net/dim: Fix fixpoint divide exception in net_dim_stats_compare

Helmut reported a bug about devision by zero while
running traffic and doing physical cable pull test.

When the cable unplugged the ppms become zero, so when
dividing the current ppms by the previous ppms in the
next dim iteration there is devision by zero.

This patch prevent this division for both ppms and epms.

Fixes: c3164d2f ("net/mlx5e: Added BW check for DIM decision mechanism")
Fixes: 4c4dbb4a ("net/mlx5e: Move dynamic interrupt coalescing code to include/linux")
Reported-by: Helmut Grauer <helmut.grauer@de.ibm.com>
Signed-off-by: Talat Batheesh <talatb@mellanox.com>
Signed-off-by: Tal Gilboa <talgi@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/net_dim.h

@@ -244,10 +244,17 @@ static inline int net_dim_stats_compare(struct net_dim_stats *curr,
 		return (curr->bpms > prev->bpms) ? NET_DIM_STATS_BETTER :
 						   NET_DIM_STATS_WORSE;
 
+	if (!prev->ppms)
+		return curr->ppms ? NET_DIM_STATS_BETTER :
+				    NET_DIM_STATS_SAME;
+
 	if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
 		return (curr->ppms > prev->ppms) ? NET_DIM_STATS_BETTER :
 						   NET_DIM_STATS_WORSE;
 
+	if (!prev->epms)
+		return NET_DIM_STATS_SAME;
+
 	if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
 		return (curr->epms < prev->epms) ? NET_DIM_STATS_BETTER :
 						   NET_DIM_STATS_WORSE;