* [PATCH 5/9] ipvs: fix buffer overflow with sync daemon and service
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Julian Anastasov <ja@ssi.bg>
syzkaller reports for buffer overflow for interface name
when starting sync daemons [1]
What we do is that we copy user structure into larger stack
buffer but later we search NUL past the stack buffer.
The same happens for sched_name when adding/editing virtual server.
We are restricted by IP_VS_SCHEDNAME_MAXLEN and IP_VS_IFNAME_MAXLEN
being used as size in include/uapi/linux/ip_vs.h, so they
include the space for NUL.
As using strlcpy is wrong for unsafe source, replace it with
strscpy and add checks to return EINVAL if source string is not
NUL-terminated. The incomplete strlcpy fix comes from 2.6.13.
For the netlink interface reduce the len parameter for
IPVS_DAEMON_ATTR_MCAST_IFN and IPVS_SVC_ATTR_SCHED_NAME,
so that we get proper EINVAL.
[1]
kernel BUG at lib/string.c:1052!
invalid opcode: 0000 [#1] SMP KASAN
Dumping ftrace buffer:
(ftrace buffer empty)
Modules linked in:
CPU: 1 PID: 373 Comm: syz-executor936 Not tainted 4.17.0-rc4+ #45
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
RIP: 0010:fortify_panic+0x13/0x20 lib/string.c:1051
RSP: 0018:ffff8801c976f800 EFLAGS: 00010282
RAX: 0000000000000022 RBX: 0000000000000040 RCX: 0000000000000000
RDX: 0000000000000022 RSI: ffffffff8160f6f1 RDI: ffffed00392edef6
RBP: ffff8801c976f800 R08: ffff8801cf4c62c0 R09: ffffed003b5e4fb0
R10: ffffed003b5e4fb0 R11: ffff8801daf27d87 R12: ffff8801c976fa20
R13: ffff8801c976fae4 R14: ffff8801c976fae0 R15: 000000000000048b
FS: 00007fd99f75e700(0000) GS:ffff8801daf00000(0000)
knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000200001c0 CR3: 00000001d6843000 CR4: 00000000001406e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
strlen include/linux/string.h:270 [inline]
strlcpy include/linux/string.h:293 [inline]
do_ip_vs_set_ctl+0x31c/0x1d00 net/netfilter/ipvs/ip_vs_ctl.c:2388
nf_sockopt net/netfilter/nf_sockopt.c:106 [inline]
nf_setsockopt+0x7d/0xd0 net/netfilter/nf_sockopt.c:115
ip_setsockopt+0xd8/0xf0 net/ipv4/ip_sockglue.c:1253
udp_setsockopt+0x62/0xa0 net/ipv4/udp.c:2487
ipv6_setsockopt+0x149/0x170 net/ipv6/ipv6_sockglue.c:917
tcp_setsockopt+0x93/0xe0 net/ipv4/tcp.c:3057
sock_common_setsockopt+0x9a/0xe0 net/core/sock.c:3046
__sys_setsockopt+0x1bd/0x390 net/socket.c:1903
__do_sys_setsockopt net/socket.c:1914 [inline]
__se_sys_setsockopt net/socket.c:1911 [inline]
__x64_sys_setsockopt+0xbe/0x150 net/socket.c:1911
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x447369
RSP: 002b:00007fd99f75dda8 EFLAGS: 00000246 ORIG_RAX: 0000000000000036
RAX: ffffffffffffffda RBX: 00000000006e39e4 RCX: 0000000000447369
RDX: 000000000000048b RSI: 0000000000000000 RDI: 0000000000000003
RBP: 0000000000000000 R08: 0000000000000018 R09: 0000000000000000
R10: 00000000200001c0 R11: 0000000000000246 R12: 00000000006e39e0
R13: 75a1ff93f0896195 R14: 6f745f3168746576 R15: 0000000000000001
Code: 08 5b 41 5c 41 5d 41 5e 41 5f 5d c3 0f 0b 48 89 df e8 d2 8f 48 fa eb
de 55 48 89 fe 48 c7 c7 60 65 64 88 48 89 e5 e8 91 dd f3 f9 <0f> 0b 90 90
90 90 90 90 90 90 90 90 90 55 48 89 e5 41 57 41 56
RIP: fortify_panic+0x13/0x20 lib/string.c:1051 RSP: ffff8801c976f800
Reported-and-tested-by: syzbot+aac887f77319868646df@syzkaller.appspotmail.com
Fixes: e4ff67513096 ("ipvs: add sync_maxlen parameter for the sync daemon")
Fixes: 4da62fc70d7c ("[IPVS]: Fix for overflows")
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/ipvs/ip_vs_ctl.c | 21 +++++++++++++++------
1 file changed, 15 insertions(+), 6 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index f36098887ad0..3ecca0616d8c 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -2381,8 +2381,10 @@ do_ip_vs_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
struct ipvs_sync_daemon_cfg cfg;
memset(&cfg, 0, sizeof(cfg));
- strlcpy(cfg.mcast_ifn, dm->mcast_ifn,
- sizeof(cfg.mcast_ifn));
+ ret = -EINVAL;
+ if (strscpy(cfg.mcast_ifn, dm->mcast_ifn,
+ sizeof(cfg.mcast_ifn)) <= 0)
+ goto out_dec;
cfg.syncid = dm->syncid;
ret = start_sync_thread(ipvs, &cfg, dm->state);
} else {
@@ -2420,12 +2422,19 @@ do_ip_vs_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
}
}
+ if ((cmd == IP_VS_SO_SET_ADD || cmd == IP_VS_SO_SET_EDIT) &&
+ strnlen(usvc.sched_name, IP_VS_SCHEDNAME_MAXLEN) ==
+ IP_VS_SCHEDNAME_MAXLEN) {
+ ret = -EINVAL;
+ goto out_unlock;
+ }
+
/* Check for valid protocol: TCP or UDP or SCTP, even for fwmark!=0 */
if (usvc.protocol != IPPROTO_TCP && usvc.protocol != IPPROTO_UDP &&
usvc.protocol != IPPROTO_SCTP) {
- pr_err("set_ctl: invalid protocol: %d %pI4:%d %s\n",
+ pr_err("set_ctl: invalid protocol: %d %pI4:%d\n",
usvc.protocol, &usvc.addr.ip,
- ntohs(usvc.port), usvc.sched_name);
+ ntohs(usvc.port));
ret = -EFAULT;
goto out_unlock;
}
@@ -2847,7 +2856,7 @@ static const struct nla_policy ip_vs_cmd_policy[IPVS_CMD_ATTR_MAX + 1] = {
static const struct nla_policy ip_vs_daemon_policy[IPVS_DAEMON_ATTR_MAX + 1] = {
[IPVS_DAEMON_ATTR_STATE] = { .type = NLA_U32 },
[IPVS_DAEMON_ATTR_MCAST_IFN] = { .type = NLA_NUL_STRING,
- .len = IP_VS_IFNAME_MAXLEN },
+ .len = IP_VS_IFNAME_MAXLEN - 1 },
[IPVS_DAEMON_ATTR_SYNC_ID] = { .type = NLA_U32 },
[IPVS_DAEMON_ATTR_SYNC_MAXLEN] = { .type = NLA_U16 },
[IPVS_DAEMON_ATTR_MCAST_GROUP] = { .type = NLA_U32 },
@@ -2865,7 +2874,7 @@ static const struct nla_policy ip_vs_svc_policy[IPVS_SVC_ATTR_MAX + 1] = {
[IPVS_SVC_ATTR_PORT] = { .type = NLA_U16 },
[IPVS_SVC_ATTR_FWMARK] = { .type = NLA_U32 },
[IPVS_SVC_ATTR_SCHED_NAME] = { .type = NLA_NUL_STRING,
- .len = IP_VS_SCHEDNAME_MAXLEN },
+ .len = IP_VS_SCHEDNAME_MAXLEN - 1 },
[IPVS_SVC_ATTR_PE_NAME] = { .type = NLA_NUL_STRING,
.len = IP_VS_PENAME_MAXLEN },
[IPVS_SVC_ATTR_FLAGS] = { .type = NLA_BINARY,
--
2.11.0
^ permalink raw reply related
* [PATCH 9/9] netfilter: nf_tables: increase nft_counters_enabled in nft_chain_stats_replace()
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Taehee Yoo <ap420073@gmail.com>
When a chain is updated, a counter can be attached. if so,
the nft_counters_enabled should be increased.
test commands:
%nft add table ip filter
%nft add chain ip filter input { type filter hook input priority 4\; }
%iptables-compat -Z input
%nft delete chain ip filter input
we can see below messages.
[ 286.443720] jump label: negative count!
[ 286.448278] WARNING: CPU: 0 PID: 1459 at kernel/jump_label.c:197 __static_key_slow_dec_cpuslocked+0x6f/0xf0
[ 286.449144] Modules linked in: nf_tables nfnetlink ip_tables x_tables
[ 286.449144] CPU: 0 PID: 1459 Comm: nft Tainted: G W 4.17.0-rc2+ #12
[ 286.449144] RIP: 0010:__static_key_slow_dec_cpuslocked+0x6f/0xf0
[ 286.449144] RSP: 0018:ffff88010e5176f0 EFLAGS: 00010286
[ 286.449144] RAX: 000000000000001b RBX: ffffffffc0179500 RCX: ffffffffb8a82522
[ 286.449144] RDX: 0000000000000001 RSI: 0000000000000008 RDI: ffff88011b7e5eac
[ 286.449144] RBP: 0000000000000000 R08: ffffed00236fce5c R09: ffffed00236fce5b
[ 286.449144] R10: ffffffffc0179503 R11: ffffed00236fce5c R12: 0000000000000000
[ 286.449144] R13: ffff88011a28e448 R14: ffff88011a28e470 R15: dffffc0000000000
[ 286.449144] FS: 00007f0384328700(0000) GS:ffff88011b600000(0000) knlGS:0000000000000000
[ 286.449144] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 286.449144] CR2: 00007f038394bf10 CR3: 0000000104a86000 CR4: 00000000001006f0
[ 286.449144] Call Trace:
[ 286.449144] static_key_slow_dec+0x6a/0x70
[ 286.449144] nf_tables_chain_destroy+0x19d/0x210 [nf_tables]
[ 286.449144] nf_tables_commit+0x1891/0x1c50 [nf_tables]
[ 286.449144] nfnetlink_rcv+0x1148/0x13d0 [nfnetlink]
[ ... ]
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nf_tables_api.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 2bdc8767aa40..501e48a7965b 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1298,8 +1298,10 @@ static void nft_chain_stats_replace(struct nft_base_chain *chain,
rcu_assign_pointer(chain->stats, newstats);
synchronize_rcu();
free_percpu(oldstats);
- } else
+ } else {
rcu_assign_pointer(chain->stats, newstats);
+ static_branch_inc(&nft_counters_enabled);
+ }
}
static void nf_tables_chain_destroy(struct nft_ctx *ctx)
--
2.11.0
^ permalink raw reply related
* [PATCH 6/9] netfilter: provide correct argument to nla_strlcpy()
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Eric Dumazet <edumazet@google.com>
Recent patch forgot to remove nla_data(), upsetting syzkaller a bit.
BUG: KASAN: slab-out-of-bounds in nla_strlcpy+0x13d/0x150 lib/nlattr.c:314
Read of size 1 at addr ffff8801ad1f4fdd by task syz-executor189/4509
CPU: 1 PID: 4509 Comm: syz-executor189 Not tainted 4.17.0-rc6+ #62
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x1b9/0x294 lib/dump_stack.c:113
print_address_description+0x6c/0x20b mm/kasan/report.c:256
kasan_report_error mm/kasan/report.c:354 [inline]
kasan_report.cold.7+0x242/0x2fe mm/kasan/report.c:412
__asan_report_load1_noabort+0x14/0x20 mm/kasan/report.c:430
nla_strlcpy+0x13d/0x150 lib/nlattr.c:314
nfnl_acct_new+0x574/0xc50 net/netfilter/nfnetlink_acct.c:118
nfnetlink_rcv_msg+0xdb5/0xff0 net/netfilter/nfnetlink.c:212
netlink_rcv_skb+0x172/0x440 net/netlink/af_netlink.c:2448
nfnetlink_rcv+0x1fe/0x1ba0 net/netfilter/nfnetlink.c:513
netlink_unicast_kernel net/netlink/af_netlink.c:1310 [inline]
netlink_unicast+0x58b/0x740 net/netlink/af_netlink.c:1336
netlink_sendmsg+0x9f0/0xfa0 net/netlink/af_netlink.c:1901
sock_sendmsg_nosec net/socket.c:629 [inline]
sock_sendmsg+0xd5/0x120 net/socket.c:639
sock_write_iter+0x35a/0x5a0 net/socket.c:908
call_write_iter include/linux/fs.h:1784 [inline]
new_sync_write fs/read_write.c:474 [inline]
__vfs_write+0x64d/0x960 fs/read_write.c:487
vfs_write+0x1f8/0x560 fs/read_write.c:549
ksys_write+0xf9/0x250 fs/read_write.c:598
__do_sys_write fs/read_write.c:610 [inline]
__se_sys_write fs/read_write.c:607 [inline]
__x64_sys_write+0x73/0xb0 fs/read_write.c:607
Fixes: 4e09fc873d92 ("netfilter: prefer nla_strlcpy for dealing with NLA_STRING attributes")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Florian Westphal <fw@strlen.de>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nfnetlink_acct.c | 2 +-
net/netfilter/nfnetlink_cthelper.c | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/net/netfilter/nfnetlink_acct.c b/net/netfilter/nfnetlink_acct.c
index 6ddf89183e7b..a0e5adf0b3b6 100644
--- a/net/netfilter/nfnetlink_acct.c
+++ b/net/netfilter/nfnetlink_acct.c
@@ -115,7 +115,7 @@ static int nfnl_acct_new(struct net *net, struct sock *nfnl,
nfacct->flags = flags;
}
- nla_strlcpy(nfacct->name, nla_data(tb[NFACCT_NAME]), NFACCT_NAME_MAX);
+ nla_strlcpy(nfacct->name, tb[NFACCT_NAME], NFACCT_NAME_MAX);
if (tb[NFACCT_BYTES]) {
atomic64_set(&nfacct->bytes,
diff --git a/net/netfilter/nfnetlink_cthelper.c b/net/netfilter/nfnetlink_cthelper.c
index fa026b269b36..cb5b5f207777 100644
--- a/net/netfilter/nfnetlink_cthelper.c
+++ b/net/netfilter/nfnetlink_cthelper.c
@@ -150,7 +150,7 @@ nfnl_cthelper_expect_policy(struct nf_conntrack_expect_policy *expect_policy,
return -EINVAL;
nla_strlcpy(expect_policy->name,
- nla_data(tb[NFCTH_POLICY_NAME]), NF_CT_HELPER_NAME_LEN);
+ tb[NFCTH_POLICY_NAME], NF_CT_HELPER_NAME_LEN);
expect_policy->max_expected =
ntohl(nla_get_be32(tb[NFCTH_POLICY_EXPECT_MAX]));
if (expect_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
@@ -235,7 +235,7 @@ nfnl_cthelper_create(const struct nlattr * const tb[],
goto err1;
nla_strlcpy(helper->name,
- nla_data(tb[NFCTH_NAME]), NF_CT_HELPER_NAME_LEN);
+ tb[NFCTH_NAME], NF_CT_HELPER_NAME_LEN);
size = ntohl(nla_get_be32(tb[NFCTH_PRIV_DATA_LEN]));
if (size > FIELD_SIZEOF(struct nf_conn_help, data)) {
ret = -ENOMEM;
--
2.11.0
^ permalink raw reply related
* [PATCH 8/9] netfilter: nf_tables: fix NULL-ptr in nf_tables_dump_obj()
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Taehee Yoo <ap420073@gmail.com>
The table field in nft_obj_filter is not an array. In order to check
tablename, we should check if the pointer is set.
Test commands:
%nft add table ip filter
%nft add counter ip filter ct1
%nft reset counters
Splat looks like:
[ 306.510504] kasan: CONFIG_KASAN_INLINE enabled
[ 306.516184] kasan: GPF could be caused by NULL-ptr deref or user memory access
[ 306.524775] general protection fault: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
[ 306.528284] Modules linked in: nft_objref nft_counter nf_tables nfnetlink ip_tables x_tables
[ 306.528284] CPU: 0 PID: 1488 Comm: nft Not tainted 4.17.0-rc4+ #17
[ 306.528284] Hardware name: To be filled by O.E.M. To be filled by O.E.M./Aptio CRB, BIOS 5.6.5 07/08/2015
[ 306.528284] RIP: 0010:nf_tables_dump_obj+0x52c/0xa70 [nf_tables]
[ 306.528284] RSP: 0018:ffff8800b6cb7520 EFLAGS: 00010246
[ 306.528284] RAX: 0000000000000000 RBX: ffff8800b6c49820 RCX: 0000000000000000
[ 306.528284] RDX: 0000000000000000 RSI: dffffc0000000000 RDI: ffffed0016d96e9a
[ 306.528284] RBP: ffff8800b6cb75c0 R08: ffffed00236fce7c R09: ffffed00236fce7b
[ 306.528284] R10: ffffffff9f6241e8 R11: ffffed00236fce7c R12: ffff880111365108
[ 306.528284] R13: 0000000000000000 R14: ffff8800b6c49860 R15: ffff8800b6c49860
[ 306.528284] FS: 00007f838b007700(0000) GS:ffff88011b600000(0000) knlGS:0000000000000000
[ 306.528284] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 306.528284] CR2: 00007ffeafabcf78 CR3: 00000000b6cbe000 CR4: 00000000001006f0
[ 306.528284] Call Trace:
[ 306.528284] netlink_dump+0x470/0xa20
[ 306.528284] __netlink_dump_start+0x5ae/0x690
[ 306.528284] ? nf_tables_getobj+0x1b3/0x740 [nf_tables]
[ 306.528284] nf_tables_getobj+0x2f5/0x740 [nf_tables]
[ 306.528284] ? nft_obj_notify+0x100/0x100 [nf_tables]
[ 306.528284] ? nf_tables_getobj+0x740/0x740 [nf_tables]
[ 306.528284] ? nf_tables_dump_flowtable_done+0x70/0x70 [nf_tables]
[ 306.528284] ? nft_obj_notify+0x100/0x100 [nf_tables]
[ 306.528284] nfnetlink_rcv_msg+0x8ff/0x932 [nfnetlink]
[ 306.528284] ? nfnetlink_rcv_msg+0x216/0x932 [nfnetlink]
[ 306.528284] netlink_rcv_skb+0x1c9/0x2f0
[ 306.528284] ? nfnetlink_bind+0x1d0/0x1d0 [nfnetlink]
[ 306.528284] ? debug_check_no_locks_freed+0x270/0x270
[ 306.528284] ? netlink_ack+0x7a0/0x7a0
[ 306.528284] ? ns_capable_common+0x6e/0x110
[ ... ]
Fixes: e46abbcc05aa8 ("netfilter: nf_tables: Allow table names of up to 255 chars")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nf_tables_api.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 91e80aa852d6..2bdc8767aa40 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -4706,7 +4706,7 @@ static int nf_tables_dump_obj(struct sk_buff *skb, struct netlink_callback *cb)
if (idx > s_idx)
memset(&cb->args[1], 0,
sizeof(cb->args) - sizeof(cb->args[0]));
- if (filter && filter->table[0] &&
+ if (filter && filter->table &&
strcmp(filter->table, table->name))
goto cont;
if (filter &&
@@ -5380,7 +5380,7 @@ static int nf_tables_dump_flowtable(struct sk_buff *skb,
if (idx > s_idx)
memset(&cb->args[1], 0,
sizeof(cb->args) - sizeof(cb->args[0]));
- if (filter && filter->table[0] &&
+ if (filter && filter->table &&
strcmp(filter->table, table->name))
goto cont;
--
2.11.0
^ permalink raw reply related
* [PATCH 7/9] netfilter: nf_tables: disable preemption in nft_update_chain_stats()
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
This patch fixes the following splat.
[118709.054937] BUG: using smp_processor_id() in preemptible [00000000] code: test/1571
[118709.054970] caller is nft_update_chain_stats.isra.4+0x53/0x97 [nf_tables]
[118709.054980] CPU: 2 PID: 1571 Comm: test Not tainted 4.17.0-rc6+ #335
[...]
[118709.054992] Call Trace:
[118709.055011] dump_stack+0x5f/0x86
[118709.055026] check_preemption_disabled+0xd4/0xe4
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nf_tables_core.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/netfilter/nf_tables_core.c b/net/netfilter/nf_tables_core.c
index 942702a2776f..40e744572283 100644
--- a/net/netfilter/nf_tables_core.c
+++ b/net/netfilter/nf_tables_core.c
@@ -126,15 +126,15 @@ static noinline void nft_update_chain_stats(const struct nft_chain *chain,
if (!base_chain->stats)
return;
+ local_bh_disable();
stats = this_cpu_ptr(rcu_dereference(base_chain->stats));
if (stats) {
- local_bh_disable();
u64_stats_update_begin(&stats->syncp);
stats->pkts++;
stats->bytes += pkt->skb->len;
u64_stats_update_end(&stats->syncp);
- local_bh_enable();
}
+ local_bh_enable();
}
struct nft_jumpstack {
--
2.11.0
^ permalink raw reply related
* [PATCH 4/9] netfilter: nft_limit: fix packet ratelimiting
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
Credit calculations for the packet ratelimiting are not correct, as per
the applied ratelimit of 25/second and burst 8, a total of 33 packets
should have been accepted. This is true in iptables(33) but not in
nftables (~65). For packet ratelimiting, use:
div_u64(limit->nsecs, limit->rate) * limit->burst;
to calculate credit, just like in iptables' xt_limit does.
Moreover, use default burst in iptables, users are expecting similar
behaviour.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nft_limit.c | 38 ++++++++++++++++++++++++--------------
1 file changed, 24 insertions(+), 14 deletions(-)
diff --git a/net/netfilter/nft_limit.c b/net/netfilter/nft_limit.c
index a9fc298ef4c3..72f13a1144dd 100644
--- a/net/netfilter/nft_limit.c
+++ b/net/netfilter/nft_limit.c
@@ -51,10 +51,13 @@ static inline bool nft_limit_eval(struct nft_limit *limit, u64 cost)
return !limit->invert;
}
+/* Use same default as in iptables. */
+#define NFT_LIMIT_PKT_BURST_DEFAULT 5
+
static int nft_limit_init(struct nft_limit *limit,
- const struct nlattr * const tb[])
+ const struct nlattr * const tb[], bool pkts)
{
- u64 unit;
+ u64 unit, tokens;
if (tb[NFTA_LIMIT_RATE] == NULL ||
tb[NFTA_LIMIT_UNIT] == NULL)
@@ -68,18 +71,25 @@ static int nft_limit_init(struct nft_limit *limit,
if (tb[NFTA_LIMIT_BURST])
limit->burst = ntohl(nla_get_be32(tb[NFTA_LIMIT_BURST]));
- else
- limit->burst = 0;
+
+ if (pkts && limit->burst == 0)
+ limit->burst = NFT_LIMIT_PKT_BURST_DEFAULT;
if (limit->rate + limit->burst < limit->rate)
return -EOVERFLOW;
- /* The token bucket size limits the number of tokens can be
- * accumulated. tokens_max specifies the bucket size.
- * tokens_max = unit * (rate + burst) / rate.
- */
- limit->tokens = div_u64(limit->nsecs * (limit->rate + limit->burst),
- limit->rate);
+ if (pkts) {
+ tokens = div_u64(limit->nsecs, limit->rate) * limit->burst;
+ } else {
+ /* The token bucket size limits the number of tokens can be
+ * accumulated. tokens_max specifies the bucket size.
+ * tokens_max = unit * (rate + burst) / rate.
+ */
+ tokens = div_u64(limit->nsecs * (limit->rate + limit->burst),
+ limit->rate);
+ }
+
+ limit->tokens = tokens;
limit->tokens_max = limit->tokens;
if (tb[NFTA_LIMIT_FLAGS]) {
@@ -144,7 +154,7 @@ static int nft_limit_pkts_init(const struct nft_ctx *ctx,
struct nft_limit_pkts *priv = nft_expr_priv(expr);
int err;
- err = nft_limit_init(&priv->limit, tb);
+ err = nft_limit_init(&priv->limit, tb, true);
if (err < 0)
return err;
@@ -185,7 +195,7 @@ static int nft_limit_bytes_init(const struct nft_ctx *ctx,
{
struct nft_limit *priv = nft_expr_priv(expr);
- return nft_limit_init(priv, tb);
+ return nft_limit_init(priv, tb, false);
}
static int nft_limit_bytes_dump(struct sk_buff *skb,
@@ -246,7 +256,7 @@ static int nft_limit_obj_pkts_init(const struct nft_ctx *ctx,
struct nft_limit_pkts *priv = nft_obj_data(obj);
int err;
- err = nft_limit_init(&priv->limit, tb);
+ err = nft_limit_init(&priv->limit, tb, true);
if (err < 0)
return err;
@@ -289,7 +299,7 @@ static int nft_limit_obj_bytes_init(const struct nft_ctx *ctx,
{
struct nft_limit *priv = nft_obj_data(obj);
- return nft_limit_init(priv, tb);
+ return nft_limit_init(priv, tb, false);
}
static int nft_limit_obj_bytes_dump(struct sk_buff *skb,
--
2.11.0
^ permalink raw reply related
* [PATCH 3/9] netfilter: nft_meta: fix wrong value dereference in nft_meta_set_eval
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Taehee Yoo <ap420073@gmail.com>
In the nft_meta_set_eval, nftrace value is dereferenced as u32 from sreg.
But correct type is u8. so that sometimes incorrect value is dereferenced.
