ppc: Fix migration issue with quiesced state
v3 cover ppc: Fix migration issue with quiesced state
Changes in v3:
- removed extra include
- changed the test to set the machine earlier
CI run: https://gitlab.com/farosas/qemu/-/pipelines/2252213960
v2:
https://lore.kernel.org/r/[email protected]
A while back, Fabian reported two issues with migration on ppc64le, one of which was already fixed. There remains this one from the previous series:
pseries migration causes a guest hang after migration due to a new variable used to track the stopped state of vcpus, which is not migrated. Fix is to migrate the new variable. To avoid breaking backward migration, a compat property is added. For forward migration, a heuritics based on the known QEMU code is used.
Thanks
CI run: https://gitlab.com/farosas/qemu/-/pipelines/2220496446
v1:
https://lore.kernel.org/r/[email protected]
Fabiano Rosas (2):
target/ppc: Fix env->quiesced migration tests/functional: Add a OS level migration test for pseries
hw/ppc/spapr.c | 6 +++ target/ppc/cpu.h | 1 + target/ppc/cpu_init.c | 7 +++ target/ppc/machine.c | 62 ++++++++++++++++++++++++ tests/functional/ppc64/test_migration.py | 12 +++++ tests/functional/ppc64/test_pseries.py | 35 +++++++++++++ 6 files changed, 123 insertions(+)
Signature · 1 line
2.51.0
PATCH v3 1/2 target/ppc: Fix env->quiesced migration
The commit referenced (from QEMU 10.0) has changed the way the pseries machine marks a cpu as quiesced. Previously, the cpu->halted value from QEMU common cpu code was (incorrectly) used. With the fix, the env->quiesced variable starts being used, which improves on the original situation, but also causes a side effect after migration:
The env->quiesced is set at reset and never migrated, which causes the destination QEMU to stop delivering interrupts and hang the machine.
To fix the issue from this point on, start migrating the env->quiesced value.
For QEMU versions < 10.0, sending the new element on the stream would cause migration to be aborted, so add the appropriate compatibility property to omit the new subsection.
Independently of this patch, all migrations from QEMU versions < 10.0 would result in a hang since the older QEMU never migrates env->quiesced. This is bad because it leaves machines already running on the old QEMU without a migration path into newer versions.
As a workaround, use a few heuristics to infer the new value of env->quiesced based on cpu->halted, LPCR and PSSCR bits that are usually set/cleared along with quiesced.
Note that this was tested with -cpu power9 and -machine ic-mode=xive due to another bug affecting migration of XICS guests. Tested both forward and backward migration and savevm/loadvm from 9.2 and 10.0.
Also tested loadvm of a savevm image that contains a mix of cpus both halted and not halted.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3079
--- hw/ppc/spapr.c | 6 +++++ target/ppc/cpu.h | 1 + target/ppc/cpu_init.c | 7 +++++ target/ppc/machine.c | 62 +++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 76 insertions(+)
hw/ppc/spapr.c +6 -0
@@ -4788,8 +4788,14 @@ DEFINE_SPAPR_MACHINE(10, 1);*/ static void spapr_machine_10_0_class_options (MachineClass *mc) { +static GlobalProperty spapr_compat_10_0[] = { + {TYPE_POWERPC_CPU ,"rtas-stopped-state" ,"false" }, + }; +spapr_machine_10_1_class_options (mc);compat_props_add (mc->compat_props, hw_compat_10_0, hw_compat_10_0_len); +compat_props_add (mc->compat_props, spapr_compat_10_0, +G_N_ELEMENTS (spapr_compat_10_0)); }DEFINE_SPAPR_MACHINE (10 ,0 );
target/ppc/cpu.h +1 -0
@@ -1529,6 +1529,7 @@ struct ArchCPU {void *machine_data;int32_t node_id;/* NUMA node this CPU belongs to */ PPCHash64Options *hash64_opts; +bool rtas_stopped_state;/* Those resources are used only during code translation */ /* opcode handlers */
target/ppc/cpu_init.c +7 -0
@@ -55,6 +55,11 @@/* #define PPC_DEBUG_SPR */ /* #define USE_APPLE_GDB */ +static const Property powerpc_cpu_properties[] = { +DEFINE_PROP_BOOL ("rtas-stopped-state" , PowerPCCPU, + rtas_stopped_state, true), +}; +static inline void vscr_init (CPUPPCState *env,uint32_t val) {/* Altivec always uses round-to-nearest */
@@ -7529,6 +7534,8 @@ static void ppc_cpu_class_init(ObjectClass *oc, const void *data) &pcc->parent_unrealize); pcc->pvr_match = ppc_pvr_match_default; +device_class_set_props (dc, powerpc_cpu_properties); +resettable_class_set_parent_phases (rc,NULL , ppc_cpu_reset_hold,NULL , &pcc->parent_phases);
target/ppc/machine.c +62 -0
80 lines not on this page · raw message
Signature · 1 line
2.51.0
Re: [PATCH v3 1/2] target/ppc: Fix env->quiesced migration
Quoted 29 lines · Fabiano Rosas
The commit referenced (from QEMU 10.0) has changed the way the pseries machine marks a cpu as quiesced. Previously, the
cpu->haltedvalue from QEMU common cpu code was (incorrectly) used. With the fix, theenv->quiescedvariable starts being used, which improves on the original situation, but also causes a side effect after migration:The
env->quiescedis set at reset and never migrated, which causes the destination QEMU to stop delivering interrupts and hang the machine.To fix the issue from this point on, start migrating the
env->quiescedvalue.For QEMU versions < 10.0, sending the new element on the stream would cause migration to be aborted, so add the appropriate compatibility property to omit the new subsection.
