[v2] drm/i915/selftests: Implement frequency check for energy reading validation
Pottumuttu, Sai Teja
sai.teja.pottumuttu at intel.com
Wed Sep 18 11:45:47 UTC 2024
On 17-09-2024 15:10, Sk Anirban wrote:
> This commit introduces a frequency check mechanism aimed at ensuring
> the accuracy of energy readings.
>
> v2 : Improved commit message
>
> Signed-off-by: Anirban Sk <sk.anirban at intel.com>
> ---
> drivers/gpu/drm/i915/gt/selftest_rc6.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/gt/selftest_rc6.c b/drivers/gpu/drm/i915/gt/selftest_rc6.c
> index 1aa1446c8fb0..b3602328c832 100644
> --- a/drivers/gpu/drm/i915/gt/selftest_rc6.c
> +++ b/drivers/gpu/drm/i915/gt/selftest_rc6.c
> @@ -12,6 +12,7 @@
>
> #include "selftests/i915_random.h"
> #include "selftests/librapl.h"
> +#include "intel_rps.h"
Just a nitpick, we should keep the headers sorted.
>
> static u64 rc6_residency(struct intel_rc6 *rc6)
> {
> @@ -38,6 +39,8 @@ int live_rc6_manual(void *arg)
> ktime_t dt;
> u64 res[2];
> int err = 0;
> + u32 rc0_freq, rc6_freq;
> + struct intel_rps *rps = >->rps;
>
> /*
> * Our claim is that we can "encourage" the GPU to enter rc6 at will.
> @@ -66,6 +69,7 @@ int live_rc6_manual(void *arg)
> rc0_power = librapl_energy_uJ() - rc0_power;
> dt = ktime_sub(ktime_get(), dt);
> res[1] = rc6_residency(rc6);
> + rc0_freq = intel_rps_read_actual_frequency(rps);
> if ((res[1] - res[0]) >> 10) {
> pr_err("RC6 residency increased by %lldus while disabled for 1000ms!\n",
> (res[1] - res[0]) >> 10);
> @@ -91,6 +95,7 @@ int live_rc6_manual(void *arg)
> dt = ktime_get();
> rc6_power = librapl_energy_uJ();
> msleep(100);
> + rc6_freq = intel_rps_read_actual_frequency(rps);
> rc6_power = librapl_energy_uJ() - rc6_power;
> dt = ktime_sub(ktime_get(), dt);
> res[1] = rc6_residency(rc6);
> @@ -109,6 +114,7 @@ int live_rc6_manual(void *arg)
> rc0_power, rc6_power);
> if (2 * rc6_power > rc0_power) {
> pr_err("GPU leaked energy while in RC6!\n");
> + pr_info("GPU Freq: %u in RC6 and %u in RC0\n", rc6_freq, rc0_freq);
> err = -EINVAL;
> goto out_unlock;
> }
More information about the Intel-gfx
mailing list