reciprocal_divide: update/correction of the algorithm

Jakub Zawadzki noticed that some divisions by reciprocal_divide()
were not correct [1][2], which he could also show with BPF code
after divisions are transformed into reciprocal_value() for runtime
invariance which can be passed to reciprocal_divide() later on;
reverse in BPF dump ended up with a different, off-by-one K in
some situations.

This has been fixed by Eric Dumazet in commit aee636c4809fa5
("bpf: do not use reciprocal divide"). This follow-up patch
improves reciprocal_value() and reciprocal_divide() to work in
all cases by using Granlund and Montgomery method, so that also
future use is safe and without any non-obvious side-effects.
Known problems with the old implementation were that division by 1
always returned 0 and some off-by-ones when the dividend and divisor
where very large. This seemed to not be problematic with its
current users, as far as we can tell. Eric Dumazet checked for
the slab usage, we cannot surely say so in the case of flex_array.
Still, in order to fix that, we propose an extension from the
original implementation from commit 6a2d7a955d8d resp. [3][4],
by using the algorithm proposed in "Division by Invariant Integers
Using Multiplication" [5], Torbjörn Granlund and Peter L.
Montgomery, that is, pseudocode for q = n/d where q, n, d is in
u32 universe:

1) Initialization:

  int l = ceil(log_2 d)
  uword m' = floor((1<<32)*((1<<l)-d)/d)+1
  int sh_1 = min(l,1)
  int sh_2 = max(l-1,0)

2) For q = n/d, all uword:

  uword t = (n*m')>>32
  q = (t+((n-t)>>sh_1))>>sh_2

The assembler implementation from Agner Fog [6] also helped a lot
while implementing. We have tested the implementation on x86_64,
ppc64, i686, s390x; on x86_64/haswell we're still half the latency
compared to normal divide.

Joint work with Daniel Borkmann.

  [1] http://www.wireshark.org/~darkjames/reciprocal-buggy.c
  [2] http://www.wireshark.org/~darkjames/set-and-dump-filter-k-bug.c
  [3] https://gmplib.org/~tege/division-paper.pdf
  [4] http://homepage.cs.uiowa.edu/~jones/bcd/divide.html
  [5] http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.1.2556
  [6] http://www.agner.org/optimize/asmlib.zip

Reported-by: Jakub Zawadzki <darkjames-ws@darkjames.pl>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Austin S Hemmelgarn <ahferroin7@gmail.com>
Cc: linux-kernel@vger.kernel.org
Cc: Jesse Gross <jesse@nicira.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andy Gospodarek <andy@greyhouse.net>
Cc: Veaceslav Falico <vfalico@redhat.com>
Cc: Jay Vosburgh <fubar@us.ibm.com>
Cc: Jakub Zawadzki <darkjames-ws@darkjames.pl>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 30 insertions, 21 deletions

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 3220b488dd1..f100bd958b8 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -79,7 +79,6 @@
 #include <net/pkt_sched.h>
 #include <linux/rculist.h>
 #include <net/flow_keys.h>
-#include <linux/reciprocal_div.h>
 #include "bonding.h"
 #include "bond_3ad.h"
 #include "bond_alb.h"
@@ -3596,8 +3595,9 @@ static void bond_xmit_slave_id(struct bonding *bond, struct sk_buff *skb, int sl
  */
 static u32 bond_rr_gen_slave_id(struct bonding *bond)
 {
-	int packets_per_slave = bond->params.packets_per_slave;
 	u32 slave_id;
+	struct reciprocal_value reciprocal_packets_per_slave;
+	int packets_per_slave = bond->params.packets_per_slave;
 
 	switch (packets_per_slave) {
 	case 0:
@@ -3607,8 +3607,10 @@ static u32 bond_rr_gen_slave_id(struct bonding *bond)
 		slave_id = bond->rr_tx_counter;
 		break;
 	default:
+		reciprocal_packets_per_slave =
+			bond->params.reciprocal_packets_per_slave;
 		slave_id = reciprocal_divide(bond->rr_tx_counter,
-					     packets_per_slave);
+					     reciprocal_packets_per_slave);
 		break;
 	}
 	bond->rr_tx_counter++;
@@ -4343,10 +4345,18 @@ static int bond_check_params(struct bond_params *params)
 	params->resend_igmp = resend_igmp;
 	params->min_links = min_links;
 	params->lp_interval = lp_interval;
-	if (packets_per_slave > 1)
-		params->packets_per_slave = reciprocal_value(packets_per_slave);
-	else
-		params->packets_per_slave = packets_per_slave;
+	params->packets_per_slave = packets_per_slave;
+	if (packets_per_slave > 0) {
+		params->reciprocal_packets_per_slave =
+			reciprocal_value(packets_per_slave);
+	} else {
+		/* reciprocal_packets_per_slave is unused if
+		 * packets_per_slave is 0 or 1, just initialize it
+		 */
+		params->reciprocal_packets_per_slave =
+			(struct reciprocal_value) { 0 };
+	}
+
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
 		params->primary[IFNAMSIZ - 1] = 0;
diff --git a/drivers/net/bonding/bond_netlink.c b/drivers/net/bonding/bond_netlink.c
index 21c648854a8..e8526552790 100644
--- a/drivers/net/bonding/bond_netlink.c
+++ b/drivers/net/bonding/bond_netlink.c
@@ -19,7 +19,6 @@
 #include <linux/if_ether.h>
 #include <net/netlink.h>
 #include <net/rtnetlink.h>
-#include <linux/reciprocal_div.h>
 #include "bonding.h"
 
 int bond_get_slave(struct net_device *slave_dev, struct sk_buff *skb)
@@ -452,9 +451,6 @@ static int bond_fill_info(struct sk_buff *skb,
 		goto nla_put_failure;
 
 	packets_per_slave = bond->params.packets_per_slave;
-	if (packets_per_slave > 1)
-		packets_per_slave = reciprocal_value(packets_per_slave);
-
 	if (nla_put_u32(skb, IFLA_BOND_PACKETS_PER_SLAVE,
 			packets_per_slave))
 		goto nla_put_failure;
diff --git a/drivers/net/bonding/bond_options.c b/drivers/net/bonding/bond_options.c
index 945a6668da8..85e434886f2 100644
--- a/drivers/net/bonding/bond_options.c
+++ b/drivers/net/bonding/bond_options.c
@@ -16,7 +16,6 @@
 #include <linux/netdevice.h>
 #include <linux/rwlock.h>
 #include <linux/rcupdate.h>
-#include <linux/reciprocal_div.h>
 #include "bonding.h"
 
 int bond_option_mode_set(struct bonding *bond, int mode)
@@ -671,11 +670,17 @@ int bond_option_packets_per_slave_set(struct bonding *bond,
 		pr_warn("%s: Warning: packets_per_slave has effect only in balance-rr mode\n",
 			bond->dev->name);
 
-	if (packets_per_slave > 1)
-		bond->params.packets_per_slave =
+	bond->params.packets_per_slave = packets_per_slave;
+	if (packets_per_slave > 0) {
+		bond->params.reciprocal_packets_per_slave =
 			reciprocal_value(packets_per_slave);
-	else
-		bond->params.packets_per_slave = packets_per_slave;
+	} else {
+		/* reciprocal_packets_per_slave is unused if
+		 * packets_per_slave is 0 or 1, just initialize it
+		 */
+		bond->params.reciprocal_packets_per_slave =
+			(struct reciprocal_value) { 0 };
+	}
 
 	return 0;
 }
diff --git a/drivers/net/bonding/bond_sysfs.c b/drivers/net/bonding/bond_sysfs.c
index 011f163c2c6..c083e9a66ec 100644
--- a/drivers/net/bonding/bond_sysfs.c
+++ b/drivers/net/bonding/bond_sysfs.c
@@ -39,7 +39,6 @@
 #include <net/net_namespace.h>
 #include <net/netns/generic.h>
 #include <linux/nsproxy.h>
-#include <linux/reciprocal_div.h>
 
 #include "bonding.h"
 
@@ -1374,10 +1373,6 @@ static ssize_t bonding_show_packets_per_slave(struct device *d,
 {
 	struct bonding *bond = to_bond(d);
 	unsigned int packets_per_slave = bond->params.packets_per_slave;
-
-	if (packets_per_slave > 1)
-		packets_per_slave = reciprocal_value(packets_per_slave);
-
 	return sprintf(buf, "%u\n", packets_per_slave);
 }
 
diff --git a/drivers/net/bonding/bonding.h b/drivers/net/bonding/bonding.h
index 8a935f8f2b3..0a616c41dc9 100644
--- a/drivers/net/bonding/bonding.h
+++ b/drivers/net/bonding/bonding.h
@@ -23,6 +23,8 @@
 #include <linux/netpoll.h>
 #include <linux/inetdevice.h>
 #include <linux/etherdevice.h>
+#include <linux/reciprocal_div.h>
+
 #include "bond_3ad.h"
 #include "bond_alb.h"
 
@@ -171,6 +173,7 @@ struct bond_params {
 	int resend_igmp;
 	int lp_interval;
 	int packets_per_slave;
+	struct reciprocal_value reciprocal_packets_per_slave;
 };
 
 struct bond_parm_tbl {