Re: [igt-dev] [PATCH v3 01/24] drm-uapi/xe_drm: Align with new PMU interface

[Aravind Iddamsetty <aravind.iddamsetty@linux.intel.com>]

On 27/09/23 22:25, Rodrigo Vivi wrote:
> On Tue, Sep 26, 2023 at 05:50:53PM +0100, Tvrtko Ursulin wrote:
>> On 26/09/2023 14:00, Francois Dugast wrote:
>>> From: Rodrigo Vivi <rodrigo.vivi@intel.com>
>>>
>>> Align with commit ("drm/xe/pmu: Enable PMU interface")
>>>
>>> Cc: Francois Dugast <francois.dugast@intel.com>
>>> Cc: Aravind Iddamsetty <aravind.iddamsetty@linux.intel.com>
>>> Signed-off-by: Francois Dugast <francois.dugast@intel.com>
>>> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
>>> ---
>>>   include/drm-uapi/xe_drm.h | 38 ++++++++++++++++++++++++++++++++++++++
>>>   1 file changed, 38 insertions(+)
>>>
>>> diff --git a/include/drm-uapi/xe_drm.h b/include/drm-uapi/xe_drm.h
>>> index 804c02270..643eb6e82 100644
>>> --- a/include/drm-uapi/xe_drm.h
>>> +++ b/include/drm-uapi/xe_drm.h
>>> @@ -1053,6 +1053,44 @@ struct drm_xe_vm_madvise {
>>>   	__u64 reserved[2];
>>>   };
>>> +/**
>>> + * DOC: XE PMU event config IDs
>>> + *
>>> + * Check 'man perf_event_open' to use these ID's in 'struct perf_event_attr'
>>> + * as part of perf_event_open syscall to read a particular event.
>>> + *
>>> + * For example to open the XE_PMU_INTERRUPTS(0):
>>> + *
>>> + * .. code-block:: C
>>> + *	struct perf_event_attr attr;
>>> + *	long long count;
>>> + *	int cpu = 0;
>>> + *	int fd;
>>> + *
>>> + *	memset(&attr, 0, sizeof(struct perf_event_attr));
>>> + *	attr.type = type; // eg: /sys/bus/event_source/devices/xe_0000_56_00.0/type
>>> + *	attr.read_format = PERF_FORMAT_TOTAL_TIME_ENABLED;
>>> + *	attr.use_clockid = 1;
>>> + *	attr.clockid = CLOCK_MONOTONIC;
>>> + *	attr.config = XE_PMU_INTERRUPTS(0);
>>> + *
>>> + *	fd = syscall(__NR_perf_event_open, &attr, -1, cpu, -1, 0);
>>> + */
>>> +
>>> +/*
>>> + * Top bits of every counter are GT id.
>>> + */
>>> +#define __XE_PMU_GT_SHIFT (56)
>>> +
>>> +#define ___XE_PMU_OTHER(gt, x) \
>>> +	(((__u64)(x)) | ((__u64)(gt) << __XE_PMU_GT_SHIFT))
>>> +
>>> +#define XE_PMU_INTERRUPTS(gt)			___XE_PMU_OTHER(gt, 0)
>> AFAIR interrupts is probably the least useful counter and I don't remember
>> that anyone asked much about it. Therefore I'd say it could be worth seeing
>> if you could just drop it. Changes to intel_gpu_top to work with the below
>> set (no per engine, no frequencies) will have to be extensive already
>> anyway.

Ok, will drop the interrupt.

Also, this is just the initial set, per engine will follow but sometime later.

Thanks,
Aravind.
> Aravind, could you please reply to that?
>
>> Regards,
>>
>> Tvrtko
>>
>>> +#define XE_PMU_RENDER_GROUP_BUSY(gt)		___XE_PMU_OTHER(gt, 1)
>>> +#define XE_PMU_COPY_GROUP_BUSY(gt)		___XE_PMU_OTHER(gt, 2)
>>> +#define XE_PMU_MEDIA_GROUP_BUSY(gt)		___XE_PMU_OTHER(gt, 3)
>>> +#define XE_PMU_ANY_ENGINE_GROUP_BUSY(gt)	___XE_PMU_OTHER(gt, 4)
>>> +
>>>   #if defined(__cplusplus)
>>>   }
>>>   #endif