Quoting Tvrtko Ursulin (2020-04-27 10:00:14)
> From: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
>
> Recently added assert in a common helper used for calculating batch
> duration and preemption period is harmful when preemption is disabled on a
> context. Split out into low level and high level helper and use the former
> for preemption period queries.
>
> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
> ---
> benchmarks/gem_wsim.c | 15 ++++++++++++---
> 1 file changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/benchmarks/gem_wsim.c b/benchmarks/gem_wsim.c
> index 81f47b86d619..ad4edb936920 100644
> --- a/benchmarks/gem_wsim.c
> +++ b/benchmarks/gem_wsim.c
> @@ -1151,7 +1151,7 @@ __get_ctx(struct workload *wrk, const struct w_step *w)
> }
>
> static unsigned long
> -get_bb_sz(const struct w_step *w, unsigned int duration)
> +__get_bb_sz(const struct w_step *w, unsigned int duration)
> {
> enum intel_engine_id engine = w->engine;
> struct ctx *ctx = __get_ctx(w->wrk, w);
> @@ -1165,6 +1165,15 @@ get_bb_sz(const struct w_step *w, unsigned int duration)
> d = ALIGN(duration * engine_calib_map[engine] * sizeof(uint32_t) /
> nop_calibration_us,
> sizeof(uint32_t));
Preempt disabled == w->preempt_us = 0 => duration = 0.
Ok, that follows that d is expected to be 0. And the caller expands to
at least a page.
Not really convinced the assert is worth it, but
Reviewed-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
