CPU idle for Armada XP
Subsystems (5)
cover CPU idle for Armada XP
Hello,
This patch set adds the CPU idle support for Armada XP and prepares the support for Armada 370. This was based on the work of Nadav Haklai.
Most of the patches modify the mvebu code in order to prepare the support for CPU idle, hence the patches 2 to 10 should go to mvebu subsystem (and then arm-soc).
The first patch should go through ARM subsystem and should be taken by Russell King.
The 11th patch 'cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC' is the only one who should go to the cpuidle subsystem. But of course I would like that Daniel Lezcano or Rafael J. Wysocki have a look on the whole series.
The last patch should also go to mvebu subsystem (and then arm-soc) but with an Acked-by from on of the device tree maintainer.
The whole series is also available in the branch CPU-idle-ArmadaXP at
https://github.com/MISL-EBU-System-SW/mainline-public.git
Thanks,
Gregory CLEMENT (12):
ARM: PJ4B: Add cpu_suspend/cpu_resume hooks for PJ4B
ARM: mvebu: ll_set_cpu_coherent no more uses the coherency address as
parameter
ARM: mvebu: ll_set_cpu_coherent always uses the current CPU
ARM: mvebu: Remove the unused argument of set_cpu_coherent()
ARM: mvebu: Makes ll_set_cpu_coherent() more configurable
ARM: mvebu: Low level functions to disbale cache snooping
ARM: mvebu: Add a new set of registers for pmsu
ARM: mvebu: Allow to power down L2 cache controller in idle mode
ARM: mvebu: Add the PMSU related part of the cpu idle functions
ARM: mvebu: Set the start address of a CPU in a separate function
cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC
ARM: dts: mvebu: Add a new set of registers to the PMSU node
.../devicetree/bindings/arm/armada-370-xp-pmsu.txt | 12 +- arch/arm/boot/dts/armada-xp.dtsi | 2 +- arch/arm/mach-mvebu/coherency.c | 12 +- arch/arm/mach-mvebu/coherency.h | 2 +- arch/arm/mach-mvebu/coherency_ll.S | 69 +++++++++-- arch/arm/mach-mvebu/headsmp.S | 15 +-- arch/arm/mach-mvebu/platsmp.c | 2 +- arch/arm/mach-mvebu/pmsu.c | 99 +++++++++++++++- arch/arm/mm/proc-v7.S | 64 +++++++++- drivers/cpuidle/Kconfig | 6 + drivers/cpuidle/Makefile | 1 + drivers/cpuidle/cpuidle-armada-370-xp.c | 132 +++++++++++++++++++++ drivers/cpuidle/suspend-armada-370-xp.S | 91 ++++++++++++++ include/linux/armada-370-xp-pmsu.h | 19 +++ 14 files changed, 484 insertions(+), 42 deletions(-) create mode 100644 drivers/cpuidle/cpuidle-armada-370-xp.c create mode 100644 drivers/cpuidle/suspend-armada-370-xp.S create mode 100644 include/linux/armada-370-xp-pmsu.h
Signature · 1 line
1.8.1.2
PATCH 1/12 ARM: PJ4B: Add cpu_suspend/cpu_resume hooks for PJ4B
PJ4B needs extra instructions for suspend and resume, so instead of using the armv7 version, this commit introduces specific versions for PJ4B.
--- arch/arm/mm/proc-v7.S | 64 ++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 61 insertions(+), 3 deletions(-)
arch/arm/mm/proc-v7.S +61 -3
@@ -162,9 +162,67 @@ ENDPROC(cpu_pj4b_do_idle) globl_equ cpu_pj4b_do_idle, cpu_v7_do_idle #endif globl_equ cpu_pj4b_dcache_clean_area, cpu_v7_dcache_clean_area - globl_equ cpu_pj4b_do_suspend, cpu_v7_do_suspend - globl_equ cpu_pj4b_do_resume, cpu_v7_do_resume - globl_equ cpu_pj4b_suspend_size, cpu_v7_suspend_size +#ifdef CONFIG_ARM_CPU_SUSPEND +ENTRY(cpu_pj4b_do_suspend) + stmfd sp!, {r4 - r10, lr} + mrc p15, 0, r4, c13, c0, 0 @ FCSE/PID + mrc p15, 0, r5, c13, c0, 3 @ User r/o thread ID + stmia r0!, {r4 - r5} + mrc p15, 1, r6, c15, c1, 0 @ save CP15 - extra features + mrc p15, 1, r7, c15, c2, 0 @ save CP15 - Aux Func Modes Ctrl 0 + mrc p15, 1, r8, c15, c1, 2 @ save CP15 - Aux Debug Modes Ctrl 2 + mrc p15, 1, r9, c15, c1, 1 @ save CP15 - Aux Debug Modes Ctrl 1 + mrc p15, 0, r10, c9, c14, 0 @ save CP15 - PMC + stmia r0!, {r6 - r10} + mrc p15, 0, r6, c3, c0, 0 @ Domain ID + mrc p15, 0, r7, c2, c0, 1 @ TTB 1 + mrc p15, 0, r11, c2, c0, 2 @ TTB control register + mrc p15, 0, r8, c1, c0, 0 @ Control register + mrc p15, 0, r9, c1, c0, 1 @ Auxiliary control register + mrc p15, 0, r10, c1, c0, 2 @ Co-processor access control + stmia r0, {r6 - r11} + ldmfd sp!, {r4 - r10, pc} +ENDPROC(cpu_pj4b_do_suspend) + +ENTRY(cpu_pj4b_do_resume) + mov ip, #0 + mcr p15, 0, ip, c7, c5, 0 @ invalidate I cache + mcr p15, 0, ip, c13, c0, 1 @ set reserved context ID + ldmia r0!, {r4 - r5} + mcr p15, 0, r4, c13, c0, 0 @ FCSE/PID + mcr p15, 0, r5, c13, c0, 3 @ User r/o thread ID + ldmia r0!, {r6 - r10} + mcr p15, 1, r6, c15, c1, 0 @ save CP15 - extra features + mcr p15, 1, r7, c15, c2, 0 @ save CP15 - Aux Func Modes Ctrl 0 + mcr p15, 1, r8, c15, c1, 2 @ save CP15 - Aux Debug Modes Ctrl 2 + mcr p15, 1, r9, c15, c1, 1 @ save CP15 - Aux Debug Modes Ctrl 1 + mcr p15, 0, r10, c9, c14, 0 @ save CP15 - PMC + ldmia r0, {r6 - r11} + mcr p15, 0, ip, c8, c7, 0 @ invalidate TLBs + mcr p15, 0, r6, c3, c0, 0 @ Domain ID +#ifndef CONFIG_ARM_LPAE + ALT_SMP(orr r1, r1, #TTB_FLAGS_SMP) + ALT_UP(orr r1, r1, #TTB_FLAGS_UP) +#endif + mcr p15, 0, r1, c2, c0, 0 @ TTB 0 + mcr p15, 0, r7, c2, c0, 1 @ TTB 1 + mcr p15, 0, r11, c2, c0, 2 @ TTB control register + ldr r4, =PRRR @ PRRR + ldr r5, =NMRR @ NMRR + mcr p15, 0, r4, c10, c2, 0 @ write PRRR + mcr p15, 0, r5, c10, c2, 1 @ write NMRR + mrc p15, 0, r4, c1, c0, 1 @ Read Auxiliary control register + teq r4, r9 @ Is it already set? + mcrne p15, 0, r9, c1, c0, 1 @ No, so write it + mcr p15, 0, r10, c1, c0, 2 @ Co-processor access control + isb + dsb + mov r0, r8 @ control register + b cpu_resume_mmu +ENDPROC(cpu_pj4b_do_resume) +#endif +.globl cpu_pj4b_suspend_size +.equ cpu_pj4b_suspend_size, 4 * 13 #endif
Signature · 1 line
1.8.1.2
PATCH 2/12 ARM: mvebu: ll_set_cpu_coherent no more uses the coherency address as parameter
In reply to Gregory CLEMENTUntil now the calling functions of ll_set_cpu_coherent() have to know the base address of the coherency registers. This commit doesn't expose anymore this address, and replace the argument by a flag to indicate if we have to use the physical or the virtual address.
--- arch/arm/mach-mvebu/coherency.c | 6 +++--- arch/arm/mach-mvebu/coherency_ll.S | 18 +++++++++++++++++- arch/arm/mach-mvebu/headsmp.S | 10 ++-------- 3 files changed, 22 insertions(+), 12 deletions(-)
arch/arm/mach-mvebu/coherency.c +3 -3
@@ -29,7 +29,7 @@#include "armada-370-xp.h" unsigned long coherency_phys_base; -static void __iomem *coherency_base; +void __iomem *coherency_base;static void __iomem *coherency_cpu_base;/* Coherency fabric registers */
@@ -43,7 +43,7 @@ static struct of_device_id of_coherency_table[] = { };/* Function defined in coherency_ll.S */ -int ll_set_cpu_coherent (void __iomem *base_addr,unsigned int hw_cpu_id); +int ll_set_cpu_coherent (bool use_virt_addr,unsigned int hw_cpu_id);int set_cpu_coherent (unsigned int hw_cpu_id,int smp_group_id) {
@@ -53,7 +53,7 @@ int set_cpu_coherent(unsigned int hw_cpu_id, int smp_group_id)return 1 ; } -return ll_set_cpu_coherent (coherency_base, hw_cpu_id); +return ll_set_cpu_coherent (true, hw_cpu_id); }static inline void mvebu_hwcc_sync_io_barrier (void )
arch/arm/mach-mvebu/coherency_ll.S +17 -1
@@ -22,10 +22,22 @@ .text /* - * r0: Coherency fabric base register address + * r0: if r0==0 => physical addres, else virtual address * r1: HW CPU id */ ENTRY(ll_set_cpu_coherent) + cmp r0, #0 + bne 1f + /* use physical address */ + adr r0, 3f + ldr r3, [r0] + ldr r0, [r0, r3] + b 2f +1: + /* use virtual address */ + ldr r0, =(coherency_base) + ldr r0, [r0] +2: /* Create bit by cpu index */ mov r3, #(1 << 24) lsl r1, r3, r1
@@ -53,3 +65,7 @@ ENTRY(ll_set_cpu_coherent) mov r0, #0 mov pc, lr ENDPROC(ll_set_cpu_coherent) + + .align 2 +3: + .long coherency_phys_base - .
arch/arm/mach-mvebu/headsmp.S +2 -8
@@ -27,10 +27,8 @@ * startup */ ENTRY(armada_xp_secondary_startup) - /* Get coherency fabric base physical address */ - adr r0, 1f - ldr r1, [r0] - ldr r0, [r0, r1] + /* Use physical addrss */ + mov r0, #0 /* Read CPU id */ mrc p15, 0, r1, c0, c0, 5
@@ -41,7 +39,3 @@ ENTRY(armada_xp_secondary_startup) b secondary_startup ENDPROC(armada_xp_secondary_startup) - - .align 2 -1: - .long coherency_phys_base - .
