Series to implement suspend for ppc/spapr
cover Series to implement suspend for ppc/spapr
Any comments on this series would be welcome. Hopefully someone who knows i386 can give some feedback on the possible bug fix, and whether the new wakeup method will suit i386.
Thanks,
Nick
Nicholas Piggin (3):
qmp: don't emit the RESET event on wakeup
machine: Add wakeup method to MachineClass
spapr: Implement ibm,suspend-me
hw/ppc/spapr.c | 11 +++++++++++ hw/ppc/spapr_rtas.c | 32 ++++++++++++++++++++++++++++++++ include/hw/boards.h | 1 + include/hw/ppc/spapr.h | 3 ++- vl.c | 31 +++++++++++++++++++++++++++++-- 5 files changed, 75 insertions(+), 3 deletions(-)
Signature · 1 line
2.20.1
PATCH 1/3 qmp: don't emit the RESET event on wakeup
Commit 1405819637f53 ("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid calling qapi_event_send_reset. Commit 76ed4b18debfe ("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.
---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.
vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
vl.c +1 -1
@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();
Signature · 1 line
2.20.1
Re: [Qemu-devel] [PATCH 1/3] qmp: don't emit the RESET event on wakeup
Quoted 20 lines · Nicholas Piggin
Commit
1405819637f53("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid callingqapi_event_send_reset. Commit76ed4b18debfe("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.Signed-off-by Nicholas Piggin <[email protected]>---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)vl.c +0
-0
vl.c void qemu_system_reset(ShutdownCause reason) · line 1556
Quoted 8 lines
vl.c +1
-1@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();
Yes, it seems correct and I've queued it for 4.1.
Paolo
Re: [Qemu-devel] [PATCH 1/3] qmp: don't emit the RESET event on wakeup
Quoted 15 lines · Paolo Bonzini
Quoted 14 lines · Nicholas Piggin
Commit
1405819637f53("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid callingqapi_event_send_reset. Commit76ed4b18debfe("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.Signed-off-by Nicholas Piggin <[email protected]>---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines
Quoted 6 lines
vl.c +0
-0vl.c void qemu_system_reset(ShutdownCause reason) · line 1556
Quoted 8 lines
vl.c +1
-1@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();Yes, it seems correct and I've queued it for 4.1.
Yes makes sense.
Would it be better to write out if(reason) e.g. replace "reason" with "reason != SHUTDOWN_CAUSE_NONE" ?
Going even further, I am asking myself if something like
hw/s390x/ipl.c +1 -1
@@ -577,7 +577,7 @@ void s390_ipl_reset_request(CPUState *cs, enum s390_reset reset_type)if (reset_type ==S390_RESET_MODIFIED_CLEAR || reset_type ==S390_RESET_LOAD_NORMAL ) {/* ignore -no-reboot, send no event */ -qemu_system_reset_request (SHUTDOWN_CAUSE_SUBSYSTEM_RESET ); +qemu_system_reset_request (SHUTDOWN_CAUSE_NONE ); }else {qemu_system_reset_request (SHUTDOWN_CAUSE_GUEST_RESET ); }
would also work.
Re: [Qemu-devel] [PATCH 1/3] qmp: don't emit the RESET event on wakeup
Christian Borntraeger's on July 18, 2019 9:27 pm:
Quoted 35 lines · Paolo Bonzini
Quoted 15 lines · Nicholas Piggin
Quoted 14 lines
Commit
1405819637f53("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid callingqapi_event_send_reset. Commit76ed4b18debfe("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.Signed-off-by Nicholas Piggin <[email protected]>---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)Quoted 16 lines
Quoted 6 lines
vl.c +0
-0vl.c void qemu_system_reset(ShutdownCause reason) · line 1556
Quoted 8 lines
vl.c +1
-1@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();Yes, it seems correct and I've queued it for 4.1.
Quoted 2 lines
Yes makes sense.
Would it be better to write out if(reason) e.g. replace "reason" with "reason !=SHUTDOWN_CAUSE_NONE" ?
I guess it's a matter of preference so I won't weigh in :) My patch did try to revert back to the previous form but I'm happy to change it.
Quoted 16 lines
Going even further, I am asking myself if something like
hw/s390x/ipl.c +1
-1@@ -577,7 +577,7 @@ void s390_ipl_reset_request(CPUState *cs, enum s390_reset reset_type)if (reset_type ==S390_RESET_MODIFIED_CLEAR || reset_type ==S390_RESET_LOAD_NORMAL ) {/* ignore -no-reboot, send no event */ -qemu_system_reset_request (SHUTDOWN_CAUSE_SUBSYSTEM_RESET ); +qemu_system_reset_request (SHUTDOWN_CAUSE_NONE ); }else {qemu_system_reset_request (SHUTDOWN_CAUSE_GUEST_RESET ); }
Quoted 1 line
would also work.