Steps to reproduce:
%nft add table ip filter
%nft add chain ip filter input { type filter hook input priority 4\; }
%nft add rule ip filter input nftrace set 0
%nft monitor
Sometimes, we can see trace messages.
trace id 16767227 ip filter input packet: iif "enp2s0"
ether saddr xx:xx:xx:xx:xx:xx ether daddr xx:xx:xx:xx:xx:xx
ip saddr 192.168.0.1 ip daddr 255.255.255.255 ip dscp cs0
ip ecn not-ect ip
trace id 16767227 ip filter input rule nftrace set 0 (verdict continue)
trace id 16767227 ip filter input verdict continue
trace id 16767227 ip filter input
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nft_meta.c | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/net/netfilter/nft_meta.c b/net/netfilter/nft_meta.c
index 8fb91940e2e7..204af9899482 100644
--- a/net/netfilter/nft_meta.c
+++ b/net/netfilter/nft_meta.c
@@ -234,7 +234,7 @@ void nft_meta_set_eval(const struct nft_expr *expr,
struct sk_buff *skb = pkt->skb;
u32 *sreg = ®s->data[meta->sreg];
u32 value = *sreg;
- u8 pkt_type;
+ u8 value8;
switch (meta->key) {
case NFT_META_MARK:
@@ -244,15 +244,17 @@ void nft_meta_set_eval(const struct nft_expr *expr,
skb->priority = value;
break;
case NFT_META_PKTTYPE:
- pkt_type = nft_reg_load8(sreg);
+ value8 = nft_reg_load8(sreg);
- if (skb->pkt_type != pkt_type &&
- skb_pkt_type_ok(pkt_type) &&
+ if (skb->pkt_type != value8 &&
+ skb_pkt_type_ok(value8) &&
skb_pkt_type_ok(skb->pkt_type))
- skb->pkt_type = pkt_type;
+ skb->pkt_type = value8;
break;
case NFT_META_NFTRACE:
- skb->nf_trace = !!value;
+ value8 = nft_reg_load8(sreg);
+
+ skb->nf_trace = !!value8;
break;
default:
WARN_ON(1);
--
2.11.0
^ permalink raw reply related
* [PATCH 2/9] netfilter: ebtables: handle string from userspace with care
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Paolo Abeni <pabeni@redhat.com>
strlcpy() can't be safely used on a user-space provided string,
as it can try to read beyond the buffer's end, if the latter is
not NULL terminated.
Leveraging the above, syzbot has been able to trigger the following
splat:
BUG: KASAN: stack-out-of-bounds in strlcpy include/linux/string.h:300
[inline]
BUG: KASAN: stack-out-of-bounds in compat_mtw_from_user
net/bridge/netfilter/ebtables.c:1957 [inline]
BUG: KASAN: stack-out-of-bounds in ebt_size_mwt
net/bridge/netfilter/ebtables.c:2059 [inline]
BUG: KASAN: stack-out-of-bounds in size_entry_mwt
net/bridge/netfilter/ebtables.c:2155 [inline]
BUG: KASAN: stack-out-of-bounds in compat_copy_entries+0x96c/0x14a0
net/bridge/netfilter/ebtables.c:2194
Write of size 33 at addr ffff8801b0abf888 by task syz-executor0/4504
CPU: 0 PID: 4504 Comm: syz-executor0 Not tainted 4.17.0-rc2+ #40
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x1b9/0x294 lib/dump_stack.c:113
print_address_description+0x6c/0x20b mm/kasan/report.c:256
kasan_report_error mm/kasan/report.c:354 [inline]
kasan_report.cold.7+0x242/0x2fe mm/kasan/report.c:412
check_memory_region_inline mm/kasan/kasan.c:260 [inline]
check_memory_region+0x13e/0x1b0 mm/kasan/kasan.c:267
memcpy+0x37/0x50 mm/kasan/kasan.c:303
strlcpy include/linux/string.h:300 [inline]
compat_mtw_from_user net/bridge/netfilter/ebtables.c:1957 [inline]
ebt_size_mwt net/bridge/netfilter/ebtables.c:2059 [inline]
size_entry_mwt net/bridge/netfilter/ebtables.c:2155 [inline]
compat_copy_entries+0x96c/0x14a0 net/bridge/netfilter/ebtables.c:2194
compat_do_replace+0x483/0x900 net/bridge/netfilter/ebtables.c:2285
compat_do_ebt_set_ctl+0x2ac/0x324 net/bridge/netfilter/ebtables.c:2367
compat_nf_sockopt net/netfilter/nf_sockopt.c:144 [inline]
compat_nf_setsockopt+0x9b/0x140 net/netfilter/nf_sockopt.c:156
compat_ip_setsockopt+0xff/0x140 net/ipv4/ip_sockglue.c:1279
inet_csk_compat_setsockopt+0x97/0x120 net/ipv4/inet_connection_sock.c:1041
compat_tcp_setsockopt+0x49/0x80 net/ipv4/tcp.c:2901
compat_sock_common_setsockopt+0xb4/0x150 net/core/sock.c:3050
__compat_sys_setsockopt+0x1ab/0x7c0 net/compat.c:403
__do_compat_sys_setsockopt net/compat.c:416 [inline]
__se_compat_sys_setsockopt net/compat.c:413 [inline]
__ia32_compat_sys_setsockopt+0xbd/0x150 net/compat.c:413
do_syscall_32_irqs_on arch/x86/entry/common.c:323 [inline]
do_fast_syscall_32+0x345/0xf9b arch/x86/entry/common.c:394
entry_SYSENTER_compat+0x70/0x7f arch/x86/entry/entry_64_compat.S:139
RIP: 0023:0xf7fb3cb9
RSP: 002b:00000000fff0c26c EFLAGS: 00000282 ORIG_RAX: 000000000000016e
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 0000000000000000
RDX: 0000000000000080 RSI: 0000000020000300 RDI: 00000000000005f4
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
The buggy address belongs to the page:
page:ffffea0006c2afc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0
flags: 0x2fffc0000000000()
raw: 02fffc0000000000 0000000000000000 0000000000000000 00000000ffffffff
raw: 0000000000000000 ffffea0006c20101 0000000000000000 0000000000000000
page dumped because: kasan: bad access detected
Fix the issue replacing the unsafe function with strscpy() and
taking care of possible errors.
Fixes: 81e675c227ec ("netfilter: ebtables: add CONFIG_COMPAT support")
Reported-and-tested-by: syzbot+4e42a04e0bc33cb6c087@syzkaller.appspotmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/bridge/netfilter/ebtables.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/bridge/netfilter/ebtables.c b/net/bridge/netfilter/ebtables.c
index 28a4c3490359..6ba639f6c51d 100644
--- a/net/bridge/netfilter/ebtables.c
+++ b/net/bridge/netfilter/ebtables.c
@@ -1954,7 +1954,8 @@ static int compat_mtw_from_user(struct compat_ebt_entry_mwt *mwt,
int off, pad = 0;
unsigned int size_kern, match_size = mwt->match_size;
- strlcpy(name, mwt->u.name, sizeof(name));
+ if (strscpy(name, mwt->u.name, sizeof(name)) < 0)
+ return -EINVAL;
if (state->buf_kern_start)
dst = state->buf_kern_start + state->buf_kern_offset;
--
2.11.0
^ permalink raw reply related
* [PATCH 1/9] netfilter: nf_tables: fix NULL pointer dereference on nft_ct_helper_obj_dump()
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
In-Reply-To: <20180528234221.31254-1-pablo@netfilter.org>
From: Taehee Yoo <ap420073@gmail.com>
In the nft_ct_helper_obj_dump(), always priv->helper4 is dereferenced.
But if family is ipv6, priv->helper6 should be dereferenced.
Steps to reproduces:
#test.nft
table ip6 filter {
ct helper ftp {
type "ftp" protocol tcp
}
chain input {
type filter hook input priority 4;
ct helper set "ftp"
}
}
%nft -f test.nft
%nft list ruleset
we can see the below messages:
[ 916.286233] kasan: GPF could be caused by NULL-ptr deref or user memory access
[ 916.294777] general protection fault: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
[ 916.302613] Modules linked in: nft_objref nf_conntrack_sip nf_conntrack_snmp nf_conntrack_broadcast nf_conntrack_ftp nft_ct nf_conntrack nf_tables nfnetlink [last unloaded: nfnetlink]
[ 916.318758] CPU: 1 PID: 2093 Comm: nft Not tainted 4.17.0-rc4+ #181
[ 916.326772] Hardware name: To be filled by O.E.M. To be filled by O.E.M./Aptio CRB, BIOS 5.6.5 07/08/2015
[ 916.338773] RIP: 0010:strlen+0x1a/0x90
[ 916.342781] RSP: 0018:ffff88010ff0f2f8 EFLAGS: 00010292
[ 916.346773] RAX: dffffc0000000000 RBX: ffff880119b26ee8 RCX: ffff88010c150038
[ 916.354777] RDX: 0000000000000002 RSI: ffff880119b26ee8 RDI: 0000000000000010
[ 916.362773] RBP: 0000000000000010 R08: 0000000000007e88 R09: ffff88010c15003c
[ 916.370773] R10: ffff88010c150037 R11: ffffed002182a007 R12: ffff88010ff04040
[ 916.378779] R13: 0000000000000010 R14: ffff880119b26f30 R15: ffff88010ff04110
[ 916.387265] FS: 00007f57a1997700(0000) GS:ffff88011b800000(0000) knlGS:0000000000000000
[ 916.394785] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 916.402778] CR2: 00007f57a0ac80f0 CR3: 000000010ff02000 CR4: 00000000001006e0
[ 916.410772] Call Trace:
[ 916.414787] nft_ct_helper_obj_dump+0x94/0x200 [nft_ct]
[ 916.418779] ? nft_ct_set_eval+0x560/0x560 [nft_ct]
[ 916.426771] ? memset+0x1f/0x40
[ 916.426771] ? __nla_reserve+0x92/0xb0
[ 916.434774] ? memcpy+0x34/0x50
[ 916.434774] nf_tables_fill_obj_info+0x484/0x860 [nf_tables]
[ 916.442773] ? __nft_release_basechain+0x600/0x600 [nf_tables]
[ 916.450779] ? lock_acquire+0x193/0x380
[ 916.454771] ? lock_acquire+0x193/0x380
[ 916.458789] ? nf_tables_dump_obj+0x148/0xcb0 [nf_tables]
[ 916.462777] nf_tables_dump_obj+0x5f0/0xcb0 [nf_tables]
[ 916.470769] ? __alloc_skb+0x30b/0x500
[ 916.474779] netlink_dump+0x752/0xb50
[ 916.478775] __netlink_dump_start+0x4d3/0x750
[ 916.482784] nf_tables_getobj+0x27a/0x930 [nf_tables]
[ 916.490774] ? nft_obj_notify+0x100/0x100 [nf_tables]
[ 916.494772] ? nf_tables_getobj+0x930/0x930 [nf_tables]
[ 916.502579] ? nf_tables_dump_flowtable_done+0x70/0x70 [nf_tables]
[ 916.506774] ? nft_obj_notify+0x100/0x100 [nf_tables]
[ 916.514808] nfnetlink_rcv_msg+0x8ab/0xa86 [nfnetlink]
[ 916.518771] ? nfnetlink_rcv_msg+0x550/0xa86 [nfnetlink]
[ 916.526782] netlink_rcv_skb+0x23e/0x360
[ 916.530773] ? nfnetlink_bind+0x200/0x200 [nfnetlink]
[ 916.534778] ? debug_check_no_locks_freed+0x280/0x280
[ 916.542770] ? netlink_ack+0x870/0x870
[ 916.546786] ? ns_capable_common+0xf4/0x130
[ 916.550765] nfnetlink_rcv+0x172/0x16c0 [nfnetlink]
[ 916.554771] ? sched_clock_local+0xe2/0x150
[ 916.558774] ? sched_clock_cpu+0x144/0x180
[ 916.566575] ? lock_acquire+0x380/0x380
[ 916.570775] ? sched_clock_local+0xe2/0x150
[ 916.574765] ? nfnetlink_net_init+0x130/0x130 [nfnetlink]
[ 916.578763] ? sched_clock_cpu+0x144/0x180
[ 916.582770] ? lock_acquire+0x193/0x380
[ 916.590771] ? lock_acquire+0x193/0x380
[ 916.594766] ? lock_acquire+0x380/0x380
[ 916.598760] ? netlink_deliver_tap+0x262/0xa60
[ 916.602766] ? lock_acquire+0x193/0x380
[ 916.606766] netlink_unicast+0x3ef/0x5a0
[ 916.610771] ? netlink_attachskb+0x630/0x630
[ 916.614763] netlink_sendmsg+0x72a/0xb00
[ 916.618769] ? netlink_unicast+0x5a0/0x5a0
[ 916.626766] ? _copy_from_user+0x92/0xc0
[ 916.630773] __sys_sendto+0x202/0x300
[ 916.634772] ? __ia32_sys_getpeername+0xb0/0xb0
[ 916.638759] ? lock_acquire+0x380/0x380
[ 916.642769] ? lock_acquire+0x193/0x380
[ 916.646761] ? finish_task_switch+0xf4/0x560
[ 916.650763] ? __schedule+0x582/0x19a0
[ 916.655301] ? __sched_text_start+0x8/0x8
[ 916.655301] ? up_read+0x1c/0x110
[ 916.655301] ? __do_page_fault+0x48b/0xaa0
[ 916.655301] ? entry_SYSCALL_64_after_hwframe+0x59/0xbe
[ 916.655301] __x64_sys_sendto+0xdd/0x1b0
[ 916.655301] do_syscall_64+0x96/0x3d0
[ 916.655301] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 916.655301] RIP: 0033:0x7f57a0ff5e03
[ 916.655301] RSP: 002b:00007fff6367e0a8 EFLAGS: 00000246 ORIG_RAX: 000000000000002c
[ 916.655301] RAX: ffffffffffffffda RBX: 00007fff6367f1e0 RCX: 00007f57a0ff5e03
[ 916.655301] RDX: 0000000000000020 RSI: 00007fff6367e110 RDI: 0000000000000003
[ 916.655301] RBP: 00007fff6367e100 R08: 00007f57a0ce9160 R09: 000000000000000c
[ 916.655301] R10: 0000000000000000 R11: 0000000000000246 R12: 00007fff6367e110
[ 916.655301] R13: 0000000000000020 R14: 00007f57a153c610 R15: 0000562417258de0
[ 916.655301] Code: ff ff ff 0f 1f 40 00 66 2e 0f 1f 84 00 00 00 00 00 55 48 89 fa 53 48 c1 ea 03 48 b8 00 00 00 00 00 fc ff df 48 89 fd 48 83 ec 08 <0f> b6 04 02 48 89 fa 83 e2 07 38 d0 7f
[ 916.655301] RIP: strlen+0x1a/0x90 RSP: ffff88010ff0f2f8
[ 916.771929] ---[ end trace 1065e048e72479fe ]---
[ 916.777204] Kernel panic - not syncing: Fatal exception
[ 916.778158] Kernel Offset: 0x14000000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
net/netfilter/nft_ct.c | 20 ++++++++++++--------
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/net/netfilter/nft_ct.c b/net/netfilter/nft_ct.c
index ea737fd789e8..5c0de704bad5 100644
--- a/net/netfilter/nft_ct.c
+++ b/net/netfilter/nft_ct.c
@@ -880,22 +880,26 @@ static int nft_ct_helper_obj_dump(struct sk_buff *skb,
struct nft_object *obj, bool reset)
{
const struct nft_ct_helper_obj *priv = nft_obj_data(obj);
- const struct nf_conntrack_helper *helper = priv->helper4;
+ const struct nf_conntrack_helper *helper;
u16 family;
+ if (priv->helper4 && priv->helper6) {
+ family = NFPROTO_INET;
+ helper = priv->helper4;
+ } else if (priv->helper6) {
+ family = NFPROTO_IPV6;
+ helper = priv->helper6;
+ } else {
+ family = NFPROTO_IPV4;
+ helper = priv->helper4;
+ }
+
if (nla_put_string(skb, NFTA_CT_HELPER_NAME, helper->name))
return -1;
if (nla_put_u8(skb, NFTA_CT_HELPER_L4PROTO, priv->l4proto))
return -1;
- if (priv->helper4 && priv->helper6)
- family = NFPROTO_INET;
- else if (priv->helper6)
- family = NFPROTO_IPV6;
- else
- family = NFPROTO_IPV4;
-
if (nla_put_be16(skb, NFTA_CT_HELPER_L3PROTO, htons(family)))
return -1;
--
2.11.0
^ permalink raw reply related
* [PATCH 0/9] Netfilter/IPVS fixes for net
From: Pablo Neira Ayuso @ 2018-05-28 23:42 UTC (permalink / raw)
To: netfilter-devel; +Cc: davem, netdev
Hi David,
The following patchset contains Netfilter/IPVS fixes for your net tree:
1) Null pointer dereference when dumping conntrack helper configuration,
from Taehee Yoo.
2) Missing sanitization in ebtables extension name through compat,
from Paolo Abeni.
3) Broken fetch of tracing value, from Taehee Yoo.
4) Incorrect arithmetics in packet ratelimiting.
5) Buffer overflow in IPVS sync daemon, from Julian Anastasov.
6) Wrong argument to nla_strlcpy() in nfnetlink_{acct,cthelper},
from Eric Dumazet.
7) Fix splat in nft_update_chain_stats().
8) Null pointer dereference from object netlink dump path, from
Taehee Yoo.
9) Missing static_branch_inc() when enabling counters in existing
chain, from Taehee Yoo.
You can pull these changes from:
git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git
Thanks.
----------------------------------------------------------------
The following changes since commit 7063efd33bb15abc0160347f89eb5aba6b7d000e:
tuntap: fix use after free during release (2018-05-16 14:53:10 -0400)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git HEAD
for you to fetch changes up to bbb8c61f97e3a2dd91b30d3e57b7964a67569d11:
netfilter: nf_tables: increase nft_counters_enabled in nft_chain_stats_replace() (2018-05-29 00:15:12 +0200)
----------------------------------------------------------------
Eric Dumazet (1):
netfilter: provide correct argument to nla_strlcpy()
Julian Anastasov (1):
ipvs: fix buffer overflow with sync daemon and service
Pablo Neira Ayuso (2):
netfilter: nft_limit: fix packet ratelimiting
netfilter: nf_tables: disable preemption in nft_update_chain_stats()
Paolo Abeni (1):
netfilter: ebtables: handle string from userspace with care
Taehee Yoo (4):
netfilter: nf_tables: fix NULL pointer dereference on nft_ct_helper_obj_dump()
netfilter: nft_meta: fix wrong value dereference in nft_meta_set_eval
netfilter: nf_tables: fix NULL-ptr in nf_tables_dump_obj()
netfilter: nf_tables: increase nft_counters_enabled in nft_chain_stats_replace()
net/bridge/netfilter/ebtables.c | 3 ++-
net/netfilter/ipvs/ip_vs_ctl.c | 21 +++++++++++++++------
net/netfilter/nf_tables_api.c | 8 +++++---
net/netfilter/nf_tables_core.c | 4 ++--
net/netfilter/nfnetlink_acct.c | 2 +-
net/netfilter/nfnetlink_cthelper.c | 4 ++--
net/netfilter/nft_ct.c | 20 ++++++++++++--------
net/netfilter/nft_limit.c | 38 ++++++++++++++++++++++++--------------
net/netfilter/nft_meta.c | 14 ++++++++------
9 files changed, 71 insertions(+), 43 deletions(-)
^ permalink raw reply
* BUG: unable to handle kernel (3)
From: syzbot @ 2018-05-28 23:26 UTC (permalink / raw)
To: bridge, coreteam, davem, fw, kadlec, linux-kernel, netdev,
netfilter-devel, pablo, stephen, syzkaller-bugs
Hello,
syzbot found the following crash on:
HEAD commit: 861d9dd37526 Merge tag 'kbuild-fixes-v4.17-2' of git://git..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=10bffd0f800000
kernel config: https://syzkaller.appspot.com/x/.config?x=982e2df1b9e60b02
dashboard link: https://syzkaller.appspot.com/bug?extid=adfeaaee641dd4fdac43
compiler: gcc (GCC) 8.0.1 20180413 (experimental)
userspace arch: i386
syzkaller repro:https://syzkaller.appspot.com/x/repro.syz?x=1156a92f800000
IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+adfeaaee641dd4fdac43@syzkaller.appspotmail.com
IPv6: ADDRCONF(NETDEV_CHANGE): bridge0: link becomes ready
BUG: unable to handle kernel
IPVS: ftp: loaded support on port[0] = 21
paging request at ffffc90001f30003
PGD 1da946067 P4D 1da946067 PUD 1da947067 PMD 1afa9e067 PTE 80000001b7d3e163
Oops: 0000 [#1] SMP KASAN
Dumping ftrace buffer:
(ftrace buffer empty)
Modules linked in:
CPU: 0 PID: 8 Comm: ksoftirqd/0 Not tainted 4.17.0-rc6+ #95
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
RIP: 0010:ebt_do_table+0x1983/0x2140 net/bridge/netfilter/ebtables.c:283
RSP: 0018:ffff8801d9aaeb68 EFLAGS: 00010246
RAX: ffffc90001f30003 RBX: ffffc90001f30003 RCX: ffffc90001f24000
RDX: 0000000000000000 RSI: ffffffff86a8513c RDI: 0000000000000000
RBP: ffff8801d9aaed38 R08: ffff8801d9a9c200 R09: ffffed003b5c46d2
R10: ffffed003b5c46d2 R11: ffff8801dae23693 R12: ffffc90001f24000
R13: ffffc90001f201a0 R14: ffffc90001f200d0 R15: dffffc0000000000
FS: 0000000000000000(0000) GS:ffff8801dae00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffffc90001f30003 CR3: 00000001ad782000 CR4: 00000000001406f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
ebt_broute+0x1f8/0x320 net/bridge/netfilter/ebtable_broute.c:60
br_handle_frame+0x6b6/0x19f0 net/bridge/br_input.c:291
__netif_receive_skb_core+0xc6e/0x3630 net/core/dev.c:4546
__netif_receive_skb+0x2c/0x1e0 net/core/dev.c:4657
process_backlog+0x219/0x760 net/core/dev.c:5337
napi_poll net/core/dev.c:5735 [inline]
net_rx_action+0x7b7/0x1930 net/core/dev.c:5801
__do_softirq+0x2e0/0xaf5 kernel/softirq.c:285
run_ksoftirqd+0x86/0x100 kernel/softirq.c:646
smpboot_thread_fn+0x417/0x870 kernel/smpboot.c:164
kthread+0x345/0x410 kernel/kthread.c:240
ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:412
Code: 6c 24 08 48 89 d8 48 89 9d d0 fe ff ff 48 c1 e8 03 42 0f b6 04 38 84
c0 74 08 3c 03 0f 8e 3b 06 00 00 48 8b 85 d0 fe ff ff 31 ff <8b> 18 89 de
e8 54 f1 d0 fa 85 db 0f 85 a0 02 00 00 e8 37 f0 d0
RIP: ebt_do_table+0x1983/0x2140 net/bridge/netfilter/ebtables.c:283 RSP:
ffff8801d9aaeb68
CR2: ffffc90001f30003
---[ end trace d121cd1897af50a4 ]---
---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.
syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#bug-status-tracking for how to communicate with
syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches
^ permalink raw reply
* possible deadlock in sock_hash_free
From: syzbot @ 2018-05-28 23:16 UTC (permalink / raw)
To: ast, daniel, linux-kernel, netdev, syzkaller-bugs
Hello,
syzbot found the following crash on:
HEAD commit: 7a1a98c171ea Merge branch 'bpf-sendmsg-hook'
git tree: bpf-next
console output: https://syzkaller.appspot.com/x/log.txt?x=131f4067800000
kernel config: https://syzkaller.appspot.com/x/.config?x=e4078980b886800c
dashboard link: https://syzkaller.appspot.com/bug?extid=83bdee62c80cc044cb1a
compiler: gcc (GCC) 8.0.1 20180413 (experimental)
syzkaller repro:https://syzkaller.appspot.com/x/repro.syz?x=17a0be2f800000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=164cf10f800000
IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+83bdee62c80cc044cb1a@syzkaller.appspotmail.com
======================================================
WARNING: possible circular locking dependency detected
4.17.0-rc6+ #25 Not tainted
------------------------------------------------------
kworker/1:0/18 is trying to acquire lock:
00000000ef3a7ff3 (clock-AF_INET6){++..}, at: sock_hash_free+0x377/0x700
kernel/bpf/sockmap.c:2089
but task is already holding lock:
00000000989798b8 (&htab->buckets[i].lock){+...}, at:
sock_hash_free+0x1d4/0x700 kernel/bpf/sockmap.c:2083
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&htab->buckets[i].lock){+...}:
__raw_spin_lock_bh include/linux/spinlock_api_smp.h:135 [inline]
_raw_spin_lock_bh+0x31/0x40 kernel/locking/spinlock.c:168
bpf_tcp_close+0x822/0x10b0 kernel/bpf/sockmap.c:285
inet_release+0x104/0x1f0 net/ipv4/af_inet.c:427
inet6_release+0x50/0x70 net/ipv6/af_inet6.c:459
sock_release+0x96/0x1b0 net/socket.c:594
sock_close+0x16/0x20 net/socket.c:1149
__fput+0x34d/0x890 fs/file_table.c:209
____fput+0x15/0x20 fs/file_table.c:243
task_work_run+0x1e4/0x290 kernel/task_work.c:113
exit_task_work include/linux/task_work.h:22 [inline]
do_exit+0x1aee/0x2730 kernel/exit.c:865
do_group_exit+0x16f/0x430 kernel/exit.c:968
__do_sys_exit_group kernel/exit.c:979 [inline]
__se_sys_exit_group kernel/exit.c:977 [inline]
__x64_sys_exit_group+0x3e/0x50 kernel/exit.c:977
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
-> #0 (clock-AF_INET6){++..}:
lock_acquire+0x1dc/0x520 kernel/locking/lockdep.c:3920
__raw_write_lock_bh include/linux/rwlock_api_smp.h:203 [inline]
_raw_write_lock_bh+0x31/0x40 kernel/locking/spinlock.c:312
sock_hash_free+0x377/0x700 kernel/bpf/sockmap.c:2089
bpf_map_free_deferred+0xba/0xf0 kernel/bpf/syscall.c:261
process_one_work+0xc1e/0x1b50 kernel/workqueue.c:2145
worker_thread+0x1cc/0x1440 kernel/workqueue.c:2279
kthread+0x345/0x410 kernel/kthread.c:240
ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:412
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&htab->buckets[i].lock);
lock(clock-AF_INET6);
lock(&htab->buckets[i].lock);
lock(clock-AF_INET6);
*** DEADLOCK ***
4 locks held by kworker/1:0/18:
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at:
__write_once_size include/linux/compiler.h:215 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at:
arch_atomic64_set arch/x86/include/asm/atomic64_64.h:34 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at: atomic64_set
include/asm-generic/atomic-instrumented.h:40 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at: atomic_long_set
include/asm-generic/atomic-long.h:57 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at: set_work_data
kernel/workqueue.c:617 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at:
set_work_pool_and_clear_pending kernel/workqueue.c:644 [inline]
#0: 00000000b569d373 ((wq_completion)"events"){+.+.}, at:
process_one_work+0xaef/0x1b50 kernel/workqueue.c:2116
#1: 0000000041d1b332 ((work_completion)(&map->work)){+.+.}, at:
process_one_work+0xb46/0x1b50 kernel/workqueue.c:2120
#2: 00000000da1a504c (rcu_read_lock){....}, at: sock_hash_free+0x0/0x700
include/net/sock.h:2178
#3: 00000000989798b8 (&htab->buckets[i].lock){+...}, at:
sock_hash_free+0x1d4/0x700 kernel/bpf/sockmap.c:2083
stack backtrace:
CPU: 1 PID: 18 Comm: kworker/1:0 Not tainted 4.17.0-rc6+ #25
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Workqueue: events bpf_map_free_deferred
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x1b9/0x294 lib/dump_stack.c:113
print_circular_bug.isra.36.cold.54+0x1bd/0x27d
kernel/locking/lockdep.c:1223
check_prev_add kernel/locking/lockdep.c:1863 [inline]
check_prevs_add kernel/locking/lockdep.c:1976 [inline]
validate_chain kernel/locking/lockdep.c:2417 [inline]
__lock_acquire+0x343e/0x5140 kernel/locking/lockdep.c:3431
lock_acquire+0x1dc/0x520 kernel/locking/lockdep.c:3920
__raw_write_lock_bh include/linux/rwlock_api_smp.h:203 [inline]
_raw_write_lock_bh+0x31/0x40 kernel/locking/spinlock.c:312
sock_hash_free+0x377/0x700 kernel/bpf/sockmap.c:2089
bpf_map_free_deferred+0xba/0xf0 kernel/bpf/syscall.c:261
process_one_work+0xc1e/0x1b50 kernel/workqueue.c:2145
worker_thread+0x1cc/0x1440 kernel/workqueue.c:2279
kthread+0x345/0x410 kernel/kthread.c:240
ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:412
---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.
syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#bug-status-tracking for how to communicate with
syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches
^ permalink raw reply
* Re: [PATCH net-next] ifb: fix packets checksum
From: Jonathan Maxwell @ 2018-05-28 23:13 UTC (permalink / raw)
To: David Miller
Cc: David Ahern, mschiffer, zhangshengju, ktkhai, netdev,
linux-kernel, Jon Maxwell
In-Reply-To: <20180525.164337.1433478215018599363.davem@davemloft.net>
On Sat, May 26, 2018 at 6:43 AM, David Miller <davem@davemloft.net> wrote:
> From: Jon Maxwell <jmaxwell37@gmail.com>
> Date: Fri, 25 May 2018 07:38:29 +1000
>
>> Fixup the checksum for CHECKSUM_COMPLETE when pulling skbs on RX path.
>> Otherwise we get splats when tc mirred is used to redirect packets to ifb.
>>
>> Before fix:
>>
>> nic: hw csum failure
>>
>> Signed-off-by: Jon Maxwell <jmaxwell37@gmail.com>
>
> This definitely seems correct, but I am really surprised a bug like this has
> lasted as long as it has.
>
Sorry for the late reply I have been away for a few days. The customer never
saw this on bnx2x. Then they switched to the mlx5 driver and it started
happening continuously when doing iperf3 tests and also for other TCP traffic.
mlx5 uses CHECKSUM_COMPLETE. I think that bnx2x uses CHECKSUM UNNECESSARY
which avoided the code path that triggers the csum failure message in
__skb_checksum_complete(). It only logs the message for CHECKSUM_COMPLETE
skbs. Probably few Linux users are using a combination of NIC drivers that
use CHECKSUM_COMPLETE, tc mirred and ifb, which is why this has never been
reported before.
This is very similar to commit 7be709af2b65.
> So I'll let this sit for another day or two for review.
^ permalink raw reply
* Re: [PATCH net-next 05/14] nfp: abm: add simple RED offload
From: Jakub Kicinski @ 2018-05-28 23:05 UTC (permalink / raw)
To: Nogah Frankel
Cc: davem, jiri, xiyou.wangcong, john.fastabend, netdev, oss-drivers,
alexei.starovoitov, nogahf, yuvalm, gerlitz.or
In-Reply-To: <63246dc9-740d-ea72-e15e-599487b3a845@gmail.com>
Hi Nogah!
On Mon, 28 May 2018 18:49:51 +0300, Nogah Frankel wrote:
> > +static int
> > +nfp_abm_red_replace(struct net_device *netdev, struct nfp_abm_link *alink,
> > + struct tc_red_qopt_offload *opt)
> > +{
> > + struct nfp_port *port = nfp_port_from_netdev(netdev);
> > + int err;
> > +
> > + if (opt->set.min != opt->set.max || !opt->set.is_ecn) {
>
> I am a bit worried about the min == max.
> sch_red doesn't really support it. It will calculate incorrect delta
> value. (And that only if tc_red_eval_P in iproute2 won't reject it).
> You might maybe use max = min+1, because in real life it will probably
> act the same but without this problem.
I remember having a long think about this when I wrote the code.
My conclusion was that the two would operate almost the same, and
setting min == max may be most obvious to the user.
If min + 1 == max sch_red would act probabilistically for qavg == min,
which is not what the card would do.
Userspace now does this:
tc_red_eval_P() {
int i = qmax - qmin;
if (!i)
return 0;
if (i < 0)
return -1;
...
}
And you've fixed delta to be treated as 1 to avoid division by 0 in
commit 5c472203421a ("net_sched: red: Avoid devision by zero"):
red_set_parms() {
int delta = qth_max - qth_min;
u32 max_p_delta;
p->qth_min = qth_min << Wlog;
p->qth_max = qth_max << Wlog;
p->Wlog = Wlog;
p->Plog = Plog;
if (delta <= 0)
delta = 1;
p->qth_delta = delta;
...
}
So we should be safe. Targets will match. Probability adjustment for
adaptive should work correctly. Which doesn't matter anyway, since we
will never use the probabilistic action...
> Nogah Frankel
> (from a new mail address)
Noted :)
^ permalink raw reply
* Re: [PATCH net-next] netfilter: nat: make symbol nat_hook static
From: Pablo Neira Ayuso @ 2018-05-28 22:37 UTC (permalink / raw)
To: Wei Yongjun
Cc: Jozsef Kadlecsik, Florian Westphal, netfilter-devel, coreteam,
netdev, kernel-janitors
In-Reply-To: <1527328133-86305-1-git-send-email-weiyongjun1@huawei.com>
On Sat, May 26, 2018 at 09:48:53AM +0000, Wei Yongjun wrote:
> Fixes the following sparse warning:
>
> net/netfilter/nf_nat_core.c:1039:20: warning:
> symbol 'nat_hook' was not declared. Should it be static?
Applied, thanks.
^ permalink raw reply
* [PATCH net-next v17 8/8] sch_cake: Conditionally split GSO segments
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
At lower bandwidths, the transmission time of a single GSO segment can add
an unacceptable amount of latency due to HOL blocking. Furthermore, with a
software shaper, any tuning mechanism employed by the kernel to control the
maximum size of GSO segments is thrown off by the artificial limit on
bandwidth. For this reason, we split GSO segments into their individual
packets iff the shaper is active and configured to a bandwidth <= 1 Gbps.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 99 +++++++++++++++++++++++++++++++++++++-------------
1 file changed, 73 insertions(+), 26 deletions(-)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index 10a968d43d44..13256b6e2440 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -80,6 +80,7 @@
#define CAKE_QUEUES (1024)
#define CAKE_FLOW_MASK 63
#define CAKE_FLOW_NAT_FLAG 64
+#define CAKE_SPLIT_GSO_THRESHOLD (125000000) /* 1Gbps */
/* struct cobalt_params - contains codel and blue parameters
* @interval: codel initial drop rate
@@ -1638,36 +1639,73 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
if (unlikely(len > b->max_skblen))
b->max_skblen = len;
- cobalt_set_enqueue_time(skb, now);
- get_cobalt_cb(skb)->adjusted_len = cake_overhead(q, skb);
- flow_queue_add(flow, skb);
-
- if (q->ack_filter)
- ack = cake_ack_filter(q, flow);
+ if (skb_is_gso(skb) && q->rate_flags & CAKE_FLAG_SPLIT_GSO) {
+ struct sk_buff *segs, *nskb;
+ netdev_features_t features = netif_skb_features(skb);
+ unsigned int slen = 0;
+
+ segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK);
+ if (IS_ERR_OR_NULL(segs))
+ return qdisc_drop(skb, sch, to_free);
+
+ while (segs) {
+ nskb = segs->next;
+ segs->next = NULL;
+ qdisc_skb_cb(segs)->pkt_len = segs->len;
+ cobalt_set_enqueue_time(segs, now);
+ get_cobalt_cb(segs)->adjusted_len = cake_overhead(q,
+ segs);
+ flow_queue_add(flow, segs);
+
+ sch->q.qlen++;
+ slen += segs->len;
+ q->buffer_used += segs->truesize;
+ b->packets++;
+ segs = nskb;
+ }
- if (ack) {
- b->ack_drops++;
- sch->qstats.drops++;
- b->bytes += qdisc_pkt_len(ack);
- len -= qdisc_pkt_len(ack);
- q->buffer_used += skb->truesize - ack->truesize;
- if (q->rate_flags & CAKE_FLAG_INGRESS)
- cake_advance_shaper(q, b, ack, now, true);
+ /* stats */
+ b->bytes += slen;
+ b->backlogs[idx] += slen;
+ b->tin_backlog += slen;
+ sch->qstats.backlog += slen;
+ q->avg_window_bytes += slen;
- qdisc_tree_reduce_backlog(sch, 1, qdisc_pkt_len(ack));
- consume_skb(ack);
+ qdisc_tree_reduce_backlog(sch, 1, len);
+ consume_skb(skb);
} else {
- sch->q.qlen++;
- q->buffer_used += skb->truesize;
- }
+ /* not splitting */
+ cobalt_set_enqueue_time(skb, now);
+ get_cobalt_cb(skb)->adjusted_len = cake_overhead(q, skb);
+ flow_queue_add(flow, skb);
+
+ if (q->ack_filter)
+ ack = cake_ack_filter(q, flow);
+
+ if (ack) {
+ b->ack_drops++;
+ sch->qstats.drops++;
+ b->bytes += qdisc_pkt_len(ack);
+ len -= qdisc_pkt_len(ack);
+ q->buffer_used += skb->truesize - ack->truesize;
+ if (q->rate_flags & CAKE_FLAG_INGRESS)
+ cake_advance_shaper(q, b, ack, now, true);
+
+ qdisc_tree_reduce_backlog(sch, 1, qdisc_pkt_len(ack));
+ consume_skb(ack);
+ } else {
+ sch->q.qlen++;
+ q->buffer_used += skb->truesize;
+ }
- /* stats */
- b->packets++;
- b->bytes += len;
- b->backlogs[idx] += len;
- b->tin_backlog += len;
- sch->qstats.backlog += len;
- q->avg_window_bytes += len;
+ /* stats */
+ b->packets++;
+ b->bytes += len;
+ b->backlogs[idx] += len;
+ b->tin_backlog += len;
+ sch->qstats.backlog += len;
+ q->avg_window_bytes += len;
+ }
if (q->overflow_timeout)
cake_heapify_up(q, b->overflow_idx[idx]);
@@ -2499,6 +2537,11 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
if (tb[TCA_CAKE_MEMORY])
q->buffer_config_limit = nla_get_u32(tb[TCA_CAKE_MEMORY]);
+ if (q->rate_bps && q->rate_bps <= CAKE_SPLIT_GSO_THRESHOLD)
+ q->rate_flags |= CAKE_FLAG_SPLIT_GSO;
+ else
+ q->rate_flags &= ~CAKE_FLAG_SPLIT_GSO;
+
if (q->tins) {
sch_tree_lock(sch);
cake_reconfigure(sch);
@@ -2654,6 +2697,10 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
if (nla_put_u32(skb, TCA_CAKE_MPU, q->rate_mpu))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_SPLIT_GSO,
+ !!(q->rate_flags & CAKE_FLAG_SPLIT_GSO)))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
^ permalink raw reply related
* [PATCH net-next v17 3/8] sch_cake: Add optional ACK filter
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: Yuchung Cheng, Neal Cardwell, cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
The ACK filter is an optional feature of CAKE which is designed to improve
performance on links with very asymmetrical rate limits. On such links
(which are unfortunately quite prevalent, especially for DSL and cable
subscribers), the downstream throughput can be limited by the number of
ACKs capable of being transmitted in the *upstream* direction.
Filtering ACKs can, in general, have adverse effects on TCP performance
because it interferes with ACK clocking (especially in slow start), and it
reduces the flow's resiliency to ACKs being dropped further along the path.
To alleviate these drawbacks, the ACK filter in CAKE tries its best to
always keep enough ACKs queued to ensure forward progress in the TCP flow
being filtered. It does this by only filtering redundant ACKs. In its
default 'conservative' mode, the filter will always keep at least two
redundant ACKs in the queue, while in 'aggressive' mode, it will filter
down to a single ACK.
The ACK filter works by inspecting the per-flow queue on every packet
enqueue. Starting at the head of the queue, the filter looks for another
eligible packet to drop (so the ACK being dropped is always closer to the
head of the queue than the packet being enqueued). An ACK is eligible only
if it ACKs *fewer* bytes than the new packet being enqueued, including any
SACK options. This prevents duplicate ACKs from being filtered, to avoid
interfering with retransmission logic. In addition, we check TCP header
options and only drop those that are known to not interfere with sender
state. In particular, packets with unknown option codes are never dropped.
In aggressive mode, an eligible packet is always dropped, while in
conservative mode, at least two ACKs are kept in the queue. Only pure ACKs
(with no data segments) are considered eligible for dropping, but when an
ACK with data segments is enqueued, this can cause another pure ACK to
become eligible for dropping.
The approach described above ensures that this ACK filter avoids most of
the drawbacks of a naive filtering mechanism that only keeps flow state but
does not inspect the queue. This is the rationale for including the ACK
filter in CAKE itself rather than as separate module (as the TC filter, for
instance).
Our performance evaluation has shown that on a 30/1 Mbps link with a
bidirectional traffic test (RRUL), turning on the ACK filter on the
upstream link improves downstream throughput by ~20% (both modes) and
upstream throughput by ~12% in conservative mode and ~40% in aggressive
mode, at the cost of ~5ms of inter-flow latency due to the increased
congestion.
In *really* pathological cases, the effect can be a lot more; for instance,
the ACK filter increases the achievable downstream throughput on a link
with 100 Kbps in the upstream direction by an order of magnitude (from ~2.5
Mbps to ~25 Mbps).
Finally, even though we consider the ACK filter to be safer than most, we
do not recommend turning it on everywhere: on more symmetrical link
bandwidths the effect is negligible at best.
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 453 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 451 insertions(+), 2 deletions(-)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index 10e208e4255d..68ac908470f1 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -757,6 +757,432 @@ static void flow_queue_add(struct cake_flow *flow, struct sk_buff *skb)
skb->next = NULL;
}
+static struct iphdr *cake_get_iphdr(const struct sk_buff *skb,
+ struct ipv6hdr *buf)
+{
+ unsigned int offset = skb_network_offset(skb);
+ struct iphdr *iph;
+
+ iph = skb_header_pointer(skb, offset, sizeof(struct iphdr), buf);
+
+ if (!iph)
+ return NULL;
+
+ if (iph->version == 4 && iph->protocol == IPPROTO_IPV6)
+ return skb_header_pointer(skb, offset + iph->ihl * 4,
+ sizeof(struct ipv6hdr), buf);
+
+ else if (iph->version == 4)
+ return iph;
+
+ else if (iph->version == 6)
+ return skb_header_pointer(skb, offset, sizeof(struct ipv6hdr),
+ buf);
+
+ return NULL;
+}
+
+static struct tcphdr *cake_get_tcphdr(const struct sk_buff *skb,
+ void *buf, unsigned int bufsize)
+{
+ unsigned int offset = skb_network_offset(skb);
+ const struct ipv6hdr *ipv6h;
+ const struct tcphdr *tcph;
+ const struct iphdr *iph;
+ struct ipv6hdr _ipv6h;
+ struct tcphdr _tcph;
+
+ ipv6h = skb_header_pointer(skb, offset, sizeof(_ipv6h), &_ipv6h);
+
+ if (!ipv6h)
+ return NULL;
+
+ if (ipv6h->version == 4) {
+ iph = (struct iphdr *)ipv6h;
+ offset += iph->ihl * 4;
+
+ /* special-case 6in4 tunnelling, as that is a common way to get
+ * v6 connectivity in the home
+ */
+ if (iph->protocol == IPPROTO_IPV6) {
+ ipv6h = skb_header_pointer(skb, offset,
+ sizeof(_ipv6h), &_ipv6h);
+
+ if (!ipv6h || ipv6h->nexthdr != IPPROTO_TCP)
+ return NULL;
+
+ offset += sizeof(struct ipv6hdr);
+
+ } else if (iph->protocol != IPPROTO_TCP) {
+ return NULL;
+ }
+
+ } else if (ipv6h->version == 6) {
+ if (ipv6h->nexthdr != IPPROTO_TCP)
+ return NULL;
+
+ offset += sizeof(struct ipv6hdr);
+ } else {
+ return NULL;
+ }
+
+ tcph = skb_header_pointer(skb, offset, sizeof(_tcph), &_tcph);
+ if (!tcph)
+ return NULL;
+
+ return skb_header_pointer(skb, offset,
+ min(__tcp_hdrlen(tcph), bufsize), buf);
+}
+
+static const void *cake_get_tcpopt(const struct tcphdr *tcph,
+ int code, int *oplen)
+{
+ /* inspired by tcp_parse_options in tcp_input.c */
+ int length = __tcp_hdrlen(tcph) - sizeof(struct tcphdr);
+ const u8 *ptr = (const u8 *)(tcph + 1);
+
+ while (length > 0) {
+ int opcode = *ptr++;
+ int opsize;
+
+ if (opcode == TCPOPT_EOL)
+ break;
+ if (opcode == TCPOPT_NOP) {
+ length--;
+ continue;
+ }
+ opsize = *ptr++;
+ if (opsize < 2 || opsize > length)
+ break;
+
+ if (opcode == code) {
+ *oplen = opsize;
+ return ptr;
+ }
+
+ ptr += opsize - 2;
+ length -= opsize;
+ }
+
+ return NULL;
+}
+
+/* Compare two SACK sequences. A sequence is considered greater if it SACKs more
+ * bytes than the other. In the case where both sequences ACKs bytes that the
+ * other doesn't, A is considered greater. DSACKs in A also makes A be
+ * considered greater.
+ *
+ * @return -1, 0 or 1 as normal compare functions
+ */
+static int cake_tcph_sack_compare(const struct tcphdr *tcph_a,
+ const struct tcphdr *tcph_b)
+{
+ const struct tcp_sack_block_wire *sack_a, *sack_b;
+ u32 ack_seq_a = ntohl(tcph_a->ack_seq);
+ u32 bytes_a = 0, bytes_b = 0;
+ int oplen_a, oplen_b;
+ bool first = true;
+
+ sack_a = cake_get_tcpopt(tcph_a, TCPOPT_SACK, &oplen_a);
+ sack_b = cake_get_tcpopt(tcph_b, TCPOPT_SACK, &oplen_b);
+
+ /* pointers point to option contents */
+ oplen_a -= TCPOLEN_SACK_BASE;
+ oplen_b -= TCPOLEN_SACK_BASE;
+
+ if (sack_a && oplen_a >= sizeof(*sack_a) &&
+ (!sack_b || oplen_b < sizeof(*sack_b)))
+ return -1;
+ else if (sack_b && oplen_b >= sizeof(*sack_b) &&
+ (!sack_a || oplen_a < sizeof(*sack_a)))
+ return 1;
+ else if ((!sack_a || oplen_a < sizeof(*sack_a)) &&
+ (!sack_b || oplen_b < sizeof(*sack_b)))
+ return 0;
+
+ while (oplen_a >= sizeof(*sack_a)) {
+ const struct tcp_sack_block_wire *sack_tmp = sack_b;
+ u32 start_a = get_unaligned_be32(&sack_a->start_seq);
+ u32 end_a = get_unaligned_be32(&sack_a->end_seq);
+ int oplen_tmp = oplen_b;
+ bool found = false;
+
+ /* DSACK; always considered greater to prevent dropping */
+ if (before(start_a, ack_seq_a))
+ return -1;
+
+ bytes_a += end_a - start_a;
+
+ while (oplen_tmp >= sizeof(*sack_tmp)) {
+ u32 start_b = get_unaligned_be32(&sack_tmp->start_seq);
+ u32 end_b = get_unaligned_be32(&sack_tmp->end_seq);
+
+ /* first time through we count the total size */
+ if (first)
+ bytes_b += end_b - start_b;
+
+ if (!after(start_b, start_a) && !before(end_b, end_a)) {
+ found = true;
+ if (!first)
+ break;
+ }
+ oplen_tmp -= sizeof(*sack_tmp);
+ sack_tmp++;
+ }
+
+ if (!found)
+ return -1;
+
+ oplen_a -= sizeof(*sack_a);
+ sack_a++;
+ first = false;
+ }
+
+ /* If we made it this far, all ranges SACKed by A are covered by B, so
+ * either the SACKs are equal, or B SACKs more bytes.
+ */
+ return bytes_b > bytes_a ? 1 : 0;
+}
+
+static void cake_tcph_get_tstamp(const struct tcphdr *tcph,
+ u32 *tsval, u32 *tsecr)
+{
+ const u8 *ptr;
+ int opsize;
+
+ ptr = cake_get_tcpopt(tcph, TCPOPT_TIMESTAMP, &opsize);
+
+ if (ptr && opsize == TCPOLEN_TIMESTAMP) {
+ *tsval = get_unaligned_be32(ptr);
+ *tsecr = get_unaligned_be32(ptr + 4);
+ }
+}
+
+static bool cake_tcph_may_drop(const struct tcphdr *tcph,
+ u32 tstamp_new, u32 tsecr_new)
+{
+ /* inspired by tcp_parse_options in tcp_input.c */
+ int length = __tcp_hdrlen(tcph) - sizeof(struct tcphdr);
+ const u8 *ptr = (const u8 *)(tcph + 1);
+ u32 tstamp, tsecr;
+
+ /* 3 reserved flags must be unset to avoid future breakage
+ * ACK must be set
+ * ECE/CWR are handled separately
+ * All other flags URG/PSH/RST/SYN/FIN must be unset
+ * 0x0FFF0000 = all TCP flags (confirm ACK=1, others zero)
+ * 0x00C00000 = CWR/ECE (handled separately)
+ * 0x0F3F0000 = 0x0FFF0000 & ~0x00C00000
+ */
+ if (((tcp_flag_word(tcph) &
+ cpu_to_be32(0x0F3F0000)) != TCP_FLAG_ACK))
+ return false;
+
+ while (length > 0) {
+ int opcode = *ptr++;
+ int opsize;
+
+ if (opcode == TCPOPT_EOL)
+ break;
+ if (opcode == TCPOPT_NOP) {
+ length--;
+ continue;
+ }
+ opsize = *ptr++;
+ if (opsize < 2 || opsize > length)
+ break;
+
+ switch (opcode) {
+ case TCPOPT_MD5SIG: /* doesn't influence state */
+ break;
+
+ case TCPOPT_SACK: /* stricter checking performed later */
+ if (opsize % 8 != 2)
+ return false;
+ break;
+
+ case TCPOPT_TIMESTAMP:
+ /* only drop timestamps lower than new */
+ if (opsize != TCPOLEN_TIMESTAMP)
+ return false;
+ tstamp = get_unaligned_be32(ptr);
+ tsecr = get_unaligned_be32(ptr + 4);
+ if (after(tstamp, tstamp_new) ||
+ after(tsecr, tsecr_new))
+ return false;
+ break;
+
+ case TCPOPT_MSS: /* these should only be set on SYN */
+ case TCPOPT_WINDOW:
+ case TCPOPT_SACK_PERM:
+ case TCPOPT_FASTOPEN:
+ case TCPOPT_EXP:
+ default: /* don't drop if any unknown options are present */
+ return false;
+ }
+
+ ptr += opsize - 2;
+ length -= opsize;
+ }
+
+ return true;
+}
+
+static struct sk_buff *cake_ack_filter(struct cake_sched_data *q,
+ struct cake_flow *flow)
+{
+ bool aggressive = q->ack_filter == CAKE_ACK_AGGRESSIVE;
+ struct sk_buff *elig_ack = NULL, *elig_ack_prev = NULL;
+ struct sk_buff *skb_check, *skb_prev = NULL;
+ const struct ipv6hdr *ipv6h, *ipv6h_check;
+ unsigned char _tcph[64], _tcph_check[64];
+ const struct tcphdr *tcph, *tcph_check;
+ const struct iphdr *iph, *iph_check;
+ struct ipv6hdr _iph, _iph_check;
+ const struct sk_buff *skb;
+ int seglen, num_found = 0;
+ u32 tstamp = 0, tsecr = 0;
+ __be32 elig_flags = 0;
+ int sack_comp;
+
+ /* no other possible ACKs to filter */
+ if (flow->head == flow->tail)
+ return NULL;
+
+ skb = flow->tail;
+ tcph = cake_get_tcphdr(skb, _tcph, sizeof(_tcph));
+ iph = cake_get_iphdr(skb, &_iph);
+ if (!tcph)
+ return NULL;
+
+ cake_tcph_get_tstamp(tcph, &tstamp, &tsecr);
+
+ /* the 'triggering' packet need only have the ACK flag set.
+ * also check that SYN is not set, as there won't be any previous ACKs.
+ */
+ if ((tcp_flag_word(tcph) &
+ (TCP_FLAG_ACK | TCP_FLAG_SYN)) != TCP_FLAG_ACK)
+ return NULL;
+
+ /* the 'triggering' ACK is at the tail of the queue, we have already
+ * returned if it is the only packet in the flow. loop through the rest
+ * of the queue looking for pure ACKs with the same 5-tuple as the
+ * triggering one.