Signature · 1 line
1.8.1.2
PATCH 3/12 ARM: mvebu: ll_set_cpu_coherent always uses the current CPU
In reply to Gregory CLEMENTll_set_cpu_coherent is always used on the current CPU, so instead of passing the CPU id as argument, ll_set_cpu_coherent() can find it by itself.
--- arch/arm/mach-mvebu/coherency.c | 10 +++++----- arch/arm/mach-mvebu/coherency.h | 2 +- arch/arm/mach-mvebu/coherency_ll.S | 7 ++++--- arch/arm/mach-mvebu/headsmp.S | 4 ---- arch/arm/mach-mvebu/platsmp.c | 2 +- 5 files changed, 11 insertions(+), 14 deletions(-)
arch/arm/mach-mvebu/coherency.c +5 -5
@@ -43,17 +43,17 @@ static struct of_device_id of_coherency_table[] = { };/* Function defined in coherency_ll.S */ -int ll_set_cpu_coherent (bool use_virt_addr,unsigned int hw_cpu_id); +int ll_set_cpu_coherent (bool use_virt_addr); -int set_cpu_coherent (unsigned int hw_cpu_id,int smp_group_id) +int set_cpu_coherent (int smp_group_id) {if (!coherency_base) { -pr_warn ("Can't make CPU %d cache coherent.\n" , hw_cpu_id); +pr_warn ("Can't make current CPU cache coherent.\n" );pr_warn ("Coherency fabric is not initialized\n" );return 1 ; } -return ll_set_cpu_coherent (true, hw_cpu_id); +return ll_set_cpu_coherent (true); }static inline void mvebu_hwcc_sync_io_barrier (void )
@@ -139,7 +139,7 @@ int __init coherency_init(void)sync_cache_w (&coherency_phys_base); coherency_base =of_iomap (np,0 ); coherency_cpu_base =of_iomap (np,1 ); -set_cpu_coherent (cpu_logical_map (smp_processor_id ()),0 ); +set_cpu_coherent (0 ); }return 0 ;
arch/arm/mach-mvebu/coherency.h +1 -1
@@ -14,7 +14,7 @@#ifndef __MACH_370_XP_COHERENCY_H#define __MACH_370_XP_COHERENCY_H -int set_cpu_coherent (int cpu_id,int smp_group_id); +int set_cpu_coherent (int smp_group_id);int coherency_init (void );#endif /* __MACH_370_XP_COHERENCY_H */
arch/arm/mach-mvebu/coherency_ll.S +4 -3
@@ -23,7 +23,6 @@ .text /* * r0: if r0==0 => physical addres, else virtual address - * r1: HW CPU id */ ENTRY(ll_set_cpu_coherent) cmp r0, #0
@@ -39,8 +38,10 @@ ENTRY(ll_set_cpu_coherent) ldr r0, [r0] 2: /* Create bit by cpu index */ - mov r3, #(1 << 24) - lsl r1, r3, r1 + mrc 15, 0, r1, cr0, cr0, 5 + and r1, r1, #15 + mov r2, #(1 << 24) + lsl r1, r2, r1 /* Add CPU to SMP group - Atomic */ add r3, r0, #ARMADA_XP_CFB_CTL_REG_OFFSET
arch/arm/mach-mvebu/headsmp.S +0 -4
@@ -30,10 +30,6 @@ ENTRY(armada_xp_secondary_startup) /* Use physical addrss */ mov r0, #0 - /* Read CPU id */ - mrc p15, 0, r1, c0, c0, 5 - and r1, r1, #0xF - /* Add CPU to coherency fabric */ bl ll_set_cpu_coherent b secondary_startup
arch/arm/mach-mvebu/platsmp.c +1 -1
@@ -116,7 +116,7 @@ void __init armada_xp_smp_prepare_cpus(unsigned int max_cpus) {set_secondary_cpus_clock ();flush_cache_all (); -set_cpu_coherent (cpu_logical_map (smp_processor_id ()),0 ); +set_cpu_coherent (0 );mvebu_mbus_add_window ("bootrom" ,0xfff00000 ,SZ_1M ); }
Signature · 1 line
1.8.1.2
PATCH 4/12 ARM: mvebu: Remove the unused argument of set_cpu_coherent()
In reply to Gregory CLEMENTset_cpu_coherent() took the SMP group ID as parameter. But this parameter was never used, and the CPU always use the SMP group 0. So we can remove this parameter.
