* [PATCH net-next v3 4/7] dsa: port: Ignore bridge VLAN events
From: Petr Machata @ 2018-05-28 15:10 UTC (permalink / raw)
To: netdev, devel, bridge
Cc: f.fainelli, andrew, nikolay, gregkh, vivien.didelot, idosch, jiri,
razvan.stefanescu, davem
In-Reply-To: <cover.1527519997.git.petrm@mellanox.com>
Ignore VLAN events where the orig_dev is the bridge device itself.
Signed-off-by: Petr Machata <petrm@mellanox.com>
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
---
net/dsa/port.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/net/dsa/port.c b/net/dsa/port.c
index 2413beb..ed05954 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -252,6 +252,9 @@ int dsa_port_vlan_add(struct dsa_port *dp,
.vlan = vlan,
};
+ if (netif_is_bridge_master(vlan->obj.orig_dev))
+ return -EOPNOTSUPP;
+
if (br_vlan_enabled(dp->bridge_dev))
return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info);
@@ -267,6 +270,9 @@ int dsa_port_vlan_del(struct dsa_port *dp,
.vlan = vlan,
};
+ if (netif_is_bridge_master(vlan->obj.orig_dev))
+ return -EOPNOTSUPP;
+
if (br_vlan_enabled(dp->bridge_dev))
return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_DEL, &info);
--
2.4.11
^ permalink raw reply related
* [PATCH net-next v3 3/7] rocker: rocker_main: Ignore bridge VLAN events
From: Petr Machata @ 2018-05-28 15:10 UTC (permalink / raw)
To: netdev, devel, bridge
Cc: f.fainelli, andrew, nikolay, gregkh, vivien.didelot, idosch, jiri,
razvan.stefanescu, davem
In-Reply-To: <cover.1527519997.git.petrm@mellanox.com>
Ignore VLAN events where the orig_dev is the bridge device itself.
Signed-off-by: Petr Machata <petrm@mellanox.com>
---
drivers/net/ethernet/rocker/rocker_main.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/net/ethernet/rocker/rocker_main.c b/drivers/net/ethernet/rocker/rocker_main.c
index e73e4fe..aeafdb9 100644
--- a/drivers/net/ethernet/rocker/rocker_main.c
+++ b/drivers/net/ethernet/rocker/rocker_main.c
@@ -1632,6 +1632,9 @@ rocker_world_port_obj_vlan_add(struct rocker_port *rocker_port,
{
struct rocker_world_ops *wops = rocker_port->rocker->wops;
+ if (netif_is_bridge_master(vlan->obj.orig_dev))
+ return -EOPNOTSUPP;
+
if (!wops->port_obj_vlan_add)
return -EOPNOTSUPP;
@@ -1647,6 +1650,9 @@ rocker_world_port_obj_vlan_del(struct rocker_port *rocker_port,
{
struct rocker_world_ops *wops = rocker_port->rocker->wops;
+ if (netif_is_bridge_master(vlan->obj.orig_dev))
+ return -EOPNOTSUPP;
+
if (!wops->port_obj_vlan_del)
return -EOPNOTSUPP;
return wops->port_obj_vlan_del(rocker_port, vlan);
--
2.4.11
^ permalink raw reply related
* [PATCH net-next v3 2/7] mlxsw: spectrum_switchdev: Ignore bridge VLAN events
From: Petr Machata @ 2018-05-28 15:10 UTC (permalink / raw)
To: netdev, devel, bridge
Cc: f.fainelli, andrew, nikolay, gregkh, vivien.didelot, idosch, jiri,
razvan.stefanescu, davem
In-Reply-To: <cover.1527519997.git.petrm@mellanox.com>
Ignore VLAN events where the orig_dev is the bridge device itself.
