From: Raag Jadav <raag.jadav@intel.com>
To: "Anirban, Sk" <sk.anirban@intel.com>
Cc: intel-gfx@lists.freedesktop.org, krzysztof.karas@intel.com
Subject: Re: [PATCH v3] drm/i915/selftest: Add throttle reason diagnostics to RPS selftests
Date: Thu, 18 Dec 2025 12:27:43 +0100 [thread overview]
Message-ID: <aUPlL6PWbpe5RhG-@black.igk.intel.com> (raw)
In-Reply-To: <281d4ce1-bb8b-48bc-bac0-286b95ddf318@intel.com>
On Thu, Dec 18, 2025 at 11:16:29AM +0530, Anirban, Sk wrote:
> Hi Krzysztof,
>
> Could you please let me know if I can retain your review for this version of
> my patch?
Yes, unless there are major changes we usually retain the tags.
> On 18-12-2025 11:02 am, Sk Anirban wrote:
> > Report GPU throttle reasons when RPS tests fail to reach expected
> > frequencies or power levels.
> >
> > v2: Read the throttle value before the spinner ends (Raag)
> > Add a condition before printing throttle value (Krzysztof)
> > v3: Extend throttle reasons debug support (Raag)
> >
> > Signed-off-by: Sk Anirban <sk.anirban@intel.com>
> > ---
> > drivers/gpu/drm/i915/gt/selftest_rps.c | 37 ++++++++++++++++++++++----
> > 1 file changed, 32 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gt/selftest_rps.c b/drivers/gpu/drm/i915/gt/selftest_rps.c
> > index 73bc91c6ea07..b42d9b39add6 100644
> > --- a/drivers/gpu/drm/i915/gt/selftest_rps.c
> > +++ b/drivers/gpu/drm/i915/gt/selftest_rps.c
> > @@ -378,6 +378,7 @@ int live_rps_control(void *arg)
> > enum intel_engine_id id;
> > struct igt_spinner spin;
> > intel_wakeref_t wakeref;
> > + u32 throttle;
> > int err = 0;
> > /*
> > @@ -463,6 +464,10 @@ int live_rps_control(void *arg)
> > max = rps_set_check(rps, limit);
> > max_dt = ktime_sub(ktime_get(), max_dt);
> > + throttle = intel_uncore_read(gt->uncore,
> > + intel_gt_perf_limit_reasons_reg(gt));
Now that we allow 100 characters I think this can be one line.
> > + throttle &= GT0_PERF_LIMIT_REASONS_MASK;
> > +
> > min_dt = ktime_get();
> > min = rps_set_check(rps, rps->min_freq);
> > min_dt = ktime_sub(ktime_get(), min_dt);
> > @@ -478,11 +483,9 @@ int live_rps_control(void *arg)
> > min, max, ktime_to_ns(min_dt), ktime_to_ns(max_dt));
> > if (limit != rps->max_freq) {
> > - u32 throttle = intel_uncore_read(gt->uncore,
> > - intel_gt_perf_limit_reasons_reg(gt));
> > -
> > - pr_warn("%s: GPU throttled with reasons 0x%08x\n",
> > - engine->name, throttle & GT0_PERF_LIMIT_REASONS_MASK);
> > + if (throttle)
> > + pr_warn("%s: GPU throttled with reasons 0x%08x\n",
> > + engine->name, throttle);
> > show_pstate_limits(rps);
> > }
> > @@ -1138,6 +1141,7 @@ int live_rps_power(void *arg)
> > struct intel_engine_cs *engine;
> > enum intel_engine_id id;
> > struct igt_spinner spin;
> > + u32 throttle;
> > int err = 0;
> > /*
> > @@ -1195,6 +1199,10 @@ int live_rps_power(void *arg)
> > max.freq = rps->max_freq;
> > max.power = measure_power_at(rps, &max.freq);
> > + throttle = intel_uncore_read(gt->uncore,
> > + intel_gt_perf_limit_reasons_reg(gt));
Ditto.
> > + throttle &= GT0_PERF_LIMIT_REASONS_MASK;
> > +
> > min.freq = rps->min_freq;
> > min.power = measure_power_at(rps, &min.freq);
> > @@ -1210,12 +1218,21 @@ int live_rps_power(void *arg)
> > pr_notice("Could not control frequency, ran at [%d:%uMHz, %d:%uMhz]\n",
> > min.freq, intel_gpu_freq(rps, min.freq),
> > max.freq, intel_gpu_freq(rps, max.freq));
> > +
> > + if (throttle)
> > + pr_warn("%s: GPU throttled with reasons 0x%08x\n",
> > + engine->name, throttle);
> > continue;
> > }
> > if (11 * min.power > 10 * max.power) {
> > pr_err("%s: did not conserve power when setting lower frequency!\n",
> > engine->name);
> > +
> > + if (throttle)
> > + pr_warn("%s: GPU throttled with reasons 0x%08x\n",
> > + engine->name, throttle);
> > +
> > err = -EINVAL;
> > break;
> > }
> > @@ -1241,6 +1258,7 @@ int live_rps_dynamic(void *arg)
> > struct intel_engine_cs *engine;
> > enum intel_engine_id id;
> > struct igt_spinner spin;
> > + u32 throttle;
> > int err = 0;
> > /*
> > @@ -1293,6 +1311,10 @@ int live_rps_dynamic(void *arg)
> > max.freq = wait_for_freq(rps, rps->max_freq, 500);
> > max.dt = ktime_sub(ktime_get(), max.dt);
> > + throttle = intel_uncore_read(gt->uncore,
> > + intel_gt_perf_limit_reasons_reg(gt));
Ditto. And with that,
Reviewed-by: Raag Jadav <raag.jadav@intel.com>
> > + throttle &= GT0_PERF_LIMIT_REASONS_MASK;
> > +
> > igt_spinner_end(&spin);
> > min.dt = ktime_get();
> > @@ -1308,6 +1330,11 @@ int live_rps_dynamic(void *arg)
> > if (min.freq >= max.freq) {
> > pr_err("%s: dynamic reclocking of spinner failed\n!",
> > engine->name);
> > +
> > + if (throttle)
> > + pr_warn("%s: GPU throttled with reasons 0x%08x\n",
> > + engine->name, throttle);
> > +
> > err = -EINVAL;
> > }
next prev parent reply other threads:[~2025-12-18 11:27 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-12-18 5:32 [PATCH v3] drm/i915/selftest: Add throttle reason diagnostics to RPS selftests Sk Anirban
2025-12-18 5:46 ` Anirban, Sk
2025-12-18 11:27 ` Raag Jadav [this message]
2025-12-18 6:31 ` ✓ i915.CI.BAT: success for drm/i915/selftest: Add throttle reason diagnostics to RPS selftests (rev3) Patchwork
2025-12-18 8:25 ` ✗ i915.CI.Full: failure " Patchwork
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=aUPlL6PWbpe5RhG-@black.igk.intel.com \
--to=raag.jadav@intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=krzysztof.karas@intel.com \
--cc=sk.anirban@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