Re: [Intel-gfx] [PATCH 3/7] drm/i915/guc: Add GuC <-> kernel time stamp translation information

[Jani Nikula <jani.nikula@linux.intel.com>]

On Fri, 19 Aug 2022, John Harrison <john.c.harrison@intel.com> wrote:
> On 8/19/2022 03:45, Jani Nikula wrote:
>> On Wed, 27 Jul 2022, John.C.Harrison@Intel.com wrote:
>>> From: John Harrison <John.C.Harrison@Intel.com>
>>>
>>> It is useful to be able to match GuC events to kernel events when
>>> looking at the GuC log. That requires being able to convert GuC
>>> timestamps to kernel time. So, when dumping error captures and/or GuC
>>> logs, include a stamp in both time zones plus the clock frequency.
>>>
>>> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
>>> ---
>>>   drivers/gpu/drm/i915/gt/intel_gt_regs.h    |  2 ++
>>>   drivers/gpu/drm/i915/gt/uc/intel_guc.c     | 19 +++++++++++++++++++
>>>   drivers/gpu/drm/i915/gt/uc/intel_guc.h     |  2 ++
>>>   drivers/gpu/drm/i915/gt/uc/intel_guc_log.c |  2 ++
>>>   drivers/gpu/drm/i915/i915_gpu_error.c      | 12 ++++++++++++
>>>   drivers/gpu/drm/i915/i915_gpu_error.h      |  3 +++
>>>   6 files changed, 40 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/i915/gt/intel_gt_regs.h b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
>>> index 60d6eb5f245b7..fc7979bd91db5 100644
>>> --- a/drivers/gpu/drm/i915/gt/intel_gt_regs.h
>>> +++ b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
>>> @@ -1007,6 +1007,8 @@
>>>   #define   GEN11_LSN_UNSLCVC_GAFS_HALF_CL2_MAXALLOC	(1 << 9)
>>>   #define   GEN11_LSN_UNSLCVC_GAFS_HALF_SF_MAXALLOC	(1 << 7)
>>>   
>>> +#define GUCPMTIMESTAMP				_MMIO(0xc3e8)
>>> +
>>>   #define __GEN9_RCS0_MOCS0			0xc800
>>>   #define GEN9_GFX_MOCS(i)			_MMIO(__GEN9_RCS0_MOCS0 + (i) * 4)
>>>   #define __GEN9_VCS0_MOCS0			0xc900
>>> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.c b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
>>> index 2706a8c650900..ab4aacc516aa4 100644
>>> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.c
>>> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
>>> @@ -389,6 +389,25 @@ void intel_guc_write_params(struct intel_guc *guc)
>>>   	intel_uncore_forcewake_put(uncore, FORCEWAKE_GT);
>>>   }
>>>   
>>> +void intel_guc_dump_time_info(struct intel_guc *guc, struct drm_printer *p)
>>> +{
>>> +	struct intel_gt *gt = guc_to_gt(guc);
>>> +	intel_wakeref_t wakeref;
>>> +	u32 stamp = 0;
>>> +	u64 ktime;
>>> +
>>> +	intel_device_info_print_runtime(RUNTIME_INFO(gt->i915), p);
>> Why does this function print runtime info alone? Seems kind of random,
>> considering what intel_device_info_print_runtime() actually
>> prints. Should it print both device info and runtime info, or nothing at
>> all?
> Hmm. That was added specifically to print the rawclk value (and only for 
> the rawclk value) because that was the frequency that the GuC time stamp 
> register ticks at. I think we later worked out that the CS frequency was 
> more correct. Hence printing gt->clock_frequency lower down. I guess I 
> forgot to go back and remove the rawclk print, though.
>
> So yeah, it can be removed.

Could you r-b the patch [1] doing just that please?

BR,
Jani.

[1] https://patchwork.freedesktop.org/patch/msgid/b395ac4c909042f5daabf29959d8733993545aa2.1660910433.git.jani.nikula@intel.com


-- 
Jani Nikula, Intel Open Source Graphics Center