From: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
To: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>,
igt-dev@lists.freedesktop.org,
Chris Wilson <chris.p.wilson@intel.com>
Subject: Re: [igt-dev] [PATCH i-g-t] test/perf: Add test to verify OA TLB invalidation
Date: Mon, 16 Mar 2020 11:52:47 +0200 [thread overview]
Message-ID: <85b5ba2a-e818-69de-e720-89d62acfce3d@intel.com> (raw)
In-Reply-To: <20200313174235.19440-1-umesh.nerlige.ramappa@intel.com>
On 13/03/2020 19:42, Umesh Nerlige Ramappa wrote:
> Run 2 blocking tests back to back and compare the number of OA reports
> captured. Make sure the number of reports are within an acceptable range
> of the expected number of reports.
>
> v2: Drop poll and use blocking IO (Lionel)
> v3: Verify number of expected reports for both tests (Lionel)
>
> Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
> ---
> tests/perf.c | 92 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 92 insertions(+)
>
> diff --git a/tests/perf.c b/tests/perf.c
> index 5e818030..b757e307 100644
> --- a/tests/perf.c
> +++ b/tests/perf.c
> @@ -2265,6 +2265,92 @@ test_polling(void)
> __perf_close(stream_fd);
> }
>
> +static int
> +num_valid_reports_captured(struct drm_i915_perf_open_param *param,
> + int64_t *duration_ns)
> +{
> + uint8_t buf[1024 * 1024];
> + int64_t start, end;
> + int num_reports = 0;
> +
> + igt_debug("Expected duration = %lu\n", *duration_ns);
> +
> + stream_fd = __perf_open(drm_fd, param, true);
> +
> + start = get_time();
> + do_ioctl(stream_fd, I915_PERF_IOCTL_ENABLE, 0);
> + for (/* nop */; ((end = get_time()) - start) < *duration_ns; /* nop */) {
> + struct drm_i915_perf_record_header *header;
> + int ret;
> +
> + while ((ret = read(stream_fd, buf, sizeof(buf))) < 0 &&
> + errno == EINTR)
> + ;
> +
> + igt_assert(ret > 0);
> +
> + for (int offset = 0; offset < ret; offset += header->size) {
> + header = (void *)(buf + offset);
> +
> + if (header->type == DRM_I915_PERF_RECORD_SAMPLE) {
> + uint32_t *report = (void *)(header + 1);
> +
> + if ((report[0] >> OAREPORT_REASON_SHIFT) &
> + OAREPORT_REASON_TIMER)
> + num_reports++;
> + }
> + }
> + }
> + __perf_close(stream_fd);
> +
> + *duration_ns = end - start;
> +
> + igt_debug("Actual duration = %lu\n", *duration_ns);
> +
> + return num_reports;
> +}
> +
> +static void
> +gen12_test_oa_tlb_invalidate(void)
> +{
> + int oa_exponent = max_oa_exponent_for_period_lte(30000000);
> + uint64_t properties[] = {
> + DRM_I915_PERF_PROP_SAMPLE_OA, true,
> +
> + DRM_I915_PERF_PROP_OA_METRICS_SET, test_set->perf_oa_metrics_set,
> + DRM_I915_PERF_PROP_OA_FORMAT, test_set->perf_oa_format,
> + DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
> + };
> + struct drm_i915_perf_open_param param = {
> + .flags = I915_PERF_FLAG_FD_CLOEXEC |
> + I915_PERF_FLAG_DISABLED,
> + .num_properties = sizeof(properties) / 16,
> + .properties_ptr = to_user_pointer(properties),
> + };
> + int num_reports1, num_reports2, num_expected_reports;
> + int64_t duration;
> +
> + /* Capture reports for 5 seconds twice and then make sure you get around
> + * the same number of reports. In the case of failure, the number of
> + * reports will vary largely since the beginning of the OA buffer
> + * will have invalid entries.
> + */
> + duration = 5LL * NSEC_PER_SEC;
> + num_reports1 = num_valid_reports_captured(¶m, &duration);
> + num_expected_reports = duration / oa_exponent_to_ns(oa_exponent);
> + igt_debug("expected num reports = %d\n", num_expected_reports);
> + igt_debug("actual num reports = %d\n", num_reports1);
> + igt_assert(num_reports1 > 0.95 * num_expected_reports);
> +
> + duration = 5LL * NSEC_PER_SEC;
> + num_reports2 = num_valid_reports_captured(¶m, &duration);
> + num_expected_reports = duration / oa_exponent_to_ns(oa_exponent);
> + igt_debug("expected num reports = %d\n", num_expected_reports);
> + igt_debug("actual num reports = %d\n", num_reports2);
> + igt_assert(num_reports2 > 0.95 * num_expected_reports);
> +}
> +
> +
> static void
> test_buffer_fill(void)
> {
> @@ -4622,6 +4708,12 @@ igt_main
> gen8_test_single_ctx_render_target_writes_a_counter();
> }
>
> + igt_describe("Test OA TLB invalidate");
> + igt_subtest("gen12-oa-tlb-invalidate") {
> + igt_require(intel_gen(devid) >= 12);
> + gen12_test_oa_tlb_invalidate();
> + }
> +
> igt_describe("Measure performance for a specific context using OAR in Gen 12");
> igt_subtest("gen12-unprivileged-single-ctx-counters") {
> igt_require(intel_gen(devid) >= 12);
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev
next prev parent reply other threads:[~2020-03-16 9:52 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-13 17:42 [igt-dev] [PATCH i-g-t] test/perf: Add test to verify OA TLB invalidation Umesh Nerlige Ramappa
2020-03-13 17:51 ` Chris Wilson
2020-03-16 9:52 ` Lionel Landwerlin [this message]
-- strict thread matches above, loose matches on Subject: below --
2020-03-14 0:00 Umesh Nerlige Ramappa
2020-03-14 10:36 ` Lionel Landwerlin
2020-03-10 19:00 Umesh Nerlige Ramappa
2020-03-12 13:28 ` Lionel Landwerlin
2020-03-10 3:07 Umesh Nerlige Ramappa
2020-03-10 9:02 ` Lionel Landwerlin
2020-03-10 17:57 ` Umesh Nerlige Ramappa
2020-03-12 13:18 ` Lionel Landwerlin
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=85b5ba2a-e818-69de-e720-89d62acfce3d@intel.com \
--to=lionel.g.landwerlin@intel.com \
--cc=chris.p.wilson@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=umesh.nerlige.ramappa@intel.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