+ */
+ for (skb_check = flow->head;
+ skb_check && skb_check != skb;
+ skb_prev = skb_check, skb_check = skb_check->next) {
+ iph_check = cake_get_iphdr(skb_check, &_iph_check);
+ tcph_check = cake_get_tcphdr(skb_check, &_tcph_check,
+ sizeof(_tcph_check));
+
+ /* only TCP packets with matching 5-tuple are eligible, and only
+ * drop safe headers
+ */
+ if (!tcph_check || iph->version != iph_check->version ||
+ tcph_check->source != tcph->source ||
+ tcph_check->dest != tcph->dest)
+ continue;
+
+ if (iph_check->version == 4) {
+ if (iph_check->saddr != iph->saddr ||
+ iph_check->daddr != iph->daddr)
+ continue;
+
+ seglen = ntohs(iph_check->tot_len) -
+ (4 * iph_check->ihl);
+ } else if (iph_check->version == 6) {
+ ipv6h = (struct ipv6hdr *)iph;
+ ipv6h_check = (struct ipv6hdr *)iph_check;
+
+ if (ipv6_addr_cmp(&ipv6h_check->saddr, &ipv6h->saddr) ||
+ ipv6_addr_cmp(&ipv6h_check->daddr, &ipv6h->daddr))
+ continue;
+
+ seglen = ntohs(ipv6h_check->payload_len);
+ } else {
+ WARN_ON(1); /* shouldn't happen */
+ continue;
+ }
+
+ /* If the ECE/CWR flags changed from the previous eligible
+ * packet in the same flow, we should no longer be dropping that
+ * previous packet as this would lose information.
+ */
+ if (elig_ack && (tcp_flag_word(tcph_check) &
+ (TCP_FLAG_ECE | TCP_FLAG_CWR)) != elig_flags) {
+ elig_ack = NULL;
+ elig_ack_prev = NULL;
+ num_found--;
+ }
+
+ /* Check TCP options and flags, don't drop ACKs with segment
+ * data, and don't drop ACKs with a higher cumulative ACK
+ * counter than the triggering packet. Check ACK seqno here to
+ * avoid parsing SACK options of packets we are going to exclude
+ * anyway.
+ */
+ if (!cake_tcph_may_drop(tcph_check, tstamp, tsecr) ||
+ (seglen - __tcp_hdrlen(tcph_check)) != 0 ||
+ after(ntohl(tcph_check->ack_seq), ntohl(tcph->ack_seq)))
+ continue;
+
+ /* Check SACK options. The triggering packet must SACK more data
+ * than the ACK under consideration, or SACK the same range but
+ * have a larger cumulative ACK counter. The latter is a
+ * pathological case, but is contained in the following check
+ * anyway, just to be safe.
+ */
+ sack_comp = cake_tcph_sack_compare(tcph_check, tcph);
+
+ if (sack_comp < 0 ||
+ (ntohl(tcph_check->ack_seq) == ntohl(tcph->ack_seq) &&
+ sack_comp == 0))
+ continue;
+
+ /* At this point we have found an eligible pure ACK to drop; if
+ * we are in aggressive mode, we are done. Otherwise, keep
+ * searching unless this is the second eligible ACK we
+ * found.
+ *
+ * Since we want to drop ACK closest to the head of the queue,
+ * save the first eligible ACK we find, even if we need to loop
+ * again.
+ */
+ if (!elig_ack) {
+ elig_ack = skb_check;
+ elig_ack_prev = skb_prev;
+ elig_flags = (tcp_flag_word(tcph_check)
+ & (TCP_FLAG_ECE | TCP_FLAG_CWR));
+ }
+
+ if (num_found++ > 0)
+ goto found;
+ }
+
+ /* We made it through the queue without finding two eligible ACKs . If
+ * we found a single eligible ACK we can drop it in aggressive mode if
+ * we can guarantee that this does not interfere with ECN flag
+ * information. We ensure this by dropping it only if the enqueued
+ * packet is consecutive with the eligible ACK, and their flags match.
+ */
+ if (elig_ack && aggressive && elig_ack->next == skb &&
+ (elig_flags == (tcp_flag_word(tcph) & (TCP_FLAG_ECE | TCP_FLAG_CWR))))
+ goto found;
+
+ return NULL;
+
+found:
+ if (elig_ack_prev)
+ elig_ack_prev->next = elig_ack->next;
+ else
+ flow->head = elig_ack->next;
+
+ elig_ack->next = NULL;
+
+ return elig_ack;
+}
+
static u64 cake_ewma(u64 avg, u64 sample, u32 shift)
{
avg -= avg >> shift;
@@ -934,6 +1360,7 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
struct cake_sched_data *q = qdisc_priv(sch);
int len = qdisc_pkt_len(skb);
int uninitialized_var(ret);
+ struct sk_buff *ack = NULL;
ktime_t now = ktime_get();
struct cake_tin_data *b;
struct cake_flow *flow;
@@ -980,8 +1407,24 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
cobalt_set_enqueue_time(skb, now);
flow_queue_add(flow, skb);
- sch->q.qlen++;
- q->buffer_used += skb->truesize;
+ if (q->ack_filter)
+ ack = cake_ack_filter(q, flow);
+
+ if (ack) {
+ b->ack_drops++;
+ sch->qstats.drops++;
+ b->bytes += qdisc_pkt_len(ack);
+ len -= qdisc_pkt_len(ack);
+ q->buffer_used += skb->truesize - ack->truesize;
+ if (q->rate_flags & CAKE_FLAG_INGRESS)
+ cake_advance_shaper(q, b, ack, now, true);
+
+ qdisc_tree_reduce_backlog(sch, 1, qdisc_pkt_len(ack));
+ consume_skb(ack);
+ } else {
+ sch->q.qlen++;
+ q->buffer_used += skb->truesize;
+ }
/* stats */
b->packets++;
@@ -1511,6 +1954,9 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
q->rate_flags &= ~CAKE_FLAG_INGRESS;
}
+ if (tb[TCA_CAKE_ACK_FILTER])
+ q->ack_filter = nla_get_u32(tb[TCA_CAKE_ACK_FILTER]);
+
if (tb[TCA_CAKE_MEMORY])
q->buffer_config_limit = nla_get_u32(tb[TCA_CAKE_MEMORY]);
@@ -1642,6 +2088,9 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
!!(q->rate_flags & CAKE_FLAG_INGRESS)))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_ACK_FILTER, q->ack_filter))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
^ permalink raw reply related
* [PATCH net-next v17 4/8] netfilter: Add nf_ct_get_tuple_skb global lookup function
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: netfilter-devel, cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
This adds a global netfilter function to extract a conntrack tuple from an
skb. The function uses a new function added to nf_ct_hook, which will try
to get the tuple from skb->_nfct, and do a full lookup if that fails. This
makes it possible to use the lookup function before the skb has passed
through the conntrack init hooks (e.g., in an ingress qdisc). The tuple is
copied to the caller to avoid issues with reference counting.
The function returns false if conntrack is not loaded, allowing it to be
used without incurring a module dependency on conntrack. This is used by
the NAT mode in sch_cake.
Cc: netfilter-devel@vger.kernel.org
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
include/linux/netfilter.h | 11 +++++++++++
net/netfilter/core.c | 15 +++++++++++++++
net/netfilter/nf_conntrack_core.c | 36 ++++++++++++++++++++++++++++++++++++
3 files changed, 62 insertions(+)
diff --git a/include/linux/netfilter.h b/include/linux/netfilter.h
index 04551af2ff23..d7be35ab7967 100644
--- a/include/linux/netfilter.h
+++ b/include/linux/netfilter.h
@@ -388,8 +388,17 @@ nf_nat_decode_session(struct sk_buff *skb, struct flowi *fl, u_int8_t family)
extern void (*ip_ct_attach)(struct sk_buff *, const struct sk_buff *) __rcu;
void nf_ct_attach(struct sk_buff *, const struct sk_buff *);
+struct nf_conntrack_tuple;
+bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
+ const struct sk_buff *skb);
#else
static inline void nf_ct_attach(struct sk_buff *new, struct sk_buff *skb) {}
+struct nf_conntrack_tuple;
+static inline bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
+ const struct sk_buff *skb)
+{
+ return false;
+}
#endif
struct nf_conn;
@@ -398,6 +407,8 @@ enum ip_conntrack_info;
struct nf_ct_hook {
int (*update)(struct net *net, struct sk_buff *skb);
void (*destroy)(struct nf_conntrack *);
+ bool (*get_tuple_skb)(struct nf_conntrack_tuple *,
+ const struct sk_buff *);
};
extern struct nf_ct_hook __rcu *nf_ct_hook;
diff --git a/net/netfilter/core.c b/net/netfilter/core.c
index 168af54db975..dc240cb47ddf 100644
--- a/net/netfilter/core.c
+++ b/net/netfilter/core.c
@@ -603,6 +603,21 @@ void nf_conntrack_destroy(struct nf_conntrack *nfct)
}
EXPORT_SYMBOL(nf_conntrack_destroy);
+bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
+ const struct sk_buff *skb)
+{
+ struct nf_ct_hook *ct_hook;
+ bool ret = false;
+
+ rcu_read_lock();
+ ct_hook = rcu_dereference(nf_ct_hook);
+ if (ct_hook)
+ ret = ct_hook->get_tuple_skb(dst_tuple, skb);
+ rcu_read_unlock();
+ return ret;
+}
+EXPORT_SYMBOL(nf_ct_get_tuple_skb);
+
/* Built-in default zone used e.g. by modules. */
const struct nf_conntrack_zone nf_ct_zone_dflt = {
.id = NF_CT_DEFAULT_ZONE_ID,
diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 3465da2a98bd..85ab2fd6a665 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -1683,6 +1683,41 @@ static int nf_conntrack_update(struct net *net, struct sk_buff *skb)
return 0;
}
+static bool nf_conntrack_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
+ const struct sk_buff *skb)
+{
+ const struct nf_conntrack_tuple *src_tuple;
+ const struct nf_conntrack_tuple_hash *hash;
+ struct nf_conntrack_tuple srctuple;
+ enum ip_conntrack_info ctinfo;
+ struct nf_conn *ct;
+
+ ct = nf_ct_get(skb, &ctinfo);
+ if (ct) {
+ src_tuple = nf_ct_tuple(ct, CTINFO2DIR(ctinfo));
+ memcpy(dst_tuple, src_tuple, sizeof(*dst_tuple));
+ return true;
+ }
+
+ if (!nf_ct_get_tuplepr(skb, skb_network_offset(skb),
+ NFPROTO_IPV4, dev_net(skb->dev),
+ &srctuple))
+ return false;
+
+ hash = nf_conntrack_find_get(dev_net(skb->dev),
+ &nf_ct_zone_dflt,
+ &srctuple);
+ if (!hash)
+ return false;
+
+ ct = nf_ct_tuplehash_to_ctrack(hash);
+ src_tuple = nf_ct_tuple(ct, !hash->tuple.dst.dir);
+ memcpy(dst_tuple, src_tuple, sizeof(*dst_tuple));
+ nf_ct_put(ct);
+
+ return true;
+}
+
/* Bring out ya dead! */
static struct nf_conn *
get_next_corpse(int (*iter)(struct nf_conn *i, void *data),
@@ -2204,6 +2239,7 @@ int nf_conntrack_init_start(void)
static struct nf_ct_hook nf_conntrack_hook = {
.update = nf_conntrack_update,
.destroy = destroy_conntrack,
+ .get_tuple_skb = nf_conntrack_get_tuple_skb,
};
void nf_conntrack_init_end(void)
^ permalink raw reply related
* [PATCH net-next v17 5/8] sch_cake: Add NAT awareness to packet classifier
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: netfilter-devel, cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
When CAKE is deployed on a gateway that also performs NAT (which is a
common deployment mode), the host fairness mechanism cannot distinguish
internal hosts from each other, and so fails to work correctly.
To fix this, we add an optional NAT awareness mode, which will query the
kernel conntrack mechanism to obtain the pre-NAT addresses for each packet
and use that in the flow and host hashing.
When the shaper is enabled and the host is already performing NAT, the cost
of this lookup is negligible. However, in unlimited mode with no NAT being
performed, there is a significant CPU cost at higher bandwidths. For this
reason, the feature is turned off by default.
Cc: netfilter-devel@vger.kernel.org
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 46 insertions(+)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index 68ac908470f1..c2cba5f32351 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -71,6 +71,10 @@
#include <net/tcp.h>
#include <net/flow_dissector.h>
+#if IS_ENABLED(CONFIG_NF_CONNTRACK)
+#include <net/netfilter/nf_conntrack_core.h>
+#endif
+
#define CAKE_SET_WAYS (8)
#define CAKE_MAX_TINS (8)
#define CAKE_QUEUES (1024)
@@ -516,6 +520,29 @@ static bool cobalt_should_drop(struct cobalt_vars *vars,
return drop;
}
+static void cake_update_flowkeys(struct flow_keys *keys,
+ const struct sk_buff *skb)
+{
+#if IS_ENABLED(CONFIG_NF_CONNTRACK)
+ struct nf_conntrack_tuple tuple = {};
+ bool rev = !skb->_nfct;
+
+ if (tc_skb_protocol(skb) != htons(ETH_P_IP))
+ return;
+
+ if (!nf_ct_get_tuple_skb(&tuple, skb))
+ return;
+
+ keys->addrs.v4addrs.src = rev ? tuple.dst.u3.ip : tuple.src.u3.ip;
+ keys->addrs.v4addrs.dst = rev ? tuple.src.u3.ip : tuple.dst.u3.ip;
+
+ if (keys->ports.ports) {
+ keys->ports.src = rev ? tuple.dst.u.all : tuple.src.u.all;
+ keys->ports.dst = rev ? tuple.src.u.all : tuple.dst.u.all;
+ }
+#endif
+}
+
/* Cake has several subtle multiple bit settings. In these cases you
* would be matching triple isolate mode as well.
*/
@@ -543,6 +570,9 @@ static u32 cake_hash(struct cake_tin_data *q, const struct sk_buff *skb,
skb_flow_dissect_flow_keys(skb, &keys,
FLOW_DISSECTOR_F_STOP_AT_FLOW_LABEL);
+ if (flow_mode & CAKE_FLOW_NAT_FLAG)
+ cake_update_flowkeys(&keys, skb);
+
/* flow_hash_from_keys() sorts the addresses by value, so we have
* to preserve their order in a separate data structure to treat
* src and dst host addresses as independently selectable.
@@ -1919,6 +1949,18 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
if (err < 0)
return err;
+ if (tb[TCA_CAKE_NAT]) {
+#if IS_ENABLED(CONFIG_NF_CONNTRACK)
+ q->flow_mode &= ~CAKE_FLOW_NAT_FLAG;
+ q->flow_mode |= CAKE_FLOW_NAT_FLAG *
+ !!nla_get_u32(tb[TCA_CAKE_NAT]);
+#else
+ NL_SET_ERR_MSG_ATTR(extack, tb[TCA_CAKE_NAT],
+ "No conntrack support in kernel");
+ return -EOPNOTSUPP;
+#endif
+ }
+
if (tb[TCA_CAKE_BASE_RATE64])
q->rate_bps = nla_get_u64(tb[TCA_CAKE_BASE_RATE64]);
@@ -2091,6 +2133,10 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
if (nla_put_u32(skb, TCA_CAKE_ACK_FILTER, q->ack_filter))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_NAT,
+ !!(q->flow_mode & CAKE_FLOW_NAT_FLAG)))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
^ permalink raw reply related
* [PATCH net-next v17 6/8] sch_cake: Add DiffServ handling
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
This adds support for DiffServ-based priority queueing to CAKE. If the
shaper is in use, each priority tier gets its own virtual clock, which
limits that tier's rate to a fraction of the overall shaped rate, to
discourage trying to game the priority mechanism.
CAKE defaults to a simple, three-tier mode that interprets most code points
as "best effort", but places CS1 traffic into a low-priority "bulk" tier
which is assigned 1/16 of the total rate, and a few code points indicating
latency-sensitive or control traffic (specifically TOS4, VA, EF, CS6, CS7)
into a "latency sensitive" high-priority tier, which is assigned 1/4 rate.
The other supported DiffServ modes are a 4-tier mode matching the 802.11e
precedence rules, as well as two 8-tier modes, one of which implements
strict precedence of the eight priority levels.
This commit also adds an optional DiffServ 'wash' mode, which will zero out
the DSCP fields of any packet passing through CAKE. While this can
technically be done with other mechanisms in the kernel, having the feature
available in CAKE significantly decreases configuration complexity; and the
implementation cost is low on top of the other DiffServ-handling code.
Filters and applications can set the skb->priority field to override the
DSCP-based classification into tiers. If TC_H_MAJ(skb->priority) matches
CAKE's qdisc handle, the minor number will be interpreted as a priority
tier if it is less than or equal to the number of configured priority
tiers.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 412 +++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 404 insertions(+), 8 deletions(-)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index c2cba5f32351..b2760bc1a34f 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -296,6 +296,68 @@ static void cobalt_set_enqueue_time(struct sk_buff *skb,
static u16 quantum_div[CAKE_QUEUES + 1] = {0};
+/* Diffserv lookup tables */
+
+static const u8 precedence[] = {
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 2, 2, 2, 2, 2, 2, 2, 2,
+ 3, 3, 3, 3, 3, 3, 3, 3,
+ 4, 4, 4, 4, 4, 4, 4, 4,
+ 5, 5, 5, 5, 5, 5, 5, 5,
+ 6, 6, 6, 6, 6, 6, 6, 6,
+ 7, 7, 7, 7, 7, 7, 7, 7,
+};
+
+static const u8 diffserv8[] = {
+ 2, 5, 1, 2, 4, 2, 2, 2,
+ 0, 2, 1, 2, 1, 2, 1, 2,
+ 5, 2, 4, 2, 4, 2, 4, 2,
+ 3, 2, 3, 2, 3, 2, 3, 2,
+ 6, 2, 3, 2, 3, 2, 3, 2,
+ 6, 2, 2, 2, 6, 2, 6, 2,
+ 7, 2, 2, 2, 2, 2, 2, 2,
+ 7, 2, 2, 2, 2, 2, 2, 2,
+};
+
+static const u8 diffserv4[] = {
+ 0, 2, 0, 0, 2, 0, 0, 0,
+ 1, 0, 0, 0, 0, 0, 0, 0,
+ 2, 0, 2, 0, 2, 0, 2, 0,
+ 2, 0, 2, 0, 2, 0, 2, 0,
+ 3, 0, 2, 0, 2, 0, 2, 0,
+ 3, 0, 0, 0, 3, 0, 3, 0,
+ 3, 0, 0, 0, 0, 0, 0, 0,
+ 3, 0, 0, 0, 0, 0, 0, 0,
+};
+
+static const u8 diffserv3[] = {
+ 0, 0, 0, 0, 2, 0, 0, 0,
+ 1, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 2, 0, 2, 0,
+ 2, 0, 0, 0, 0, 0, 0, 0,
+ 2, 0, 0, 0, 0, 0, 0, 0,
+};
+
+static const u8 besteffort[] = {
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+};
+
+/* tin priority order for stats dumping */
+
+static const u8 normal_order[] = {0, 1, 2, 3, 4, 5, 6, 7};
+static const u8 bulk_order[] = {1, 0, 2, 3};
+
#define REC_INV_SQRT_CACHE (16)
static u32 cobalt_rec_inv_sqrt_cache[REC_INV_SQRT_CACHE] = {0};
@@ -1382,6 +1444,46 @@ static unsigned int cake_drop(struct Qdisc *sch, struct sk_buff **to_free)
return idx + (tin << 16);
}
+static void cake_wash_diffserv(struct sk_buff *skb)
+{
+ switch (skb->protocol) {
+ case htons(ETH_P_IP):
+ ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, 0);
+ break;
+ case htons(ETH_P_IPV6):
+ ipv6_change_dsfield(ipv6_hdr(skb), INET_ECN_MASK, 0);
+ break;
+ default:
+ break;
+ }
+}
+
+static u8 cake_handle_diffserv(struct sk_buff *skb, u16 wash)
+{
+ u8 dscp;
+
+ switch (skb->protocol) {
+ case htons(ETH_P_IP):
+ dscp = ipv4_get_dsfield(ip_hdr(skb)) >> 2;
+ if (wash && dscp)
+ ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, 0);
+ return dscp;
+
+ case htons(ETH_P_IPV6):
+ dscp = ipv6_get_dsfield(ipv6_hdr(skb)) >> 2;
+ if (wash && dscp)
+ ipv6_change_dsfield(ipv6_hdr(skb), INET_ECN_MASK, 0);
+ return dscp;
+
+ case htons(ETH_P_ARP):
+ return 0x38; /* CS7 - Net Control */
+
+ default:
+ /* If there is no Diffserv field, treat as best-effort */
+ return 0;
+ }
+}
+
static void cake_reconfigure(struct Qdisc *sch);
static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
@@ -1396,7 +1498,26 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
struct cake_flow *flow;
u32 idx, tin;
- tin = 0;
+ if (TC_H_MAJ(skb->priority) == sch->handle &&
+ TC_H_MIN(skb->priority) > 0 &&
+ TC_H_MIN(skb->priority) <= q->tin_cnt) {
+ tin = TC_H_MIN(skb->priority) - 1;
+
+ if (q->rate_flags & CAKE_FLAG_WASH)
+ cake_wash_diffserv(skb);
+ } else if (q->tin_mode != CAKE_DIFFSERV_BESTEFFORT) {
+ /* extract the Diffserv Precedence field, if it exists */
+ /* and clear DSCP bits if washing */
+ tin = q->tin_index[cake_handle_diffserv(skb,
+ q->rate_flags & CAKE_FLAG_WASH)];
+ if (unlikely(tin >= q->tin_cnt))
+ tin = 0;
+ } else {
+ tin = 0;
+ if (q->rate_flags & CAKE_FLAG_WASH)
+ cake_wash_diffserv(skb);
+ }
+
b = &q->tins[tin];
/* choose flow to insert into */
@@ -1897,18 +2018,275 @@ static void cake_set_rate(struct cake_tin_data *b, u64 rate, u32 mtu,
b->cparams.p_dec = 1 << 20; /* 1/4096 */
}
-static void cake_reconfigure(struct Qdisc *sch)
+static int cake_config_besteffort(struct Qdisc *sch)
{
struct cake_sched_data *q = qdisc_priv(sch);
struct cake_tin_data *b = &q->tins[0];
- int c, ft = 0;
+ u32 mtu = psched_mtu(qdisc_dev(sch));
+ u64 rate = q->rate_bps;
q->tin_cnt = 1;
- cake_set_rate(b, q->rate_bps, psched_mtu(qdisc_dev(sch)),
+
+ q->tin_index = besteffort;
+ q->tin_order = normal_order;
+
+ cake_set_rate(b, rate, mtu,
us_to_ns(q->target), us_to_ns(q->interval));
b->tin_quantum_band = 65535;
b->tin_quantum_prio = 65535;
+ return 0;
+}
+
+static int cake_config_precedence(struct Qdisc *sch)
+{
+ /* convert high-level (user visible) parameters into internal format */
+ struct cake_sched_data *q = qdisc_priv(sch);
+ u32 mtu = psched_mtu(qdisc_dev(sch));
+ u64 rate = q->rate_bps;
+ u32 quantum1 = 256;
+ u32 quantum2 = 256;
+ u32 i;
+
+ q->tin_cnt = 8;
+ q->tin_index = precedence;
+ q->tin_order = normal_order;
+
+ for (i = 0; i < q->tin_cnt; i++) {
+ struct cake_tin_data *b = &q->tins[i];
+
+ cake_set_rate(b, rate, mtu, us_to_ns(q->target),
+ us_to_ns(q->interval));
+
+ b->tin_quantum_prio = max_t(u16, 1U, quantum1);
+ b->tin_quantum_band = max_t(u16, 1U, quantum2);
+
+ /* calculate next class's parameters */
+ rate *= 7;
+ rate >>= 3;
+
+ quantum1 *= 3;
+ quantum1 >>= 1;
+
+ quantum2 *= 7;
+ quantum2 >>= 3;
+ }
+
+ return 0;
+}
+
+/* List of known Diffserv codepoints:
+ *
+ * Least Effort (CS1)
+ * Best Effort (CS0)
+ * Max Reliability & LLT "Lo" (TOS1)
+ * Max Throughput (TOS2)
+ * Min Delay (TOS4)
+ * LLT "La" (TOS5)
+ * Assured Forwarding 1 (AF1x) - x3
+ * Assured Forwarding 2 (AF2x) - x3
+ * Assured Forwarding 3 (AF3x) - x3
+ * Assured Forwarding 4 (AF4x) - x3
+ * Precedence Class 2 (CS2)
+ * Precedence Class 3 (CS3)
+ * Precedence Class 4 (CS4)
+ * Precedence Class 5 (CS5)
+ * Precedence Class 6 (CS6)
+ * Precedence Class 7 (CS7)
+ * Voice Admit (VA)
+ * Expedited Forwarding (EF)
+
+ * Total 25 codepoints.
+ */
+
+/* List of traffic classes in RFC 4594:
+ * (roughly descending order of contended priority)
+ * (roughly ascending order of uncontended throughput)
+ *
+ * Network Control (CS6,CS7) - routing traffic
+ * Telephony (EF,VA) - aka. VoIP streams
+ * Signalling (CS5) - VoIP setup
+ * Multimedia Conferencing (AF4x) - aka. video calls
+ * Realtime Interactive (CS4) - eg. games
+ * Multimedia Streaming (AF3x) - eg. YouTube, NetFlix, Twitch
+ * Broadcast Video (CS3)
+ * Low Latency Data (AF2x,TOS4) - eg. database
+ * Ops, Admin, Management (CS2,TOS1) - eg. ssh
+ * Standard Service (CS0 & unrecognised codepoints)
+ * High Throughput Data (AF1x,TOS2) - eg. web traffic
+ * Low Priority Data (CS1) - eg. BitTorrent
+
+ * Total 12 traffic classes.
+ */
+
+static int cake_config_diffserv8(struct Qdisc *sch)
+{
+/* Pruned list of traffic classes for typical applications:
+ *
+ * Network Control (CS6, CS7)
+ * Minimum Latency (EF, VA, CS5, CS4)
+ * Interactive Shell (CS2, TOS1)
+ * Low Latency Transactions (AF2x, TOS4)
+ * Video Streaming (AF4x, AF3x, CS3)
+ * Bog Standard (CS0 etc.)
+ * High Throughput (AF1x, TOS2)
+ * Background Traffic (CS1)
+ *
+ * Total 8 traffic classes.