--- arch/arm/mach-mvebu/coherency.c | 4 ++-- arch/arm/mach-mvebu/coherency.h | 2 +- arch/arm/mach-mvebu/platsmp.c | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-)
arch/arm/mach-mvebu/coherency.c +2 -2
@@ -45,7 +45,7 @@ static struct of_device_id of_coherency_table[] = {/* Function defined in coherency_ll.S */ int ll_set_cpu_coherent (bool use_virt_addr); -int set_cpu_coherent (int smp_group_id) +int set_cpu_coherent (void ) {if (!coherency_base) {pr_warn ("Can't make current CPU cache coherent.\n" );
@@ -139,7 +139,7 @@ int __init coherency_init(void)sync_cache_w (&coherency_phys_base); coherency_base =of_iomap (np,0 ); coherency_cpu_base =of_iomap (np,1 ); -set_cpu_coherent (0 ); +set_cpu_coherent (); }return 0 ;
arch/arm/mach-mvebu/coherency.h +1 -1
@@ -14,7 +14,7 @@#ifndef __MACH_370_XP_COHERENCY_H#define __MACH_370_XP_COHERENCY_H -int set_cpu_coherent (int smp_group_id); +int set_cpu_coherent (void );int coherency_init (void );#endif /* __MACH_370_XP_COHERENCY_H */
arch/arm/mach-mvebu/platsmp.c +1 -1
@@ -116,7 +116,7 @@ void __init armada_xp_smp_prepare_cpus(unsigned int max_cpus) {set_secondary_cpus_clock ();flush_cache_all (); -set_cpu_coherent (0 ); +set_cpu_coherent ();mvebu_mbus_add_window ("bootrom" ,0xfff00000 ,SZ_1M ); }
Signature · 1 line
1.8.1.2
PATCH 5/12 ARM: mvebu: Makes ll_set_cpu_coherent() more configurable
In reply to Gregory CLEMENTll_set_cpu_coherent does two things to set the coherency for a CPU:
- Adding the CPU to the SMP group 0
- Enabling the snooping on the CPU
Only the second part is needed when coming back from a CPU Idle. This commit allows to choose to do the first part or not.
--- arch/arm/mach-mvebu/coherency.c | 4 ++-- arch/arm/mach-mvebu/coherency_ll.S | 30 ++++++++++++++++++------------ arch/arm/mach-mvebu/headsmp.S | 3 +++ 3 files changed, 23 insertions(+), 14 deletions(-)
arch/arm/mach-mvebu/coherency.c +2 -2
@@ -43,7 +43,7 @@ static struct of_device_id of_coherency_table[] = { };/* Function defined in coherency_ll.S */ -int ll_set_cpu_coherent (bool use_virt_addr); +int ll_set_cpu_coherent (bool use_virt_addr,bool use_smp_group);int set_cpu_coherent (void ) {
@@ -53,7 +53,7 @@ int set_cpu_coherent(void)return 1 ; } -return ll_set_cpu_coherent (true); +return ll_set_cpu_coherent (true, true); }static inline void mvebu_hwcc_sync_io_barrier (void )
arch/arm/mach-mvebu/coherency_ll.S +18 -12
@@ -22,7 +22,8 @@ .text /* - * r0: if r0==0 => physical addres, else virtual address + * r0: Use virtual address if r0 != 0, else physical address + * r1: Add CPU to the SMP group if r1 != 0 */ ENTRY(ll_set_cpu_coherent) cmp r0, #0
@@ -38,26 +39,31 @@ ENTRY(ll_set_cpu_coherent) ldr r0, [r0] 2: /* Create bit by cpu index */ - mrc 15, 0, r1, cr0, cr0, 5 - and r1, r1, #15 + mrc 15, 0, r3, cr0, cr0, 5 + and r3, r3, #15 mov r2, #(1 << 24) - lsl r1, r2, r1 + lsl r3, r2, r3 + + /* If r1=!0 then just enable the snooping */ + cmp r1, #0 + beq 2f /* Add CPU to SMP group - Atomic */ - add r3, r0, #ARMADA_XP_CFB_CTL_REG_OFFSET + add r1, r0, #ARMADA_XP_CFB_CTL_REG_OFFSET 1: - ldrex r2, [r3] - orr r2, r2, r1 - strex r0, r2, [r3] + ldrex r2, [r1] + orr r2, r2, r3 + strex r0, r2, [r1] cmp r0, #0 bne 1b +2: /* Enable coherency on CPU - Atomic */ - add r3, r3, #ARMADA_XP_CFB_CFG_REG_OFFSET + add r1, r1, #ARMADA_XP_CFB_CFG_REG_OFFSET 1: - ldrex r2, [r3] - orr r2, r2, r1 - strex r0, r2, [r3] + ldrex r2, [r1] + orr r2, r2, r3 + strex r0, r2, [r1] cmp r0, #0 bne 1b
arch/arm/mach-mvebu/headsmp.S +3 -0
@@ -30,6 +30,9 @@ ENTRY(armada_xp_secondary_startup) /* Use physical addrss */ mov r0, #0 + /* Add CPU to the SMP group*/ + mov r1, #1 + /* Add CPU to coherency fabric */ bl ll_set_cpu_coherent b secondary_startup
Signature · 1 line
1.8.1.2
PATCH 6/12 ARM: mvebu: Low level functions to disbale cache snooping
In reply to Gregory CLEMENTWhen going to deep idle we need to disable the SoC snooping by "hand". Playing with the coherency fabric requires to use assembly code to be sure that the compiler doesn't reorder the instructions nor do wrong optimization.
This function will be called by the low level (in assembly) part of the CPU idle functions.
--- arch/arm/mach-mvebu/coherency_ll.S | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-)
arch/arm/mach-mvebu/coherency_ll.S +23 -1
@@ -73,6 +73,28 @@ ENTRY(ll_set_cpu_coherent) mov pc, lr ENDPROC(ll_set_cpu_coherent) +/* + * r0: if r0==0 => physical addres, else virtual address + */ +ENTRY(armada_370_xp_disable_snoop_ena) + ldr r0, =(coherency_base) + ldr r0, [r0] + /* Enable SnoopEna - Exclusive */ + mrc 15, 0, r1, cr0, cr0, 5 + and r1, r1, #15 + mov r2, #(1 << 24) + lsl r2, r2, r1 + +1: + ldrex r1, [r0] + bic r1, r1, r2 + strex r3, r1, [r0] + cmp r3, #0 + bne 1b + + mov pc, lr +ENDPROC(armada_370_xp_disable_snoop_ena) + .align 2 3: - .long coherency_phys_base - . + .long coherency_phys_base - .
