From: Andrew Jones <drjones@redhat.com>
To: Eric Auger <eric.auger@redhat.com>
Cc: eric.auger.pro@gmail.com, maz@kernel.org,
kvmarm@lists.cs.columbia.edu, kvm@vger.kernel.org,
qemu-devel@nongnu.org, qemu-arm@nongnu.org,
peter.maydell@linaro.org, andre.przywara@arm.com,
thuth@redhat.com, yuzenghui@huawei.com, alexandru.elisei@arm.com
Subject: Re: [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test
Date: Fri, 7 Feb 2020 15:06:27 +0100 [thread overview]
Message-ID: <20200207140627.rikkw5coe7naxssb@kamzik.brq.redhat.com> (raw)
In-Reply-To: <20200128103459.19413-15-eric.auger@redhat.com>
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
next prev parent reply other threads:[~2020-02-07 14:06 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-28 10:34 [kvm-unit-tests PATCH v3 00/14] arm/arm64: Add ITS tests Eric Auger
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 01/14] libcflat: Add other size defines Eric Auger
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 02/14] arm: gic: Provide per-IRQ helper functions Eric Auger
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 03/14] arm/arm64: gic: Introduce setup_irq() helper Eric Auger
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 04/14] arm/arm64: gicv3: Add some re-distributor defines Eric Auger
2020-02-06 14:35 ` Zenghui Yu
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 05/14] arm/arm64: ITS: Introspection tests Eric Auger
2020-02-06 15:12 ` Zenghui Yu
2020-02-07 10:19 ` Andrew Jones
2020-03-04 14:20 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 06/14] arm/arm64: gicv3: Set the LPI config and pending tables Eric Auger
2020-02-07 2:12 ` Zenghui Yu
2020-03-05 19:40 ` Auger Eric
2020-02-07 12:11 ` Andrew Jones
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 07/14] arm/arm64: gicv3: Enable/Disable LPIs at re-distributor level Eric Auger
2020-02-07 12:14 ` Andrew Jones
2020-02-07 12:19 ` Andrew Jones
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults Eric Auger
2020-02-07 3:20 ` Zenghui Yu
2020-03-04 14:26 ` Auger Eric
2020-03-05 6:30 ` Zenghui Yu
2020-02-07 12:41 ` Andrew Jones
2020-03-05 17:59 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 09/14] arm/arm64: ITS: Device and collection Initialization Eric Auger
2020-02-07 5:41 ` Zenghui Yu
2020-03-05 19:42 ` Auger Eric
2020-02-07 12:51 ` Andrew Jones
2020-03-06 8:47 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 10/14] arm/arm64: ITS: commands Eric Auger
2020-02-07 13:37 ` Andrew Jones
2020-03-06 9:13 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 11/14] arm/arm64: ITS: INT functional tests Eric Auger
2020-02-07 13:15 ` Andrew Jones
2020-03-06 12:55 ` Auger Eric
2020-03-06 13:29 ` Andrew Jones
2020-03-06 13:40 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 12/14] arm/run: Allow Migration tests Eric Auger
2020-01-29 8:07 ` Thomas Huth
2020-01-29 9:29 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 13/14] arm/arm64: ITS: migration tests Eric Auger
2020-02-07 13:49 ` Andrew Jones
2020-03-06 13:06 ` Auger Eric
2020-01-28 10:34 ` [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test Eric Auger
2020-02-07 14:06 ` Andrew Jones [this message]
2020-03-06 13:21 ` Auger Eric
2020-03-06 13:36 ` Andrew Jones
2020-03-06 13:41 ` Auger Eric
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200207140627.rikkw5coe7naxssb@kamzik.brq.redhat.com \
--to=drjones@redhat.com \
--cc=alexandru.elisei@arm.com \
--cc=andre.przywara@arm.com \
--cc=eric.auger.pro@gmail.com \
--cc=eric.auger@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=maz@kernel.org \
--cc=peter.maydell@linaro.org \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=thuth@redhat.com \
--cc=yuzenghui@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).