cpuidle support working with omap5
> and dra7. Not sure if there are still some unsorted issues on enabling
> this on dra7, but at least omap5 has been behving for me for few
> weeks with these.

Seems to work on Pyra, but I could not really test.
At least it has no adverse effect.

BR,
Nikolaus