hw/arm/exynos: Fix poweroff
Krzysztof Kozlowski
maintainercover hw/arm/exynos: Fix poweroff
Hi,
Beside two minor cleanups, the patchset implements proper Exynos power off. Still the GIC [1] fix is needed because the kernel will not reach power off handlers without it.
[1] http://patchwork.ozlabs.org/patch/738320/
Best regards,
Krzysztof
Krzysztof Kozlowski (3):
hw/arm/exynos: Use type define instead of hard-coded a9mpcore_priv
string
hw/intc/exynos: Constify array of combiner interrupts hw/arm/exynos: Add support for system poweroff
hw/arm/exynos4210.c | 3 ++- hw/intc/exynos4210_gic.c | 2 +- hw/misc/exynos4210_pmu.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+), 2 deletions(-)
Signature · 1 line
2.9.3
Krzysztof Kozlowski
maintainerPATCH 1/3 hw/arm/exynos: Use type define instead of hard-coded a9mpcore_priv string
Use a define for a9mpcore_priv device type name instead of hard-coded string.
--- hw/arm/exynos4210.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
hw/arm/exynos4210.c +2 -1
@@ -26,6 +26,7 @@#include "qemu-common.h" #include "qemu/log.h" #include "cpu.h" +#include "hw/cpu/a9mpcore.h" #include "hw/boards.h" #include "sysemu/sysemu.h" #include "hw/sysbus.h"
@@ -212,7 +213,7 @@ static void exynos4210_init(Object *obj) }/* Private memory region and Internal GIC */ - dev =qdev_create (NULL ,"a9mpcore_priv" ); + dev =qdev_create (NULL ,TYPE_A9MPCORE_PRIV );qdev_prop_set_uint32 (dev,"num-cpu" ,EXYNOS4210_NCPUS );qdev_init_nofail (dev); busdev =SYS_BUS_DEVICE (dev);
Signature · 1 line
2.9.3
Re: [Qemu-devel] [Qemu-arm] [PATCH 1/3] hw/arm/exynos: Use type define instead of hard-coded a9mpcore_priv string
Quoted 4 lines · Krzysztof Kozlowski
Use a define for
a9mpcore_privdevice type name instead of hard-coded string.Signed-off-by Krzysztof Kozlowski <[email protected]>
Quoted 26 lines
--- hw/arm/exynos4210.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)hw/arm/exynos4210.c +2
-1@@ -26,6 +26,7 @@#include "qemu-common.h" #include "qemu/log.h" #include "cpu.h" +#include "hw/cpu/a9mpcore.h" #include "hw/boards.h" #include "sysemu/sysemu.h" #include "hw/sysbus.h" @@ -212,7 +213,7 @@ static void exynos4210_init(Object *obj) }/* Private memory region and Internal GIC */ - dev =qdev_create (NULL ,"a9mpcore_priv" ); + dev =qdev_create (NULL ,TYPE_A9MPCORE_PRIV );qdev_prop_set_uint32 (dev,"num-cpu" ,EXYNOS4210_NCPUS );qdev_init_nofail (dev); busdev =SYS_BUS_DEVICE (dev);
Krzysztof Kozlowski
maintainerPATCH 2/3 hw/intc/exynos: Constify array of combiner interrupts
In reply to Krzysztof KozlowskiThe static array of interrupt combiner mappings is not modified so it can be made const for code safeness.
--- hw/intc/exynos4210_gic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
hw/intc/exynos4210_gic.c +1 -1
@@ -116,7 +116,7 @@ enum ExtInt {* which is INTG16 in Internal Interrupt Combiner. */ -static uint32_t +static const uint32_t combiner_grp_to_gic_id[64 -EXYNOS4210_MAX_EXT_COMBINER_OUT_IRQ ][8 ] = {/* int combiner groups 16-19 */ { }, { }, { }, { },
Signature · 1 line
2.9.3
Re: [Qemu-devel] [Qemu-arm] [PATCH 2/3] hw/intc/exynos: Constify array of combiner interrupts
Quoted 4 lines · Krzysztof Kozlowski
The static array of interrupt combiner mappings is not modified so it can be made const for code safeness.
Signed-off-by Krzysztof Kozlowski <[email protected]>
Quoted 18 lines
--- hw/intc/exynos4210_gic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)hw/intc/exynos4210_gic.c +1
-1@@ -116,7 +116,7 @@ enum ExtInt {* which is INTG16 in Internal Interrupt Combiner. */ -static uint32_t +static const uint32_t combiner_grp_to_gic_id[64 -EXYNOS4210_MAX_EXT_COMBINER_OUT_IRQ ][8 ] = {/* int combiner groups 16-19 */ { }, { }, { }, { },
Krzysztof Kozlowski
maintainerPATCH 3/3 hw/arm/exynos: Add support for system poweroff
In reply to Krzysztof KozlowskiOn all Exynos-based boards, the system powers down itself by driving PS_HOLD signal low. Handle writing to respective PMU register to fix power off failure:
reboot: Power down
Unable to poweroff system
shutdown: 31 output lines suppressed due to ratelimiting
Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000000
CPU: 0 PID: 1 Comm: shutdown Not tainted 4.11.0-rc8 #846
Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
[<c031050c>] (unwind_backtrace) from [<c030ba6c>] (show_stack+0x10/0x14)
[<c030ba6c>] (show_stack) from [<c05b2800>] (dump_stack+0x88/0x9c)
[<c05b2800>] (dump_stack) from [<c03d3140>] (panic+0xdc/0x268)
[<c03d3140>] (panic) from [<c0343614>] (do_exit+0xa90/0xab4)
[<c0343614>] (do_exit) from [<c035f2dc>] (SyS_reboot+0x164/0x1d0)
[<c035f2dc>] (SyS_reboot) from [<c0307c80>] (ret_fast_syscall+0x0/0x3c)--- hw/misc/exynos4210_pmu.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
hw/misc/exynos4210_pmu.c +14 -0
@@ -26,6 +26,7 @@#include "qemu/osdep.h" #include "hw/sysbus.h" +#include "sysemu/sysemu.h" #ifndef DEBUG_PMU #define DEBUG_PMU 0
@@ -397,6 +398,12 @@ typedef struct Exynos4210PmuState {uint32_t reg[PMU_NUM_OF_REGISTERS ]; } Exynos4210PmuState; +static void exynos4210_pmu_poweroff (void ) +{ +PRINT_DEBUG ("QEMU PMU: PS_HOLD bit down, powering off\n" ); +qemu_system_shutdown_request (); +} +static uint64_t exynos4210_pmu_read (void *opaque,hwaddr offset,unsigned size) {
@@ -428,6 +435,13 @@ static void exynos4210_pmu_write(void *opaque, hwaddr offset,PRINT_DEBUG_EXTEND ("%s <0x%04x> <- 0x%04x\n" , reg_p->name, (uint32_t )offset, (uint32_t )val); s->reg[i] = val; +if ((offset ==PS_HOLD_CONTROL ) && ((val &BIT (8 )) ==0 )) { +/* +* We are interested only in setting data bit +* of PS_HOLD_CONTROL register to indicate power off request. +*/ +exynos4210_pmu_poweroff (); + }return ; } reg_p++;
Signature · 1 line
2.9.3
Re: [Qemu-arm] [Qemu-devel] [PATCH 3/3] hw/arm/exynos: Add support for system poweroff
Quoted 24 lines · Krzysztof Kozlowski
On all Exynos-based boards, the system powers down itself by driving
PS_HOLDsignal low. Handle writing to respective PMU register to fix power off failure:reboot: Power down Unable to poweroff system shutdown: 31 output lines suppressed due to ratelimiting Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000000 CPU: 0 PID: 1 Comm: shutdown Not tainted 4.11.0-rc8 #846 Hardware name: SAMSUNG EXYNOS (Flattened Device Tree) [<c031050c>] (unwind_backtrace) from [<c030ba6c>] (show_stack+0x10/0x14) [<c030ba6c>] (show_stack) from [<c05b2800>] (dump_stack+0x88/0x9c) [<c05b2800>] (dump_stack) from [<c03d3140>] (panic+0xdc/0x268) [<c03d3140>] (panic) from [<c0343614>] (do_exit+0xa90/0xab4) [<c0343614>] (do_exit) from [<c035f2dc>] (SyS_reboot+0x164/0x1d0) [<c035f2dc>] (SyS_reboot) from [<c0307c80>] (ret_fast_syscall+0x0/0x3c)Signed-off-by Krzysztof Kozlowski <[email protected]>--- hw/misc/exynos4210_pmu.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
hw/misc/exynos4210_pmu.c typedef struct Exynos4210PmuState { · line 404
Quoted 7 lines
hw/misc/exynos4210_pmu.c +4
-0@@ -397,6 +398,12 @@ typedef struct Exynos4210PmuState {uint32_t reg[PMU_NUM_OF_REGISTERS ]; } Exynos4210PmuState; +static void exynos4210_pmu_poweroff (void ) +{ +PRINT_DEBUG ("QEMU PMU: PS_HOLD bit down, powering off\n" ); +qemu_system_shutdown_request ();
Depending on whether your patch lands before or after mine, we'll have to update this. It sounds like it is a guest-triggered reset, so my patch would have this use SHUTDOWN_CAUSE_GUEST_SHUTDOWN.
https://lists.gnu.org/archive/html/qemu-devel/2017-05/msg01380.html
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Krzysztof Kozlowski
maintainerRe: [Qemu-arm] [Qemu-devel] [PATCH 3/3] hw/arm/exynos: Add support for system poweroff
Quoted 37 lines · Eric Blake
Quoted 24 lines · Krzysztof Kozlowski
On all Exynos-based boards, the system powers down itself by driving
PS_HOLDsignal low. Handle writing to respective PMU register to fix power off failure:reboot: Power down Unable to poweroff system shutdown: 31 output lines suppressed due to ratelimiting Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000000 CPU: 0 PID: 1 Comm: shutdown Not tainted 4.11.0-rc8 #846 Hardware name: SAMSUNG EXYNOS (Flattened Device Tree) [<c031050c>] (unwind_backtrace) from [<c030ba6c>] (show_stack+0x10/0x14) [<c030ba6c>] (show_stack) from [<c05b2800>] (dump_stack+0x88/0x9c) [<c05b2800>] (dump_stack) from [<c03d3140>] (panic+0xdc/0x268) [<c03d3140>] (panic) from [<c0343614>] (do_exit+0xa90/0xab4) [<c0343614>] (do_exit) from [<c035f2dc>] (SyS_reboot+0x164/0x1d0) [<c035f2dc>] (SyS_reboot) from [<c0307c80>] (ret_fast_syscall+0x0/0x3c)Signed-off-by Krzysztof Kozlowski <[email protected]>--- hw/misc/exynos4210_pmu.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)hw/misc/exynos4210_pmu.c typedef struct Exynos4210PmuState { · line 404
Quoted 7 lines
hw/misc/exynos4210_pmu.c +4
-0@@ -397,6 +398,12 @@ typedef struct Exynos4210PmuState {uint32_t reg[PMU_NUM_OF_REGISTERS ]; } Exynos4210PmuState; +static void exynos4210_pmu_poweroff (void ) +{ +PRINT_DEBUG ("QEMU PMU: PS_HOLD bit down, powering off\n" ); +qemu_system_shutdown_request ();Depending on whether your patch lands before or after mine, we'll have to update this. It sounds like it is a guest-triggered reset, so my patch would have this use
SHUTDOWN_CAUSE_GUEST_SHUTDOWN.
Sure, I can rebase and fix the difference once your patchset gets in.
Best regards,
Krzysztof
Re: [Qemu-arm] [PATCH 3/3] hw/arm/exynos: Add support for system poweroff
In reply to Krzysztof KozlowskiQuoted 18 lines · Krzysztof Kozlowski
On all Exynos-based boards, the system powers down itself by driving
PS_HOLDsignal low. Handle writing to respective PMU register to fix power off failure:reboot: Power down Unable to poweroff system shutdown: 31 output lines suppressed due to ratelimiting Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000000 CPU: 0 PID: 1 Comm: shutdown Not tainted 4.11.0-rc8 #846 Hardware name: SAMSUNG EXYNOS (Flattened Device Tree) [<c031050c>] (unwind_backtrace) from [<c030ba6c>] (show_stack+0x10/0x14) [<c030ba6c>] (show_stack) from [<c05b2800>] (dump_stack+0x88/0x9c) [<c05b2800>] (dump_stack) from [<c03d3140>] (panic+0xdc/0x268) [<c03d3140>] (panic) from [<c0343614>] (do_exit+0xa90/0xab4) [<c0343614>] (do_exit) from [<c035f2dc>] (SyS_reboot+0x164/0x1d0) [<c035f2dc>] (SyS_reboot) from [<c0307c80>] (ret_fast_syscall+0x0/0x3c)
Quoted 1 line
Signed-off-by Krzysztof Kozlowski <[email protected]>
give or take the minor change mentioned by Eric if needed.
thanks
-- PMM