Signed-off-by: Petr Machata <petrm@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
index 8c9cf8e..cbc8fab 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
@@ -1144,6 +1144,9 @@ static int mlxsw_sp_port_vlans_add(struct mlxsw_sp_port *mlxsw_sp_port,
struct mlxsw_sp_bridge_port *bridge_port;
u16 vid;
+ if (netif_is_bridge_master(orig_dev))
+ return -EOPNOTSUPP;
+
if (switchdev_trans_ph_prepare(trans))
return 0;
@@ -1741,6 +1744,9 @@ static int mlxsw_sp_port_vlans_del(struct mlxsw_sp_port *mlxsw_sp_port,
struct mlxsw_sp_bridge_port *bridge_port;
u16 vid;
+ if (netif_is_bridge_master(orig_dev))
+ return -EOPNOTSUPP;
+
bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
if (WARN_ON(!bridge_port))
return -EINVAL;
--
2.4.11
^ permalink raw reply related
* [PATCH net-next v3 1/7] net: bridge: Extract boilerplate around switchdev_port_obj_*()
From: Petr Machata @ 2018-05-28 15:10 UTC (permalink / raw)
To: netdev, devel, bridge
Cc: f.fainelli, andrew, nikolay, gregkh, vivien.didelot, idosch, jiri,
razvan.stefanescu, davem
In-Reply-To: <cover.1527519997.git.petrm@mellanox.com>
A call to switchdev_port_obj_add() or switchdev_port_obj_del() involves
initializing a struct switchdev_obj_port_vlan, a piece of code that
repeats on each call site almost verbatim. While in the current codebase
there is just one duplicated add call, the follow-up patches add more of
both add and del calls.
Thus to remove the duplication, extract the repetition into named
functions and reuse.
Signed-off-by: Petr Machata <petrm@mellanox.com>
---
net/bridge/br_private.h | 13 +++++++++++++
net/bridge/br_switchdev.c | 25 +++++++++++++++++++++++++
net/bridge/br_vlan.c | 31 ++++---------------------------
3 files changed, 42 insertions(+), 27 deletions(-)
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 11520ed..5216a52 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -1139,6 +1139,8 @@ int br_switchdev_set_port_flag(struct net_bridge_port *p,
unsigned long mask);
void br_switchdev_fdb_notify(const struct net_bridge_fdb_entry *fdb,
int type);
+int br_switchdev_port_vlan_add(struct net_device *dev, u16 vid, u16 flags);
+int br_switchdev_port_vlan_del(struct net_device *dev, u16 vid);
static inline void br_switchdev_frame_unmark(struct sk_buff *skb)
{
@@ -1168,6 +1170,17 @@ static inline int br_switchdev_set_port_flag(struct net_bridge_port *p,
return 0;
}
+static inline int br_switchdev_port_vlan_add(struct net_device *dev,
+ u16 vid, u16 flags)
+{
+ return -EOPNOTSUPP;
+}
+
+static inline int br_switchdev_port_vlan_del(struct net_device *dev, u16 vid)
+{
+ return -EOPNOTSUPP;
+}
+
static inline void
br_switchdev_fdb_notify(const struct net_bridge_fdb_entry *fdb, int type)
{
diff --git a/net/bridge/br_switchdev.c b/net/bridge/br_switchdev.c
index 35474d4..d77f807 100644
--- a/net/bridge/br_switchdev.c
+++ b/net/bridge/br_switchdev.c
@@ -136,3 +136,28 @@ br_switchdev_fdb_notify(const struct net_bridge_fdb_entry *fdb, int type)
break;
}
}
+
+int br_switchdev_port_vlan_add(struct net_device *dev, u16 vid, u16 flags)
+{
+ struct switchdev_obj_port_vlan v = {
+ .obj.orig_dev = dev,
+ .obj.id = SWITCHDEV_OBJ_ID_PORT_VLAN,
+ .flags = flags,
+ .vid_begin = vid,
+ .vid_end = vid,
+ };
+
+ return switchdev_port_obj_add(dev, &v.obj);
+}
+
+int br_switchdev_port_vlan_del(struct net_device *dev, u16 vid)
+{
+ struct switchdev_obj_port_vlan v = {
+ .obj.orig_dev = dev,
+ .obj.id = SWITCHDEV_OBJ_ID_PORT_VLAN,
+ .vid_begin = vid,
+ .vid_end = vid,
+ };
+
+ return switchdev_port_obj_del(dev, &v.obj);
+}
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index dc832c09..8ad5756 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -82,19 +82,10 @@ static bool __vlan_add_flags(struct net_bridge_vlan *v, u16 flags)
static int __vlan_vid_add(struct net_device *dev, struct net_bridge *br,
u16 vid, u16 flags)
{
- struct switchdev_obj_port_vlan v = {
- .obj.orig_dev = dev,
- .obj.id = SWITCHDEV_OBJ_ID_PORT_VLAN,
- .flags = flags,
- .vid_begin = vid,
- .vid_end = vid,
- };
- int err;
-
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
- err = switchdev_port_obj_add(dev, &v.obj);
+ int err = br_switchdev_port_vlan_add(dev, vid, flags);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, vid);
return err;
@@ -130,18 +121,11 @@ static void __vlan_del_list(struct net_bridge_vlan *v)
static int __vlan_vid_del(struct net_device *dev, struct net_bridge *br,
u16 vid)
{
- struct switchdev_obj_port_vlan v = {
- .obj.orig_dev = dev,
- .obj.id = SWITCHDEV_OBJ_ID_PORT_VLAN,
- .vid_begin = vid,
- .vid_end = vid,
- };
- int err;
-
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q del.
