drm/amdgpu: fix fence calculation (v2)

The proper metric for fence utilization over several
contexts is an harmonic mean, but such calculation is
prohibitive in kernel space, so the code approximates it.

Because the approximation diverges when one context has a
very small ratio compared with the other context, this change
filter out ratios smaller that 0.01%

v2: make the fence calculation static and initialize variables
within that function

v3: Fix warnings (Alex)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: David M Nieto <david.nieto@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210513174539.27409-2-david.nieto@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c

@@ -652,12 +652,14 @@ void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
 	mutex_destroy(&mgr->lock);
 }
 
-void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, struct amdgpu_ctx_entity *centity,
-		ktime_t *total, ktime_t *max)
+static void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx,
+		struct amdgpu_ctx_entity *centity, ktime_t *total, ktime_t *max)
 {
 	ktime_t now, t1;
 	uint32_t i;
 
+	*total = *max = 0;
+
 	now = ktime_get();
 	for (i = 0; i < amdgpu_sched_jobs; i++) {
 		struct dma_fence *fence;
@@ -703,11 +705,22 @@ ktime_t amdgpu_ctx_mgr_fence_usage(struct amdgpu_ctx_mgr *mgr, uint32_t hwip,
 	idp = &mgr->ctx_handles;
 	mutex_lock(&mgr->lock);
 	idr_for_each_entry(idp, ctx, id) {
+		ktime_t ttotal, tmax;
+
 		if (!ctx->entities[hwip][idx])
 			continue;
 
 		centity = ctx->entities[hwip][idx];
-		amdgpu_ctx_fence_time(ctx, centity, &total, &max);
+		amdgpu_ctx_fence_time(ctx, centity, &ttotal, &tmax);
+
+		/* Harmonic mean approximation diverges for very small
+		 * values. If ratio < 0.01% ignore
+		 */
+		if (AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(tmax, ttotal))
+			continue;
+
+		total = ktime_add(total, ttotal);
+		max = ktime_after(tmax, max) ? tmax : max;
 	}
 
 	mutex_unlock(&mgr->lock);

drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h

@@ -30,6 +30,7 @@ struct drm_file;
 struct amdgpu_fpriv;
 
 #define AMDGPU_MAX_ENTITY_NUM 4
+#define AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(max, total) ((max) > 16384ULL*(total))
 
 struct amdgpu_ctx_entity {
 	uint64_t		sequence;
@@ -89,6 +90,4 @@ long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout);
 void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
 ktime_t amdgpu_ctx_mgr_fence_usage(struct amdgpu_ctx_mgr *mgr, uint32_t hwip,
 		uint32_t idx, uint64_t *elapsed);
-void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, struct amdgpu_ctx_entity *centity,
-		ktime_t *total, ktime_t *max);
 #endif