Signature · 1 line
1.8.1.2
PATCH 7/12 ARM: mvebu: Add a new set of registers for pmsu
In reply to Gregory CLEMENTThe Power Management Unit Service block also controls the Coherency Fabric subsystem. This new set of registers is needed for the CPU idle implementation for the Armada 370/XP, it allows to enter a deep CPU idle state where the Coherency Fabric and the L2 cache are power down.
This patch also adds warnings if one of the base registers set can't be ioremapped.
--- arch/arm/mach-mvebu/pmsu.c | 5 +++++ 1 file changed, 5 insertions(+)
arch/arm/mach-mvebu/pmsu.c +5 -0
@@ -25,6 +25,7 @@static void __iomem *pmsu_mp_base;static void __iomem *pmsu_reset_base; +static void __iomem *pmsu_fabric_base;#define PMSU_BOOT_ADDR_REDIRECT_OFFSET (cpu) ((cpu * 0x100) + 0x24)#define PMSU_RESET_CTL_OFFSET (cpu) (cpu * 0x8)
@@ -66,7 +67,11 @@ int __init armada_370_xp_pmsu_init(void)if (np) {pr_info ("Initializing Power Management Service Unit\n" ); pmsu_mp_base =of_iomap (np,0 ); +WARN_ON (!pmsu_mp_base); pmsu_reset_base =of_iomap (np,1 ); +WARN_ON (!pmsu_reset_base); + pmsu_fabric_base =of_iomap (np,2 ); +WARN_ON (!pmsu_fabric_base); }return 0 ;
Signature · 1 line
1.8.1.2
PATCH 8/12 ARM: mvebu: Allow to power down L2 cache controller in idle mode
In reply to Gregory CLEMENTThis commit adds an exported function which allows to power down the L2 cache controller and the Coherency Fabric when the CPU goes into deep idle mode.
This feature is part of the Power Management Service Unit of the Armada 370 and Armada XP SoCs.
--- arch/arm/mach-mvebu/pmsu.c | 14 ++++++++++++++ include/linux/armada-370-xp-pmsu.h | 16 ++++++++++++++++ 2 files changed, 30 insertions(+) create mode 100644 include/linux/armada-370-xp-pmsu.h
arch/arm/mach-mvebu/pmsu.c +14 -0
@@ -30,6 +30,9 @@ static void __iomem *pmsu_fabric_base;#define PMSU_BOOT_ADDR_REDIRECT_OFFSET (cpu) ((cpu * 0x100) + 0x24)#define PMSU_RESET_CTL_OFFSET (cpu) (cpu * 0x8) +#define L2C_NFABRIC_PM_CTL 0x4 +#define L2C_NFABRIC_PM_CTL_PWR_DOWN (1 << 20) +static struct of_device_id of_pmsu_table[] = { {.compatible ="marvell,armada-370-xp-pmsu" }, {/* end of list */ },
@@ -77,4 +80,15 @@ int __init armada_370_xp_pmsu_init(void)return 0 ; } +void armada_370_xp_pmsu_enable_l2_powerdown_onidle (void ) +{ +int reg; + +/* Enable L2 & Fabric powerdown in Deep-Idle mode - Fabric */ + reg =readl (pmsu_fabric_base +L2C_NFABRIC_PM_CTL ); + reg |=L2C_NFABRIC_PM_CTL_PWR_DOWN ; +writel (reg, pmsu_fabric_base +L2C_NFABRIC_PM_CTL ); +} +EXPORT_SYMBOL_GPL (armada_370_xp_pmsu_enable_l2_powerdown_onidle); +early_initcall (armada_370_xp_pmsu_init);
include/linux/armada-370-xp-pmsu.h new file +16 -0
@@ -0,0 +1,16 @@ +/* +* Power Management Service Unit (PMSU) support for Armada 370/XP platforms. +* +* Copyright (C) 2013 Marvell +* +* This file is licensed under the terms of the GNU General Public +* License version 2. This program is licensed "as is" without any +* warranty of any kind, whether express or implied. +*/ + +#ifndef __ARMADA_370_XP__PMSU_H +#define __ARMADA_370_XP__PMSU_H + +void armada_370_xp_pmsu_enable_l2_powerdown_onidle (void ); + +#endif /* __ARMADA_370_XP__PMSU_H */
Signature · 1 line
1.8.1.2
PATCH 9/12 ARM: mvebu: Add the PMSU related part of the cpu idle functions
In reply to Gregory CLEMENTThe cpu idle support will need to acces to the Power Mangement Service Unit. This commit adds and exports the prepare and restore functions that will be used by the CPU idle.