+ */
+
+ struct cake_sched_data *q = qdisc_priv(sch);
+ u32 mtu = psched_mtu(qdisc_dev(sch));
+ u64 rate = q->rate_bps;
+ u32 quantum1 = 256;
+ u32 quantum2 = 256;
+ u32 i;
+
+ q->tin_cnt = 8;
+
+ /* codepoint to class mapping */
+ q->tin_index = diffserv8;
+ q->tin_order = normal_order;
+
+ /* class characteristics */
+ for (i = 0; i < q->tin_cnt; i++) {
+ struct cake_tin_data *b = &q->tins[i];
+
+ cake_set_rate(b, rate, mtu, us_to_ns(q->target),
+ us_to_ns(q->interval));
+
+ b->tin_quantum_prio = max_t(u16, 1U, quantum1);
+ b->tin_quantum_band = max_t(u16, 1U, quantum2);
+
+ /* calculate next class's parameters */
+ rate *= 7;
+ rate >>= 3;
+
+ quantum1 *= 3;
+ quantum1 >>= 1;
+
+ quantum2 *= 7;
+ quantum2 >>= 3;
+ }
+
+ return 0;
+}
+
+static int cake_config_diffserv4(struct Qdisc *sch)
+{
+/* Further pruned list of traffic classes for four-class system:
+ *
+ * Latency Sensitive (CS7, CS6, EF, VA, CS5, CS4)
+ * Streaming Media (AF4x, AF3x, CS3, AF2x, TOS4, CS2, TOS1)
+ * Best Effort (CS0, AF1x, TOS2, and those not specified)
+ * Background Traffic (CS1)
+ *
+ * Total 4 traffic classes.
+ */
+
+ struct cake_sched_data *q = qdisc_priv(sch);
+ u32 mtu = psched_mtu(qdisc_dev(sch));
+ u64 rate = q->rate_bps;
+ u32 quantum = 1024;
+
+ q->tin_cnt = 4;
+
+ /* codepoint to class mapping */
+ q->tin_index = diffserv4;
+ q->tin_order = bulk_order;
+
+ /* class characteristics */
+ cake_set_rate(&q->tins[0], rate, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+ cake_set_rate(&q->tins[1], rate >> 4, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+ cake_set_rate(&q->tins[2], rate >> 1, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+ cake_set_rate(&q->tins[3], rate >> 2, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+
+ /* priority weights */
+ q->tins[0].tin_quantum_prio = quantum;
+ q->tins[1].tin_quantum_prio = quantum >> 4;
+ q->tins[2].tin_quantum_prio = quantum << 2;
+ q->tins[3].tin_quantum_prio = quantum << 4;
+
+ /* bandwidth-sharing weights */
+ q->tins[0].tin_quantum_band = quantum;
+ q->tins[1].tin_quantum_band = quantum >> 4;
+ q->tins[2].tin_quantum_band = quantum >> 1;
+ q->tins[3].tin_quantum_band = quantum >> 2;
+
+ return 0;
+}
+
+static int cake_config_diffserv3(struct Qdisc *sch)
+{
+/* Simplified Diffserv structure with 3 tins.
+ * Low Priority (CS1)
+ * Best Effort
+ * Latency Sensitive (TOS4, VA, EF, CS6, CS7)
+ */
+ struct cake_sched_data *q = qdisc_priv(sch);
+ u32 mtu = psched_mtu(qdisc_dev(sch));
+ u64 rate = q->rate_bps;
+ u32 quantum = 1024;
+
+ q->tin_cnt = 3;
+
+ /* codepoint to class mapping */
+ q->tin_index = diffserv3;
+ q->tin_order = bulk_order;
+
+ /* class characteristics */
+ cake_set_rate(&q->tins[0], rate, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+ cake_set_rate(&q->tins[1], rate >> 4, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+ cake_set_rate(&q->tins[2], rate >> 2, mtu,
+ us_to_ns(q->target), us_to_ns(q->interval));
+
+ /* priority weights */
+ q->tins[0].tin_quantum_prio = quantum;
+ q->tins[1].tin_quantum_prio = quantum >> 4;
+ q->tins[2].tin_quantum_prio = quantum << 4;
+
+ /* bandwidth-sharing weights */
+ q->tins[0].tin_quantum_band = quantum;
+ q->tins[1].tin_quantum_band = quantum >> 4;
+ q->tins[2].tin_quantum_band = quantum >> 2;
+
+ return 0;
+}
+
+static void cake_reconfigure(struct Qdisc *sch)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ int c, ft;
+
+ switch (q->tin_mode) {
+ case CAKE_DIFFSERV_BESTEFFORT:
+ ft = cake_config_besteffort(sch);
+ break;
+
+ case CAKE_DIFFSERV_PRECEDENCE:
+ ft = cake_config_precedence(sch);
+ break;
+
+ case CAKE_DIFFSERV_DIFFSERV8:
+ ft = cake_config_diffserv8(sch);
+ break;
+
+ case CAKE_DIFFSERV_DIFFSERV4:
+ ft = cake_config_diffserv4(sch);
+ break;
+
+ case CAKE_DIFFSERV_DIFFSERV3:
+ default:
+ ft = cake_config_diffserv3(sch);
+ break;
+ }
+
for (c = q->tin_cnt; c < CAKE_MAX_TINS; c++) {
cake_clear_tin(sch, c);
q->tins[c].cparams.mtu_time = q->tins[ft].cparams.mtu_time;
@@ -1964,6 +2342,16 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
if (tb[TCA_CAKE_BASE_RATE64])
q->rate_bps = nla_get_u64(tb[TCA_CAKE_BASE_RATE64]);
+ if (tb[TCA_CAKE_DIFFSERV_MODE])
+ q->tin_mode = nla_get_u32(tb[TCA_CAKE_DIFFSERV_MODE]);
+
+ if (tb[TCA_CAKE_WASH]) {
+ if (!!nla_get_u32(tb[TCA_CAKE_WASH]))
+ q->rate_flags |= CAKE_FLAG_WASH;
+ else
+ q->rate_flags &= ~CAKE_FLAG_WASH;
+ }
+
if (tb[TCA_CAKE_FLOW_MODE])
q->flow_mode = (nla_get_u32(tb[TCA_CAKE_FLOW_MODE]) &
CAKE_FLOW_MASK);
@@ -2027,7 +2415,7 @@ static int cake_init(struct Qdisc *sch, struct nlattr *opt,
int i, j, err;
sch->limit = 10240;
- q->tin_mode = CAKE_DIFFSERV_BESTEFFORT;
+ q->tin_mode = CAKE_DIFFSERV_DIFFSERV3;
q->flow_mode = CAKE_FLOW_TRIPLE;
q->rate_bps = 0; /* unlimited by default */
@@ -2137,6 +2525,13 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
!!(q->flow_mode & CAKE_FLOW_NAT_FLAG)))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_DIFFSERV_MODE, q->tin_mode))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_WASH,
+ !!(q->rate_flags & CAKE_FLAG_WASH)))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
@@ -2190,7 +2585,7 @@ static int cake_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
} while (0)
for (i = 0; i < q->tin_cnt; i++) {
- struct cake_tin_data *b = &q->tins[i];
+ struct cake_tin_data *b = &q->tins[q->tin_order[i]];
ts = nla_nest_start(d->skb, i + 1);
if (!ts)
@@ -2289,7 +2684,8 @@ static int cake_dump_class_stats(struct Qdisc *sch, unsigned long cl,
u32 idx = cl - 1;
if (idx < CAKE_QUEUES * q->tin_cnt) {
- const struct cake_tin_data *b = &q->tins[idx / CAKE_QUEUES];
+ const struct cake_tin_data *b = \
+ &q->tins[q->tin_order[idx / CAKE_QUEUES]];
const struct sk_buff *skb;
flow = &b->flows[idx % CAKE_QUEUES];
@@ -2361,7 +2757,7 @@ static void cake_walk(struct Qdisc *sch, struct qdisc_walker *arg)
return;
for (i = 0; i < q->tin_cnt; i++) {
- struct cake_tin_data *b = &q->tins[i];
+ struct cake_tin_data *b = &q->tins[q->tin_order[i]];
for (j = 0; j < CAKE_QUEUES; j++) {
if (list_empty(&b->flows[j].flowchain) ||
^ permalink raw reply related
* [PATCH net-next v17 1/8] sched: Add Common Applications Kept Enhanced (cake) qdisc
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: Pete Heist, Georgios Amanakis, Dave Taht, cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
sch_cake targets the home router use case and is intended to squeeze the
most bandwidth and latency out of even the slowest ISP links and routers,
while presenting an API simple enough that even an ISP can configure it.
Example of use on a cable ISP uplink:
tc qdisc add dev eth0 cake bandwidth 20Mbit nat docsis ack-filter
To shape a cable download link (ifb and tc-mirred setup elided)
tc qdisc add dev ifb0 cake bandwidth 200mbit nat docsis ingress wash
CAKE is filled with:
* A hybrid Codel/Blue AQM algorithm, "Cobalt", tied to an FQ_Codel
derived Flow Queuing system, which autoconfigures based on the bandwidth.
* A novel "triple-isolate" mode (the default) which balances per-host
and per-flow FQ even through NAT.
* An deficit based shaper, that can also be used in an unlimited mode.
* 8 way set associative hashing to reduce flow collisions to a minimum.
* A reasonable interpretation of various diffserv latency/loss tradeoffs.
* Support for zeroing diffserv markings for entering and exiting traffic.
* Support for interacting well with Docsis 3.0 shaper framing.
* Extensive support for DSL framing types.
* Support for ack filtering.
* Extensive statistics for measuring, loss, ecn markings, latency
variation.
A paper describing the design of CAKE is available at
https://arxiv.org/abs/1804.07617, and will be published at the 2018 IEEE
International Symposium on Local and Metropolitan Area Networks (LANMAN).
This patch adds the base shaper and packet scheduler, while subsequent
commits add the optional (configurable) features. The full userspace API
and most data structures are included in this commit, but options not
understood in the base version will be ignored.
Various versions baking have been available as an out of tree build for
kernel versions going back to 3.10, as the embedded router world has been
running a few years behind mainline Linux. A stable version has been
generally available on lede-17.01 and later.
sch_cake replaces a combination of iptables, tc filter, htb and fq_codel
in the sqm-scripts, with sane defaults and vastly simpler configuration.
CAKE's principal author is Jonathan Morton, with contributions from
Kevin Darbyshire-Bryant, Toke Høiland-Jørgensen, Sebastian Moeller,
Ryan Mounce, Guido Sarducci, Dean Scarff, Nils Andreas Svee, Dave Täht,
and Loganaden Velvindron.
Testing from Pete Heist, Georgios Amanakis, and the many other members of
the cake@lists.bufferbloat.net mailing list.
tc -s qdisc show dev eth2
qdisc cake 1: root refcnt 2 bandwidth 100Mbit diffserv3 triple-isolate rtt 100.0ms raw overhead 0
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 0b of 5000000b
capacity estimate: 100Mbit
min/max network layer size: 65535 / 0
min/max overhead-adjusted size: 65535 / 0
average network hdr offset: 0
Bulk Best Effort Voice
thresh 6250Kbit 100Mbit 25Mbit
target 5.0ms 5.0ms 5.0ms
interval 100.0ms 100.0ms 100.0ms
pk_delay 0us 0us 0us
av_delay 0us 0us 0us
sp_delay 0us 0us 0us
pkts 0 0 0
bytes 0 0 0
way_inds 0 0 0
way_miss 0 0 0
way_cols 0 0 0
drops 0 0 0
marks 0 0 0
ack_drop 0 0 0
sp_flows 0 0 0
bk_flows 0 0 0
un_flows 0 0 0
max_len 0 0 0
quantum 300 1514 762
Tested-by: Pete Heist <peteheist@gmail.com>
Tested-by: Georgios Amanakis <gamanakis@gmail.com>
Signed-off-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
include/uapi/linux/pkt_sched.h | 113 ++
net/sched/Kconfig | 11
net/sched/Makefile | 1
net/sched/sch_cake.c | 1850 ++++++++++++++++++++++++++++++++++++++++
4 files changed, 1975 insertions(+)
create mode 100644 net/sched/sch_cake.c
diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index 37b5096ae97b..07648e6ea569 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h
@@ -934,4 +934,117 @@ enum {
#define TCA_CBS_MAX (__TCA_CBS_MAX - 1)
+/* CAKE */
+enum {
+ TCA_CAKE_UNSPEC,
+ TCA_CAKE_PAD,
+ TCA_CAKE_BASE_RATE64,
+ TCA_CAKE_DIFFSERV_MODE,
+ TCA_CAKE_ATM,
+ TCA_CAKE_FLOW_MODE,
+ TCA_CAKE_OVERHEAD,
+ TCA_CAKE_RTT,
+ TCA_CAKE_TARGET,
+ TCA_CAKE_AUTORATE,
+ TCA_CAKE_MEMORY,
+ TCA_CAKE_NAT,
+ TCA_CAKE_RAW,
+ TCA_CAKE_WASH,
+ TCA_CAKE_MPU,
+ TCA_CAKE_INGRESS,
+ TCA_CAKE_ACK_FILTER,
+ TCA_CAKE_SPLIT_GSO,
+ __TCA_CAKE_MAX
+};
+#define TCA_CAKE_MAX (__TCA_CAKE_MAX - 1)
+
+enum {
+ __TCA_CAKE_STATS_INVALID,
+ TCA_CAKE_STATS_PAD,
+ TCA_CAKE_STATS_CAPACITY_ESTIMATE64,
+ TCA_CAKE_STATS_MEMORY_LIMIT,
+ TCA_CAKE_STATS_MEMORY_USED,
+ TCA_CAKE_STATS_AVG_NETOFF,
+ TCA_CAKE_STATS_MIN_NETLEN,
+ TCA_CAKE_STATS_MAX_NETLEN,
+ TCA_CAKE_STATS_MIN_ADJLEN,
+ TCA_CAKE_STATS_MAX_ADJLEN,
+ TCA_CAKE_STATS_TIN_STATS,
+ TCA_CAKE_STATS_DEFICIT,
+ TCA_CAKE_STATS_COBALT_COUNT,
+ TCA_CAKE_STATS_DROPPING,
+ TCA_CAKE_STATS_DROP_NEXT_US,
+ TCA_CAKE_STATS_P_DROP,
+ TCA_CAKE_STATS_BLUE_TIMER_US,
+ __TCA_CAKE_STATS_MAX
+};
+#define TCA_CAKE_STATS_MAX (__TCA_CAKE_STATS_MAX - 1)
+
+enum {
+ __TCA_CAKE_TIN_STATS_INVALID,
+ TCA_CAKE_TIN_STATS_PAD,
+ TCA_CAKE_TIN_STATS_SENT_PACKETS,
+ TCA_CAKE_TIN_STATS_SENT_BYTES64,
+ TCA_CAKE_TIN_STATS_DROPPED_PACKETS,
+ TCA_CAKE_TIN_STATS_DROPPED_BYTES64,
+ TCA_CAKE_TIN_STATS_ACKS_DROPPED_PACKETS,
+ TCA_CAKE_TIN_STATS_ACKS_DROPPED_BYTES64,
+ TCA_CAKE_TIN_STATS_ECN_MARKED_PACKETS,
+ TCA_CAKE_TIN_STATS_ECN_MARKED_BYTES64,
+ TCA_CAKE_TIN_STATS_BACKLOG_PACKETS,
+ TCA_CAKE_TIN_STATS_BACKLOG_BYTES,
+ TCA_CAKE_TIN_STATS_THRESHOLD_RATE64,
+ TCA_CAKE_TIN_STATS_TARGET_US,
+ TCA_CAKE_TIN_STATS_INTERVAL_US,
+ TCA_CAKE_TIN_STATS_WAY_INDIRECT_HITS,
+ TCA_CAKE_TIN_STATS_WAY_MISSES,
+ TCA_CAKE_TIN_STATS_WAY_COLLISIONS,
+ TCA_CAKE_TIN_STATS_PEAK_DELAY_US,
+ TCA_CAKE_TIN_STATS_AVG_DELAY_US,
+ TCA_CAKE_TIN_STATS_BASE_DELAY_US,
+ TCA_CAKE_TIN_STATS_SPARSE_FLOWS,
+ TCA_CAKE_TIN_STATS_BULK_FLOWS,
+ TCA_CAKE_TIN_STATS_UNRESPONSIVE_FLOWS,
+ TCA_CAKE_TIN_STATS_MAX_SKBLEN,
+ TCA_CAKE_TIN_STATS_FLOW_QUANTUM,
+ __TCA_CAKE_TIN_STATS_MAX
+};
+#define TCA_CAKE_TIN_STATS_MAX (__TCA_CAKE_TIN_STATS_MAX - 1)
+#define TC_CAKE_MAX_TINS (8)
+
+enum {
+ CAKE_FLOW_NONE = 0,
+ CAKE_FLOW_SRC_IP,
+ CAKE_FLOW_DST_IP,
+ CAKE_FLOW_HOSTS, /* = CAKE_FLOW_SRC_IP | CAKE_FLOW_DST_IP */
+ CAKE_FLOW_FLOWS,
+ CAKE_FLOW_DUAL_SRC, /* = CAKE_FLOW_SRC_IP | CAKE_FLOW_FLOWS */
+ CAKE_FLOW_DUAL_DST, /* = CAKE_FLOW_DST_IP | CAKE_FLOW_FLOWS */
+ CAKE_FLOW_TRIPLE, /* = CAKE_FLOW_HOSTS | CAKE_FLOW_FLOWS */
+ CAKE_FLOW_MAX,
+};
+
+enum {
+ CAKE_DIFFSERV_DIFFSERV3 = 0,
+ CAKE_DIFFSERV_DIFFSERV4,
+ CAKE_DIFFSERV_DIFFSERV8,
+ CAKE_DIFFSERV_BESTEFFORT,
+ CAKE_DIFFSERV_PRECEDENCE,
+ CAKE_DIFFSERV_MAX
+};
+
+enum {
+ CAKE_ACK_NONE = 0,
+ CAKE_ACK_FILTER,
+ CAKE_ACK_AGGRESSIVE,
+ CAKE_ACK_MAX
+};
+
+enum {
+ CAKE_ATM_NONE = 0,
+ CAKE_ATM_ATM,
+ CAKE_ATM_PTM,
+ CAKE_ATM_MAX
+};
+
#endif
diff --git a/net/sched/Kconfig b/net/sched/Kconfig
index a01169fb5325..6e7d614b5757 100644
--- a/net/sched/Kconfig
+++ b/net/sched/Kconfig
@@ -284,6 +284,17 @@ config NET_SCH_FQ_CODEL
If unsure, say N.
+config NET_SCH_CAKE
+ tristate "Common Applications Kept Enhanced (CAKE)"
+ help
+ Say Y here if you want to use the Common Applications Kept Enhanced
+ (CAKE) queue management algorithm.
+
+ To compile this driver as a module, choose M here: the module
+ will be called sch_cake.
+
+ If unsure, say N.
+
config NET_SCH_FQ
tristate "Fair Queue"
help
diff --git a/net/sched/Makefile b/net/sched/Makefile
index 8811d3804878..435054cee32c 100644
--- a/net/sched/Makefile
+++ b/net/sched/Makefile
@@ -50,6 +50,7 @@ obj-$(CONFIG_NET_SCH_CHOKE) += sch_choke.o
obj-$(CONFIG_NET_SCH_QFQ) += sch_qfq.o
obj-$(CONFIG_NET_SCH_CODEL) += sch_codel.o
obj-$(CONFIG_NET_SCH_FQ_CODEL) += sch_fq_codel.o
+obj-$(CONFIG_NET_SCH_CAKE) += sch_cake.o
obj-$(CONFIG_NET_SCH_FQ) += sch_fq.o
obj-$(CONFIG_NET_SCH_HHF) += sch_hhf.o
obj-$(CONFIG_NET_SCH_PIE) += sch_pie.o
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
new file mode 100644
index 000000000000..7ea4aa261cec
--- /dev/null
+++ b/net/sched/sch_cake.c
@@ -0,0 +1,1850 @@
+// SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
+
+/* COMMON Applications Kept Enhanced (CAKE) discipline
+ *
+ * Copyright (C) 2014-2018 Jonathan Morton <chromatix99@gmail.com>
+ * Copyright (C) 2015-2018 Toke Høiland-Jørgensen <toke@toke.dk>
+ * Copyright (C) 2014-2018 Dave Täht <dave.taht@gmail.com>
+ * Copyright (C) 2015-2018 Sebastian Moeller <moeller0@gmx.de>
+ * (C) 2015-2018 Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
+ * Copyright (C) 2017-2018 Ryan Mounce <ryan@mounce.com.au>
+ *
+ * The CAKE Principles:
+ * (or, how to have your cake and eat it too)
+ *
+ * This is a combination of several shaping, AQM and FQ techniques into one
+ * easy-to-use package:
+ *
+ * - An overall bandwidth shaper, to move the bottleneck away from dumb CPE
+ * equipment and bloated MACs. This operates in deficit mode (as in sch_fq),
+ * eliminating the need for any sort of burst parameter (eg. token bucket
+ * depth). Burst support is limited to that necessary to overcome scheduling
+ * latency.
+ *
+ * - A Diffserv-aware priority queue, giving more priority to certain classes,
+ * up to a specified fraction of bandwidth. Above that bandwidth threshold,
+ * the priority is reduced to avoid starving other tins.
+ *
+ * - Each priority tin has a separate Flow Queue system, to isolate traffic
+ * flows from each other. This prevents a burst on one flow from increasing
+ * the delay to another. Flows are distributed to queues using a
+ * set-associative hash function.
+ *
+ * - Each queue is actively managed by Cobalt, which is a combination of the
+ * Codel and Blue AQM algorithms. This serves flows fairly, and signals
+ * congestion early via ECN (if available) and/or packet drops, to keep
+ * latency low. The codel parameters are auto-tuned based on the bandwidth
+ * setting, as is necessary at low bandwidths.
+ *
+ * The configuration parameters are kept deliberately simple for ease of use.
+ * Everything has sane defaults. Complete generality of configuration is *not*
+ * a goal.
+ *
+ * The priority queue operates according to a weighted DRR scheme, combined with
+ * a bandwidth tracker which reuses the shaper logic to detect which side of the
+ * bandwidth sharing threshold the tin is operating. This determines whether a
+ * priority-based weight (high) or a bandwidth-based weight (low) is used for
+ * that tin in the current pass.
+ *
+ * This qdisc was inspired by Eric Dumazet's fq_codel code, which he kindly
+ * granted us permission to leverage.
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/jiffies.h>
+#include <linux/string.h>
+#include <linux/in.h>
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/skbuff.h>
+#include <linux/jhash.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/reciprocal_div.h>
+#include <net/netlink.h>
+#include <linux/version.h>
+#include <linux/if_vlan.h>
+#include <net/pkt_sched.h>
+#include <net/pkt_cls.h>
+#include <net/tcp.h>
+#include <net/flow_dissector.h>
+
+#define CAKE_SET_WAYS (8)
+#define CAKE_MAX_TINS (8)
+#define CAKE_QUEUES (1024)
+#define CAKE_FLOW_MASK 63
+#define CAKE_FLOW_NAT_FLAG 64
+
+/* struct cobalt_params - contains codel and blue parameters
+ * @interval: codel initial drop rate
+ * @target: maximum persistent sojourn time & blue update rate
+ * @mtu_time: serialisation delay of maximum-size packet
+ * @p_inc: increment of blue drop probability (0.32 fxp)
+ * @p_dec: decrement of blue drop probability (0.32 fxp)
+ */
+struct cobalt_params {
+ u64 interval;
+ u64 target;
+ u64 mtu_time;
+ u32 p_inc;
+ u32 p_dec;
+};
+
+/* struct cobalt_vars - contains codel and blue variables
+ * @count: codel dropping frequency
+ * @rec_inv_sqrt: reciprocal value of sqrt(count) >> 1
+ * @drop_next: time to drop next packet, or when we dropped last
+ * @blue_timer: Blue time to next drop
+ * @p_drop: BLUE drop probability (0.32 fxp)
+ * @dropping: set if in dropping state
+ * @ecn_marked: set if marked
+ */
+struct cobalt_vars {
+ u32 count;
+ u32 rec_inv_sqrt;
+ ktime_t drop_next;
+ ktime_t blue_timer;
+ u32 p_drop;
+ bool dropping;
+ bool ecn_marked;
+};
+
+enum {
+ CAKE_SET_NONE = 0,
+ CAKE_SET_SPARSE,
+ CAKE_SET_SPARSE_WAIT, /* counted in SPARSE, actually in BULK */
+ CAKE_SET_BULK,
+ CAKE_SET_DECAYING
+};
+
+struct cake_flow {
+ /* this stuff is all needed per-flow at dequeue time */
+ struct sk_buff *head;
+ struct sk_buff *tail;
+ struct list_head flowchain;
+ s32 deficit;
+ u32 dropped;
+ struct cobalt_vars cvars;
+ u16 srchost; /* index into cake_host table */
+ u16 dsthost;
+ u8 set;
+}; /* please try to keep this structure <= 64 bytes */
+
+struct cake_host {
+ u32 srchost_tag;
+ u32 dsthost_tag;
+ u16 srchost_refcnt;
+ u16 dsthost_refcnt;
+};
+
+struct cake_heap_entry {
+ u16 t:3, b:10;
+};
+
+struct cake_tin_data {
+ struct cake_flow flows[CAKE_QUEUES];
+ u32 backlogs[CAKE_QUEUES];
+ u32 tags[CAKE_QUEUES]; /* for set association */
+ u16 overflow_idx[CAKE_QUEUES];
+ struct cake_host hosts[CAKE_QUEUES]; /* for triple isolation */
+ u16 flow_quantum;
+
+ struct cobalt_params cparams;
+ u32 drop_overlimit;
+ u16 bulk_flow_count;
+ u16 sparse_flow_count;
+ u16 decaying_flow_count;
+ u16 unresponsive_flow_count;
+
+ u32 max_skblen;
+
+ struct list_head new_flows;
+ struct list_head old_flows;
+ struct list_head decaying_flows;
+
+ /* time_next = time_this + ((len * rate_ns) >> rate_shft) */
+ ktime_t time_next_packet;
+ u64 tin_rate_ns;
+ u64 tin_rate_bps;
+ u16 tin_rate_shft;
+
+ u16 tin_quantum_prio;
+ u16 tin_quantum_band;
+ s32 tin_deficit;
+ u32 tin_backlog;
+ u32 tin_dropped;
+ u32 tin_ecn_mark;
+
+ u32 packets;
+ u64 bytes;
+
+ u32 ack_drops;
+
+ /* moving averages */
+ u64 avge_delay;
+ u64 peak_delay;
+ u64 base_delay;
+
+ /* hash function stats */
+ u32 way_directs;
+ u32 way_hits;
+ u32 way_misses;
+ u32 way_collisions;
+}; /* number of tins is small, so size of this struct doesn't matter much */
+
+struct cake_sched_data {
+ struct tcf_proto __rcu *filter_list; /* optional external classifier */
+ struct tcf_block *block;
+ struct cake_tin_data *tins;
+
+ struct cake_heap_entry overflow_heap[CAKE_QUEUES * CAKE_MAX_TINS];
+ u16 overflow_timeout;
+
+ u16 tin_cnt;
+ u8 tin_mode;
+ u8 flow_mode;
+ u8 ack_filter;
+ u8 atm_mode;
+
+ /* time_next = time_this + ((len * rate_ns) >> rate_shft) */
+ u16 rate_shft;
+ ktime_t time_next_packet;
+ ktime_t failsafe_next_packet;
+ u64 rate_ns;
+ u64 rate_bps;
+ u16 rate_flags;
+ s16 rate_overhead;
+ u16 rate_mpu;
+ u64 interval;
+ u64 target;
+
+ /* resource tracking */
+ u32 buffer_used;
+ u32 buffer_max_used;
+ u32 buffer_limit;
+ u32 buffer_config_limit;
+
+ /* indices for dequeue */
+ u16 cur_tin;
+ u16 cur_flow;
+
+ struct qdisc_watchdog watchdog;
+ const u8 *tin_index;
+ const u8 *tin_order;
+
+ /* bandwidth capacity estimate */
+ ktime_t last_packet_time;
+ ktime_t avg_window_begin;
+ u64 avg_packet_interval;
+ u64 avg_window_bytes;
+ u64 avg_peak_bandwidth;
+ ktime_t last_reconfig_time;
+
+ /* packet length stats */
+ u32 avg_netoff;
+ u16 max_netlen;
+ u16 max_adjlen;
+ u16 min_netlen;
+ u16 min_adjlen;
+};
+
+enum {
+ CAKE_FLAG_OVERHEAD = BIT(0),
+ CAKE_FLAG_AUTORATE_INGRESS = BIT(1),
+ CAKE_FLAG_INGRESS = BIT(2),
+ CAKE_FLAG_WASH = BIT(3),
+ CAKE_FLAG_SPLIT_GSO = BIT(4)
+};
+
+/* COBALT operates the Codel and BLUE algorithms in parallel, in order to
+ * obtain the best features of each. Codel is excellent on flows which
+ * respond to congestion signals in a TCP-like way. BLUE is more effective on
+ * unresponsive flows.