*/
- err = switchdev_port_obj_del(dev, &v.obj);
+ int err = br_switchdev_port_vlan_del(dev, vid);
+
if (err == -EOPNOTSUPP) {
vlan_vid_del(dev, br->vlan_proto, vid);
return 0;
@@ -1053,13 +1037,6 @@ int nbp_vlan_init(struct net_bridge_port *p)
int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags,
bool *changed)
{
- struct switchdev_obj_port_vlan v = {
- .obj.orig_dev = port->dev,
- .obj.id = SWITCHDEV_OBJ_ID_PORT_VLAN,
- .flags = flags,
- .vid_begin = vid,
- .vid_end = vid,
- };
struct net_bridge_vlan *vlan;
int ret;
@@ -1069,7 +1046,7 @@ int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags,
vlan = br_vlan_find(nbp_vlan_group(port), vid);
if (vlan) {
/* Pass the flags to the hardware bridge */
- ret = switchdev_port_obj_add(port->dev, &v.obj);
+ ret = br_switchdev_port_vlan_add(port->dev, vid, flags);
if (ret && ret != -EOPNOTSUPP)
return ret;
*changed = __vlan_add_flags(vlan, flags);
--
2.4.11
^ permalink raw reply related
* [PATCH net-next v3 0/7] net: bridge: Notify about bridge VLANs
From: Petr Machata @ 2018-05-28 15:10 UTC (permalink / raw)
To: netdev, devel, bridge
Cc: f.fainelli, andrew, nikolay, gregkh, vivien.didelot, idosch, jiri,
razvan.stefanescu, davem
In commit 946a11e7408e ("mlxsw: spectrum_span: Allow bridge for gretap
mirror"), mlxsw got support for offloading mirror-to-gretap such that
the underlay packet path involves a bridge. In that case, the offload is
also influenced by PVID setting of said bridge. However, changes to VLAN
configuration of the bridge itself do not generate switchdev
notifications, so there's no mechanism to prod mlxsw to update the
offload when these settings change.
In this patchset, the problem is resolved by distributing the switchdev
notification SWITCHDEV_OBJ_ID_PORT_VLAN also for configuration changes
on bridge VLANs. Since stacked devices distribute the notification to
lower devices, such event eventually reaches the driver, which can
determine whether it's a bridge or port VLAN by inspecting orig_dev.
To keep things consistent, the newly-distributed notifications observe
the same protocol as the existing ones: dual prepare/commit, with
-EOPNOTSUPP indicating lack of support, even though there's currently
nothing to prepare for and nothing to support. Correspondingly, all
switchdev drivers have been updated to return -EOPNOTSUPP for bridge
VLAN notifications.
In patch #1, the code to send notifications for adding and deleting is
factored out into two named functions.
In patches #2-#5, respectively for mlxsw, rocker, DSA and DPAA2 ethsw,
the new notifications (which are not enabled yet) are ignored to
maintain the current behavior.
In patch #6, the notification is actually enabled.
In patch #7, mlxsw is changed to update offloads of mirror-to-gre also
for bridge-related notifications.
Changes from v2 to v3:
- Add a fallback definition for br_switchdev_port_obj_add() and
br_switchdev_port_obj_del() when !CONFIG_NET_SWITCHDEV.
Changes from v1 to v2:
- Rename br_switchdev_port_obj_add() and br_switchdev_port_obj_del() to
br_switchdev_port_vlan_add() and br_switchdev_port_vlan_del(), and
move from br_vlan.c to br_switchdev.c.
Petr Machata (7):
net: bridge: Extract boilerplate around switchdev_port_obj_*()
mlxsw: spectrum_switchdev: Ignore bridge VLAN events
rocker: rocker_main: Ignore bridge VLAN events
dsa: port: Ignore bridge VLAN events
staging: fsl-dpaa2: ethsw: Ignore bridge VLAN events
net: bridge: Notify about bridge VLANs
mlxsw: spectrum_switchdev: Schedule respin during trans prepare
.../ethernet/mellanox/mlxsw/spectrum_switchdev.c | 8 +++-
drivers/net/ethernet/rocker/rocker_main.c | 6 +++
drivers/staging/fsl-dpaa2/ethsw/ethsw.c | 6 +++
net/bridge/br_private.h | 13 +++++++
net/bridge/br_switchdev.c | 25 ++++++++++++
net/bridge/br_vlan.c | 45 +++++++++-------------
net/dsa/port.c | 6 +++
7 files changed, 81 insertions(+), 28 deletions(-)
--
2.4.11
^ permalink raw reply
* Re: [PATCH v3 09/11] net: sched: use reference counting action init
From: Jiri Pirko @ 2018-05-28 15:08 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-10-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:27PM CEST, vladbu@mellanox.com wrote:
>Change action API to assume that action init function always takes
>reference to action, even when overwriting existing action. This is
>necessary because action API continues to use action pointer after init
>function is done. At this point action becomes accessible for concurrent
>modifications, so user must always hold reference to it.
>
>Implement helper put list function to atomically release list of actions
>after action API init code is done using them.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* Re: [PATCH v4 net-next 00/19] inet: frags: bring rhashtables to IP defrag
From: Alexander Aring @ 2018-05-28 14:52 UTC (permalink / raw)
To: Tariq Toukan
Cc: David Miller, edumazet, netdev, fw, herbert, tgraf, brouer,
alex.aring, stefan, ktkhai, eric.dumazet, Moshe Shemesh,
Eran Ben Elisha
In-Reply-To: <9329021e-2d77-7e90-b0e2-8b391508f6cb@mellanox.com>
Hi,
On Mon, May 28, 2018 at 12:12:42PM +0300, Tariq Toukan wrote:
>
>
> On 01/04/2018 6:25 AM, David Miller wrote:
> > From: Eric Dumazet <edumazet@google.com>
> > Date: Sat, 31 Mar 2018 12:58:41 -0700
> >
> > > IP defrag processing is one of the remaining problematic layer in linux.
> > >
> > > It uses static hash tables of 1024 buckets, and up to 128 items per bucket.
> > >
> > > A work queue is supposed to garbage collect items when host is under memory
> > > pressure, and doing a hash rebuild, changing seed used in hash computations.
> > >
> > > This work queue blocks softirqs for up to 25 ms when doing a hash rebuild,
> > > occurring every 5 seconds if host is under fire.
> > >
> > > Then there is the problem of sharing this hash table for all netns.
> > >
> > > It is time to switch to rhashtables, and allocate one of them per netns
> > > to speedup netns dismantle, since this is a critical metric these days.
> > >
> > > Lookup is now using RCU, and 64bit hosts can now provision whatever amount
> > > of memory needed to handle the expected workloads.
> > ...
> >
> > Series applied, thanks Eric.
> >
>
> Hi Eric,
>
> Recently my colleague (Moshe Shemesh) got a failure in upstream regression,
> which is related to this patchset. We don’t see the failure before it was
> merged.
> We checked again on net-next (from May 24th), it still reproduces.
>
> The test case runs netperf with ipv6 udp single stream (64K message size).
> After the change we see huge packet loss:
> 145,134 messages failed out of 145,419 (only 285 fully received)
>
as somebody who had similar issues with this patch series I can tell you
about what happened for the 6LoWPAN fragmentation.
The issue sounds similar, but there is too much missing information here
to say something about if you have exactly the issue which we had.
Our problem:
The patch series uses memcmp() to compare hash keys, we had some padding
bytes in our hash key and it occurs that we had sometimes random bytes
in this structure when it's put on stack. We solved it by a struct
foo_key bar = {}, which in case of gcc it _seems_ it makes a whole
memset(bar, 0, ..) on the structure.
I asked on the netdev mailinglist how to deal with this problem in
general, because = {} works in case of gcc, others compilers may have a
different handling or even gcc will changes this behaviour in future.
I got no reply so I did what it works for me. :-)
At least maybe a memcmp() on structures should never be used, it should
be compared by field. I would recommend this way when the compiler is
always clever enough to optimize it in some cases, but I am not so a
compiler expert to say anything about that.
I checked the hash key structures for x86_64 and pahole, so far I didn't
find any padding bytes there, but it might be different on
architectures or ?compiler?.
Additional useful information to check if you running into the same problem
would be:
- Which architecture do you use?
- