Re: [PATCH] drm/xe/oa: Wa_14026633728

[Shekhar Chauhan <shekhar.chauhan@intel.com>]

On 4/24/2026 13:43, Shekhar Chauhan wrote:
> For MERT_OA in CRI, the OA_Buffer is in device memory. The minimum
> exponent needed for it is 8. Bump it up.
>
> Signed-off-by: Shekhar Chauhan <shekhar.chauhan@intel.com>
> ---
>   tests/intel/xe_oa.c | 6 ++++++
>   1 file changed, 6 insertions(+)
>
> diff --git a/tests/intel/xe_oa.c b/tests/intel/xe_oa.c
> index 988c46df6..7bd9d17c4 100644
> --- a/tests/intel/xe_oa.c
> +++ b/tests/intel/xe_oa.c
> @@ -2645,6 +2645,12 @@ test_non_zero_reason(const struct drm_xe_oa_unit *oau, size_t oa_buffer_size)
>   {
>   	/* ~20 micro second period */
>   	int oa_exponent = max_oa_exponent_for_period_lte(20000);

While testing on BMG, if we use 26670us, we get the oa_exponent as 8. 
Either we can hardcode something like 27000 (for a round figure) or 
carry on with my approach down below, since OA_MERT is on CRI, so why to 
increase the timeout for other platforms, was my thought.

Now that you have the exact 'us' count for the oa_exponent, I'll let you 
decide which way to move forward.

-shekhar

> +
> +	/* Wa_14026633728: MERTOA buffer on CRI is in device memory,
> +	 * minimum exponent to avoid overflow is 8 */
> +	if (oau->oa_unit_type == DRM_XE_OA_UNIT_TYPE_MERT && IS_CRESCENTISLAND(devid))
> +		oa_exponent = max(oa_exponent, 8);
> +
>   	struct intel_xe_perf_metric_set *test_set = oa_unit_metric_set(oau);
>   	uint64_t fmt = test_set->perf_oa_format;
>   	size_t report_size = get_oa_format(fmt).size;

-- 
Shekhar Chauhan
Linux Graphics Software Engineer
Intel Corporation