Do not use or count or use the getworks submitted which are simply testing that pools are still up. This was increasing share leakage and making stats not reflect real work.

cgminer.c

@@ -5936,30 +5936,17 @@ static bool pool_getswork(struct pool *pool)
 	gettimeofday(&tv_getwork_reply, NULL);
 
 	if (val) {
-		struct work *work = make_work();
-		bool rc = work_decode(pool, work, val);
+		struct work work;
+		bool rc = work_decode(pool, &work, val);
 
 		if (rc) {
 			applog(LOG_DEBUG, "Successfully retrieved and deciphered work from pool %u %s",
 			       pool->pool_no, pool->rpc_url);
-			work->pool = pool;
-			work->rolltime = rolltime;
-			memcpy(&(work->tv_getwork), &tv_getwork, sizeof(struct timeval));
-			memcpy(&(work->tv_getwork_reply), &tv_getwork_reply, sizeof(struct timeval));
-			work->getwork_mode = GETWORK_MODE_TESTPOOL;
-			calc_diff(work, 0);
-			applog(LOG_DEBUG, "Pushing pool_getswork work for pool %d", pool->pool_no);
-
-			tq_push(thr_info[stage_thr_id].q, work);
-			total_getworks++;
-			pool->getwork_requested++;
-			ret = true;
 			gettimeofday(&pool->tv_idle, NULL);
 			ret = true;
 		} else {
 			applog(LOG_DEBUG, "Successfully retrieved but FAILED to decipher work from pool %u %s",
 			       pool->pool_no, pool->rpc_url);
-			free_work(work);
 		}
 		json_decref(val);
 	} else {