Quoting Tvrtko Ursulin (2020-01-10 13:30:48)
> From: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
>
> GPU saves accumulated context runtime (in CS timestamp units) in PPHWSP
> which will be useful for us in cases when we are not able to track context
> busyness ourselves (like with GuC). Keep a copy of this in struct
> intel_context from where it can be easily read even if the context is not
> pinned.
>
> QQQ: Do we want to make this accounting conditional / able to turn on/off?
>
> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
> ---
> drivers/gpu/drm/i915/gt/intel_context.h | 7 +++++++
> drivers/gpu/drm/i915/gt/intel_context_types.h | 5 +++++
> drivers/gpu/drm/i915/gt/intel_lrc.c | 9 +++++++++
> drivers/gpu/drm/i915/intel_device_info.c | 2 ++
> drivers/gpu/drm/i915/intel_device_info.h | 1 +
> 5 files changed, 24 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/gt/intel_context.h b/drivers/gpu/drm/i915/gt/intel_context.h
> index 30f0268fcc9a..389a05736fc7 100644
> --- a/drivers/gpu/drm/i915/gt/intel_context.h
> +++ b/drivers/gpu/drm/i915/gt/intel_context.h
> @@ -12,6 +12,7 @@
> #include <linux/types.h>
>
> #include "i915_active.h"
> +#include "i915_drv.h"
> #include "intel_context_types.h"
> #include "intel_engine_types.h"
> #include "intel_ring_types.h"
> @@ -235,4 +236,10 @@ __intel_context_stats_start(struct intel_context_stats *stats, ktime_t now)
>
> ktime_t intel_context_get_busy_time(struct intel_context *ce);
>
> +static inline u64 intel_context_get_hw_runtime_ns(struct intel_context *ce)
> +{
> + return ce->total_runtime *
> + RUNTIME_INFO(ce->engine->i915)->cs_timestamp_period_ns;
> +}
> +
> #endif /* __INTEL_CONTEXT_H__ */
> diff --git a/drivers/gpu/drm/i915/gt/intel_context_types.h b/drivers/gpu/drm/i915/gt/intel_context_types.h
> index 963d33dc5289..7b08bf87fb82 100644
> --- a/drivers/gpu/drm/i915/gt/intel_context_types.h
> +++ b/drivers/gpu/drm/i915/gt/intel_context_types.h
> @@ -69,6 +69,11 @@ struct intel_context {
> u64 lrc_desc;
> u32 tag; /* cookie passed to HW to track this context on submission */
>
> + /* Time on GPU as tracked by the hw. */
> + u32 last_runtime;
> + u64 total_runtime;
> + u32 *pphwsp;
I wouldn't bother with keeping pphwsp, we know it's the page before the
reg state. At least for the foreseeable future.
> unsigned int active_count; /* protected by timeline->mutex */
>
> atomic_t pin_count;
> diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> index dd559547500f..26999b43e5a1 100644
> --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> @@ -1289,6 +1289,7 @@ __execlists_schedule_out(struct i915_request *rq,
> struct intel_engine_cs * const engine)
> {
> struct intel_context *ce = rq->context;
> + u32 old, new;
>
> /*
> * NB process_csb() is not under the engine->active.lock and hence
> @@ -1309,6 +1310,13 @@ __execlists_schedule_out(struct i915_request *rq,
> execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_OUT);
> intel_gt_pm_put_async(engine->gt);
>
> + old = ce->last_runtime;
> + ce->last_runtime = new = ce->pphwsp[16];
> + if (new > old)
> + ce->total_runtime += new - old;
> + else
> + ce->total_runtime += (~0UL - old) + new + 1;
It's u32, unsigned wrap-around arithmetic is defined, so just
ce->total_runtime += new - old;
> +
> /*
> * If this is part of a virtual engine, its next request may
> * have been blocked waiting for access to the active context.
> @@ -2608,6 +2616,7 @@ __execlists_context_pin(struct intel_context *ce,
>
> ce->lrc_desc = lrc_descriptor(ce, engine) | CTX_DESC_FORCE_RESTORE;
> ce->lrc_reg_state = vaddr + LRC_STATE_PN * PAGE_SIZE;
> + ce->pphwsp = vaddr + LRC_PPHWSP_PN * PAGE_SIZE;
> __execlists_update_reg_state(ce, engine);
>
> return 0;
> diff --git a/drivers/gpu/drm/i915/intel_device_info.c b/drivers/gpu/drm/i915/intel_device_info.c
> index 6670a0763be2..7732748e1939 100644
> --- a/drivers/gpu/drm/i915/intel_device_info.c
> +++ b/drivers/gpu/drm/i915/intel_device_info.c
> @@ -1042,6 +1042,8 @@ void intel_device_info_runtime_init(struct drm_i915_private *dev_priv)
>
> /* Initialize command stream timestamp frequency */
> runtime->cs_timestamp_frequency_khz = read_timestamp_frequency(dev_priv);
> + runtime->cs_timestamp_period_ns =
> + div_u64(1e6, runtime->cs_timestamp_frequency_khz);
drm_debug(&dev_priv->drm, "CS timestamp wraparound in %lld\n",
div_u64(U32_MAX * runtime->cs_timestamp_period_ns, NSEC_PER_SEC);
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