--- arch/arm/mach-mvebu/pmsu.c | 70 ++++++++++++++++++++++++++++++++++++++ include/linux/armada-370-xp-pmsu.h | 2 ++ 2 files changed, 72 insertions(+)
arch/arm/mach-mvebu/pmsu.c +70 -0
@@ -30,6 +30,24 @@ static void __iomem *pmsu_fabric_base;#define PMSU_BOOT_ADDR_REDIRECT_OFFSET (cpu) ((cpu * 0x100) + 0x24)#define PMSU_RESET_CTL_OFFSET (cpu) (cpu * 0x8) +#define PM_CONTROL_AND_CONFIG (cpu) ((cpu * 0x100) + 0x4) +#define PM_CONTROL_AND_CONFIG_DFS_REQ (1 << 18) +#define PM_CONTROL_AND_CONFIG_PWDDN_REQ (1 << 16) +#define PM_CONTROL_AND_CONFIG_L2_PWDDN (1 << 20) + +#define PM_CPU_POWER_DOWN_CONTROL (cpu) ((cpu * 0x100) + 0x8) + +#define PM_CPU_POWER_DOWN_DIS_SNP_Q_SKIP (1 << 0) + +#define PM_STATUS_AND_MASK (cpu) ((cpu * 0x100) + 0xc) +#define PM_STATUS_AND_MASK_CPU_IDLE_WAIT (1 << 16) +#define PM_STATUS_AND_MASK_SNP_Q_EMPTY_WAIT (1 << 17) +#define PM_STATUS_AND_MASK_IRQ_WAKEUP (1 << 20) +#define PM_STATUS_AND_MASK_FIQ_WAKEUP (1 << 21) +#define PM_STATUS_AND_MASK_DBG_WAKEUP (1 << 22) +#define PM_STATUS_AND_MASK_IRQ_MASK (1 << 24) +#define PM_STATUS_AND_MASK_FIQ_MASK (1 << 25) +#define L2C_NFABRIC_PM_CTL 0x4#define L2C_NFABRIC_PM_CTL_PWR_DOWN (1 << 20)
@@ -91,4 +109,56 @@ void armada_370_xp_pmsu_enable_l2_powerdown_onidle(void) }EXPORT_SYMBOL_GPL (armada_370_xp_pmsu_enable_l2_powerdown_onidle); +void armada_370_xp_pmsu_idle_prepare (bool deepidle) +{ +unsigned int hw_cpu =cpu_logical_map (smp_processor_id ()); +int reg; + + reg =readl (pmsu_mp_base +PM_STATUS_AND_MASK (hw_cpu)); +/* set WaitMask fields - wait for WFI signal */ + reg |=PM_STATUS_AND_MASK_CPU_IDLE_WAIT ; +/* enable wakeup events */ + reg |=PM_STATUS_AND_MASK_IRQ_WAKEUP |PM_STATUS_AND_MASK_FIQ_WAKEUP ; +/* set wait for snoop queue empty indication */ + reg |=PM_STATUS_AND_MASK_SNP_Q_EMPTY_WAIT ; +/* mask IRQ and FIQ from CPU */ + reg |=PM_STATUS_AND_MASK_IRQ_MASK |PM_STATUS_AND_MASK_FIQ_MASK ; +writel (reg, pmsu_mp_base +PM_STATUS_AND_MASK (hw_cpu)); + + reg =readl (pmsu_mp_base +PM_CONTROL_AND_CONFIG (hw_cpu)); +/* ask HW to power down the L2 Cache if needed */ +if (deepidle) + reg |=PM_CONTROL_AND_CONFIG_L2_PWDDN ; +/* request power down */ + reg |=PM_CONTROL_AND_CONFIG_PWDDN_REQ ; +writel (reg, pmsu_mp_base +PM_CONTROL_AND_CONFIG (hw_cpu)); + +/* Disable snoop disable by HW - SW is taking care of it */ + reg =readl (pmsu_mp_base +PM_CPU_POWER_DOWN_CONTROL (hw_cpu)); + reg |=PM_CPU_POWER_DOWN_DIS_SNP_Q_SKIP ; +writel (reg, pmsu_mp_base +PM_CPU_POWER_DOWN_CONTROL (hw_cpu)); +} +EXPORT_SYMBOL_GPL (armada_370_xp_pmsu_idle_prepare); + +noinline void armada_370_xp_pmsu_idle_restore(void) +{ +unsigned int hw_cpu =cpu_logical_map (smp_processor_id ()); +int reg; + +/* cancel ask HW to power down the L2 Cache if possible */ + reg =readl (pmsu_mp_base +PM_CONTROL_AND_CONFIG (hw_cpu)); + reg &= ~PM_CONTROL_AND_CONFIG_L2_PWDDN ; +writel (reg, pmsu_mp_base +PM_CONTROL_AND_CONFIG (hw_cpu)); + +/* cancel Enable wakeup events */ + reg =readl (pmsu_mp_base +PM_STATUS_AND_MASK (hw_cpu)); + reg &= ~(PM_STATUS_AND_MASK_IRQ_WAKEUP |PM_STATUS_AND_MASK_FIQ_WAKEUP ); + reg &= ~PM_STATUS_AND_MASK_CPU_IDLE_WAIT ; + reg &= ~PM_STATUS_AND_MASK_SNP_Q_EMPTY_WAIT ; +/* Mask interrupts */ + reg &= ~(PM_STATUS_AND_MASK_IRQ_MASK |PM_STATUS_AND_MASK_FIQ_MASK ); +writel (reg, pmsu_mp_base +PM_STATUS_AND_MASK (hw_cpu)); +} +EXPORT_SYMBOL_GPL (armada_370_xp_pmsu_idle_restore); +early_initcall (armada_370_xp_pmsu_init);
include/linux/armada-370-xp-pmsu.h +2 -0
@@ -12,5 +12,7 @@#define __ARMADA_370_XP__PMSU_Hvoid armada_370_xp_pmsu_enable_l2_powerdown_onidle (void ); +void armada_370_xp_pmsu_idle_prepare (bool deepidle); +void armada_370_xp_pmsu_idle_restore (void );#endif /* __ARMADA_370_XP__PMSU_H */
Signature · 1 line
1.8.1.2
PATCH 10/12 ARM: mvebu: Set the start address of a CPU in a separate function
In reply to Gregory CLEMENTSetting the start (or boot) address of A CPU is no more used only during SMP bring up, but will also be used by the CPU idle functions or later by the CPU hot plug ones.
