Skip to contentSkip to conversation

nexthop: Fix performance regression in nexthop deletion

PATCH Applied Ido Schimmel 5 participants 5 messages Oct 16, 2020 – Oct 20, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nexthop: Fix performance regression in nexthop deletion

From: Ido Schimmel <[email protected]>

While insertion of 16k nexthops all using the same netdev ('dummy10') takes less than a second, deletion takes about 130 seconds:

# time -p ip -b nexthop.batch
real 0.29
user 0.01
sys 0.15
# time -p ip link set dev dummy10 down
real 131.03
user 0.06
sys 0.52

This is because of repeated calls to synchronize_rcu() whenever a nexthop is removed from a nexthop group:

# /usr/share/bcc/tools/offcputime -p `pgrep -nx ip` -K
...
    b'finish_task_switch'
    b'schedule'
    b'schedule_timeout'
    b'wait_for_completion'
    b'__wait_rcu_gp'
    b'synchronize_rcu.part.0'
    b'synchronize_rcu'
    b'__remove_nexthop'
    b'remove_nexthop'
    b'nexthop_flush_dev'
    b'nh_netdev_event'
    b'raw_notifier_call_chain'
    b'call_netdevice_notifiers_info'
    b'__dev_notify_flags'
    b'dev_change_flags'
    b'do_setlink'
    b'__rtnl_newlink'
    b'rtnl_newlink'
    b'rtnetlink_rcv_msg'
    b'netlink_rcv_skb'
    b'rtnetlink_rcv'
    b'netlink_unicast'
    b'netlink_sendmsg'
    b'____sys_sendmsg'
    b'___sys_sendmsg'
    b'__sys_sendmsg'
    b'__x64_sys_sendmsg'
    b'do_syscall_64'
    b'entry_SYSCALL_64_after_hwframe'
    -                ip (277)
        126554955

Since nexthops are always deleted under RTNL, synchronize_net() can be used instead. It will call synchronize_rcu_expedited() which only blocks for several microseconds as opposed to multiple milliseconds like synchronize_rcu().

With this patch deletion of 16k nexthops takes less than a second:

# time -p ip link set dev dummy10 down
real 0.12
user 0.00
sys 0.04

Tested with fib_nexthops.sh which includes torture tests that prompted the initial change:

# ./fib_nexthops.sh
...
Tests passed: 134
Tests failed:   0
Fixes 90f33bffa382 ("nexthops: don't modify published nexthop groups")
---
 net/ipv4/nexthop.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv4/nexthop.c +1 -1
index 8c0f17c6863c..0dc43ad28eb9 100644
@@ -845,7 +845,7 @@ static void remove_nexthop_from_groups(struct net *net, struct nexthop *nh,
 		remove_nh_grp_entry(net, nhge, nlinfo);
 
 	/* make sure all see the newly published array before releasing rtnl */
-	synchronize_rcu();
+	synchronize_net();
 }
 
 static void remove_nexthop_group(struct nexthop *nh, struct nl_info *nlinfo)
Signature · 1 line
2.26.2
Reviewed-by on v1 1/1
#1 lore raw
Quoted 14 lines · Ido Schimmel

From: Ido Schimmel <[email protected]>

While insertion of 16k nexthops all using the same netdev ('dummy10') takes less than a second, deletion takes about 130 seconds:

# time -p ip -b nexthop.batch
real 0.29
user 0.01
sys 0.15
# time -p ip link set dev dummy10 down
real 131.03
user 0.06
sys 0.52

snip...

Quoted 11 lines

Since nexthops are always deleted under RTNL, synchronize_net() can be used instead. It will call synchronize_rcu_expedited() which only blocks for several microseconds as opposed to multiple milliseconds like synchronize_rcu().

With this patch deletion of 16k nexthops takes less than a second:

# time -p ip link set dev dummy10 down
real 0.12
user 0.00
sys 0.04

That's a nice result! Well done! I can't really speak to whether or not there is any horrible side effect of using synchronize_rcu_expedited(), but FWIW:

Quoted 11 lines

Tested with fib_nexthops.sh which includes torture tests that prompted the initial change:

# ./fib_nexthops.sh
...
Tests passed: 134
Tests failed:   0
Fixes 90f33bffa382 ("nexthops: don't modify published nexthop groups")

Nice fix, good commit message, thanks!