+ */
+
+struct cobalt_skb_cb {
+ ktime_t enqueue_time;
+};
+
+static u64 us_to_ns(u64 us)
+{
+ return us * NSEC_PER_USEC;
+}
+
+static struct cobalt_skb_cb *get_cobalt_cb(const struct sk_buff *skb)
+{
+ qdisc_cb_private_validate(skb, sizeof(struct cobalt_skb_cb));
+ return (struct cobalt_skb_cb *)qdisc_skb_cb(skb)->data;
+}
+
+static ktime_t cobalt_get_enqueue_time(const struct sk_buff *skb)
+{
+ return get_cobalt_cb(skb)->enqueue_time;
+}
+
+static void cobalt_set_enqueue_time(struct sk_buff *skb,
+ ktime_t now)
+{
+ get_cobalt_cb(skb)->enqueue_time = now;
+}
+
+static u16 quantum_div[CAKE_QUEUES + 1] = {0};
+
+#define REC_INV_SQRT_CACHE (16)
+static u32 cobalt_rec_inv_sqrt_cache[REC_INV_SQRT_CACHE] = {0};
+
+/* http://en.wikipedia.org/wiki/Methods_of_computing_square_roots
+ * new_invsqrt = (invsqrt / 2) * (3 - count * invsqrt^2)
+ *
+ * Here, invsqrt is a fixed point number (< 1.0), 32bit mantissa, aka Q0.32
+ */
+
+static void cobalt_newton_step(struct cobalt_vars *vars)
+{
+ u32 invsqrt, invsqrt2;
+ u64 val;
+
+ invsqrt = vars->rec_inv_sqrt;
+ invsqrt2 = ((u64)invsqrt * invsqrt) >> 32;
+ val = (3LL << 32) - ((u64)vars->count * invsqrt2);
+
+ val >>= 2; /* avoid overflow in following multiply */
+ val = (val * invsqrt) >> (32 - 2 + 1);
+
+ vars->rec_inv_sqrt = val;
+}
+
+static void cobalt_invsqrt(struct cobalt_vars *vars)
+{
+ if (vars->count < REC_INV_SQRT_CACHE)
+ vars->rec_inv_sqrt = cobalt_rec_inv_sqrt_cache[vars->count];
+ else
+ cobalt_newton_step(vars);
+}
+
+/* There is a big difference in timing between the accurate values placed in
+ * the cache and the approximations given by a single Newton step for small
+ * count values, particularly when stepping from count 1 to 2 or vice versa.
+ * Above 16, a single Newton step gives sufficient accuracy in either
+ * direction, given the precision stored.
+ *
+ * The magnitude of the error when stepping up to count 2 is such as to give
+ * the value that *should* have been produced at count 4.
+ */
+
+static void cobalt_cache_init(void)
+{
+ struct cobalt_vars v;
+
+ memset(&v, 0, sizeof(v));
+ v.rec_inv_sqrt = ~0U;
+ cobalt_rec_inv_sqrt_cache[0] = v.rec_inv_sqrt;
+
+ for (v.count = 1; v.count < REC_INV_SQRT_CACHE; v.count++) {
+ cobalt_newton_step(&v);
+ cobalt_newton_step(&v);
+ cobalt_newton_step(&v);
+ cobalt_newton_step(&v);
+
+ cobalt_rec_inv_sqrt_cache[v.count] = v.rec_inv_sqrt;
+ }
+}
+
+static void cobalt_vars_init(struct cobalt_vars *vars)
+{
+ memset(vars, 0, sizeof(*vars));
+
+ if (!cobalt_rec_inv_sqrt_cache[0]) {
+ cobalt_cache_init();
+ cobalt_rec_inv_sqrt_cache[0] = ~0;
+ }
+}
+
+/* CoDel control_law is t + interval/sqrt(count)
+ * We maintain in rec_inv_sqrt the reciprocal value of sqrt(count) to avoid
+ * both sqrt() and divide operation.
+ */
+static ktime_t cobalt_control(ktime_t t,
+ u64 interval,
+ u32 rec_inv_sqrt)
+{
+ return ktime_add_ns(t, reciprocal_scale(interval,
+ rec_inv_sqrt));
+}
+
+/* Call this when a packet had to be dropped due to queue overflow. Returns
+ * true if the BLUE state was quiescent before but active after this call.
+ */
+static bool cobalt_queue_full(struct cobalt_vars *vars,
+ struct cobalt_params *p,
+ ktime_t now)
+{
+ bool up = false;
+
+ if (ktime_to_ns(ktime_sub(now, vars->blue_timer)) > p->target) {
+ up = !vars->p_drop;
+ vars->p_drop += p->p_inc;
+ if (vars->p_drop < p->p_inc)
+ vars->p_drop = ~0;
+ vars->blue_timer = now;
+ }
+ vars->dropping = true;
+ vars->drop_next = now;
+ if (!vars->count)
+ vars->count = 1;
+
+ return up;
+}
+
+/* Call this when the queue was serviced but turned out to be empty. Returns
+ * true if the BLUE state was active before but quiescent after this call.
+ */
+static bool cobalt_queue_empty(struct cobalt_vars *vars,
+ struct cobalt_params *p,
+ ktime_t now)
+{
+ bool down = false;
+
+ if (vars->p_drop &&
+ ktime_to_ns(ktime_sub(now, vars->blue_timer)) > p->target) {
+ if (vars->p_drop < p->p_dec)
+ vars->p_drop = 0;
+ else
+ vars->p_drop -= p->p_dec;
+ vars->blue_timer = now;
+ down = !vars->p_drop;
+ }
+ vars->dropping = false;
+
+ if (vars->count && ktime_to_ns(ktime_sub(now, vars->drop_next)) >= 0) {
+ vars->count--;
+ cobalt_invsqrt(vars);
+ vars->drop_next = cobalt_control(vars->drop_next,
+ p->interval,
+ vars->rec_inv_sqrt);
+ }
+
+ return down;
+}
+
+/* Call this with a freshly dequeued packet for possible congestion marking.
+ * Returns true as an instruction to drop the packet, false for delivery.
+ */
+static bool cobalt_should_drop(struct cobalt_vars *vars,
+ struct cobalt_params *p,
+ ktime_t now,
+ struct sk_buff *skb)
+{
+ bool next_due, over_target, drop = false;
+ ktime_t schedule;
+ u64 sojourn;
+
+/* The 'schedule' variable records, in its sign, whether 'now' is before or
+ * after 'drop_next'. This allows 'drop_next' to be updated before the next
+ * scheduling decision is actually branched, without destroying that
+ * information. Similarly, the first 'schedule' value calculated is preserved
+ * in the boolean 'next_due'.
+ *
+ * As for 'drop_next', we take advantage of the fact that 'interval' is both
+ * the delay between first exceeding 'target' and the first signalling event,
+ * *and* the scaling factor for the signalling frequency. It's therefore very
+ * natural to use a single mechanism for both purposes, and eliminates a
+ * significant amount of reference Codel's spaghetti code. To help with this,
+ * both the '0' and '1' entries in the invsqrt cache are 0xFFFFFFFF, as close
+ * as possible to 1.0 in fixed-point.
+ */
+
+ sojourn = ktime_to_ns(ktime_sub(now, cobalt_get_enqueue_time(skb)));
+ schedule = ktime_sub(now, vars->drop_next);
+ over_target = sojourn > p->target &&
+ sojourn > p->mtu_time * 4;
+ next_due = vars->count && ktime_to_ns(schedule) >= 0;
+
+ vars->ecn_marked = false;
+
+ if (over_target) {
+ if (!vars->dropping) {
+ vars->dropping = true;
+ vars->drop_next = cobalt_control(now,
+ p->interval,
+ vars->rec_inv_sqrt);
+ }
+ if (!vars->count)
+ vars->count = 1;
+ } else if (vars->dropping) {
+ vars->dropping = false;
+ }
+
+ if (next_due && vars->dropping) {
+ /* Use ECN mark if possible, otherwise drop */
+ drop = !(vars->ecn_marked = INET_ECN_set_ce(skb));
+
+ vars->count++;
+ if (!vars->count)
+ vars->count--;
+ cobalt_invsqrt(vars);
+ vars->drop_next = cobalt_control(vars->drop_next,
+ p->interval,
+ vars->rec_inv_sqrt);
+ schedule = ktime_sub(now, vars->drop_next);
+ } else {
+ while (next_due) {
+ vars->count--;
+ cobalt_invsqrt(vars);
+ vars->drop_next = cobalt_control(vars->drop_next,
+ p->interval,
+ vars->rec_inv_sqrt);
+ schedule = ktime_sub(now, vars->drop_next);
+ next_due = vars->count && ktime_to_ns(schedule) >= 0;
+ }
+ }
+
+ /* Simple BLUE implementation. Lack of ECN is deliberate. */
+ if (vars->p_drop)
+ drop |= (prandom_u32() < vars->p_drop);
+
+ /* Overload the drop_next field as an activity timeout */
+ if (!vars->count)
+ vars->drop_next = ktime_add_ns(now, p->interval);
+ else if (ktime_to_ns(schedule) > 0 && !drop)
+ vars->drop_next = now;
+
+ return drop;
+}
+
+/* Cake has several subtle multiple bit settings. In these cases you
+ * would be matching triple isolate mode as well.
+ */
+
+static bool cake_dsrc(int flow_mode)
+{
+ return (flow_mode & CAKE_FLOW_DUAL_SRC) == CAKE_FLOW_DUAL_SRC;
+}
+
+static bool cake_ddst(int flow_mode)
+{
+ return (flow_mode & CAKE_FLOW_DUAL_DST) == CAKE_FLOW_DUAL_DST;
+}
+
+static u32 cake_hash(struct cake_tin_data *q, const struct sk_buff *skb,
+ int flow_mode)
+{
+ u32 flow_hash = 0, srchost_hash, dsthost_hash;
+ u16 reduced_hash, srchost_idx, dsthost_idx;
+ struct flow_keys keys, host_keys;
+
+ if (unlikely(flow_mode == CAKE_FLOW_NONE))
+ return 0;
+
+ skb_flow_dissect_flow_keys(skb, &keys,
+ FLOW_DISSECTOR_F_STOP_AT_FLOW_LABEL);
+
+ /* flow_hash_from_keys() sorts the addresses by value, so we have
+ * to preserve their order in a separate data structure to treat
+ * src and dst host addresses as independently selectable.
+ */
+ host_keys = keys;
+ host_keys.ports.ports = 0;
+ host_keys.basic.ip_proto = 0;
+ host_keys.keyid.keyid = 0;
+ host_keys.tags.flow_label = 0;
+
+ switch (host_keys.control.addr_type) {
+ case FLOW_DISSECTOR_KEY_IPV4_ADDRS:
+ host_keys.addrs.v4addrs.src = 0;
+ dsthost_hash = flow_hash_from_keys(&host_keys);
+ host_keys.addrs.v4addrs.src = keys.addrs.v4addrs.src;
+ host_keys.addrs.v4addrs.dst = 0;
+ srchost_hash = flow_hash_from_keys(&host_keys);
+ break;
+
+ case FLOW_DISSECTOR_KEY_IPV6_ADDRS:
+ memset(&host_keys.addrs.v6addrs.src, 0,
+ sizeof(host_keys.addrs.v6addrs.src));
+ dsthost_hash = flow_hash_from_keys(&host_keys);
+ host_keys.addrs.v6addrs.src = keys.addrs.v6addrs.src;
+ memset(&host_keys.addrs.v6addrs.dst, 0,
+ sizeof(host_keys.addrs.v6addrs.dst));
+ srchost_hash = flow_hash_from_keys(&host_keys);
+ break;
+
+ default:
+ dsthost_hash = 0;
+ srchost_hash = 0;
+ }
+
+ /* This *must* be after the above switch, since as a
+ * side-effect it sorts the src and dst addresses.
+ */
+ if (flow_mode & CAKE_FLOW_FLOWS)
+ flow_hash = flow_hash_from_keys(&keys);
+
+ if (!(flow_mode & CAKE_FLOW_FLOWS)) {
+ if (flow_mode & CAKE_FLOW_SRC_IP)
+ flow_hash ^= srchost_hash;
+
+ if (flow_mode & CAKE_FLOW_DST_IP)
+ flow_hash ^= dsthost_hash;
+ }
+
+ reduced_hash = flow_hash % CAKE_QUEUES;
+
+ /* set-associative hashing */
+ /* fast path if no hash collision (direct lookup succeeds) */
+ if (likely(q->tags[reduced_hash] == flow_hash &&
+ q->flows[reduced_hash].set)) {
+ q->way_directs++;
+ } else {
+ u32 inner_hash = reduced_hash % CAKE_SET_WAYS;
+ u32 outer_hash = reduced_hash - inner_hash;
+ bool allocate_src = false;
+ bool allocate_dst = false;
+ u32 i, k;
+
+ /* check if any active queue in the set is reserved for
+ * this flow.
+ */
+ for (i = 0, k = inner_hash; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (q->tags[outer_hash + k] == flow_hash) {
+ if (i)
+ q->way_hits++;
+
+ if (!q->flows[outer_hash + k].set) {
+ /* need to increment host refcnts */
+ allocate_src = cake_dsrc(flow_mode);
+ allocate_dst = cake_ddst(flow_mode);
+ }
+
+ goto found;
+ }
+ }
+
+ /* no queue is reserved for this flow, look for an
+ * empty one.
+ */
+ for (i = 0; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (!q->flows[outer_hash + k].set) {
+ q->way_misses++;
+ allocate_src = cake_dsrc(flow_mode);
+ allocate_dst = cake_ddst(flow_mode);
+ goto found;
+ }
+ }
+
+ /* With no empty queues, default to the original
+ * queue, accept the collision, update the host tags.
+ */
+ q->way_collisions++;
+ q->hosts[q->flows[reduced_hash].srchost].srchost_refcnt--;
+ q->hosts[q->flows[reduced_hash].dsthost].dsthost_refcnt--;
+ allocate_src = cake_dsrc(flow_mode);
+ allocate_dst = cake_ddst(flow_mode);
+found:
+ /* reserve queue for future packets in same flow */
+ reduced_hash = outer_hash + k;
+ q->tags[reduced_hash] = flow_hash;
+
+ if (allocate_src) {
+ srchost_idx = srchost_hash % CAKE_QUEUES;
+ inner_hash = srchost_idx % CAKE_SET_WAYS;
+ outer_hash = srchost_idx - inner_hash;
+ for (i = 0, k = inner_hash; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (q->hosts[outer_hash + k].srchost_tag ==
+ srchost_hash)
+ goto found_src;
+ }
+ for (i = 0; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (!q->hosts[outer_hash + k].srchost_refcnt)
+ break;
+ }
+ q->hosts[outer_hash + k].srchost_tag = srchost_hash;
+found_src:
+ srchost_idx = outer_hash + k;
+ q->hosts[srchost_idx].srchost_refcnt++;
+ q->flows[reduced_hash].srchost = srchost_idx;
+ }
+
+ if (allocate_dst) {
+ dsthost_idx = dsthost_hash % CAKE_QUEUES;
+ inner_hash = dsthost_idx % CAKE_SET_WAYS;
+ outer_hash = dsthost_idx - inner_hash;
+ for (i = 0, k = inner_hash; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (q->hosts[outer_hash + k].dsthost_tag ==
+ dsthost_hash)
+ goto found_dst;
+ }
+ for (i = 0; i < CAKE_SET_WAYS;
+ i++, k = (k + 1) % CAKE_SET_WAYS) {
+ if (!q->hosts[outer_hash + k].dsthost_refcnt)
+ break;
+ }
+ q->hosts[outer_hash + k].dsthost_tag = dsthost_hash;
+found_dst:
+ dsthost_idx = outer_hash + k;
+ q->hosts[dsthost_idx].dsthost_refcnt++;
+ q->flows[reduced_hash].dsthost = dsthost_idx;
+ }
+ }
+
+ return reduced_hash;
+}
+
+static u32 cake_classify(struct Qdisc *sch, struct cake_tin_data *t,
+ struct sk_buff *skb, int flow_mode, int *qerr)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct tcf_proto *filter;
+ struct tcf_result res;
+ int result;
+
+ filter = rcu_dereference_bh(q->filter_list);
+ if (!filter)
+ return cake_hash(t, skb, flow_mode) + 1;
+
+ *qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
+ result = tcf_classify(skb, filter, &res, false);
+ if (result >= 0) {
+#ifdef CONFIG_NET_CLS_ACT
+ switch (result) {
+ case TC_ACT_STOLEN:
+ case TC_ACT_QUEUED:
+ case TC_ACT_TRAP:
+ *qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
+ /* fall through */
+ case TC_ACT_SHOT:
+ return 0;
+ }
+#endif
+ if (TC_H_MIN(res.classid) <= CAKE_QUEUES)
+ return TC_H_MIN(res.classid);
+ }
+ return 0;
+}
+
+/* helper functions : might be changed when/if skb use a standard list_head */
+/* remove one skb from head of slot queue */
+
+static struct sk_buff *dequeue_head(struct cake_flow *flow)
+{
+ struct sk_buff *skb = flow->head;
+
+ if (skb) {
+ flow->head = skb->next;
+ skb->next = NULL;
+ }
+
+ return skb;
+}
+
+/* add skb to flow queue (tail add) */
+
+static void flow_queue_add(struct cake_flow *flow, struct sk_buff *skb)
+{
+ if (!flow->head)
+ flow->head = skb;
+ else
+ flow->tail->next = skb;
+ flow->tail = skb;
+ skb->next = NULL;
+}
+
+static u64 cake_ewma(u64 avg, u64 sample, u32 shift)
+{
+ avg -= avg >> shift;
+ avg += sample >> shift;
+ return avg;
+}
+
+static void cake_heap_swap(struct cake_sched_data *q, u16 i, u16 j)
+{
+ struct cake_heap_entry ii = q->overflow_heap[i];
+ struct cake_heap_entry jj = q->overflow_heap[j];
+
+ q->overflow_heap[i] = jj;
+ q->overflow_heap[j] = ii;
+
+ q->tins[ii.t].overflow_idx[ii.b] = j;
+ q->tins[jj.t].overflow_idx[jj.b] = i;
+}
+
+static u32 cake_heap_get_backlog(const struct cake_sched_data *q, u16 i)
+{
+ struct cake_heap_entry ii = q->overflow_heap[i];
+
+ return q->tins[ii.t].backlogs[ii.b];
+}
+
+static void cake_heapify(struct cake_sched_data *q, u16 i)
+{
+ static const u32 a = CAKE_MAX_TINS * CAKE_QUEUES;
+ u32 mb = cake_heap_get_backlog(q, i);
+ u32 m = i;
+
+ while (m < a) {
+ u32 l = m + m + 1;
+ u32 r = l + 1;
+
+ if (l < a) {
+ u32 lb = cake_heap_get_backlog(q, l);
+
+ if (lb > mb) {
+ m = l;
+ mb = lb;
+ }
+ }
+
+ if (r < a) {
+ u32 rb = cake_heap_get_backlog(q, r);
+
+ if (rb > mb) {
+ m = r;
+ mb = rb;
+ }
+ }
+
+ if (m != i) {
+ cake_heap_swap(q, i, m);
+ i = m;
+ } else {
+ break;
+ }
+ }
+}
+
+static void cake_heapify_up(struct cake_sched_data *q, u16 i)
+{
+ while (i > 0 && i < CAKE_MAX_TINS * CAKE_QUEUES) {
+ u16 p = (i - 1) >> 1;
+ u32 ib = cake_heap_get_backlog(q, i);
+ u32 pb = cake_heap_get_backlog(q, p);
+
+ if (ib > pb) {
+ cake_heap_swap(q, i, p);
+ i = p;
+ } else {
+ break;
+ }
+ }
+}
+
+static int cake_advance_shaper(struct cake_sched_data *q,
+ struct cake_tin_data *b,
+ struct sk_buff *skb,
+ ktime_t now, bool drop)
+{
+ u32 len = qdisc_pkt_len(skb);
+
+ /* charge packet bandwidth to this tin
+ * and to the global shaper.
+ */
+ if (q->rate_ns) {
+ u64 tin_dur = (len * b->tin_rate_ns) >> b->tin_rate_shft;
+ u64 global_dur = (len * q->rate_ns) >> q->rate_shft;
+ u64 failsafe_dur = global_dur + (global_dur >> 1);
+
+ if (ktime_before(b->time_next_packet, now))
+ b->time_next_packet = ktime_add_ns(b->time_next_packet,
+ tin_dur);
+
+ else if (ktime_before(b->time_next_packet,
+ ktime_add_ns(now, tin_dur)))
+ b->time_next_packet = ktime_add_ns(now, tin_dur);
+
+ q->time_next_packet = ktime_add_ns(q->time_next_packet,
+ global_dur);
+ if (!drop)
+ q->failsafe_next_packet = \
+ ktime_add_ns(q->failsafe_next_packet,
+ failsafe_dur);
+ }
+ return len;
+}
+
+static unsigned int cake_drop(struct Qdisc *sch, struct sk_buff **to_free)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ ktime_t now = ktime_get();
+ u32 idx = 0, tin = 0, len;
+ struct cake_heap_entry qq;
+ struct cake_tin_data *b;
+ struct cake_flow *flow;
+ struct sk_buff *skb;
+
+ if (!q->overflow_timeout) {
+ int i;
+ /* Build fresh max-heap */
+ for (i = CAKE_MAX_TINS * CAKE_QUEUES / 2; i >= 0; i--)
+ cake_heapify(q, i);
+ }
+ q->overflow_timeout = 65535;
+
+ /* select longest queue for pruning */
+ qq = q->overflow_heap[0];
+ tin = qq.t;
+ idx = qq.b;
+
+ b = &q->tins[tin];
+ flow = &b->flows[idx];
+ skb = dequeue_head(flow);
+ if (unlikely(!skb)) {
+ /* heap has gone wrong, rebuild it next time */
+ q->overflow_timeout = 0;
+ return idx + (tin << 16);
+ }
+
+ if (cobalt_queue_full(&flow->cvars, &b->cparams, now))
+ b->unresponsive_flow_count++;
+
+ len = qdisc_pkt_len(skb);
+ q->buffer_used -= skb->truesize;
+ b->backlogs[idx] -= len;
+ b->tin_backlog -= len;
+ sch->qstats.backlog -= len;
+ qdisc_tree_reduce_backlog(sch, 1, len);
+
+ flow->dropped++;
+ b->tin_dropped++;
+ sch->qstats.drops++;
+
+ __qdisc_drop(skb, to_free);
+ sch->q.qlen--;
+
+ cake_heapify(q, 0);
+
+ return idx + (tin << 16);
+}
+
+static void cake_reconfigure(struct Qdisc *sch);
+
+static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
+ struct sk_buff **to_free)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ int len = qdisc_pkt_len(skb);
+ int uninitialized_var(ret);
+ ktime_t now = ktime_get();
+ struct cake_tin_data *b;
+ struct cake_flow *flow;
+ u32 idx, tin;
+
+ tin = 0;
+ b = &q->tins[tin];
+
+ /* choose flow to insert into */
+ idx = cake_classify(sch, b, skb, q->flow_mode, &ret);
+ if (idx == 0) {
+ if (ret & __NET_XMIT_BYPASS)
+ qdisc_qstats_drop(sch);
+ __qdisc_drop(skb, to_free);
+ return ret;
+ }
+ idx--;
+ flow = &b->flows[idx];
+
+ /* ensure shaper state isn't stale */
+ if (!b->tin_backlog) {
+ if (ktime_before(b->time_next_packet, now))
+ b->time_next_packet = now;
+
+ if (!sch->q.qlen) {
+ if (ktime_before(q->time_next_packet, now)) {
+ q->failsafe_next_packet = now;
+ q->time_next_packet = now;
+ } else if (ktime_after(q->time_next_packet, now) &&
+ ktime_after(q->failsafe_next_packet, now)) {
+ u64 next = \
+ min(ktime_to_ns(q->time_next_packet),
+ ktime_to_ns(
+ q->failsafe_next_packet));
+ sch->qstats.overlimits++;
+ qdisc_watchdog_schedule_ns(&q->watchdog, next);
+ }
+ }
+ }
+
+ if (unlikely(len > b->max_skblen))
+ b->max_skblen = len;
+
+ cobalt_set_enqueue_time(skb, now);
+ flow_queue_add(flow, skb);
+
+ sch->q.qlen++;
+ q->buffer_used += skb->truesize;
+
+ /* stats */
+ b->packets++;
+ b->bytes += len;
+ b->backlogs[idx] += len;
+ b->tin_backlog += len;
+ sch->qstats.backlog += len;
+ q->avg_window_bytes += len;
+
+ if (q->overflow_timeout)
+ cake_heapify_up(q, b->overflow_idx[idx]);
+
+ /* incoming bandwidth capacity estimate */
+ q->avg_window_bytes = 0;
+ q->last_packet_time = now;
+
+ /* flowchain */
+ if (!flow->set || flow->set == CAKE_SET_DECAYING) {
+ struct cake_host *srchost = &b->hosts[flow->srchost];
+ struct cake_host *dsthost = &b->hosts[flow->dsthost];
+ u16 host_load = 1;
+
+ if (!flow->set) {
+ list_add_tail(&flow->flowchain, &b->new_flows);
+ } else {
+ b->decaying_flow_count--;
+ list_move_tail(&flow->flowchain, &b->new_flows);
+ }
+ flow->set = CAKE_SET_SPARSE;
+ b->sparse_flow_count++;
+
+ if (cake_dsrc(q->flow_mode))
+ host_load = max(host_load, srchost->srchost_refcnt);
+
+ if (cake_ddst(q->flow_mode))
+ host_load = max(host_load, dsthost->dsthost_refcnt);
+
+ flow->deficit = (b->flow_quantum *
+ quantum_div[host_load]) >> 16;
+ } else if (flow->set == CAKE_SET_SPARSE_WAIT) {
+ /* this flow was empty, accounted as a sparse flow, but actually
+ * in the bulk rotation.