This commit moves it in a separate function and exports it.
--- arch/arm/mach-mvebu/pmsu.c | 10 ++++++++-- include/linux/armada-370-xp-pmsu.h | 1 + 2 files changed, 9 insertions(+), 2 deletions(-)
arch/arm/mach-mvebu/pmsu.c +8 -2
@@ -56,6 +56,13 @@ static struct of_device_id of_pmsu_table[] = { { /* end of list */ }, }; +void armada_370_xp_pmsu_set_start_addr (void *start_addr,int hw_cpu) +{ +writel (virt_to_phys (start_addr), pmsu_mp_base + +PMSU_BOOT_ADDR_REDIRECT_OFFSET (hw_cpu)); +} +EXPORT_SYMBOL_GPL (armada_370_xp_pmsu_set_start_addr); +#ifdef CONFIG_SMP int armada_xp_boot_cpu (unsigned int cpu_id,void *boot_addr) {
@@ -68,8 +75,7 @@ int armada_xp_boot_cpu(unsigned int cpu_id, void *boot_addr) hw_cpu =cpu_logical_map (cpu_id); -writel (virt_to_phys (boot_addr), pmsu_mp_base + -PMSU_BOOT_ADDR_REDIRECT_OFFSET (hw_cpu)); +armada_370_xp_pmsu_set_start_addr (boot_addr, hw_cpu);/* Release CPU from reset by clearing reset bit*/ reg =readl (pmsu_reset_base +PMSU_RESET_CTL_OFFSET (hw_cpu));
include/linux/armada-370-xp-pmsu.h +1 -0
@@ -14,5 +14,6 @@void armada_370_xp_pmsu_enable_l2_powerdown_onidle (void );void armada_370_xp_pmsu_idle_prepare (bool deepidle);void armada_370_xp_pmsu_idle_restore (void ); +void armada_370_xp_pmsu_set_start_addr (void *start_addr,int hw_cpu);#endif /* __ARMADA_370_XP__PMSU_H */
Signature · 1 line
1.8.1.2
PATCH 11/12 cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC
In reply to Gregory CLEMENTAdd wfi/cpu idle/cpu deep idle power states support for Armada XP SoC.
All the latencies and the power consumption values used at the "armada_370_xp_idle_driver" structure are preliminary and will be modified in the future after running some measurements and analysis.
Based on the work of Nadav Haklai.
--- drivers/cpuidle/Kconfig | 6 ++ drivers/cpuidle/Makefile | 1 + drivers/cpuidle/cpuidle-armada-370-xp.c | 132 ++++++++++++++++++++++++++++++++ drivers/cpuidle/suspend-armada-370-xp.S | 91 ++++++++++++++++++++++ 4 files changed, 230 insertions(+) create mode 100644 drivers/cpuidle/cpuidle-armada-370-xp.c create mode 100644 drivers/cpuidle/suspend-armada-370-xp.S
drivers/cpuidle/Kconfig +6 -0
@@ -42,6 +42,12 @@ config CPU_IDLE_ZYNQ help Select this to enable cpuidle on Xilinx Zynq processors. +config CPU_IDLE_ARMADA_370_XP + bool "CPU Idle Driver for Armada 370/XP family processors" + depends on ARCH_MVEBU + help + Select this to enable cpuidle on Armada 370/XP processors. + endif config ARCH_NEEDS_CPU_IDLE_COUPLED
drivers/cpuidle/Makefile +1 -0
@@ -5,6 +5,7 @@ obj-y += cpuidle.o driver.o governor.o sysfs.o governors/ obj-$(CONFIG_ARCH_NEEDS_CPU_IDLE_COUPLED) += coupled.o +obj-$(CONFIG_CPU_IDLE_ARMADA_370_XP) += cpuidle-armada-370-xp.o suspend-armada-370-xp.o obj-$(CONFIG_CPU_IDLE_CALXEDA) += cpuidle-calxeda.o obj-$(CONFIG_ARCH_KIRKWOOD) += cpuidle-kirkwood.o obj-$(CONFIG_CPU_IDLE_ZYNQ) += cpuidle-zynq.o
drivers/cpuidle/cpuidle-armada-370-xp.c new file +132 -0
132 lines not on this page · raw message
drivers/cpuidle/suspend-armada-370-xp.S new file +91 -0
91 lines not on this page · raw message
Signature · 1 line
1.8.1.2
PATCH 11/12 cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC
Quoted 7 lines · Gregory CLEMENT
Add wfi/cpu idle/cpu deep idle power states support for Armada XP SoC.
All the latencies and the power consumption values used at the "
armada_370_xp_idle_driver" structure are preliminary and will be modified in the future after running some measurements and analysis.Based on the work of Nadav Haklai.
Atsushi Yamagata reported me an kernel crash when pm_support=off. Indeed I have lost a small chunk of code during my code reshaping. The following code will fix the issue. I will add it in the next version of this series.
[...]
drivers/cpuidle/cpuidle-armada-370-xp.c line 123
Quoted 11 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +9
-0+ +if (pm_support ==WFI ) +pr_info (" (WFI)\n" ); +else if (pm_support ==MV_CPU_IDLE ) +pr_info (" (CPU_IDLE)\n" ); +else if (pm_support ==MV_CPU_DEEP_IDLE ) +pr_info (" (CPU_DEEP_IDLE)\n" ); +else +pr_info (" (DISABLED)\n" );
Will be replaced by
+ else {
+ pr_info(" (DISABLED)\n");
+ return; /* do not register the cpuidle device */
+ }
Quoted 6 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +6
-0+ + armada_370_xp_idle_driver.state_count = pm_support +1 ; + +return cpuidle_register (&armada_370_xp_idle_driver,NULL ); +} +
Signature · 4 lines
Gregory Clement, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
PATCH 11/12 cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC
In reply to Gregory CLEMENTHi Gregory,
Quoted 10 lines · Gregory CLEMENT
Add wfi/cpu idle/cpu deep idle power states support for Armada XP SoC.
All the latencies and the power consumption values used at the "
armada_370_xp_idle_driver" structure are preliminary and will be modified in the future after running some measurements and analysis.Based on the work of Nadav Haklai.
Signed-off-by Nadav Haklai <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>
[...]
Quoted 5 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +5
-0+int pm_support =WFI ; +static int __init pm_enable_setup (char *str) +{ +if (!strncmp (str,"wfi" ,3 )) + pm_support =WFI ;
drivers/cpuidle/cpuidle-armada-370-xp.c line 56
Quoted 10 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +10
-0+else if (!strncmp (str,"idle" ,4 )) + pm_support =MV_CPU_IDLE ; +else if (!strncmp (str,"deepidle" ,6 )) + pm_support =MV_CPU_DEEP_IDLE ; +else if (!strncmp (str,"off" ,3 )) + pm_support =DISABLED ; + +return 1 ; +} +__setup ("pm_level=" , pm_enable_setup);
Why is this new (but undocumented) kernel commandline needed when there is already a way to configure the deepest C state from userspace? (c.f. the 'disable' file under /sys/devices/system/cpu/cpuX/cpuidle/stateY)
Kevin
PATCH 11/12 cpuidle: mvebu: Add initial cpu idle support for Armada 370/XP SoC
Hi Kevin,
Quoted 28 lines · Kevin Hilman
Hi Gregory,
Quoted 10 lines · Gregory CLEMENT
Add wfi/cpu idle/cpu deep idle power states support for Armada XP SoC.
All the latencies and the power consumption values used at the "
armada_370_xp_idle_driver" structure are preliminary and will be modified in the future after running some measurements and analysis.Based on the work of Nadav Haklai.
Signed-off-by Nadav Haklai <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>[...]
Quoted 11 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +11
-0+int pm_support =WFI ; +static int __init pm_enable_setup (char *str) +{ +if (!strncmp (str,"wfi" ,3 )) + pm_support =WFI ; +else if (!strncmp (str,"idle" ,4 )) + pm_support =MV_CPU_IDLE ; +else if (!strncmp (str,"deepidle" ,6 )) + pm_support =MV_CPU_DEEP_IDLE ; +else if (!strncmp (str,"off" ,3 )) + pm_support =DISABLED ;
Quoted 8 lines
drivers/cpuidle/cpuidle-armada-370-xp.c line 56
Quoted 4 lines
drivers/cpuidle/cpuidle-armada-370-xp.c new file +4
-0+ +return 1 ; +} +__setup ("pm_level=" , pm_enable_setup);Why is this new (but undocumented) kernel commandline needed when there is already a way to configure the deepest C state from userspace? (c.f. the 'disable' file under
/sys/devices/system/cpu/cpuX/cpuidle/stateY)
You're right I should remove this. I wanted look see if we really need it the first time I started to work on this series and then I forgot!
Thanks,
Gregory
Signature · 4 lines
Gregory Clement, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
PATCH 12/12 ARM: dts: mvebu: Add a new set of registers to the PMSU node
In reply to Gregory CLEMENTThe Power Management Unit Service block also controls the Coherency Fabric subsystem. This new set of registers is needed for the CPU idle implementation for the Armada XP, it allows to enter in a deep CPU idle state where the Coherency Fabric and the L2 cache are powerdown.
--- Documentation/devicetree/bindings/arm/armada-370-xp-pmsu.txt | 12 +++++++----- arch/arm/boot/dts/armada-xp.dtsi | 2 +- 2 files changed, 8 insertions(+), 6 deletions(-)
Documentation/devicetree/bindings/arm/armada-370-xp-pmsu.txt +7 -5
@@ -7,14 +7,16 @@ Required properties: - compatible: "marvell,armada-370-xp-pmsu" - reg: Should contain PMSU registers location and length. First pair - for the per-CPU SW Reset Control registers, second pair for the - Power Management Service Unit. + for the per-CPU SW Reset Control registers, second pair for the CPU + Power Management Service Unit registers, third pair for the Fabric Power + Management Service Unit registers. Example: -armada-370-xp-pmsu at d0022000 { +armada-370-xp-pmsu at 22000 { compatible = "marvell,armada-370-xp-pmsu"; - reg = <0xd0022100 0x430>, - <0xd0020800 0x20>; + reg = <0x22100 0x430>, + <0x20800 0x20>, + <0x22000 0x24>; };
arch/arm/boot/dts/armada-xp.dtsi +1 -1
@@ -41,7 +41,7 @@ armada-370-xp-pmsu at 22000 { compatible = "marvell,armada-370-xp-pmsu"; - reg = <0x22100 0x430>, <0x20800 0x20>; + reg = <0x22100 0x430>, <0x20800 0x20>, <0x22000 0x24>; }; serial at 12200 {
Signature · 1 line
1.8.1.2