Independently of this patch, all migrations from QEMU versions < 10.0 would result in a hang since the older QEMU never migrates
env->quiesced. This is bad because it leaves machines already running on the old QEMU without a migration path into newer versions.As a workaround, use a few heuristics to infer the new value of
env->quiescedbased oncpu->halted, LPCR and PSSCR bits that are usually set/cleared along with quiesced.Note that this was tested with -cpu power9 and -machine ic-mode=xive due to another bug affecting migration of XICS guests. Tested both forward and backward migration and savevm/loadvm from 9.2 and 10.0.
Quoted 7 lines
Also tested loadvm of a savevm image that contains a mix of cpus both halted and not halted.
Reported-by Fabian Vogt <[email protected]>Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3079
Fixes fb802acdc8b ("ppc/spapr: Fix RTAS stopped state")Signed-off-by Fabiano Rosas <[email protected]>
Quoted 150 lines
--- hw/ppc/spapr.c | 6 +++++ target/ppc/cpu.h | 1 + target/ppc/cpu_init.c | 7 +++++ target/ppc/machine.c | 62 +++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 76 insertions(+)hw/ppc/spapr.c +6
-0@@ -4788,8 +4788,14 @@ DEFINE_SPAPR_MACHINE(10, 1);*/ static void spapr_machine_10_0_class_options (MachineClass *mc) { +static GlobalProperty spapr_compat_10_0[] = { + {TYPE_POWERPC_CPU ,"rtas-stopped-state" ,"false" }, + }; +spapr_machine_10_1_class_options (mc);compat_props_add (mc->compat_props, hw_compat_10_0, hw_compat_10_0_len); +compat_props_add (mc->compat_props, spapr_compat_10_0, +G_N_ELEMENTS (spapr_compat_10_0)); }DEFINE_SPAPR_MACHINE (10 ,0 );target/ppc/cpu.h +1
-0@@ -1529,6 +1529,7 @@ struct ArchCPU {void *machine_data;int32_t node_id;/* NUMA node this CPU belongs to */ PPCHash64Options *hash64_opts; +bool rtas_stopped_state;/* Those resources are used only during code translation */ /* opcode handlers */ target/ppc/cpu_init.c +7
-0@@ -55,6 +55,11 @@/* #define PPC_DEBUG_SPR */ /* #define USE_APPLE_GDB */ +static const Property powerpc_cpu_properties[] = { +DEFINE_PROP_BOOL ("rtas-stopped-state" , PowerPCCPU, + rtas_stopped_state, true), +}; +static inline void vscr_init (CPUPPCState *env,uint32_t val) {/* Altivec always uses round-to-nearest */ @@ -7529,6 +7534,8 @@ static void ppc_cpu_class_init(ObjectClass *oc, const void *data) &pcc->parent_unrealize); pcc->pvr_match = ppc_pvr_match_default; +device_class_set_props (dc, powerpc_cpu_properties); +resettable_class_set_parent_phases (rc,NULL , ppc_cpu_reset_hold,NULL , &pcc->parent_phases);target/ppc/machine.c +62
-0@@ -257,6 +257,45 @@ static int cpu_post_load(void *opaque, int version_id)ppc_store_sdr1 (env, env->spr[SPR_SDR1 ]); } +if (!cpu->rtas_stopped_state) { +/* +* The source QEMU doesn't have fb802acdc8 and still uses halt + +* PM bits in LPCR to implement RTAS stopped state. The new (this) +* QEMU will have put the secondary vcpus in stopped state, +* waiting for the start-cpu RTAS call. That call will never come +* if the source cpus were already running. Try to infer the cpus +* state and set env->quiesced accordingly. +* +* env->quiesced = true ==> the cpu is waiting to start +* env->quiesced = false ==> the cpu is running (unless halted) +*/ + +/* +* Halted _could_ mean quiesced, but it could also be cede, +* confer_self, power management, etc. +*/ +if (CPU (cpu)->halted) { +PowerPCCPUClass *pcc =POWERPC_CPU_GET_CLASS (cpu); +/* +* Both the PSSCR_EC bit and LPCR PM bits set