diff mbox series

[kvm-unit-tests,v3,14/14] arm/arm64: ITS: pending table migration test

Message ID 20200128103459.19413-15-eric.auger@redhat.com (mailing list archive)
State New, archived
Headers show
Series arm/arm64: Add ITS tests | expand

Commit Message

Eric Auger Jan. 28, 2020, 10:34 a.m. UTC
Add two new migration tests. One testing the migration of
a topology where collection were unmapped. The second test
checks the migration of the pending table.

Signed-off-by: Eric Auger <eric.auger@redhat.com>

---

v2 -> v3:
- tests belong to both its and migration groups
---
 arm/gic.c         | 150 ++++++++++++++++++++++++++++++++++++++++++++++
 arm/unittests.cfg |  16 +++++
 2 files changed, 166 insertions(+)

Comments

Andrew Jones Feb. 7, 2020, 2:06 p.m. UTC | #1
On Tue, Jan 28, 2020 at 11:34:59AM +0100, Eric Auger wrote:
> Add two new migration tests. One testing the migration of
> a topology where collection were unmapped. The second test
> checks the migration of the pending table.
> 
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> 
> ---
> 
> v2 -> v3:
> - tests belong to both its and migration groups
> ---
>  arm/gic.c         | 150 ++++++++++++++++++++++++++++++++++++++++++++++
>  arm/unittests.cfg |  16 +++++
>  2 files changed, 166 insertions(+)
> 
> diff --git a/arm/gic.c b/arm/gic.c
> index fa8626a..ec3dd3a 100644
> --- a/arm/gic.c
> +++ b/arm/gic.c
> @@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)
>  	smp_rmb(); /* pairs with wmb in lpi_stats_expect */
>  	lpi_stats.observed.cpu_id = smp_processor_id();
>  	lpi_stats.observed.lpi_id = irqnr;
> +	acked[lpi_stats.observed.cpu_id]++;
>  	smp_wmb(); /* pairs with rmb in check_lpi_stats */
>  }
>  
> @@ -239,6 +240,18 @@ static void secondary_lpi_test(void)
>  	while (1)
>  		wfi();
>  }
> +
> +static void check_lpi_hits(int *expected)
> +{
> +	int i;
> +
> +	for (i = 0; i < nr_cpus; i++) {
> +		if (acked[i] != expected[i])
> +			report(false, "expected %d LPIs on PE #%d, %d observed",
> +			       expected[i], i, acked[i]);

report_info
pass = false

> +		}
> +	report(true, "check LPI on all vcpus");

report(pass, ...);

> +}
>  #endif
>  
>  static void gicv2_ipi_send_self(void)
> @@ -594,6 +607,8 @@ static void gic_test_mmio(void)
>  static void test_its_introspection(void) {}
>  static void test_its_trigger(void) {}
>  static void test_its_migration(void) {}
> +static void test_migrate_unmapped_collection(void) {}
> +static void test_its_pending_migration(void) {}

I'm not sure what's worse. This pile of stubs or one #ifdef in main()
wrapping all the calls.

>  
>  #else /* __arch64__ */
>  
> @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)
>  	return false;
>  }
>  
> +static void set_lpi(struct its_device *dev, u32 eventid, u32 physid,
> +		    struct its_collection *col)
> +{
> +	if (!dev || !col)

I don't think col can be null, and this doesn't look like the right place
to check if dev is null.  If we're bothiner to call set_lpi, then I
think we should already expect dev to be good to go.

> +		report_abort("wrong device or collection");
> +
> +	its_send_mapti(dev, physid, eventid, col);
> +
> +	gicv3_lpi_set_config(physid, LPI_PROP_DEFAULT);
> +	its_send_invall(col);
> +}
> +
>  /*
>   * Setup the configuration for those mappings:
>   * dev_id=2 event=20 -> vcpu 3, intid=8195
> @@ -806,6 +833,121 @@ static void test_its_migration(void)
>  	its_send_int(dev7, 255);
>  	check_lpi_stats();
>  }
> +
> +static void test_migrate_unmapped_collection(void)
> +{
> +	struct its_collection *col;
> +	struct its_device *dev2, *dev7;
> +	u8 config;
> +
> +	if (its_setup1())
> +		return;
> +
> +	col = its_create_collection(nr_cpus - 1, nr_cpus - 1);
> +	dev2 = its_get_device(2);
> +	dev7 = its_get_device(7);
> +
> +	/* MAPTI with the collection unmapped */
> +	set_lpi(dev2, 0, 8192, col);
> +
> +	puts("Now migrate the VM, then press a key to continue...\n");
> +	(void)getchar();
> +	report(true, "Migration complete");

report_info

> +
> +	/* on the destination, map the collection */
> +	its_send_mapc(col, true);
> +
> +	lpi_stats_expect(2, 8196);
> +	its_send_int(dev7, 255);
> +	check_lpi_stats();
> +
> +	config = gicv3_lpi_get_config(8192);
> +	report(config == LPI_PROP_DEFAULT,
> +	       "Config of LPI 8192 was properly migrated");
> +
> +	lpi_stats_expect(nr_cpus - 1, 8192);
> +	its_send_int(dev2, 0);
> +	check_lpi_stats();
> +
> +	/* unmap the collection */
> +	its_send_mapc(col, false);
> +
> +	lpi_stats_expect(-1, -1);
> +	its_send_int(dev2, 0);
> +	check_lpi_stats();
> +
> +	/* remap event 0 onto lpiid 8193 */
> +	set_lpi(dev2, 0, 8193, col);
> +	lpi_stats_expect(-1, -1);
> +	its_send_int(dev2, 0);
> +	check_lpi_stats();
> +
> +	/* remap the collection */
> +	its_send_mapc(col, true);
> +	lpi_stats_expect(nr_cpus - 1, 8193);
> +}
> +
> +static void test_its_pending_migration(void)
> +{
> +	struct its_device *dev;
> +	struct its_collection *collection[2];
> +	int expected[NR_CPUS];

expected = malloc(nr_cpus * sizeof(int));

I know there are other places using NR_CPUS right now that don't have to,
but we shouldn't add more. Eventually I'll change the other places too.