+ */
+ flow->set = CAKE_SET_BULK;
+ b->sparse_flow_count--;
+ b->bulk_flow_count++;
+ }
+
+ if (q->buffer_used > q->buffer_max_used)
+ q->buffer_max_used = q->buffer_used;
+
+ if (q->buffer_used > q->buffer_limit) {
+ u32 dropped = 0;
+
+ while (q->buffer_used > q->buffer_limit) {
+ dropped++;
+ cake_drop(sch, to_free);
+ }
+ b->drop_overlimit += dropped;
+ }
+ return NET_XMIT_SUCCESS;
+}
+
+static struct sk_buff *cake_dequeue_one(struct Qdisc *sch)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct cake_tin_data *b = &q->tins[q->cur_tin];
+ struct cake_flow *flow = &b->flows[q->cur_flow];
+ struct sk_buff *skb = NULL;
+ u32 len;
+
+ if (flow->head) {
+ skb = dequeue_head(flow);
+ len = qdisc_pkt_len(skb);
+ b->backlogs[q->cur_flow] -= len;
+ b->tin_backlog -= len;
+ sch->qstats.backlog -= len;
+ q->buffer_used -= skb->truesize;
+ sch->q.qlen--;
+
+ if (q->overflow_timeout)
+ cake_heapify(q, b->overflow_idx[q->cur_flow]);
+ }
+ return skb;
+}
+
+/* Discard leftover packets from a tin no longer in use. */
+static void cake_clear_tin(struct Qdisc *sch, u16 tin)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct sk_buff *skb;
+
+ q->cur_tin = tin;
+ for (q->cur_flow = 0; q->cur_flow < CAKE_QUEUES; q->cur_flow++)
+ while (!!(skb = cake_dequeue_one(sch)))
+ kfree_skb(skb);
+}
+
+static struct sk_buff *cake_dequeue(struct Qdisc *sch)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct cake_tin_data *b = &q->tins[q->cur_tin];
+ struct cake_host *srchost, *dsthost;
+ ktime_t now = ktime_get();
+ struct cake_flow *flow;
+ struct list_head *head;
+ bool first_flow = true;
+ struct sk_buff *skb;
+ u16 host_load;
+ u64 delay;
+ u32 len;
+
+begin:
+ if (!sch->q.qlen)
+ return NULL;
+
+ /* global hard shaper */
+ if (ktime_after(q->time_next_packet, now) &&
+ ktime_after(q->failsafe_next_packet, now)) {
+ u64 next = min(ktime_to_ns(q->time_next_packet),
+ ktime_to_ns(q->failsafe_next_packet));
+
+ sch->qstats.overlimits++;
+ qdisc_watchdog_schedule_ns(&q->watchdog, next);
+ return NULL;
+ }
+
+ /* Choose a class to work on. */
+ if (!q->rate_ns) {
+ /* In unlimited mode, can't rely on shaper timings, just balance
+ * with DRR
+ */
+ while (b->tin_deficit < 0 ||
+ !(b->sparse_flow_count + b->bulk_flow_count)) {
+ if (b->tin_deficit <= 0)
+ b->tin_deficit += b->tin_quantum_band;
+
+ q->cur_tin++;
+ b++;
+ if (q->cur_tin >= q->tin_cnt) {
+ q->cur_tin = 0;
+ b = q->tins;
+ }
+ }
+ } else {
+ /* In shaped mode, choose:
+ * - Highest-priority tin with queue and meeting schedule, or
+ * - The earliest-scheduled tin with queue.
+ */
+ ktime_t best_time = KTIME_MAX;
+ int tin, best_tin = 0;
+
+ for (tin = 0; tin < q->tin_cnt; tin++) {
+ b = q->tins + tin;
+ if ((b->sparse_flow_count + b->bulk_flow_count) > 0) {
+ ktime_t time_to_pkt = \
+ ktime_sub(b->time_next_packet, now);
+
+ if (ktime_to_ns(time_to_pkt) <= 0 ||
+ ktime_compare(time_to_pkt,
+ best_time) <= 0) {
+ best_time = time_to_pkt;
+ best_tin = tin;
+ }
+ }
+ }
+
+ q->cur_tin = best_tin;
+ b = q->tins + best_tin;
+ }
+
+retry:
+ /* service this class */
+ head = &b->decaying_flows;
+ if (!first_flow || list_empty(head)) {
+ head = &b->new_flows;
+ if (list_empty(head)) {
+ head = &b->old_flows;
+ if (unlikely(list_empty(head))) {
+ head = &b->decaying_flows;
+ if (unlikely(list_empty(head)))
+ goto begin;
+ }
+ }
+ }
+ flow = list_first_entry(head, struct cake_flow, flowchain);
+ q->cur_flow = flow - b->flows;
+ first_flow = false;
+
+ /* triple isolation (modified DRR++) */
+ srchost = &b->hosts[flow->srchost];
+ dsthost = &b->hosts[flow->dsthost];
+ host_load = 1;
+
+ if (cake_dsrc(q->flow_mode))
+ host_load = max(host_load, srchost->srchost_refcnt);
+
+ if (cake_ddst(q->flow_mode))
+ host_load = max(host_load, dsthost->dsthost_refcnt);
+
+ WARN_ON(host_load > CAKE_QUEUES);
+
+ /* flow isolation (DRR++) */
+ if (flow->deficit <= 0) {
+ /* The shifted prandom_u32() is a way to apply dithering to
+ * avoid accumulating roundoff errors
+ */
+ flow->deficit += (b->flow_quantum * quantum_div[host_load] +
+ (prandom_u32() >> 16)) >> 16;
+ list_move_tail(&flow->flowchain, &b->old_flows);
+
+ /* Keep all flows with deficits out of the sparse and decaying
+ * rotations. No non-empty flow can go into the decaying
+ * rotation, so they can't get deficits
+ */
+ if (flow->set == CAKE_SET_SPARSE) {
+ if (flow->head) {
+ b->sparse_flow_count--;
+ b->bulk_flow_count++;
+ flow->set = CAKE_SET_BULK;
+ } else {
+ /* we've moved it to the bulk rotation for
+ * correct deficit accounting but we still want
+ * to count it as a sparse flow, not a bulk one.
+ */
+ flow->set = CAKE_SET_SPARSE_WAIT;
+ }
+ }
+ goto retry;
+ }
+
+ /* Retrieve a packet via the AQM */
+ while (1) {
+ skb = cake_dequeue_one(sch);
+ if (!skb) {
+ /* this queue was actually empty */
+ if (cobalt_queue_empty(&flow->cvars, &b->cparams, now))
+ b->unresponsive_flow_count--;
+
+ if (flow->cvars.p_drop || flow->cvars.count ||
+ ktime_before(now, flow->cvars.drop_next)) {
+ /* keep in the flowchain until the state has
+ * decayed to rest
+ */
+ list_move_tail(&flow->flowchain,
+ &b->decaying_flows);
+ if (flow->set == CAKE_SET_BULK) {
+ b->bulk_flow_count--;
+ b->decaying_flow_count++;
+ } else if (flow->set == CAKE_SET_SPARSE ||
+ flow->set == CAKE_SET_SPARSE_WAIT) {
+ b->sparse_flow_count--;
+ b->decaying_flow_count++;
+ }
+ flow->set = CAKE_SET_DECAYING;
+ } else {
+ /* remove empty queue from the flowchain */
+ list_del_init(&flow->flowchain);
+ if (flow->set == CAKE_SET_SPARSE ||
+ flow->set == CAKE_SET_SPARSE_WAIT)
+ b->sparse_flow_count--;
+ else if (flow->set == CAKE_SET_BULK)
+ b->bulk_flow_count--;
+ else
+ b->decaying_flow_count--;
+
+ flow->set = CAKE_SET_NONE;
+ srchost->srchost_refcnt--;
+ dsthost->dsthost_refcnt--;
+ }
+ goto begin;
+ }
+
+ /* Last packet in queue may be marked, shouldn't be dropped */
+ if (!cobalt_should_drop(&flow->cvars, &b->cparams, now, skb) ||
+ !flow->head)
+ break;
+
+ flow->dropped++;
+ b->tin_dropped++;
+ qdisc_tree_reduce_backlog(sch, 1, qdisc_pkt_len(skb));
+ qdisc_qstats_drop(sch);
+ kfree_skb(skb);
+ }
+
+ b->tin_ecn_mark += !!flow->cvars.ecn_marked;
+ qdisc_bstats_update(sch, skb);
+
+ /* collect delay stats */
+ delay = ktime_to_ns(ktime_sub(now, cobalt_get_enqueue_time(skb)));
+ b->avge_delay = cake_ewma(b->avge_delay, delay, 8);
+ b->peak_delay = cake_ewma(b->peak_delay, delay,
+ delay > b->peak_delay ? 2 : 8);
+ b->base_delay = cake_ewma(b->base_delay, delay,
+ delay < b->base_delay ? 2 : 8);
+
+ len = cake_advance_shaper(q, b, skb, now, false);
+ flow->deficit -= len;
+ b->tin_deficit -= len;
+
+ if (ktime_after(q->time_next_packet, now) && sch->q.qlen) {
+ u64 next = min(ktime_to_ns(q->time_next_packet),
+ ktime_to_ns(q->failsafe_next_packet));
+
+ qdisc_watchdog_schedule_ns(&q->watchdog, next);
+ } else if (!sch->q.qlen) {
+ int i;
+
+ for (i = 0; i < q->tin_cnt; i++) {
+ if (q->tins[i].decaying_flow_count) {
+ ktime_t next = \
+ ktime_add_ns(now,
+ q->tins[i].cparams.target);
+
+ qdisc_watchdog_schedule_ns(&q->watchdog,
+ ktime_to_ns(next));
+ break;
+ }
+ }
+ }
+
+ if (q->overflow_timeout)
+ q->overflow_timeout--;
+
+ return skb;
+}
+
+static void cake_reset(struct Qdisc *sch)
+{
+ u32 c;
+
+ for (c = 0; c < CAKE_MAX_TINS; c++)
+ cake_clear_tin(sch, c);
+}
+
+static const struct nla_policy cake_policy[TCA_CAKE_MAX + 1] = {
+ [TCA_CAKE_BASE_RATE64] = { .type = NLA_U64 },
+ [TCA_CAKE_DIFFSERV_MODE] = { .type = NLA_U32 },
+ [TCA_CAKE_ATM] = { .type = NLA_U32 },
+ [TCA_CAKE_FLOW_MODE] = { .type = NLA_U32 },
+ [TCA_CAKE_OVERHEAD] = { .type = NLA_S32 },
+ [TCA_CAKE_RTT] = { .type = NLA_U32 },
+ [TCA_CAKE_TARGET] = { .type = NLA_U32 },
+ [TCA_CAKE_AUTORATE] = { .type = NLA_U32 },
+ [TCA_CAKE_MEMORY] = { .type = NLA_U32 },
+ [TCA_CAKE_NAT] = { .type = NLA_U32 },
+ [TCA_CAKE_RAW] = { .type = NLA_U32 },
+ [TCA_CAKE_WASH] = { .type = NLA_U32 },
+ [TCA_CAKE_MPU] = { .type = NLA_U32 },
+ [TCA_CAKE_INGRESS] = { .type = NLA_U32 },
+ [TCA_CAKE_ACK_FILTER] = { .type = NLA_U32 },
+};
+
+static void cake_set_rate(struct cake_tin_data *b, u64 rate, u32 mtu,
+ u64 target_ns, u64 rtt_est_ns)
+{
+ /* convert byte-rate into time-per-byte
+ * so it will always unwedge in reasonable time.
+ */
+ static const u64 MIN_RATE = 64;
+ u32 byte_target = mtu;
+ u64 byte_target_ns;
+ u8 rate_shft = 0;
+ u64 rate_ns = 0;
+
+ b->flow_quantum = 1514;
+ if (rate) {
+ b->flow_quantum = max(min(rate >> 12, 1514ULL), 300ULL);
+ rate_shft = 34;
+ rate_ns = ((u64)NSEC_PER_SEC) << rate_shft;
+ rate_ns = div64_u64(rate_ns, max(MIN_RATE, rate));
+ while (!!(rate_ns >> 34)) {
+ rate_ns >>= 1;
+ rate_shft--;
+ }
+ } /* else unlimited, ie. zero delay */
+
+ b->tin_rate_bps = rate;
+ b->tin_rate_ns = rate_ns;
+ b->tin_rate_shft = rate_shft;
+
+ byte_target_ns = (byte_target * rate_ns) >> rate_shft;
+
+ b->cparams.target = max((byte_target_ns * 3) / 2, target_ns);
+ b->cparams.interval = max(rtt_est_ns +
+ b->cparams.target - target_ns,
+ b->cparams.target * 2);
+ b->cparams.mtu_time = byte_target_ns;
+ b->cparams.p_inc = 1 << 24; /* 1/256 */
+ b->cparams.p_dec = 1 << 20; /* 1/4096 */
+}
+
+static void cake_reconfigure(struct Qdisc *sch)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct cake_tin_data *b = &q->tins[0];
+ int c, ft = 0;
+
+ q->tin_cnt = 1;
+ cake_set_rate(b, q->rate_bps, psched_mtu(qdisc_dev(sch)),
+ us_to_ns(q->target), us_to_ns(q->interval));
+ b->tin_quantum_band = 65535;
+ b->tin_quantum_prio = 65535;
+
+ for (c = q->tin_cnt; c < CAKE_MAX_TINS; c++) {
+ cake_clear_tin(sch, c);
+ q->tins[c].cparams.mtu_time = q->tins[ft].cparams.mtu_time;
+ }
+
+ q->rate_ns = q->tins[ft].tin_rate_ns;
+ q->rate_shft = q->tins[ft].tin_rate_shft;
+
+ if (q->buffer_config_limit) {
+ q->buffer_limit = q->buffer_config_limit;
+ } else if (q->rate_bps) {
+ u64 t = q->rate_bps * q->interval;
+
+ do_div(t, USEC_PER_SEC / 4);
+ q->buffer_limit = max_t(u32, t, 4U << 20);
+ } else {
+ q->buffer_limit = ~0;
+ }
+
+ sch->flags &= ~TCQ_F_CAN_BYPASS;
+
+ q->buffer_limit = min(q->buffer_limit,
+ max(sch->limit * psched_mtu(qdisc_dev(sch)),
+ q->buffer_config_limit));
+}
+
+static int cake_change(struct Qdisc *sch, struct nlattr *opt,
+ struct netlink_ext_ack *extack)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct nlattr *tb[TCA_CAKE_MAX + 1];
+ int err;
+
+ if (!opt)
+ return -EINVAL;
+
+ err = nla_parse_nested(tb, TCA_CAKE_MAX, opt, cake_policy, extack);
+ if (err < 0)
+ return err;
+
+ if (tb[TCA_CAKE_BASE_RATE64])
+ q->rate_bps = nla_get_u64(tb[TCA_CAKE_BASE_RATE64]);
+
+ if (tb[TCA_CAKE_FLOW_MODE])
+ q->flow_mode = (nla_get_u32(tb[TCA_CAKE_FLOW_MODE]) &
+ CAKE_FLOW_MASK);
+
+ if (tb[TCA_CAKE_RTT]) {
+ q->interval = nla_get_u32(tb[TCA_CAKE_RTT]);
+
+ if (!q->interval)
+ q->interval = 1;
+ }
+
+ if (tb[TCA_CAKE_TARGET]) {
+ q->target = nla_get_u32(tb[TCA_CAKE_TARGET]);
+
+ if (!q->target)
+ q->target = 1;
+ }
+
+ if (tb[TCA_CAKE_MEMORY])
+ q->buffer_config_limit = nla_get_u32(tb[TCA_CAKE_MEMORY]);
+
+ if (q->tins) {
+ sch_tree_lock(sch);
+ cake_reconfigure(sch);
+ sch_tree_unlock(sch);
+ }
+
+ return 0;
+}
+
+static void cake_destroy(struct Qdisc *sch)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+
+ qdisc_watchdog_cancel(&q->watchdog);
+ tcf_block_put(q->block);
+ kvfree(q->tins);
+}
+
+static int cake_init(struct Qdisc *sch, struct nlattr *opt,
+ struct netlink_ext_ack *extack)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ int i, j, err;
+
+ sch->limit = 10240;
+ q->tin_mode = CAKE_DIFFSERV_BESTEFFORT;
+ q->flow_mode = CAKE_FLOW_TRIPLE;
+
+ q->rate_bps = 0; /* unlimited by default */
+
+ q->interval = 100000; /* 100ms default */
+ q->target = 5000; /* 5ms: codel RFC argues
+ * for 5 to 10% of interval
+ */
+
+ q->cur_tin = 0;
+ q->cur_flow = 0;
+
+ if (opt) {
+ int err = cake_change(sch, opt, extack);
+
+ if (err)
+ return err;
+ }
+
+ err = tcf_block_get(&q->block, &q->filter_list, sch, extack);
+ if (err)
+ return err;
+
+ qdisc_watchdog_init(&q->watchdog, sch);
+
+ quantum_div[0] = ~0;
+ for (i = 1; i <= CAKE_QUEUES; i++)
+ quantum_div[i] = 65535 / i;
+
+ q->tins = kvzalloc(CAKE_MAX_TINS * sizeof(struct cake_tin_data),
+ GFP_KERNEL);
+ if (!q->tins)
+ goto nomem;
+
+ for (i = 0; i < CAKE_MAX_TINS; i++) {
+ struct cake_tin_data *b = q->tins + i;
+
+ INIT_LIST_HEAD(&b->new_flows);
+ INIT_LIST_HEAD(&b->old_flows);
+ INIT_LIST_HEAD(&b->decaying_flows);
+ b->sparse_flow_count = 0;
+ b->bulk_flow_count = 0;
+ b->decaying_flow_count = 0;
+
+ for (j = 0; j < CAKE_QUEUES; j++) {
+ struct cake_flow *flow = b->flows + j;
+ u32 k = j * CAKE_MAX_TINS + i;
+
+ INIT_LIST_HEAD(&flow->flowchain);
+ cobalt_vars_init(&flow->cvars);
+
+ q->overflow_heap[k].t = i;
+ q->overflow_heap[k].b = j;
+ b->overflow_idx[j] = k;
+ }
+ }
+
+ cake_reconfigure(sch);
+ q->avg_peak_bandwidth = q->rate_bps;
+ q->min_netlen = ~0;
+ q->min_adjlen = ~0;
+ return 0;
+
+nomem:
+ cake_destroy(sch);
+ return -ENOMEM;
+}
+
+static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct nlattr *opts;
+
+ opts = nla_nest_start(skb, TCA_OPTIONS);
+ if (!opts)
+ goto nla_put_failure;
+
+ if (nla_put_u64_64bit(skb, TCA_CAKE_BASE_RATE64, q->rate_bps,
+ TCA_CAKE_PAD))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_FLOW_MODE,
+ q->flow_mode & CAKE_FLOW_MASK))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_RTT, q->interval))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_TARGET, q->target))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_MEMORY, q->buffer_config_limit))
+ goto nla_put_failure;
+
+ return nla_nest_end(skb, opts);
+
+nla_put_failure:
+ return -1;
+}
+
+static int cake_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
+{
+ struct nlattr *stats = nla_nest_start(d->skb, TCA_STATS_APP);
+ struct cake_sched_data *q = qdisc_priv(sch);
+ struct nlattr *tstats, *ts;
+ int i;
+
+ if (!stats)
+ return -1;
+
+#define PUT_STAT_U32(attr, data) do { \
+ if (nla_put_u32(d->skb, TCA_CAKE_STATS_ ## attr, data)) \
+ goto nla_put_failure; \
+ } while (0)
+#define PUT_STAT_U64(attr, data) do { \
+ if (nla_put_u64_64bit(d->skb, TCA_CAKE_STATS_ ## attr, \
+ data, TCA_CAKE_STATS_PAD)) \
+ goto nla_put_failure; \
+ } while (0)
+
+ PUT_STAT_U64(CAPACITY_ESTIMATE64, q->avg_peak_bandwidth);
+ PUT_STAT_U32(MEMORY_LIMIT, q->buffer_limit);
+ PUT_STAT_U32(MEMORY_USED, q->buffer_max_used);
+ PUT_STAT_U32(AVG_NETOFF, ((q->avg_netoff + 0x8000) >> 16));
+ PUT_STAT_U32(MAX_NETLEN, q->max_netlen);
+ PUT_STAT_U32(MAX_ADJLEN, q->max_adjlen);
+ PUT_STAT_U32(MIN_NETLEN, q->min_netlen);
+ PUT_STAT_U32(MIN_ADJLEN, q->min_adjlen);
+
+#undef PUT_STAT_U32
+#undef PUT_STAT_U64
+
+ tstats = nla_nest_start(d->skb, TCA_CAKE_STATS_TIN_STATS);
+ if (!tstats)
+ goto nla_put_failure;
+
+#define PUT_TSTAT_U32(attr, data) do { \
+ if (nla_put_u32(d->skb, TCA_CAKE_TIN_STATS_ ## attr, data)) \
+ goto nla_put_failure; \
+ } while (0)
+#define PUT_TSTAT_U64(attr, data) do { \
+ if (nla_put_u64_64bit(d->skb, TCA_CAKE_TIN_STATS_ ## attr, \
+ data, TCA_CAKE_TIN_STATS_PAD)) \
+ goto nla_put_failure; \
+ } while (0)
+
+ for (i = 0; i < q->tin_cnt; i++) {
+ struct cake_tin_data *b = &q->tins[i];
+
+ ts = nla_nest_start(d->skb, i + 1);
+ if (!ts)
+ goto nla_put_failure;
+
+ PUT_TSTAT_U64(THRESHOLD_RATE64, b->tin_rate_bps);
+ PUT_TSTAT_U64(SENT_BYTES64, b->bytes);
+ PUT_TSTAT_U32(BACKLOG_BYTES, b->tin_backlog);
+
+ PUT_TSTAT_U32(TARGET_US,
+ ktime_to_us(ns_to_ktime(b->cparams.target)));
+ PUT_TSTAT_U32(INTERVAL_US,
+ ktime_to_us(ns_to_ktime(b->cparams.interval)));
+
+ PUT_TSTAT_U32(SENT_PACKETS, b->packets);
+ PUT_TSTAT_U32(DROPPED_PACKETS, b->tin_dropped);
+ PUT_TSTAT_U32(ECN_MARKED_PACKETS, b->tin_ecn_mark);
+ PUT_TSTAT_U32(ACKS_DROPPED_PACKETS, b->ack_drops);
+
+ PUT_TSTAT_U32(PEAK_DELAY_US,
+ ktime_to_us(ns_to_ktime(b->peak_delay)));
+ PUT_TSTAT_U32(AVG_DELAY_US,
+ ktime_to_us(ns_to_ktime(b->avge_delay)));
+ PUT_TSTAT_U32(BASE_DELAY_US,
+ ktime_to_us(ns_to_ktime(b->base_delay)));
+
+ PUT_TSTAT_U32(WAY_INDIRECT_HITS, b->way_hits);
+ PUT_TSTAT_U32(WAY_MISSES, b->way_misses);
+ PUT_TSTAT_U32(WAY_COLLISIONS, b->way_collisions);
+
+ PUT_TSTAT_U32(SPARSE_FLOWS, b->sparse_flow_count +
+ b->decaying_flow_count);
+ PUT_TSTAT_U32(BULK_FLOWS, b->bulk_flow_count);
+ PUT_TSTAT_U32(UNRESPONSIVE_FLOWS, b->unresponsive_flow_count);
+ PUT_TSTAT_U32(MAX_SKBLEN, b->max_skblen);
+
+ PUT_TSTAT_U32(FLOW_QUANTUM, b->flow_quantum);
+ nla_nest_end(d->skb, ts);
+ }
+
+#undef PUT_TSTAT_U32
+#undef PUT_TSTAT_U64
+
+ nla_nest_end(d->skb, tstats);
+ return nla_nest_end(d->skb, stats);
+
+nla_put_failure:
+ nla_nest_cancel(d->skb, stats);
+ return -1;
+}
+
+static struct Qdisc *cake_leaf(struct Qdisc *sch, unsigned long arg)
+{
+ return NULL;
+}
+
+static unsigned long cake_find(struct Qdisc *sch, u32 classid)
+{
+ return 0;
+}
+
+static unsigned long cake_bind(struct Qdisc *sch, unsigned long parent,
+ u32 classid)
+{
+ return 0;
+}
+
+static void cake_unbind(struct Qdisc *q, unsigned long cl)
+{
+}
+
+static struct tcf_block *cake_tcf_block(struct Qdisc *sch, unsigned long cl,
+ struct netlink_ext_ack *extack)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+
+ if (cl)
+ return NULL;
+ return q->block;
+}
+
+static int cake_dump_class(struct Qdisc *sch, unsigned long cl,
+ struct sk_buff *skb, struct tcmsg *tcm)
+{
+ tcm->tcm_handle |= TC_H_MIN(cl);
+ return 0;
+}
+
+static int cake_dump_class_stats(struct Qdisc *sch, unsigned long cl,
+ struct gnet_dump *d)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ const struct cake_flow *flow = NULL;
+ struct gnet_stats_queue qs = { 0 };
+ struct nlattr *stats;
+ u32 idx = cl - 1;
+
+ if (idx < CAKE_QUEUES * q->tin_cnt) {
+ const struct cake_tin_data *b = &q->tins[idx / CAKE_QUEUES];
+ const struct sk_buff *skb;
+
+ flow = &b->flows[idx % CAKE_QUEUES];
+
+ if (flow->head) {
+ sch_tree_lock(sch);
+ skb = flow->head;
+ while (skb) {
+ qs.qlen++;
+ skb = skb->next;
+ }
+ sch_tree_unlock(sch);
+ }
+ qs.backlog = b->backlogs[idx % CAKE_QUEUES];
+ qs.drops = flow->dropped;
+ }
+ if (gnet_stats_copy_queue(d, NULL, &qs, qs.qlen) < 0)
+ return -1;
+ if (flow) {
+ ktime_t now = ktime_get();
+
+ stats = nla_nest_start(d->skb, TCA_STATS_APP);
+ if (!stats)
+ return -1;
+
+#define PUT_STAT_U32(attr, data) do { \
+ if (nla_put_u32(d->skb, TCA_CAKE_STATS_ ## attr, data)) \
+ goto nla_put_failure; \
+ } while (0)
+#define PUT_STAT_S32(attr, data) do { \
+ if (nla_put_s32(d->skb, TCA_CAKE_STATS_ ## attr, data)) \
+ goto nla_put_failure; \
+ } while (0)
+
+ PUT_STAT_S32(DEFICIT, flow->deficit);
+ PUT_STAT_U32(DROPPING, flow->cvars.dropping);
+ PUT_STAT_U32(COBALT_COUNT, flow->cvars.count);
+ PUT_STAT_U32(P_DROP, flow->cvars.p_drop);
+ if (flow->cvars.p_drop) {
+ PUT_STAT_S32(BLUE_TIMER_US,
+ ktime_to_us(
+ ktime_sub(now,
+ flow->cvars.blue_timer)));
+ }
+ if (flow->cvars.dropping) {
+ PUT_STAT_S32(DROP_NEXT_US,
+ ktime_to_us(
+ ktime_sub(now,
+ flow->cvars.drop_next)));
+ }
+
+ if (nla_nest_end(d->skb, stats) < 0)
+ return -1;
+ }
+
+ return 0;
+
+nla_put_failure:
+ nla_nest_cancel(d->skb, stats);
+ return -1;
+}
+
+static void cake_walk(struct Qdisc *sch, struct qdisc_walker *arg)
+{
+ struct cake_sched_data *q = qdisc_priv(sch);
+ unsigned int i, j;
+
+ if (arg->stop)
+ return;
+
+ for (i = 0; i < q->tin_cnt; i++) {
+ struct cake_tin_data *b = &q->tins[i];
+
+ for (j = 0; j < CAKE_QUEUES; j++) {
+ if (list_empty(&b->flows[j].flowchain) ||
+ arg->count < arg->skip) {
+ arg->count++;
+ continue;
+ }
+ if (arg->fn(sch, i * CAKE_QUEUES + j + 1, arg) < 0) {
+ arg->stop = 1;
+ break;
+ }
+ arg->count++;
+ }
+ }
+}
+
+static const struct Qdisc_class_ops cake_class_ops = {
+ .leaf = cake_leaf,
+ .find = cake_find,
+ .tcf_block = cake_tcf_block,
+ .bind_tcf = cake_bind,
+ .unbind_tcf = cake_unbind,
+ .dump = cake_dump_class,
+ .dump_stats = cake_dump_class_stats,
+ .walk = cake_walk,
+};
+
+static struct Qdisc_ops cake_qdisc_ops __read_mostly = {
+ .cl_ops = &cake_class_ops,
+ .id = "cake",
+ .priv_size = sizeof(struct cake_sched_data),
+ .enqueue = cake_enqueue,
+ .dequeue = cake_dequeue,
+ .peek = qdisc_peek_dequeued,
+ .init = cake_init,
+ .reset = cake_reset,
+ .destroy = cake_destroy,
+ .change = cake_change,
+ .dump = cake_dump,
+ .dump_stats = cake_dump_stats,
+ .owner = THIS_MODULE,
+};
+
+static int __init cake_module_init(void)
+{
+ return register_qdisc(&cake_qdisc_ops);
+}
+
+static void __exit cake_module_exit(void)
+{
+ unregister_qdisc(&cake_qdisc_ops);
+}
+
+module_init(cake_module_init)
+module_exit(cake_module_exit)
+MODULE_AUTHOR("Jonathan Morton");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("The CAKE shaper.");
^ permalink raw reply related
* [PATCH net-next v17 0/8] sched: Add Common Applications Kept Enhanced (cake) qdisc
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev
Cc: Georgios Amanakis, Pete Heist, Yuchung Cheng, Neal Cardwell,
Dave Taht, netfilter-devel, cake
This patch series adds the CAKE qdisc, and has been split up to ease
review.