Quoted 17 lines
---
 net/ipv4/nexthop.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv4/nexthop.c +1 -1
index 8c0f17c6863c..0dc43ad28eb9 100644
@@ -845,7 +845,7 @@ static void remove_nexthop_from_groups(struct net *net, struct nexthop *nh,
 		remove_nh_grp_entry(net, nhge, nlinfo);
 
 	/* make sure all see the newly published array before releasing rtnl */
-	synchronize_rcu();
+	synchronize_net();
 }
 
 static void remove_nexthop_group(struct nexthop *nh, struct nl_info *nlinfo)
Reviewed-by on v1 1/1
#2 lore raw
In reply to Ido Schimmel
Quoted 72 lines · Ido Schimmel

From: Ido Schimmel <[email protected]>

While insertion of 16k nexthops all using the same netdev ('dummy10') takes less than a second, deletion takes about 130 seconds:

# time -p ip -b nexthop.batch
real 0.29
user 0.01
sys 0.15
# time -p ip link set dev dummy10 down
real 131.03
user 0.06
sys 0.52

This is because of repeated calls to synchronize_rcu() whenever a nexthop is removed from a nexthop group:

# /usr/share/bcc/tools/offcputime -p `pgrep -nx ip` -K
...
    b'finish_task_switch'
    b'schedule'
    b'schedule_timeout'
    b'wait_for_completion'
    b'__wait_rcu_gp'
    b'synchronize_rcu.part.0'
    b'synchronize_rcu'
    b'__remove_nexthop'
    b'remove_nexthop'
    b'nexthop_flush_dev'
    b'nh_netdev_event'
    b'raw_notifier_call_chain'
    b'call_netdevice_notifiers_info'
    b'__dev_notify_flags'
    b'dev_change_flags'
    b'do_setlink'
    b'__rtnl_newlink'
    b'rtnl_newlink'
    b'rtnetlink_rcv_msg'
    b'netlink_rcv_skb'
    b'rtnetlink_rcv'
    b'netlink_unicast'
    b'netlink_sendmsg'
    b'____sys_sendmsg'
    b'___sys_sendmsg'
    b'__sys_sendmsg'
    b'__x64_sys_sendmsg'
    b'do_syscall_64'
    b'entry_SYSCALL_64_after_hwframe'
    -                ip (277)
        126554955

Since nexthops are always deleted under RTNL, synchronize_net() can be used instead. It will call synchronize_rcu_expedited() which only blocks for several microseconds as opposed to multiple milliseconds like synchronize_rcu().

With this patch deletion of 16k nexthops takes less than a second:

# time -p ip link set dev dummy10 down
real 0.12
user 0.00
sys 0.04

Tested with fib_nexthops.sh which includes torture tests that prompted the initial change:

# ./fib_nexthops.sh
...
Tests passed: 134
Tests failed:   0
Quoted 6 lines
Fixes 90f33bffa382 ("nexthops: don't modify published nexthop groups")
---
 net/ipv4/nexthop.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Thanks for finding this, Ido.

Acked-by on v1 1/1
#3 lore raw
In reply to Ido Schimmel
Quoted 72 lines · Ido Schimmel

From: Ido Schimmel <[email protected]>

While insertion of 16k nexthops all using the same netdev ('dummy10') takes less than a second, deletion takes about 130 seconds:

# time -p ip -b nexthop.batch
real 0.29
user 0.01
sys 0.15
# time -p ip link set dev dummy10 down
real 131.03
user 0.06
sys 0.52

This is because of repeated calls to synchronize_rcu() whenever a nexthop is removed from a nexthop group:

# /usr/share/bcc/tools/offcputime -p `pgrep -nx ip` -K
...
    b'finish_task_switch'
    b'schedule'
    b'schedule_timeout'
    b'wait_for_completion'
    b'__wait_rcu_gp'
    b'synchronize_rcu.part.0'
    b'synchronize_rcu'
    b'__remove_nexthop'
    b'remove_nexthop'
    b'nexthop_flush_dev'
    b'nh_netdev_event'
    b'raw_notifier_call_chain'
    b'call_netdevice_notifiers_info'
    b'__dev_notify_flags'
    b'dev_change_flags'
    b'do_setlink'
    b'__rtnl_newlink'
    b'rtnl_newlink'
    b'rtnetlink_rcv_msg'
    b'netlink_rcv_skb'
    b'rtnetlink_rcv'
    b'netlink_unicast'
    b'netlink_sendmsg'
    b'____sys_sendmsg'
    b'___sys_sendmsg'
    b'__sys_sendmsg'
    b'__x64_sys_sendmsg'
    b'do_syscall_64'
    b'entry_SYSCALL_64_after_hwframe'
    -                ip (277)
        126554955

Since nexthops are always deleted under RTNL, synchronize_net() can be used instead. It will call synchronize_rcu_expedited() which only blocks for several microseconds as opposed to multiple milliseconds like synchronize_rcu().

With this patch deletion of 16k nexthops takes less than a second:

# time -p ip link set dev dummy10 down
real 0.12
user 0.00
sys 0.04

Tested with fib_nexthops.sh which includes torture tests that prompted the initial change:

# ./fib_nexthops.sh
...
Tests passed: 134
Tests failed:   0
Quoted 7 lines
Fixes 90f33bffa382 ("nexthops: don't modify published nexthop groups")
---
 net/ipv4/nexthop.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Looks good.

maintainer Applied
#4 lore raw
In reply to Ido Schimmel
Quoted 4 lines · On Fri, 16 Oct 2020 20:29:14 +0300 Ido Schimmel wrote:

From: Ido Schimmel <[email protected]>

While insertion of 16k nexthops all using the same netdev ('dummy10') takes less than a second, deletion takes about 130 seconds:

Applied, thank you!