at cpu reset +* and rtas_stop and cleared at rtas_start, it's a good +* heuristic. +*/ +if ((env->spr[SPR_PSSCR ] &PSSCR_EC ) && + (env->spr[SPR_LPCR ] & pcc->lpcr_pm)) { + env->quiesced = true; + }else { + env->quiesced = false; + } + }else { +/* +* Old QEMU sets halted during rtas_stop_self. Not halted, +* therefore definitely not quiesced. +*/ + env->quiesced = false; + } + } +post_load_update_msr (env);if (tcg_enabled ()) {@@ -649,6 +688,28 @@ static const VMStateDescription vmstate_reservation = { } }; +static bool rtas_stopped_needed (void *opaque) +{ +PowerPCCPU *cpu = opaque; + +return cpu->rtas_stopped_state; +} + +static const VMStateDescription vmstate_rtas_stopped = { + .name ="cpu/rtas_stopped" , + .version_id =1 , + .minimum_version_id =1 , + .needed = rtas_stopped_needed, + .fields = (const VMStateField []) { +/* +* "RTAS stopped" state, independent of halted state. For QEMU +* < 10.0, this is taken from cpu->halted at cpu_post_load() +*/ +VMSTATE_BOOL (env.quiesced, PowerPCCPU), +VMSTATE_END_OF_LIST () + } +}; +#ifdef TARGET_PPC64 static bool bhrb_needed (void *opaque) {@@ -715,6 +776,7 @@ const VMStateDescription vmstate_ppc_cpu = { &vmstate_tlbmas, &vmstate_compat, &vmstate_reservation, + &vmstate_rtas_stopped, NULL } };
Re: [PATCH v3 1/2] target/ppc: Fix env->quiesced migration
In reply to Fabiano RosasQuoted 29 lines · Fabiano Rosas
The commit referenced (from QEMU 10.0) has changed the way the pseries machine marks a cpu as quiesced. Previously, the
cpu->haltedvalue from QEMU common cpu code was (incorrectly) used. With the fix, theenv->quiescedvariable starts being used, which improves on the original situation, but also causes a side effect after migration:The
env->quiescedis set at reset and never migrated, which causes the destination QEMU to stop delivering interrupts and hang the machine.To fix the issue from this point on, start migrating the
env->quiescedvalue.For QEMU versions < 10.0, sending the new element on the stream would cause migration to be aborted, so add the appropriate compatibility property to omit the new subsection.
Independently of this patch, all migrations from QEMU versions < 10.0 would result in a hang since the older QEMU never migrates
env->quiesced. This is bad because it leaves machines already running on the old QEMU without a migration path into newer versions.As a workaround, use a few heuristics to infer the new value of
env->quiescedbased oncpu->halted, LPCR and PSSCR bits that are usually set/cleared along with quiesced.Note that this was tested with -cpu power9 and -machine ic-mode=xive due to another bug affecting migration of XICS guests. Tested both forward and backward migration and savevm/loadvm from 9.2 and 10.0.
Quoted 7 lines
Also tested loadvm of a savevm image that contains a mix of cpus both halted and not halted.
Reported-by Fabian Vogt<[email protected]>Resolves:https://gitlab.com/qemu-project/qemu/-/issues/3079
Fixes fb802acdc8b ("ppc/spapr: Fix RTAS stopped state")Signed-off-by Fabiano Rosas<[email protected]>
and Queued.
PATCH v3 2/2 tests/functional: Add a OS level migration test for pseries
In reply to Fabiano RosasThere's currently no OS level test for ppc64le. Add one such test by reusing the boot level tests that are already present.
The test boots the source machine, waits for it to reach a mid-boot message, migrates and checks that the destination has reached the final boot message (VFS error due to no disk).
--- tests/functional/ppc64/test_migration.py | 12 ++++++++ tests/functional/ppc64/test_pseries.py | 35 ++++++++++++++++++++++++ 2 files changed, 47 insertions(+)
tests/functional/ppc64/test_migration.py +12 -0
@@ -4,6 +4,7 @@ # # ppc migration test +from qemu_test.ports import Ports from migration import MigrationTest
@@ -21,6 +22,17 @@ def test_migration_with_exec(self): self.set_machine('mac99') self.migration_with_exec() + def do_migrate_ppc64_linux(self, source_vm, dest_vm): + with Ports() as ports: + port = ports.find_free_port() + if port is None: + self.skipTest('Failed to find a free port') + uri = 'tcp:localhost:%u' % port + + dest_vm.qmp('migrate-incoming', uri=uri) + source_vm.qmp('migrate', uri=uri) + self.assert_migration(source_vm, dest_vm) + if __name__ == '__main__': MigrationTest.main()
tests/functional/ppc64/test_pseries.py +35 -0
@@ -9,6 +9,7 @@ from qemu_test import QemuSystemTest, Asset from qemu_test import wait_for_console_pattern +from test_migration import PpcMigrationTest class PseriesMachine(QemuSystemTest):
@@ -87,5 +88,39 @@ def test_ppc64_linux_big_boot(self): wait_for_console_pattern(self, console_pattern, self.panic_message) wait_for_console_pattern(self, self.good_message, self.panic_message) + def test_ppc64_linux_migration(self): + self.set_machine('pseries') + + kernel_path = self.ASSET_KERNEL.fetch() + kernel_command_line = self.KERNEL_COMMON_COMMAND_LINE + + dest_vm = self.get_vm(name="dest-qemu") + dest_vm.add_args('-incoming', 'defer') + dest_vm.add_args('-smp', '4') + dest_vm.add_args('-nodefaults') + dest_vm.add_args('-kernel', kernel_path, + '-append', kernel_command_line) + dest_vm.set_console() + dest_vm.launch() + + source_vm = self.get_vm(name="source-qemu") + source_vm.add_args('-smp', '4') + source_vm.add_args('-nodefaults') + source_vm.add_args('-kernel', kernel_path, + '-append', kernel_command_line) + source_vm.set_console() + source_vm.launch() + + # ensure the boot has reached Linux + console_pattern = 'smp: Brought up 1 node, 4 CPUs' + wait_for_console_pattern(self, console_pattern, self.panic_message, + vm=source_vm) + + PpcMigrationTest().do_migrate_ppc64_linux(source_vm, dest_vm); + + # ensure the boot proceeds after migration + wait_for_console_pattern(self, self.good_message, self.panic_message, + vm=dest_vm) + if __name__ == '__main__': QemuSystemTest.main()
Signature · 1 line
2.51.0
Re: [PATCH v3 2/2] tests/functional: Add a OS level migration test for pseries
Quoted 27 lines · Fabiano Rosas
There's currently no OS level test for ppc64le. Add one such test by reusing the boot level tests that are already present.
The test boots the source machine, waits for it to reach a mid-boot message, migrates and checks that the destination has reached the final boot message (VFS error due to no disk).
Reviewed-by Thomas Huth <[email protected]>Signed-off-by Fabiano Rosas <[email protected]>--- tests/functional/ppc64/test_migration.py | 12 ++++++++ tests/functional/ppc64/test_pseries.py | 35 ++++++++++++++++++++++++ 2 files changed, 47 insertions(+)tests/functional/ppc64/test_migration.py +1
-0@@ -4,6 +4,7 @@ # # ppc migration test +from qemu_test.ports import Ports from migration import MigrationTest
tests/functional/ppc64/test_migration.py def test_migration_with_exec(self): · line 30
Quoted 9 lines
tests/functional/ppc64/test_migration.py +6
-0@@ -21,6 +22,17 @@ def test_migration_with_exec(self): self.set_machine('mac99') self.migration_with_exec() + def do_migrate_ppc64_linux(self, source_vm, dest_vm): + with Ports() as ports: + port = ports.find_free_port() + if port is None: + self.skipTest('Failed to find a free port') + uri = 'tcp:localhost:%u' % port
* When port is None, shouldn't it return after the skipTest() call? With port = None, uri will become -> 'tcp:localhost:None' OR maybe port should have default value?
Quoted 50 lines
tests/functional/ppc64/test_migration.py +4
-0+ dest_vm.qmp('migrate-incoming', uri=uri) + source_vm.qmp('migrate', uri=uri) + self.assert_migration(source_vm, dest_vm) + if __name__ == '__main__': MigrationTest.main()tests/functional/ppc64/test_pseries.py +28
-0@@ -9,6 +9,7 @@ from qemu_test import QemuSystemTest, Asset from qemu_test import wait_for_console_pattern +from test_migration import PpcMigrationTest class PseriesMachine(QemuSystemTest):@@ -87,5 +88,39 @@ def test_ppc64_linux_big_boot(self): wait_for_console_pattern(self, console_pattern, self.panic_message) wait_for_console_pattern(self, self.good_message, self.panic_message) + def test_ppc64_linux_migration(self): + self.set_machine('pseries') + + kernel_path = self.ASSET_KERNEL.fetch() + kernel_command_line = self.KERNEL_COMMON_COMMAND_LINE + + dest_vm = self.get_vm(name="dest-qemu") + dest_vm.add_args('-incoming', 'defer') + dest_vm.add_args('-smp', '4') + dest_vm.add_args('-nodefaults') + dest_vm.add_args('-kernel', kernel_path, + '-append', kernel_command_line) + dest_vm.set_console() + dest_vm.launch() + + source_vm = self.get_vm(name="source-qemu") + source_vm.add_args('-smp', '4') + source_vm.add_args('-nodefaults') + source_vm.add_args('-kernel', kernel_path, + '-append', kernel_command_line) + source_vm.set_console() + source_vm.launch() + + # ensure the boot has reached Linux + console_pattern = 'smp: Brought up 1 node, 4 CPUs' + wait_for_console_pattern(self, console_pattern, self.panic_message, + vm=source_vm)
tests/functional/ppc64/test_pseries.py def test_ppc64_linux_big_boot(self):
Quoted 10 lines
tests/functional/ppc64/test_pseries.py +7
-1+ + PpcMigrationTest().do_migrate_ppc64_linux(source_vm, dest_vm); + + # ensure the boot proceeds after migration + wait_for_console_pattern(self, self.good_message, self.panic_message, + vm=dest_vm) + if __name__ == '__main__': QemuSystemTest.main() --
* Looks okay.
Thank you.
---
- Prasad
Re: [PATCH v3 2/2] tests/functional: Add a OS level migration test for pseries
Quoted 31 lines · Prasad Pandit
Quoted 30 lines · Fabiano Rosas
There's currently no OS level test for ppc64le. Add one such test by reusing the boot level tests that are already present.
The test boots the source machine, waits for it to reach a mid-boot message, migrates and checks that the destination has reached the final boot message (VFS error due to no disk).
Reviewed-by Thomas Huth <[email protected]>Signed-off-by Fabiano Rosas <[email protected]>--- tests/functional/ppc64/test_migration.py | 12 ++++++++ tests/functional/ppc64/test_pseries.py | 35 ++++++++++++++++++++++++ 2 files changed, 47 insertions(+)tests/functional/ppc64/test_migration.py +1
-0@@ -4,6 +4,7 @@ # # ppc migration test +from qemu_test.ports import Ports from migration import MigrationTest@@ -21,6 +22,17 @@ def test_migration_with_exec(self): self.set_machine('mac99') self.migration_with_exec()
Quoted 10 lines
tests/functional/ppc64/test_migration.py def test_migration_with_exec(self): · line 30
Quoted 6 lines
tests/functional/ppc64/test_migration.py +6
-0+ def do_migrate_ppc64_linux(self, source_vm, dest_vm): + with Ports() as ports: + port = ports.find_free_port() + if port is None: + self.skipTest('Failed to find a free port') + uri = 'tcp:localhost:%u' % port* When port is None, shouldn't it return after the
skipTest()call? With port = None, uri will become -> 'tcp:localhost:None' OR maybe port should have default value?
skipTest() aborts the test immediately, the remaining code after this statement is not executed anymore. (I think it internally raises an exception that is caught by the unittest code to mark the test as skipped, but you might better ask a Python wizard if you want to know the gory details. See https://docs.python.org/3/library/unittest.html#unittest-skipping for example)
HTH,
Thomas
Re: [PATCH v3 2/2] tests/functional: Add a OS level migration test for pseries
Quoted 6 lines · Thomas Huth
skipTest()aborts the test immediately, the remaining code after this statement is not executed anymore. (I think it internally raises an exception that is caught by the unittest code to mark the test as skipped, but you might better ask a Python wizard if you want to know the gory details. See
https://docs.python.org/3/library/unittest.html#unittest-skipping for example)
* I see, got it. Thank you for the link.
---
- Prasad