I have attempted to split out each configurable feature into its own patch.
The first commit adds the base shaper and packet scheduler, while
subsequent commits add the optional features. The full userspace API and
most data structures are included in this commit, but options not
understood in the base version will be ignored.
The result of applying the entire series is identical to the out of tree
version that have seen extensive testing in previous deployments, most
notably as an out of tree patch to OpenWrt. However, note that I have only
compile tested the individual patches; so the whole series should be
considered as a unit.
---
Changelog
v17:
- Rebase to newest net-next and move the conntrack callback to
nf_ct_hook
- Fix a compile error when NF_CONNTRACK is unset.
v16:
- Move conntrack lookup function into conntrack core and read it via
RCU so it is only active when the nf_conntrack module is loaded.
This avoids the module dependency on conntrack for NAT mode. Thanks
to Pablo for the idea.
v15:
- Handle ECN flags in ACK filter
v14:
- Handle seqno wraps and DSACKs in ACK filter
v13:
- Avoid ktime_t to scalar compares
- Add class dumping and basic stats
- Fail with ENOTSUPP when requesting NAT mode and conntrack is not
available.
- Parse all TCP options in ACK filter and make sure to only drop safe
ones. Also handle SACK ranges properly.
v12:
- Get rid of custom time typedefs. Use ktime_t for time and u64 for
duration instead.
v11:
- Fix overhead compensation calculation for GSO packets
- Change configured rate to be u64 (I ran out of bits before I ran out
of CPU when testing the effects of the above)
v10:
- Christmas tree gardening (fix variable declarations to be in reverse
line length order)
v9:
- Remove duplicated checks around kvfree() and just call it
unconditionally.
- Don't pass __GFP_NOWARN when allocating memory
- Move options in cake_dump() that are related to optional features to
later patches implementing the features.
- Support attaching filters to the qdisc and use the classification
result to select flow queue.
- Support overriding diffserv priority tin from skb->priority
v8:
- Remove inline keyword from function definitions
- Simplify ACK filter; remove the complex state handling to make the
logic easier to follow. This will potentially be a bit less efficient,
but I have not been able to measure a difference.
v7:
- Split up patch into a series to ease review.
- Constify the ACK filter.
v6:
- Fix 6in4 encapsulation checks in ACK filter code
- Checkpatch fixes
v5:
- Refactor ACK filter code and hopefully fix the safety issues
properly this time.
v4:
- Only split GSO packets if shaping at speeds <= 1Gbps
- Fix overhead calculation code to also work for GSO packets
- Don't re-implement kvzalloc()
- Remove local header include from out-of-tree build (fixes kbuild-bot
complaint).
- Several fixes to the ACK filter:
- Check pskb_may_pull() before deref of transport headers.
- Don't run ACK filter logic on split GSO packets
- Fix TCP sequence number compare to deal with wraparounds
v3:
- Use IS_REACHABLE() macro to fix compilation when sch_cake is
built-in and conntrack is a module.
- Switch the stats output to use nested netlink attributes instead
of a versioned struct.
- Remove GPL boilerplate.
- Fix array initialisation style.
v2:
- Fix kbuild test bot complaint
- Clean up the netlink ABI
- Fix checkpatch complaints
- A few tweaks to the behaviour of cake based on testing carried out
while writing the paper.
---
Toke Høiland-Jørgensen (8):
sched: Add Common Applications Kept Enhanced (cake) qdisc
sch_cake: Add ingress mode
sch_cake: Add optional ACK filter
netfilter: Add nf_ct_get_tuple_skb global lookup function
sch_cake: Add NAT awareness to packet classifier
sch_cake: Add DiffServ handling
sch_cake: Add overhead compensation support to the rate shaper
sch_cake: Conditionally split GSO segments
include/linux/netfilter.h | 11
include/uapi/linux/pkt_sched.h | 113 +
net/netfilter/core.c | 15
net/netfilter/nf_conntrack_core.c | 36
net/sched/Kconfig | 11
net/sched/Makefile | 1
net/sched/sch_cake.c | 2987 +++++++++++++++++++++++++++++++++++++
7 files changed, 3174 insertions(+)
create mode 100644 net/sched/sch_cake.c
^ permalink raw reply
* [PATCH net-next v17 2/8] sch_cake: Add ingress mode
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
The ingress mode is meant to be enabled when CAKE runs downlink of the
actual bottleneck (such as on an IFB device). The mode changes the shaper
to also account dropped packets to the shaped rate, as these have already
traversed the bottleneck.
Enabling ingress mode will also tune the AQM to always keep at least two
packets queued *for each flow*. This is done by scaling the minimum queue
occupancy level that will disable the AQM by the number of active bulk
flows. The rationale for this is that retransmits are more expensive in
ingress mode, since dropped packets have to traverse the bottleneck again
when they are retransmitted; thus, being more lenient and keeping a minimum
number of packets queued will improve throughput in cases where the number
of active flows are so large that they saturate the bottleneck even at
their minimum window size.
This commit also adds a separate switch to enable ingress mode rate
autoscaling. If enabled, the autoscaling code will observe the actual
traffic rate and adjust the shaper rate to match it. This can help avoid
latency increases in the case where the actual bottleneck rate decreases
below the shaped rate. The scaling filters out spikes by an EWMA filter.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 85 ++++++++++++++++++++++++++++++++++++++++++++++++--
1 file changed, 81 insertions(+), 4 deletions(-)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index 7ea4aa261cec..10e208e4255d 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -435,7 +435,8 @@ static bool cobalt_queue_empty(struct cobalt_vars *vars,
static bool cobalt_should_drop(struct cobalt_vars *vars,
struct cobalt_params *p,
ktime_t now,
- struct sk_buff *skb)
+ struct sk_buff *skb,
+ u32 bulk_flows)
{
bool next_due, over_target, drop = false;
ktime_t schedule;
@@ -459,6 +460,7 @@ static bool cobalt_should_drop(struct cobalt_vars *vars,
sojourn = ktime_to_ns(ktime_sub(now, cobalt_get_enqueue_time(skb)));
schedule = ktime_sub(now, vars->drop_next);
over_target = sojourn > p->target &&
+ sojourn > p->mtu_time * bulk_flows * 2 &&
sojourn > p->mtu_time * 4;
next_due = vars->count && ktime_to_ns(schedule) >= 0;
@@ -913,6 +915,9 @@ static unsigned int cake_drop(struct Qdisc *sch, struct sk_buff **to_free)
b->tin_dropped++;
sch->qstats.drops++;
+ if (q->rate_flags & CAKE_FLAG_INGRESS)
+ cake_advance_shaper(q, b, skb, now, true);
+
__qdisc_drop(skb, to_free);
sch->q.qlen--;
@@ -990,8 +995,46 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
cake_heapify_up(q, b->overflow_idx[idx]);
/* incoming bandwidth capacity estimate */
- q->avg_window_bytes = 0;
- q->last_packet_time = now;
+ if (q->rate_flags & CAKE_FLAG_AUTORATE_INGRESS) {
+ u64 packet_interval = \
+ ktime_to_ns(ktime_sub(now, q->last_packet_time));
+
+ if (packet_interval > NSEC_PER_SEC)
+ packet_interval = NSEC_PER_SEC;
+
+ /* filter out short-term bursts, eg. wifi aggregation */
+ q->avg_packet_interval = \
+ cake_ewma(q->avg_packet_interval,
+ packet_interval,
+ (packet_interval > q->avg_packet_interval ?
+ 2 : 8));
+
+ q->last_packet_time = now;
+
+ if (packet_interval > q->avg_packet_interval) {
+ u64 window_interval = \
+ ktime_to_ns(ktime_sub(now,
+ q->avg_window_begin));
+ u64 b = q->avg_window_bytes * (u64)NSEC_PER_SEC;
+
+ do_div(b, window_interval);
+ q->avg_peak_bandwidth =
+ cake_ewma(q->avg_peak_bandwidth, b,
+ b > q->avg_peak_bandwidth ? 2 : 8);
+ q->avg_window_bytes = 0;
+ q->avg_window_begin = now;
+
+ if (ktime_after(now,
+ ktime_add_ms(q->last_reconfig_time,
+ 250))) {
+ q->rate_bps = (q->avg_peak_bandwidth * 15) >> 4;
+ cake_reconfigure(sch);
+ }
+ }
+ } else {
+ q->avg_window_bytes = 0;
+ q->last_packet_time = now;
+ }
/* flowchain */
if (!flow->set || flow->set == CAKE_SET_DECAYING) {
@@ -1251,15 +1294,27 @@ static struct sk_buff *cake_dequeue(struct Qdisc *sch)
}
/* Last packet in queue may be marked, shouldn't be dropped */
- if (!cobalt_should_drop(&flow->cvars, &b->cparams, now, skb) ||
+ if (!cobalt_should_drop(&flow->cvars, &b->cparams, now, skb,
+ (b->bulk_flow_count *
+ !!(q->rate_flags &
+ CAKE_FLAG_INGRESS))) ||
!flow->head)
break;
+ /* drop this packet, get another one */
+ if (q->rate_flags & CAKE_FLAG_INGRESS) {
+ len = cake_advance_shaper(q, b, skb,
+ now, true);
+ flow->deficit -= len;
+ b->tin_deficit -= len;
+ }
flow->dropped++;
b->tin_dropped++;
qdisc_tree_reduce_backlog(sch, 1, qdisc_pkt_len(skb));
qdisc_qstats_drop(sch);
kfree_skb(skb);
+ if (q->rate_flags & CAKE_FLAG_INGRESS)
+ goto retry;
}
b->tin_ecn_mark += !!flow->cvars.ecn_marked;
@@ -1442,6 +1497,20 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
q->target = 1;
}
+ if (tb[TCA_CAKE_AUTORATE]) {
+ if (!!nla_get_u32(tb[TCA_CAKE_AUTORATE]))
+ q->rate_flags |= CAKE_FLAG_AUTORATE_INGRESS;
+ else
+ q->rate_flags &= ~CAKE_FLAG_AUTORATE_INGRESS;
+ }
+
+ if (tb[TCA_CAKE_INGRESS]) {
+ if (!!nla_get_u32(tb[TCA_CAKE_INGRESS]))
+ q->rate_flags |= CAKE_FLAG_INGRESS;
+ else
+ q->rate_flags &= ~CAKE_FLAG_INGRESS;
+ }
+
if (tb[TCA_CAKE_MEMORY])
q->buffer_config_limit = nla_get_u32(tb[TCA_CAKE_MEMORY]);
@@ -1565,6 +1634,14 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
if (nla_put_u32(skb, TCA_CAKE_MEMORY, q->buffer_config_limit))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_AUTORATE,
+ !!(q->rate_flags & CAKE_FLAG_AUTORATE_INGRESS)))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_INGRESS,
+ !!(q->rate_flags & CAKE_FLAG_INGRESS)))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
^ permalink raw reply related
* [PATCH net-next v17 7/8] sch_cake: Add overhead compensation support to the rate shaper
From: Toke Høiland-Jørgensen @ 2018-05-28 22:28 UTC (permalink / raw)
To: netdev; +Cc: cake
In-Reply-To: <152754646847.18169.14016621463603092505.stgit@alrua-kau>
This commit adds configurable overhead compensation support to the rate
shaper. With this feature, userspace can configure the actual bottleneck
link overhead and encapsulation mode used, which will be used by the shaper
to calculate the precise duration of each packet on the wire.
This feature is needed because CAKE is often deployed one or two hops
upstream of the actual bottleneck (which can be, e.g., inside a DSL or
cable modem). In this case, the link layer characteristics and overhead
reported by the kernel does not match the actual bottleneck. Being able to
set the actual values in use makes it possible to configure the shaper rate
much closer to the actual bottleneck rate (our experience shows it is
possible to get with 0.1% of the actual physical bottleneck rate), thus
keeping latency low without sacrificing bandwidth.
The overhead compensation has three tunables: A fixed per-packet overhead
size (which, if set, will be accounted from the IP packet header), a
minimum packet size (MPU) and a framing mode supporting either ATM or PTM
framing. We include a set of common keywords in TC to help users configure
the right parameters. If no overhead value is set, the value reported by
the kernel is used.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
---
net/sched/sch_cake.c | 124 ++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 123 insertions(+), 1 deletion(-)
diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
index b2760bc1a34f..10a968d43d44 100644
--- a/net/sched/sch_cake.c
+++ b/net/sched/sch_cake.c
@@ -270,6 +270,7 @@ enum {
struct cobalt_skb_cb {
ktime_t enqueue_time;
+ u32 adjusted_len;
};
static u64 us_to_ns(u64 us)
@@ -1282,6 +1283,88 @@ static u64 cake_ewma(u64 avg, u64 sample, u32 shift)
return avg;
}
+static u32 cake_calc_overhead(struct cake_sched_data *q, u32 len, u32 off)
+{
+ if (q->rate_flags & CAKE_FLAG_OVERHEAD)
+ len -= off;
+
+ if (q->max_netlen < len)
+ q->max_netlen = len;
+ if (q->min_netlen > len)
+ q->min_netlen = len;
+
+ len += q->rate_overhead;
+
+ if (len < q->rate_mpu)
+ len = q->rate_mpu;
+
+ if (q->atm_mode == CAKE_ATM_ATM) {
+ len += 47;
+ len /= 48;
+ len *= 53;
+ } else if (q->atm_mode == CAKE_ATM_PTM) {
+ /* Add one byte per 64 bytes or part thereof.
+ * This is conservative and easier to calculate than the
+ * precise value.
+ */
+ len += (len + 63) / 64;
+ }
+
+ if (q->max_adjlen < len)
+ q->max_adjlen = len;
+ if (q->min_adjlen > len)
+ q->min_adjlen = len;
+
+ return len;
+}
+
+static u32 cake_overhead(struct cake_sched_data *q, const struct sk_buff *skb)
+{
+ const struct skb_shared_info *shinfo = skb_shinfo(skb);
+ unsigned int hdr_len, last_len = 0;
+ u32 off = skb_network_offset(skb);
+ u32 len = qdisc_pkt_len(skb);
+ u16 segs = 1;
+
+ q->avg_netoff = cake_ewma(q->avg_netoff, off << 16, 8);
+
+ if (!shinfo->gso_size)
+ return cake_calc_overhead(q, len, off);
+
+ /* borrowed from qdisc_pkt_len_init() */
+ hdr_len = skb_transport_header(skb) - skb_mac_header(skb);
+
+ /* + transport layer */
+ if (likely(shinfo->gso_type & (SKB_GSO_TCPV4 |
+ SKB_GSO_TCPV6))) {
+ const struct tcphdr *th;
+ struct tcphdr _tcphdr;
+
+ th = skb_header_pointer(skb, skb_transport_offset(skb),
+ sizeof(_tcphdr), &_tcphdr);
+ if (likely(th))
+ hdr_len += __tcp_hdrlen(th);
+ } else {
+ struct udphdr _udphdr;
+
+ if (skb_header_pointer(skb, skb_transport_offset(skb),
+ sizeof(_udphdr), &_udphdr))
+ hdr_len += sizeof(struct udphdr);
+ }
+
+ if (unlikely(shinfo->gso_type & SKB_GSO_DODGY))
+ segs = DIV_ROUND_UP(skb->len - hdr_len,
+ shinfo->gso_size);
+ else
+ segs = shinfo->gso_segs;
+
+ len = shinfo->gso_size + hdr_len;
+ last_len = skb->len - shinfo->gso_size * (segs - 1);
+
+ return (cake_calc_overhead(q, len, off) * (segs - 1) +
+ cake_calc_overhead(q, last_len, off));
+}
+
static void cake_heap_swap(struct cake_sched_data *q, u16 i, u16 j)
{
struct cake_heap_entry ii = q->overflow_heap[i];
@@ -1359,7 +1442,7 @@ static int cake_advance_shaper(struct cake_sched_data *q,
struct sk_buff *skb,
ktime_t now, bool drop)
{
- u32 len = qdisc_pkt_len(skb);
+ u32 len = get_cobalt_cb(skb)->adjusted_len;
/* charge packet bandwidth to this tin
* and to the global shaper.
@@ -1556,6 +1639,7 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
b->max_skblen = len;
cobalt_set_enqueue_time(skb, now);
+ get_cobalt_cb(skb)->adjusted_len = cake_overhead(q, skb);
flow_queue_add(flow, skb);
if (q->ack_filter)
@@ -2356,6 +2440,31 @@ static int cake_change(struct Qdisc *sch, struct nlattr *opt,
q->flow_mode = (nla_get_u32(tb[TCA_CAKE_FLOW_MODE]) &
CAKE_FLOW_MASK);
+ if (tb[TCA_CAKE_ATM])
+ q->atm_mode = nla_get_u32(tb[TCA_CAKE_ATM]);
+
+ if (tb[TCA_CAKE_OVERHEAD]) {
+ q->rate_overhead = nla_get_s32(tb[TCA_CAKE_OVERHEAD]);
+ q->rate_flags |= CAKE_FLAG_OVERHEAD;
+
+ q->max_netlen = 0;
+ q->max_adjlen = 0;
+ q->min_netlen = ~0;
+ q->min_adjlen = ~0;
+ }
+
+ if (tb[TCA_CAKE_RAW]) {
+ q->rate_flags &= ~CAKE_FLAG_OVERHEAD;
+
+ q->max_netlen = 0;
+ q->max_adjlen = 0;
+ q->min_netlen = ~0;
+ q->min_adjlen = ~0;
+ }
+
+ if (tb[TCA_CAKE_MPU])
+ q->rate_mpu = nla_get_u32(tb[TCA_CAKE_MPU]);
+
if (tb[TCA_CAKE_RTT]) {
q->interval = nla_get_u32(tb[TCA_CAKE_RTT]);
@@ -2532,6 +2641,19 @@ static int cake_dump(struct Qdisc *sch, struct sk_buff *skb)
!!(q->rate_flags & CAKE_FLAG_WASH)))
goto nla_put_failure;
+ if (nla_put_u32(skb, TCA_CAKE_OVERHEAD, q->rate_overhead))
+ goto nla_put_failure;
+
+ if (!(q->rate_flags & CAKE_FLAG_OVERHEAD))
+ if (nla_put_u32(skb, TCA_CAKE_RAW, 0))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_ATM, q->atm_mode))
+ goto nla_put_failure;
+
+ if (nla_put_u32(skb, TCA_CAKE_MPU, q->rate_mpu))
+ goto nla_put_failure;
+
return nla_nest_end(skb, opts);
nla_put_failure:
^ permalink raw reply related
* Re: [PATCH net-next v16 5/8] sch_cake: Add NAT awareness to packet classifier
From: Toke Høiland-Jørgensen @ 2018-05-28 22:19 UTC (permalink / raw)
To: Pablo Neira Ayuso; +Cc: netdev, cake, netfilter-devel
In-Reply-To: <20180528195101.7mceghxngz5jhawa@salvia>
Pablo Neira Ayuso <pablo@netfilter.org> writes:
> On Mon, May 28, 2018 at 04:27:46PM +0200, Toke Høiland-Jørgensen wrote:
>> When CAKE is deployed on a gateway that also performs NAT (which is a
>> common deployment mode), the host fairness mechanism cannot distinguish
>> internal hosts from each other, and so fails to work correctly.
>>
>> To fix this, we add an optional NAT awareness mode, which will query the
>> kernel conntrack mechanism to obtain the pre-NAT addresses for each packet
>> and use that in the flow and host hashing.
>>
>> When the shaper is enabled and the host is already performing NAT, the cost
>> of this lookup is negligible. However, in unlimited mode with no NAT being
>> performed, there is a significant CPU cost at higher bandwidths. For this
>> reason, the feature is turned off by default.
>>
>> Cc: netfilter-devel@vger.kernel.org
>> Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
>> ---
>> net/sched/sch_cake.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 46 insertions(+)
>>
>> diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
>> index 68ac908470f1..fecd9caac0cc 100644
>> --- a/net/sched/sch_cake.c
>> +++ b/net/sched/sch_cake.c
>> @@ -71,6 +71,10 @@
>> #include <net/tcp.h>
>> #include <net/flow_dissector.h>
>>
>> +#if IS_ENABLED(CONFIG_NF_CONNTRACK)
>> +#include <net/netfilter/nf_conntrack_core.h>
>> +#endif
>> +
>> #define CAKE_SET_WAYS (8)
>> #define CAKE_MAX_TINS (8)
>> #define CAKE_QUEUES (1024)
>> @@ -516,6 +520,29 @@ static bool cobalt_should_drop(struct cobalt_vars *vars,
>> return drop;
>> }
>>
>> +static void cake_update_flowkeys(struct flow_keys *keys,
>> + const struct sk_buff *skb)
>> +{
>> +#if IS_ENABLED(CONFIG_NF_CONNTRACK)
>
> I would remove the ifdef, not really needed, it will simplify things.
>
> But I leave it to you to decide, this is not I deal breaker.
If I remove it I get a bunch of 'incomplete type' errors when compiling.
Besides, we use it to report an error to userspace when conntrack is
disabled anyway, so might as well keep the whole thing ifdef'ed.
-Toke
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox