net_sched: gen_estimator: complete rewrite of rate estimators

1) Old code was hard to maintain, due to complex lock chains.
   (We probably will be able to remove some kfree_rcu() in callers)

2) Using a single timer to update all estimators does not scale.

3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity
   is not supposed to work well)

In this rewrite :

- I removed the RB tree that had to be scanned in
  gen_estimator_active(). qdisc dumps should be much faster.

- Each estimator has its own timer.

- Estimations are maintained in net_rate_estimator structure,
  instead of dirtying the qdisc. Minor, but part of the simplification.

- Reading the estimator uses RCU and a seqcount to provide proper
  support for 32bit kernels.

- We reduce memory need when estimators are not used, since
  we store a pointer, instead of the bytes/packets counters.

- xt_rateest_mt() no longer has to grab a spinlock.
  (In the future, xt_rateest_tg() could be switched to per cpu counters)

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 9 insertions, 11 deletions

diff --git a/net/core/gen_stats.c b/net/core/gen_stats.c
index 508e051304fb..87f28557b329 100644
--- a/net/core/gen_stats.c
+++ b/net/core/gen_stats.c
@@ -194,8 +194,7 @@ EXPORT_SYMBOL(gnet_stats_copy_basic);
 /**
  * gnet_stats_copy_rate_est - copy rate estimator statistics into statistics TLV
  * @d: dumping handle
- * @b: basic statistics
- * @r: rate estimator statistics
+ * @rate_est: rate estimator
  *
  * Appends the rate estimator statistics to the top level TLV created by
  * gnet_stats_start_copy().
@@ -205,18 +204,17 @@ EXPORT_SYMBOL(gnet_stats_copy_basic);
  */
 int
 gnet_stats_copy_rate_est(struct gnet_dump *d,
-			 const struct gnet_stats_basic_packed *b,
-			 struct gnet_stats_rate_est64 *r)
+			 struct net_rate_estimator __rcu **rate_est)
 {
+	struct gnet_stats_rate_est64 sample;
 	struct gnet_stats_rate_est est;
 	int res;
 
-	if (b && !gen_estimator_active(b, r))
+	if (!gen_estimator_read(rate_est, &sample))
 		return 0;
-
-	est.bps = min_t(u64, UINT_MAX, r->bps);
+	est.bps = min_t(u64, UINT_MAX, sample.bps);
 	/* we have some time before reaching 2^32 packets per second */
-	est.pps = r->pps;
+	est.pps = sample.pps;
 
 	if (d->compat_tc_stats) {
 		d->tc_stats.bps = est.bps;
@@ -226,11 +224,11 @@ gnet_stats_copy_rate_est(struct gnet_dump *d,
 	if (d->tail) {
 		res = gnet_stats_copy(d, TCA_STATS_RATE_EST, &est, sizeof(est),
 				      TCA_STATS_PAD);
-		if (res < 0 || est.bps == r->bps)
+		if (res < 0 || est.bps == sample.bps)
 			return res;
 		/* emit 64bit stats only if needed */
-		return gnet_stats_copy(d, TCA_STATS_RATE_EST64, r, sizeof(*r),
-				       TCA_STATS_PAD);
+		return gnet_stats_copy(d, TCA_STATS_RATE_EST64, &sample,
+				       sizeof(sample), TCA_STATS_PAD);
 	}
 
 	return 0;