If that works for you, then you could take out the test in the reset code. You would have to modify qemu_system_reset_request too of course.
But it seems a bit unsatisfactory to change the reason for the request so as to influence behaviour. Either the requests should ask for particular behaviour, or the logic for determining how to handle the type of request should remain in the reset logic, I would say.
Thanks,
Nick
Christian Borntraeger
Acked-byRe: [Qemu-devel] [qemu-s390x] [PATCH 1/3] qmp: don't emit the RESET event on wakeup
Quoted 64 lines · Nicholas Piggin
Christian Borntraeger's on July 18, 2019 9:27 pm:
Quoted 35 lines · Paolo Bonzini
Quoted 15 lines · Nicholas Piggin
Quoted 14 lines
Commit
1405819637f53("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid callingqapi_event_send_reset. Commit76ed4b18debfe("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.Signed-off-by Nicholas Piggin <[email protected]>---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)Quoted 16 lines
Quoted 6 lines
vl.c +0
-0vl.c void qemu_system_reset(ShutdownCause reason) · line 1556
Quoted 8 lines
vl.c +1
-1@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();Yes, it seems correct and I've queued it for 4.1.
Quoted 2 lines
Yes makes sense.
Would it be better to write out if(reason) e.g. replace "reason" with "reason !=SHUTDOWN_CAUSE_NONE" ?I guess it's a matter of preference so I won't weigh in :) My patch did try to revert back to the previous form but I'm happy to change it.
Quoted 16 lines
Going even further, I am asking myself if something like
hw/s390x/ipl.c +1
-1@@ -577,7 +577,7 @@ void s390_ipl_reset_request(CPUState *cs, enum s390_reset reset_type)if (reset_type ==S390_RESET_MODIFIED_CLEAR || reset_type ==S390_RESET_LOAD_NORMAL ) {/* ignore -no-reboot, send no event */ -qemu_system_reset_request (SHUTDOWN_CAUSE_SUBSYSTEM_RESET ); +qemu_system_reset_request (SHUTDOWN_CAUSE_NONE ); }else {qemu_system_reset_request (SHUTDOWN_CAUSE_GUEST_RESET ); }Quoted 1 line
would also work.
If that works for you, then you could take out the test in the reset code. You would have to modify
qemu_system_reset_requesttoo of course.
Quoted 4 lines
But it seems a bit unsatisfactory to change the reason for the request so as to influence behaviour. Either the requests should ask for particular behaviour, or the logic for determining how to handle the type of request should remain in the reset logic, I would say.
I agree.
Anyway I think your patch is good as is.
Cornelia Huck
Acked-byRe: [Qemu-devel] [PATCH 1/3] qmp: don't emit the RESET event on wakeup
In reply to Nicholas PigginQuoted 35 lines · On Fri, 19 Jul 2019 09:24:20 +1000 Nicholas Piggin <[email protected]> wrote:
Christian Borntraeger's on July 18, 2019 9:27 pm:
Quoted 34 lines · Paolo Bonzini
Quoted 31 lines · Nicholas Piggin
Quoted 20 lines
Commit
1405819637f53("qmp: don't emit the RESET event on wakeup from S3") changed system wakeup to avoid callingqapi_event_send_reset. Commit76ed4b18debfe("s390/ipl: fix ipl with -no-reboot") appears to have inadvertently broken that logic.Signed-off-by Nicholas Piggin <[email protected]>---
I'm not quite sure if this patch is correct and haven't tested it, I found it by inspection. If this patch is incorrect, I will have to adjust patch 2.vl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)vl.c +0
-0vl.c void qemu_system_reset(ShutdownCause reason) · line 1556
Quoted 8 lines
vl.c +1
-1@@ -1550,7 +1550,7 @@ void qemu_system_reset(ShutdownCause reason) }else {qemu_devices_reset (); } -if (reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) { +if (reason && reason !=SHUTDOWN_CAUSE_SUBSYSTEM_RESET ) {qapi_event_send_reset (shutdown_caused_by_guest (reason), reason); }cpu_synchronize_all_post_reset ();Yes, it seems correct and I've queued it for 4.1.
FWIW,
Quoted 24 lines
Quoted 3 lines
Yes makes sense.
Would it be better to write out if(reason) e.g. replace "reason" with "reason !=SHUTDOWN_CAUSE_NONE" ?I guess it's a matter of preference so I won't weigh in :) My patch did try to revert back to the previous form but I'm happy to change it.