> +	u64 pendbaser;
> +	void *ptr;
> +	int i;
> +
> +	if (its_prerequisites(4))
> +		return;
> +
> +	dev = its_create_device(2 /* dev id */, 8 /* nb_ites */);
> +	its_send_mapd(dev, true);
> +
> +	collection[0] = its_create_collection(nr_cpus - 1, nr_cpus - 1);
> +	collection[1] = its_create_collection(nr_cpus - 2, nr_cpus - 2);
> +	its_send_mapc(collection[0], true);
> +	its_send_mapc(collection[1], true);
> +
> +	/* disable lpi at redist level */
> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, false);
> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, false);
> +
> +	/* even lpis are assigned to even cpu */
> +	for (i = 0; i < 256; i++) {
> +		struct its_collection *col = i % 2 ? collection[0] :
> +						     collection[1];
> +		int vcpu = col->target_address >> 16;

I'm lost with the even/odd (nr_cpus - 1)/(nr_cpus - 2) stuff, and won't
it swap if nr_cpus is odd vs. even?

Shouldn't we just have something like

  pe1 = nr_cpus - 1;
  pe2 = nr_cpus - 2;
  col1 = its_create_collection(pe1, pe1);
  col2 = its_create_collection(pe2, pe2);

without mentioning even and odd?

> +
> +		its_send_mapti(dev, 8192 + i, i, col);
> +		gicv3_lpi_set_config(8192 + i, LPI_PROP_DEFAULT);
> +		gicv3_lpi_set_pending_table_bit(vcpu, 8192 + i, true);
> +	}
> +	its_send_invall(collection[0]);
> +	its_send_invall(collection[1]);
> +
> +	/* Set the PTZ bit on each pendbaser */
> +
> +	expected[nr_cpus - 1] = 128;
> +	expected[nr_cpus - 2] = 128;
> +
> +	ptr = gicv3_data.redist_base[nr_cpus - 1] + GICR_PENDBASER;
> +	pendbaser = readq(ptr);
> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
> +
> +	ptr = gicv3_data.redist_base[nr_cpus - 2] + GICR_PENDBASER;
> +	pendbaser = readq(ptr);
> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
> +
> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, true);
> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, true);
> +
> +	puts("Now migrate the VM, then press a key to continue...\n");
> +	(void)getchar();
> +	report(true, "Migration complete");

report_info

> +
> +	mdelay(1000);

This delay needs a comment explaining why it's here.

> +
> +	check_lpi_hits(expected);
> +}
>  #endif
>  
>  int main(int argc, char **argv)
> @@ -847,6 +989,14 @@ int main(int argc, char **argv)
>  		report_prefix_push(argv[1]);
>  		test_its_migration();
>  		report_prefix_pop();
> +	} else if (!strcmp(argv[1], "its-pending-migration")) {
> +		report_prefix_push(argv[1]);
> +		test_its_pending_migration();
> +		report_prefix_pop();
> +	} else if (!strcmp(argv[1], "its-migrate-unmapped-collection")) {
> +		report_prefix_push(argv[1]);
> +		test_migrate_unmapped_collection();
> +		report_prefix_pop();
>  	} else if (strcmp(argv[1], "its-introspection") == 0) {
>  		report_prefix_push(argv[1]);
>  		test_its_introspection();
> diff --git a/arm/unittests.cfg b/arm/unittests.cfg
> index 8b8ec79..d917157 100644
> --- a/arm/unittests.cfg
> +++ b/arm/unittests.cfg
> @@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration'
>  groups = its migration
>  arch = arm64
>  
> +[its-pending-migration]
> +file = gic.flat
> +smp = $MAX_SMP
> +accel = kvm
> +extra_params = -machine gic-version=3 -append 'its-pending-migration'
> +groups = its migration
> +arch = arm64
> +
> +[its-migrate-unmapped-collection]
> +file = gic.flat
> +smp = $MAX_SMP
> +accel = kvm
> +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection'
> +groups = its migration
> +arch = arm64
> +
>  # Test PSCI emulation
>  [psci]
>  file = psci.flat
> -- 
> 2.20.1
> 
>

Thanks,
drew
Eric Auger March 6, 2020, 1:21 p.m. UTC | #2
Hi Drew,

On 2/7/20 3:06 PM, Andrew Jones wrote:
> On Tue, Jan 28, 2020 at 11:34:59AM +0100, Eric Auger wrote:
>> Add two new migration tests. One testing the migration of
>> a topology where collection were unmapped. The second test
>> checks the migration of the pending table.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>
>> ---
>>
>> v2 -> v3:
>> - tests belong to both its and migration groups
>> ---
>>  arm/gic.c         | 150 ++++++++++++++++++++++++++++++++++++++++++++++
>>  arm/unittests.cfg |  16 +++++
>>  2 files changed, 166 insertions(+)
>>
>> diff --git a/arm/gic.c b/arm/gic.c
>> index fa8626a..ec3dd3a 100644
>> --- a/arm/gic.c
>> +++ b/arm/gic.c
>> @@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)
>>  	smp_rmb(); /* pairs with wmb in lpi_stats_expect */
>>  	lpi_stats.observed.cpu_id = smp_processor_id();
>>  	lpi_stats.observed.lpi_id = irqnr;
>> +	acked[lpi_stats.observed.cpu_id]++;
>>  	smp_wmb(); /* pairs with rmb in check_lpi_stats */
>>  }
>>  
>> @@ -239,6 +240,18 @@ static void secondary_lpi_test(void)
>>  	while (1)
>>  		wfi();
>>  }
>> +
>> +static void check_lpi_hits(int *expected)
>> +{
>> +	int i;
>> +
>> +	for (i = 0; i < nr_cpus; i++) {
>> +		if (acked[i] != expected[i])
>> +			report(false, "expected %d LPIs on PE #%d, %d observed",
>> +			       expected[i], i, acked[i]);
> 
> report_info
> pass = false
> 
>> +		}
>> +	report(true, "check LPI on all vcpus");
> 
> report(pass, ...);
I still don't get the issue.
> 
>> +}
>>  #endif
>>  
>>  static void gicv2_ipi_send_self(void)
>> @@ -594,6 +607,8 @@ static void gic_test_mmio(void)
>>  static void test_its_introspection(void) {}
>>  static void test_its_trigger(void) {}
>>  static void test_its_migration(void) {}
>> +static void test_migrate_unmapped_collection(void) {}
>> +static void test_its_pending_migration(void) {}
> 
> I'm not sure what's worse. This pile of stubs or one #ifdef in main()
> wrapping all the calls.
Those stubs now are in the arm header.
> 
>>  
>>  #else /* __arch64__ */
>>  
>> @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)
>>  	return false;
>>  }
>>  
>> +static void set_lpi(struct its_device *dev, u32 eventid, u32 physid,
>> +		    struct its_collection *col)
>> +{
>> +	if (!dev || !col)
> 
> I don't think col can be null, and this doesn't look like the right place
> to check if dev is null.  If we're bothiner to call set_lpi, then I
> think we should already expect dev to be good to go.
put an assert() instead
> 
>> +		report_abort("wrong device or collection");
>> +
>> +	its_send_mapti(dev, physid, eventid, col);
>> +
>> +	gicv3_lpi_set_config(physid, LPI_PROP_DEFAULT);
>> +	its_send_invall(col);
>> +}
>> +
>>  /*
>>   * Setup the configuration for those mappings:
>>   * dev_id=2 event=20 -> vcpu 3, intid=8195
>> @@ -806,6 +833,121 @@ static void test_its_migration(void)
>>  	its_send_int(dev7, 255);
>>  	check_lpi_stats();
>>  }
>> +
>> +static void test_migrate_unmapped_collection(void)
>> +{
>> +	struct its_collection *col;
>> +	struct its_device *dev2, *dev7;
>> +	u8 config;
>> +
>> +	if (its_setup1())
>> +		return;
>> +
>> +	col = its_create_collection(nr_cpus - 1, nr_cpus - 1);
>> +	dev2 = its_get_device(2);
>> +	dev7 = its_get_device(7);
>> +
>> +	/* MAPTI with the collection unmapped */
>> +	set_lpi(dev2, 0, 8192, col);
>> +
>> +	puts("Now migrate the VM, then press a key to continue...\n");
>> +	(void)getchar();
>> +	report(true, "Migration complete");
> 
> report_info
yep
> 
>> +
>> +	/* on the destination, map the collection */
>> +	its_send_mapc(col, true);
>> +
>> +	lpi_stats_expect(2, 8196);
>> +	its_send_int(dev7, 255);
>> +	check_lpi_stats();
>> +
>> +	config = gicv3_lpi_get_config(8192);
>> +	report(config == LPI_PROP_DEFAULT,
>> +	       "Config of LPI 8192 was properly migrated");
>> +
>> +	lpi_stats_expect(nr_cpus - 1, 8192);
>> +	its_send_int(dev2, 0);
>> +	check_lpi_stats();
>> +
>> +	/* unmap the collection */
>> +	its_send_mapc(col, false);
>> +
>> +	lpi_stats_expect(-1, -1);
>> +	its_send_int(dev2, 0);
>> +	check_lpi_stats();
>> +
>> +	/* remap event 0 onto lpiid 8193 */
>> +	set_lpi(dev2, 0, 8193, col);
>> +	lpi_stats_expect(-1, -1);
>> +	its_send_int(dev2, 0);
>> +	check_lpi_stats();
>> +
>> +	/* remap the collection */
>> +	its_send_mapc(col, true);
>> +	lpi_stats_expect(nr_cpus - 1, 8193);
>> +}
>> +
>> +static void test_its_pending_migration(void)
>> +{
>> +	struct its_device *dev;
>> +	struct its_collection *collection[2];
>> +	int expected[NR_CPUS];
> 
> expected = malloc(nr_cpus * sizeof(int));
> 
> I know there are other places using NR_CPUS right now that don't have to,
> but we shouldn't add more. Eventually I'll change the other places too.
OK
> 
>> +	u64 pendbaser;
>> +	void *ptr;
>> +	int i;
>> +
>> +	if (its_prerequisites(4))
>> +		return;
>> +
>> +	dev = its_create_device(2 /* dev id */, 8 /* nb_ites */);
>> +	its_send_mapd(dev, true);
>> +
>> +	collection[0] = its_create_collection(nr_cpus - 1, nr_cpus - 1);
>> +	collection[1] = its_create_collection(nr_cpus - 2, nr_cpus - 2);
>> +	its_send_mapc(collection[0], true);
>> +	its_send_mapc(collection[1], true);
>> +
>> +	/* disable lpi at redist level */
>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, false);
>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, false);
>> +
>> +	/* even lpis are assigned to even cpu */
>> +	for (i = 0; i < 256; i++) {
>> +		struct its_collection *col = i % 2 ? collection[0] :
>> +						     collection[1];
>> +		int vcpu = col->target_address >> 16;
> 
> I'm lost with the even/odd (nr_cpus - 1)/(nr_cpus - 2) stuff, and won't
> it swap if nr_cpus is odd vs. even?
> 
> Shouldn't we just have something like
> 
>   pe1 = nr_cpus - 1;
>   pe2 = nr_cpus - 2;
>   col1 = its_create_collection(pe1, pe1);
>   col2 = its_create_collection(pe2, pe2);
> 
> without mentioning even and odd?
OK
> 
>> +
>> +		its_send_mapti(dev, 8192 + i, i, col);
>> +		gicv3_lpi_set_config(8192 + i, LPI_PROP_DEFAULT);
>> +		gicv3_lpi_set_pending_table_bit(vcpu, 8192 + i, true);
>> +	}
>> +	its_send_invall(collection[0]);
>> +	its_send_invall(collection[1]);
>> +
>> +	/* Set the PTZ bit on each pendbaser */
>> +
>> +	expected[nr_cpus - 1] = 128;
>> +	expected[nr_cpus - 2] = 128;
>> +
>> +	ptr = gicv3_data.redist_base[nr_cpus - 1] + GICR_PENDBASER;
>> +	pendbaser = readq(ptr);
>> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
>> +
>> +	ptr = gicv3_data.redist_base[nr_cpus - 2] + GICR_PENDBASER;
>> +	pendbaser = readq(ptr);
>> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
>> +
>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, true);
>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, true);
>> +
>> +	puts("Now migrate the VM, then press a key to continue...\n");
>> +	(void)getchar();
>> +	report(true, "Migration complete");
> 
> report_info
OK
> 
>> +
>> +	mdelay(1000);
> 
> This delay needs a comment explaining why it's here.
OK

Thanks

Eric
> 
>> +
>> +	check_lpi_hits(expected);
>> +}
>>  #endif
>>  
>>  int main(int argc, char **argv)
>> @@ -847,6 +989,14 @@ int main(int argc, char **argv)
>>  		report_prefix_push(argv[1]);
>>  		test_its_migration();
>>  		report_prefix_pop();
>> +	} else if (!strcmp(argv[1], "its-pending-migration")) {
>> +		report_prefix_push(argv[1]);
>> +		test_its_pending_migration();
>> +		report_prefix_pop();
>> +	} else if (!strcmp(argv[1], "its-migrate-unmapped-collection")) {
>> +		report_prefix_push(argv[1]);
>> +		test_migrate_unmapped_collection();
>> +		report_prefix_pop();
>>  	} else if (strcmp(argv[1], "its-introspection") == 0) {
>>  		report_prefix_push(argv[1]);
>>  		test_its_introspection();
>> diff --git a/arm/unittests.cfg b/arm/unittests.cfg
>> index 8b8ec79..d917157 100644
>> --- a/arm/unittests.cfg
>> +++ b/arm/unittests.cfg
>> @@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration'
>>  groups = its migration
>>  arch = arm64
>>  
>> +[its-pending-migration]
>> +file = gic.flat
>> +smp = $MAX_SMP
>> +accel = kvm
>> +extra_params = -machine gic-version=3 -append 'its-pending-migration'
>> +groups = its migration
>> +arch = arm64
>> +
>> +[its-migrate-unmapped-collection]
>> +file = gic.flat
>> +smp = $MAX_SMP
>> +accel = kvm
>> +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection'
>> +groups = its migration
>> +arch = arm64
>> +
>>  # Test PSCI emulation
>>  [psci]
>>  file = psci.flat
>> -- 
>> 2.20.1
>>
>>
> 
> Thanks,
> drew 
>
Andrew Jones March 6, 2020, 1:36 p.m. UTC | #3
On Fri, Mar 06, 2020 at 02:21:37PM +0100, Auger Eric wrote:
> Hi Drew,
> 
> On 2/7/20 3:06 PM, Andrew Jones wrote:
> > On Tue, Jan 28, 2020 at 11:34:59AM +0100, Eric Auger wrote:
> >> Add two new migration tests. One testing the migration of
> >> a topology where collection were unmapped. The second test
> >> checks the migration of the pending table.
> >>
> >> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> >>
> >> ---
> >>
> >> v2 -> v3:
> >> - tests belong to both its and migration groups
> >> ---
> >>  arm/gic.c         | 150 ++++++++++++++++++++++++++++++++++++++++++++++
> >>  arm/unittests.cfg |  16 +++++
> >>  2 files changed, 166 insertions(+)
> >>
> >> diff --git a/arm/gic.c b/arm/gic.c
> >> index fa8626a..ec3dd3a 100644
> >> --- a/arm/gic.c
> >> +++ b/arm/gic.c
> >> @@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)
> >>  	smp_rmb(); /* pairs with wmb in lpi_stats_expect */
> >>  	lpi_stats.observed.cpu_id = smp_processor_id();
> >>  	lpi_stats.observed.lpi_id = irqnr;
> >> +	acked[lpi_stats.observed.cpu_id]++;
> >>  	smp_wmb(); /* pairs with rmb in check_lpi_stats */
> >>  }
> >>  
> >> @@ -239,6 +240,18 @@ static void secondary_lpi_test(void)
> >>  	while (1)
> >>  		wfi();
> >>  }
> >> +
> >> +static void check_lpi_hits(int *expected)
> >> +{
> >> +	int i;
> >> +
> >> +	for (i = 0; i < nr_cpus; i++) {
> >> +		if (acked[i] != expected[i])
> >> +			report(false, "expected %d LPIs on PE #%d, %d observed",
> >> +			       expected[i], i, acked[i]);
> > 
> > report_info
> > pass = false
> > 
> >> +		}
> >> +	report(true, "check LPI on all vcpus");
> > 
> > report(pass, ...);
> I still don't get the issue.

Your messages aren't consistent and have dynamic info. You need

report_info(...); // dynamic info
if (failure_condition) {
   pass_boolean = false;
}
report(pass_boolean, single_message_for_both_pass_and_failure);

> > 
> >> +}
> >>  #endif
> >>  
> >>  static void gicv2_ipi_send_self(void)
> >> @@ -594,6 +607,8 @@ static void gic_test_mmio(void)
> >>  static void test_its_introspection(void) {}
> >>  static void test_its_trigger(void) {}
> >>  static void test_its_migration(void) {}
> >> +static void test_migrate_unmapped_collection(void) {}
> >> +static void test_its_pending_migration(void) {}
> > 
> > I'm not sure what's worse. This pile of stubs or one #ifdef in main()
> > wrapping all the calls.
> Those stubs now are in the arm header.
> > 
> >>  
> >>  #else /* __arch64__ */
> >>  
> >> @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)
> >>  	return false;
> >>  }
> >>  
> >> +static void set_lpi(struct its_device *dev, u32 eventid, u32 physid,
> >> +		    struct its_collection *col)
> >> +{
> >> +	if (!dev || !col)
> > 
> > I don't think col can be null, and this doesn't look like the right place
> > to check if dev is null.  If we're bothiner to call set_lpi, then I
> > think we should already expect dev to be good to go.
> put an assert() instead
> > 
> >> +		report_abort("wrong device or collection");
> >> +
> >> +	its_send_mapti(dev, physid, eventid, col);
> >> +
> >> +	gicv3_lpi_set_config(physid, LPI_PROP_DEFAULT);
> >> +	its_send_invall(col);
> >> +}
> >> +
> >>  /*
> >>   * Setup the configuration for those mappings:
> >>   * dev_id=2 event=20 -> vcpu 3, intid=8195
> >> @@ -806,6 +833,121 @@ static void test_its_migration(void)
> >>  	its_send_int(dev7, 255);
> >>  	check_lpi_stats();
> >>  }
> >> +
> >> +static void test_migrate_unmapped_collection(void)
> >> +{
> >> +	struct its_collection *col;
> >> +	struct its_device *dev2, *dev7;
> >> +	u8 config;
> >> +
> >> +	if (its_setup1())
> >> +		return;
> >> +
> >> +	col = its_create_collection(nr_cpus - 1, nr_cpus - 1);
> >> +	dev2 = its_get_device(2);
> >> +	dev7 = its_get_device(7);
> >> +
> >> +	/* MAPTI with the collection unmapped */
> >> +	set_lpi(dev2, 0, 8192, col);
> >> +
> >> +	puts("Now migrate the VM, then press a key to continue...\n");
> >> +	(void)getchar();
> >> +	report(true, "Migration complete");
> > 
> > report_info
> yep
> > 
> >> +
> >> +	/* on the destination, map the collection */
> >> +	its_send_mapc(col, true);
> >> +
> >> +	lpi_stats_expect(2, 8196);
> >> +	its_send_int(dev7, 255);
> >> +	check_lpi_stats();
> >> +
> >> +	config = gicv3_lpi_get_config(8192);
> >> +	report(config == LPI_PROP_DEFAULT,
> >> +	       "Config of LPI 8192 was properly migrated");
> >> +
> >> +	lpi_stats_expect(nr_cpus - 1, 8192);
> >> +	its_send_int(dev2, 0);
> >> +	check_lpi_stats();
> >> +
> >> +	/* unmap the collection */
> >> +	its_send_mapc(col, false);
> >> +
> >> +	lpi_stats_expect(-1, -1);
> >> +	its_send_int(dev2, 0);
> >> +	check_lpi_stats();
> >> +
> >> +	/* remap event 0 onto lpiid 8193 */
> >> +	set_lpi(dev2, 0, 8193, col);
> >> +	lpi_stats_expect(-1, -1);
> >> +	its_send_int(dev2, 0);
> >> +	check_lpi_stats();
> >> +
> >> +	/* remap the collection */
> >> +	its_send_mapc(col, true);
> >> +	lpi_stats_expect(nr_cpus - 1, 8193);
> >> +}
> >> +
> >> +static void test_its_pending_migration(void)
> >> +{
> >> +	struct its_device *dev;
> >> +	struct its_collection *collection[2];
> >> +	int expected[NR_CPUS];
> > 
> > expected = malloc(nr_cpus * sizeof(int));
> > 
> > I know there are other places using NR_CPUS right now that don't have to,
> > but we shouldn't add more. Eventually I'll change the other places too.
> OK
> > 
> >> +	u64 pendbaser;
> >> +	void *ptr;
> >> +	int i;
> >> +
> >> +	if (its_prerequisites(4))
> >> +		return;
> >> +
> >> +	dev = its_create_device(2 /* dev id */, 8 /* nb_ites */);
> >> +	its_send_mapd(dev, true);
> >> +
> >> +	collection[0] = its_create_collection(nr_cpus - 1, nr_cpus - 1);
> >> +	collection[1] = its_create_collection(nr_cpus - 2, nr_cpus - 2);
> >> +	its_send_mapc(collection[0], true);
> >> +	its_send_mapc(collection[1], true);
> >> +
> >> +	/* disable lpi at redist level */
> >> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, false);
> >> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, false);
> >> +
> >> +	/* even lpis are assigned to even cpu */
> >> +	for (i = 0; i < 256; i++) {
> >> +		struct its_collection *col = i % 2 ? collection[0] :
> >> +						     collection[1];
> >> +		int vcpu = col->target_address >> 16;
> > 
> > I'm lost with the even/odd (nr_cpus - 1)/(nr_cpus - 2) stuff, and won't
> > it swap if nr_cpus is odd vs. even?
> > 
> > Shouldn't we just have something like
> > 
> >   pe1 = nr_cpus - 1;
> >   pe2 = nr_cpus - 2;
> >   col1 = its_create_collection(pe1, pe1);
> >   col2 = its_create_collection(pe2, pe2);
> > 
> > without mentioning even and odd?
> OK
> > 
> >> +
> >> +		its_send_mapti(dev, 8192 + i, i, col);
> >> +		gicv3_lpi_set_config(8192 + i, LPI_PROP_DEFAULT);
> >> +		gicv3_lpi_set_pending_table_bit(vcpu, 8192 + i, true);
> >> +	}
> >> +	its_send_invall(collection[0]);
> >> +	its_send_invall(collection[1]);
> >> +
> >> +	/* Set the PTZ bit on each pendbaser */
> >> +
> >> +	expected[nr_cpus - 1] = 128;
> >> +	expected[nr_cpus - 2] = 128;
> >> +
> >> +	ptr = gicv3_data.redist_base[nr_cpus - 1] + GICR_PENDBASER;
> >> +	pendbaser = readq(ptr);
> >> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
> >> +
> >> +	ptr = gicv3_data.redist_base[nr_cpus - 2] + GICR_PENDBASER;
> >> +	pendbaser = readq(ptr);
> >> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
> >> +
> >> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, true);
> >> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, true);
> >> +
> >> +	puts("Now migrate the VM, then press a key to continue...\n");
> >> +	(void)getchar();
> >> +	report(true, "Migration complete");
> > 
> > report_info
> OK
> > 
> >> +
> >> +	mdelay(1000);
> > 
> > This delay needs a comment explaining why it's here.
> OK
> 
> Thanks
> 
> Eric
> > 
> >> +
> >> +	check_lpi_hits(expected);
> >> +}
> >>  #endif
> >>  
> >>  int main(int argc, char **argv)
> >> @@ -847,6 +989,14 @@ int main(int argc, char **argv)
> >>  		report_prefix_push(argv[1]);
> >>  		test_its_migration();
> >>  		report_prefix_pop();
> >> +	} else if (!strcmp(argv[1], "its-pending-migration")) {
> >> +		report_prefix_push(argv[1]);
> >> +		test_its_pending_migration();
> >> +		report_prefix_pop();
> >> +	} else if (!strcmp(argv[1], "its-migrate-unmapped-collection")) {
> >> +		report_prefix_push(argv[1]);
> >> +		test_migrate_unmapped_collection();
> >> +		report_prefix_pop();
> >>  	} else if (strcmp(argv[1], "its-introspection") == 0) {
> >>  		report_prefix_push(argv[1]);
> >>  		test_its_introspection();
> >> diff --git a/arm/unittests.cfg b/arm/unittests.cfg
> >> index 8b8ec79..d917157 100644
> >> --- a/arm/unittests.cfg
> >> +++ b/arm/unittests.cfg
> >> @@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration'
> >>  groups = its migration
> >>  arch = arm64
> >>  
> >> +[its-pending-migration]
> >> +file = gic.flat
> >> +smp = $MAX_SMP
> >> +accel = kvm
> >> +extra_params = -machine gic-version=3 -append 'its-pending-migration'
> >> +groups = its migration
> >> +arch = arm64
> >> +
> >> +[its-migrate-unmapped-collection]
> >> +file = gic.flat
> >> +smp = $MAX_SMP
> >> +accel = kvm
> >> +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection'
> >> +groups = its migration
> >> +arch = arm64
> >> +
> >>  # Test PSCI emulation
> >>  [psci]
> >>  file = psci.flat
> >> -- 
> >> 2.20.1
> >>
> >>
> > 
> > Thanks,
> > drew 
> > 
> 
>
Eric Auger March 6, 2020, 1:41 p.m. UTC | #4
Hi

On 3/6/20 2:36 PM, Andrew Jones wrote:
> On Fri, Mar 06, 2020 at 02:21:37PM +0100, Auger Eric wrote:
>> Hi Drew,
>>
>> On 2/7/20 3:06 PM, Andrew Jones wrote:
>>> On Tue, Jan 28, 2020 at 11:34:59AM +0100, Eric Auger wrote:
>>>> Add two new migration tests. One testing the migration of
>>>> a topology where collection were unmapped. The second test
>>>> checks the migration of the pending table.
>>>>
>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>
>>>> ---
>>>>
>>>> v2 -> v3:
>>>> - tests belong to both its and migration groups
>>>> ---
>>>>  arm/gic.c         | 150 ++++++++++++++++++++++++++++++++++++++++++++++
>>>>  arm/unittests.cfg |  16 +++++
>>>>  2 files changed, 166 insertions(+)
>>>>
>>>> diff --git a/arm/gic.c b/arm/gic.c
>>>> index fa8626a..ec3dd3a 100644
>>>> --- a/arm/gic.c
>>>> +++ b/arm/gic.c
>>>> @@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)
>>>>  	smp_rmb(); /* pairs with wmb in lpi_stats_expect */
>>>>  	lpi_stats.observed.cpu_id = smp_processor_id();
>>>>  	lpi_stats.observed.lpi_id = irqnr;
>>>> +	acked[lpi_stats.observed.cpu_id]++;
>>>>  	smp_wmb(); /* pairs with rmb in check_lpi_stats */
>>>>  }
>>>>  
>>>> @@ -239,6 +240,18 @@ static void secondary_lpi_test(void)
>>>>  	while (1)
>>>>  		wfi();
>>>>  }
>>>> +
>>>> +static void check_lpi_hits(int *expected)
>>>> +{
>>>> +	int i;
>>>> +
>>>> +	for (i = 0; i < nr_cpus; i++) {
>>>> +		if (acked[i] != expected[i])
>>>> +			report(false, "expected %d LPIs on PE #%d, %d observed",
>>>> +			       expected[i], i, acked[i]);
>>>
>>> report_info
>>> pass = false
>>>
>>>> +		}
>>>> +	report(true, "check LPI on all vcpus");
>>>
>>> report(pass, ...);
>> I still don't get the issue.
> 
> Your messages aren't consistent and have dynamic info. You need
> 
> report_info(...); // dynamic info
> if (failure_condition) {
>    pass_boolean = false;
> }
> report(pass_boolean, single_message_for_both_pass_and_failure);
OK understood.

Thanks!

Eric
> 
>>>
>>>> +}
>>>>  #endif
>>>>  
>>>>  static void gicv2_ipi_send_self(void)
>>>> @@ -594,6 +607,8 @@ static void gic_test_mmio(void)
>>>>  static void test_its_introspection(void) {}
>>>>  static void test_its_trigger(void) {}
>>>>  static void test_its_migration(void) {}
>>>> +static void test_migrate_unmapped_collection(void) {}
>>>> +static void test_its_pending_migration(void) {}
>>>
>>> I'm not sure what's worse. This pile of stubs or one #ifdef in main()
>>> wrapping all the calls.
>> Those stubs now are in the arm header.
>>>
>>>>  
>>>>  #else /* __arch64__ */
>>>>  
>>>> @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)
>>>>  	return false;
>>>>  }
>>>>  
>>>> +static void set_lpi(struct its_device *dev, u32 eventid, u32 physid,
>>>> +		    struct its_collection *col)
>>>> +{
>>>> +	if (!dev || !col)
>>>
>>> I don't think col can be null, and this doesn't look like the right place
>>> to check if dev is null.  If we're bothiner to call set_lpi, then I
>>> think we should already expect dev to be good to go.
>> put an assert() instead
>>>
>>>> +		report_abort("wrong device or collection");
>>>> +
>>>> +	its_send_mapti(dev, physid, eventid, col);
>>>> +
>>>> +	gicv3_lpi_set_config(physid, LPI_PROP_DEFAULT);
>>>> +	its_send_invall(col);
>>>> +}
>>>> +
>>>>  /*
>>>>   * Setup the configuration for those mappings:
>>>>   * dev_id=2 event=20 -> vcpu 3, intid=8195
>>>> @@ -806,6 +833,121 @@ static void test_its_migration(void)
>>>>  	its_send_int(dev7, 255);
>>>>  	check_lpi_stats();
>>>>  }
>>>> +
>>>> +static void test_migrate_unmapped_collection(void)
>>>> +{
>>>> +	struct its_collection *col;
>>>> +	struct its_device *dev2, *dev7;
>>>> +	u8 config;
>>>> +
>>>> +	if (its_setup1())
>>>> +		return;
>>>> +
>>>> +	col = its_create_collection(nr_cpus - 1, nr_cpus - 1);
>>>> +	dev2 = its_get_device(2);
>>>> +	dev7 = its_get_device(7);
>>>> +
>>>> +	/* MAPTI with the collection unmapped */
>>>> +	set_lpi(dev2, 0, 8192, col);
>>>> +
>>>> +	puts("Now migrate the VM, then press a key to continue...\n");
>>>> +	(void)getchar();
>>>> +	report(true, "Migration complete");
>>>
>>> report_info
>> yep
>>>
>>>> +
>>>> +	/* on the destination, map the collection */
>>>> +	its_send_mapc(col, true);
>>>> +
>>>> +	lpi_stats_expect(2, 8196);
>>>> +	its_send_int(dev7, 255);
>>>> +	check_lpi_stats();
>>>> +
>>>> +	config = gicv3_lpi_get_config(8192);
>>>> +	report(config == LPI_PROP_DEFAULT,
>>>> +	       "Config of LPI 8192 was properly migrated");
>>>> +
>>>> +	lpi_stats_expect(nr_cpus - 1, 8192);
>>>> +	its_send_int(dev2, 0);
>>>> +	check_lpi_stats();
>>>> +
>>>> +	/* unmap the collection */
>>>> +	its_send_mapc(col, false);
>>>> +
>>>> +	lpi_stats_expect(-1, -1);
>>>> +	its_send_int(dev2, 0);
>>>> +	check_lpi_stats();
>>>> +
>>>> +	/* remap event 0 onto lpiid 8193 */
>>>> +	set_lpi(dev2, 0, 8193, col);
>>>> +	lpi_stats_expect(-1, -1);
>>>> +	its_send_int(dev2, 0);
>>>> +	check_lpi_stats();
>>>> +
>>>> +	/* remap the collection */
>>>> +	its_send_mapc(col, true);
>>>> +	lpi_stats_expect(nr_cpus - 1, 8193);
>>>> +}
>>>> +
>>>> +static void test_its_pending_migration(void)
>>>> +{
>>>> +	struct its_device *dev;
>>>> +	struct its_collection *collection[2];
>>>> +	int expected[NR_CPUS];
>>>
>>> expected = malloc(nr_cpus * sizeof(int));
>>>
>>> I know there are other places using NR_CPUS right now that don't have to,
>>> but we shouldn't add more. Eventually I'll change the other places too.
>> OK
>>>
>>>> +	u64 pendbaser;
>>>> +	void *ptr;
>>>> +	int i;
>>>> +
>>>> +	if (its_prerequisites(4))
>>>> +		return;
>>>> +
>>>> +	dev = its_create_device(2 /* dev id */, 8 /* nb_ites */);
>>>> +	its_send_mapd(dev, true);
>>>> +
>>>> +	collection[0] = its_create_collection(nr_cpus - 1, nr_cpus - 1);
>>>> +	collection[1] = its_create_collection(nr_cpus - 2, nr_cpus - 2);
>>>> +	its_send_mapc(collection[0], true);
>>>> +	its_send_mapc(collection[1], true);
>>>> +
>>>> +	/* disable lpi at redist level */
>>>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, false);
>>>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, false);
>>>> +
>>>> +	/* even lpis are assigned to even cpu */
>>>> +	for (i = 0; i < 256; i++) {
>>>> +		struct its_collection *col = i % 2 ? collection[0] :
>>>> +						     collection[1];
>>>> +		int vcpu = col->target_address >> 16;
>>>
>>> I'm lost with the even/odd (nr_cpus - 1)/(nr_cpus - 2) stuff, and won't
>>> it swap if nr_cpus is odd vs. even?
>>>
>>> Shouldn't we just have something like
>>>
>>>   pe1 = nr_cpus - 1;
>>>   pe2 = nr_cpus - 2;
>>>   col1 = its_create_collection(pe1, pe1);
>>>   col2 = its_create_collection(pe2, pe2);
>>>
>>> without mentioning even and odd?
>> OK
>>>
>>>> +
>>>> +		its_send_mapti(dev, 8192 + i, i, col);
>>>> +		gicv3_lpi_set_config(8192 + i, LPI_PROP_DEFAULT);
>>>> +		gicv3_lpi_set_pending_table_bit(vcpu, 8192 + i, true);
>>>> +	}
>>>> +	its_send_invall(collection[0]);
>>>> +	its_send_invall(collection[1]);
>>>> +
>>>> +	/* Set the PTZ bit on each pendbaser */
>>>> +
>>>> +	expected[nr_cpus - 1] = 128;
>>>> +	expected[nr_cpus - 2] = 128;
>>>> +
>>>> +	ptr = gicv3_data.redist_base[nr_cpus - 1] + GICR_PENDBASER;
>>>> +	pendbaser = readq(ptr);
>>>> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
>>>> +
>>>> +	ptr = gicv3_data.redist_base[nr_cpus - 2] + GICR_PENDBASER;
>>>> +	pendbaser = readq(ptr);
>>>> +	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
>>>> +
>>>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 1, true);
>>>> +	gicv3_lpi_rdist_ctrl(nr_cpus - 2, true);
>>>> +
>>>> +	puts("Now migrate the VM, then press a key to continue...\n");
>>>> +	(void)getchar();
>>>> +	report(true, "Migration complete");
>>>
>>> report_info
>> OK
>>>
>>>> +
>>>> +	mdelay(1000);
>>>
>>> This delay needs a comment explaining why it's here.
>> OK
>>
>> Thanks
>>
>> Eric
>>>
>>>> +
>>>> +	check_lpi_hits(expected);
>>>> +}
>>>>  #endif
>>>>  
>>>>  int main(int argc, char **argv)
>>>> @@ -847,6 +989,14 @@ int main(int argc, char **argv)
>>>>  		report_prefix_push(argv[1]);
>>>>  		test_its_migration();
>>>>  		report_prefix_pop();
>>>> +	} else if (!strcmp(argv[1], "its-pending-migration")) {
>>>> +		report_prefix_push(argv[1]);
>>>> +		test_its_pending_migration();
>>>> +		report_prefix_pop();
>>>> +	} else if (!strcmp(argv[1], "its-migrate-unmapped-collection")) {
>>>> +		report_prefix_push(argv[1]);
>>>> +		test_migrate_unmapped_collection();
>>>> +		report_prefix_pop();
>>>>  	} else if (strcmp(argv[1], "its-introspection") == 0) {
>>>>  		report_prefix_push(argv[1]);
>>>>  		test_its_introspection();
>>>> diff --git a/arm/unittests.cfg b/arm/unittests.cfg
>>>> index 8b8ec79..d917157 100644
>>>> --- a/arm/unittests.cfg
>>>> +++ b/arm/unittests.cfg
>>>> @@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration'
>>>>  groups = its migration
>>>>  arch = arm64
>>>>  
>>>> +[its-pending-migration]
>>>> +file = gic.flat
>>>> +smp = $MAX_SMP
>>>> +accel = kvm
>>>> +extra_params = -machine gic-version=3 -append 'its-pending-migration'
>>>> +groups = its migration
>>>> +arch = arm64
>>>> +
>>>> +[its-migrate-unmapped-collection]
>>>> +file = gic.flat
>>>> +smp = $MAX_SMP
>>>> +accel = kvm
>>>> +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection'
>>>> +groups = its migration
>>>> +arch = arm64
>>>> +
>>>>  # Test PSCI emulation
>>>>  [psci]
>>>>  file = psci.flat
>>>> -- 
>>>> 2.20.1
>>>>
>>>>
>>>
>>> Thanks,
>>> drew 
>>>
>>
>>
diff mbox series

Patch

diff --git a/arm/gic.c b/arm/gic.c
index fa8626a..ec3dd3a 100644
--- a/arm/gic.c
+++ b/arm/gic.c
@@ -195,6 +195,7 @@  static void lpi_handler(struct pt_regs *regs __unused)
 	smp_rmb(); /* pairs with wmb in lpi_stats_expect */
 	lpi_stats.observed.cpu_id = smp_processor_id();
 	lpi_stats.observed.lpi_id = irqnr;
+	acked[lpi_stats.observed.cpu_id]++;
 	smp_wmb(); /* pairs with rmb in check_lpi_stats */
 }
 
@@ -239,6 +240,18 @@  static void secondary_lpi_test(void)
 	while (1)
 		wfi();
 }
+
+static void check_lpi_hits(int *expected)
+{
+	int i;
+
+	for (i = 0; i < nr_cpus; i++) {
+		if (acked[i] != expected[i])
+			report(false, "expected %d LPIs on PE #%d, %d observed",
+			       expected[i], i, acked[i]);
+		}
+	report(true, "check LPI on all vcpus");
+}
 #endif
 
 static void gicv2_ipi_send_self(void)
@@ -594,6 +607,8 @@  static void gic_test_mmio(void)
 static void test_its_introspection(void) {}
 static void test_its_trigger(void) {}
 static void test_its_migration(void) {}
+static void test_migrate_unmapped_collection(void) {}
+static void test_its_pending_migration(void) {}
 
 #else /* __arch64__ */
 
@@ -666,6 +681,18 @@  static bool its_prerequisites(int nb_cpus)
 	return false;
 }
 
+static void set_lpi(struct its_device *dev, u32 eventid, u32 physid,
+		    struct its_collection *col)
+{
+	if (!dev || !col)
+		report_abort("wrong device or collection");
+
+	its_send_mapti(dev, physid, eventid, col);
+
+	gicv3_lpi_set_config(physid, LPI_PROP_DEFAULT);
+	its_send_invall(col);
+}
+
 /*
  * Setup the configuration for those mappings:
  * dev_id=2 event=20 -> vcpu 3, intid=8195
@@ -806,6 +833,121 @@  static void test_its_migration(void)
 	its_send_int(dev7, 255);
 	check_lpi_stats();
 }
+
+static void test_migrate_unmapped_collection(void)
+{
+	struct its_collection *col;
+	struct its_device *dev2, *dev7;
+	u8 config;
+
+	if (its_setup1())
+		return;
+
+	col = its_create_collection(nr_cpus - 1, nr_cpus - 1);
+	dev2 = its_get_device(2);
+	dev7 = its_get_device(7);
+
+	/* MAPTI with the collection unmapped */
+	set_lpi(dev2, 0, 8192, col);
+
+	puts("Now migrate the VM, then press a key to continue...\n");
+	(void)getchar();
+	report(true, "Migration complete");
+
+	/* on the destination, map the collection */
+	its_send_mapc(col, true);
+
+	lpi_stats_expect(2, 8196);
+	its_send_int(dev7, 255);
+	check_lpi_stats();
+
+	config = gicv3_lpi_get_config(8192);
+	report(config == LPI_PROP_DEFAULT,
+	       "Config of LPI 8192 was properly migrated");
+
+	lpi_stats_expect(nr_cpus - 1, 8192);
+	its_send_int(dev2, 0);
+	check_lpi_stats();
+
+	/* unmap the collection */
+	its_send_mapc(col, false);
+
+	lpi_stats_expect(-1, -1);
+	its_send_int(dev2, 0);
+	check_lpi_stats();
+
+	/* remap event 0 onto lpiid 8193 */
+	set_lpi(dev2, 0, 8193, col);
+	lpi_stats_expect(-1, -1);
+	its_send_int(dev2, 0);
+	check_lpi_stats();
+
+	/* remap the collection */
+	its_send_mapc(col, true);
+	lpi_stats_expect(nr_cpus - 1, 8193);
+}
+
+static void test_its_pending_migration(void)
+{
+	struct its_device *dev;
+	struct its_collection *collection[2];
+	int expected[NR_CPUS];
+	u64 pendbaser;
+	void *ptr;
+	int i;
+
+	if (its_prerequisites(4))
+		return;
+
+	dev = its_create_device(2 /* dev id */, 8 /* nb_ites */);
+	its_send_mapd(dev, true);
+
+	collection[0] = its_create_collection(nr_cpus - 1, nr_cpus - 1);
+	collection[1] = its_create_collection(nr_cpus - 2, nr_cpus - 2);
+	its_send_mapc(collection[0], true);
+	its_send_mapc(collection[1], true);
+
+	/* disable lpi at redist level */
+	gicv3_lpi_rdist_ctrl(nr_cpus - 1, false);
+	gicv3_lpi_rdist_ctrl(nr_cpus - 2, false);
+
+	/* even lpis are assigned to even cpu */
+	for (i = 0; i < 256; i++) {
+		struct its_collection *col = i % 2 ? collection[0] :
+						     collection[1];
+		int vcpu = col->target_address >> 16;
+
+		its_send_mapti(dev, 8192 + i, i, col);
+		gicv3_lpi_set_config(8192 + i, LPI_PROP_DEFAULT);
+		gicv3_lpi_set_pending_table_bit(vcpu, 8192 + i, true);
+	}
+	its_send_invall(collection[0]);
+	its_send_invall(collection[1]);
+
+	/* Set the PTZ bit on each pendbaser */
+
+	expected[nr_cpus - 1] = 128;
+	expected[nr_cpus - 2] = 128;
+
+	ptr = gicv3_data.redist_base[nr_cpus - 1] + GICR_PENDBASER;
+	pendbaser = readq(ptr);
+	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
+
+	ptr = gicv3_data.redist_base[nr_cpus - 2] + GICR_PENDBASER;
+	pendbaser = readq(ptr);
+	writeq(pendbaser & ~GICR_PENDBASER_PTZ, ptr);
+
+	gicv3_lpi_rdist_ctrl(nr_cpus - 1, true);
+	gicv3_lpi_rdist_ctrl(nr_cpus - 2, true);
+
+	puts("Now migrate the VM, then press a key to continue...\n");
+	(void)getchar();
+	report(true, "Migration complete");
+
+	mdelay(1000);
+
+	check_lpi_hits(expected);
+}
 #endif
 
 int main(int argc, char **argv)
@@ -847,6 +989,14 @@  int main(int argc, char **argv)
 		report_prefix_push(argv[1]);
 		test_its_migration();
 		report_prefix_pop();
+	} else if (!strcmp(argv[1], "its-pending-migration")) {
+		report_prefix_push(argv[1]);
+		test_its_pending_migration();
+		report_prefix_pop();
+	} else if (!strcmp(argv[1], "its-migrate-unmapped-collection")) {
+		report_prefix_push(argv[1]);
+		test_migrate_unmapped_collection();
+		report_prefix_pop();
 	} else if (strcmp(argv[1], "its-introspection") == 0) {
 		report_prefix_push(argv[1]);
 		test_its_introspection();
diff --git a/arm/unittests.cfg b/arm/unittests.cfg
index 8b8ec79..d917157 100644
--- a/arm/unittests.cfg
+++ b/arm/unittests.cfg
@@ -144,6 +144,22 @@  extra_params = -machine gic-version=3 -append 'its-migration'
 groups = its migration
 arch = arm64
 
+[its-pending-migration]
+file = gic.flat
+smp = $MAX_SMP
+accel = kvm
+extra_params = -machine gic-version=3 -append 'its-pending-migration'
+groups = its migration
+arch = arm64
+
+[its-migrate-unmapped-collection]
+file = gic.flat
+smp = $MAX_SMP
+accel = kvm
+extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection'
+groups = its migration
+arch = arm64
+
 # Test PSCI emulation
 [psci]
 file = psci.flat