Quoted 16 lines
Going even further, I am asking myself if something like
hw/s390x/ipl.c +1
-1@@ -577,7 +577,7 @@ void s390_ipl_reset_request(CPUState *cs, enum s390_reset reset_type)if (reset_type ==S390_RESET_MODIFIED_CLEAR || reset_type ==S390_RESET_LOAD_NORMAL ) {/* ignore -no-reboot, send no event */ -qemu_system_reset_request (SHUTDOWN_CAUSE_SUBSYSTEM_RESET ); +qemu_system_reset_request (SHUTDOWN_CAUSE_NONE ); }else {qemu_system_reset_request (SHUTDOWN_CAUSE_GUEST_RESET ); }
Quoted 1 line
Quoted 1 line
would also work.
Doesn't that have side effects in qemu_system_reset_request()?
Quoted 8 lines
If that works for you, then you could take out the test in the reset code. You would have to modify
qemu_system_reset_requesttoo of course.But it seems a bit unsatisfactory to change the reason for the request so as to influence behaviour. Either the requests should ask for particular behaviour, or the logic for determining how to handle the type of request should remain in the reset logic, I would say.
Agreed.
PATCH 2/3 machine: Add wakeup method to MachineClass
In reply to Nicholas PigginWaking from suspend is not logically a machine reset on all machines, particularly in the paravirtualized case rather than hardware emulated. The ppc spapr machine for example just invokes hypervisor to suspend, and expects that call to return with the machine in the same state (modulo some possible migration and reconfiguration details).
Implement a machine ->wakeup method and use that if it exists.
--- include/hw/boards.h | 1 + vl.c | 29 ++++++++++++++++++++++++++++- 2 files changed, 29 insertions(+), 1 deletion(-)
include/hw/boards.h +1 -0
@@ -180,6 +180,7 @@ struct MachineClass {void (*init)(MachineState *state);void (*reset)(MachineState *state); +void (*wakeup)(MachineState *state);void (*hot_add_cpu)(MachineState *state,const int64_t id,Error **errp);int (*kvm_type)(MachineState *machine,const char *arg);void (*smp_parse)(MachineState *ms,QemuOpts *opts);
vl.c +28 -1
@@ -1556,6 +1556,33 @@ void qemu_system_reset(ShutdownCause reason)cpu_synchronize_all_post_reset (); } +/* +* Wake the VM after suspend. +*/ +static void qemu_system_wakeup (void ) +{ +MachineClass *mc; + + mc = current_machine ?MACHINE_GET_CLASS (current_machine) :NULL ; + +if (mc && mc->wakeup) { + mc->wakeup (current_machine); + }else { +/* +* Fall back to old reset wakeup method. Platforms supporting +* wakeup should be converted over to use ->wakeup and this +* fallback code removed. +*/ +cpu_synchronize_all_states (); +if (mc && mc->reset) { + mc->reset (current_machine); + }else { +qemu_devices_reset (); + } +cpu_synchronize_all_post_reset (); + } +} +void qemu_system_guest_panicked (GuestPanicInformation *info) {qemu_log_mask (LOG_GUEST_ERROR ,"Guest crashed" );
@@ -1764,7 +1791,7 @@ static bool main_loop_should_exit(void) }if (qemu_wakeup_requested ()) {pause_all_vcpus (); -qemu_system_reset (SHUTDOWN_CAUSE_NONE ); +qemu_system_wakeup ();notifier_list_notify (&wakeup_notifiers, &wakeup_reason); wakeup_reason =QEMU_WAKEUP_REASON_NONE ;resume_all_vcpus ();
Signature · 1 line
2.20.1
PATCH 3/3 spapr: Implement ibm,suspend-me
In reply to Nicholas PigginThis has been useful to modify and test the Linux pseries suspend code but it requires modification to the guest to call it (due to being gated by other unimplemented features). It is not otherwise used by Linux yet, but work is slowly progressing there.
This allows a (lightly modified) guest kernel to suspend with echo mem > /sys/power/state and be resumed with system_wakeup monitor command.
--- hw/ppc/spapr.c | 11 +++++++++++ hw/ppc/spapr_rtas.c | 32 ++++++++++++++++++++++++++++++++ include/hw/ppc/spapr.h | 3 ++- 3 files changed, 45 insertions(+), 1 deletion(-)
hw/ppc/spapr.c +11 -0
@@ -1701,6 +1701,14 @@ static int spapr_reset_drcs(Object *child, void *opaque)return 0 ; } +static void spapr_machine_wakeup (MachineState *machine) +{ +/* +* Nothing needs to be done to resume a suspended guest because +* suspending does not change the machine state. +*/ +} +static void spapr_machine_reset (MachineState *machine) {SpaprMachineState *spapr =SPAPR_MACHINE (machine);
@@ -3078,6 +3086,8 @@ static void spapr_machine_init(MachineState *machine)qemu_register_boot_set (spapr_boot_set, spapr); +qemu_register_wakeup_support (); +if (kvm_enabled ()) {/* to stop and start vmclock */ qemu_add_vm_change_state_handler(cpu_ppc_clock_vm_state_change,
@@ -4373,6 +4383,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data)*/ mc->init = spapr_machine_init; mc->reset = spapr_machine_reset; + mc->wakeup = spapr_machine_wakeup; mc->block_default_type =IF_SCSI ; mc->max_cpus =1024 ; mc->no_parallel =1 ;
hw/ppc/spapr_rtas.c +32 -0
@@ -216,6 +216,36 @@ static void rtas_stop_self(PowerPCCPU *cpu, SpaprMachineState *spapr,qemu_cpu_kick (cs); } +static void rtas_ibm_suspend_me (PowerPCCPU *cpu,SpaprMachineState *spapr, +uint32_t token,uint32_t nargs, +target_ulong args, +uint32_t nret,target_ulong rets) +{ +CPUState *cs; + +if (nargs !=0 || nret !=1 ) { +rtas_st (rets,0 ,RTAS_OUT_PARAM_ERROR ); +return ; + } + +CPU_FOREACH (cs) { +PowerPCCPU *c =POWERPC_CPU (cs); +CPUPPCState *e = &c->env; +if (c == cpu) { +continue ; + } + +/* See h_join */ +if (!cs->halted || (e->msr & (1ULL <<MSR_EE ))) { +rtas_st (rets,0 ,H_MULTI_THREADS_ACTIVE ); +return ; + } + } + +qemu_system_suspend_request (); +rtas_st (rets,0 ,RTAS_OUT_SUCCESS ); +} +static inline int sysparm_st (target_ulong addr,target_ulong len,const void *val,uint16_t vallen) {
@@ -483,6 +513,8 @@ static void core_rtas_register_types(void) rtas_query_cpu_stopped_state);spapr_rtas_register (RTAS_START_CPU ,"start-cpu" , rtas_start_cpu);spapr_rtas_register (RTAS_STOP_SELF ,"stop-self" , rtas_stop_self); +spapr_rtas_register (RTAS_IBM_SUSPEND_ME ,"ibm,suspend-me" , + rtas_ibm_suspend_me);spapr_rtas_register (RTAS_IBM_GET_SYSTEM_PARAMETER ,"ibm,get-system-parameter" , rtas_ibm_get_system_parameter);
include/hw/ppc/spapr.h +2 -1
@@ -631,8 +631,9 @@ target_ulong spapr_hypercall(PowerPCCPU *cpu, target_ulong opcode,#define RTAS_IBM_CREATE_PE_DMA_WINDOW (RTAS_TOKEN_BASE + 0x27)#define RTAS_IBM_REMOVE_PE_DMA_WINDOW (RTAS_TOKEN_BASE + 0x28)#define RTAS_IBM_RESET_PE_DMA_WINDOW (RTAS_TOKEN_BASE + 0x29) +#define RTAS_IBM_SUSPEND_ME (RTAS_TOKEN_BASE + 0x2A) -#define RTAS_TOKEN_MAX (RTAS_TOKEN_BASE + 0x2A) +#define RTAS_TOKEN_MAX (RTAS_TOKEN_BASE + 0x2B)/* RTAS ibm,get-system-parameter token values */ #define RTAS_SYSPARM_SPLPAR_CHARACTERISTICS 20
Signature · 1 line
2.20.1
Quoted 3 lines · Nicholas Piggin
Any comments on this series would be welcome. Hopefully someone who knows i386 can give some feedback on the possible bug fix, and whether the new wakeup method will suit i386.
Looks good, though only i386 supports wakeup so perhaps it's better to DTRT and move the reset to the PC machine's wakeup method. Then pseries need not implement mc->wakeup at all.
Paolo
Quoted 15 lines
Thanks,
NickNicholas Piggin (3):
qmp: don't emit the RESET event on wakeup
machine: Add wakeup method to MachineClass
spapr: Implement ibm,suspend-mehw/ppc/spapr.c | 11 +++++++++++ hw/ppc/spapr_rtas.c | 32 ++++++++++++++++++++++++++++++++ include/hw/boards.h | 1 + include/hw/ppc/spapr.h | 3 ++- vl.c | 31 +++++++++++++++++++++++++++++-- 5 files changed, 75 insertions(+), 3 deletions(-)
Paolo Bonzini's on July 18, 2019 9:08 pm:
Quoted 8 lines · Nicholas Piggin
Quoted 3 lines
Any comments on this series would be welcome. Hopefully someone who knows i386 can give some feedback on the possible bug fix, and whether the new wakeup method will suit i386.
Looks good, though only i386 supports wakeup so perhaps it's better to DTRT and move the reset to the PC machine's wakeup method. Then pseries need not implement
mc->wakeupat all.
Yeah that probably makes more sense because the i386 patch should be quite trivial. I will try that and re-send.
Thanks,
Nick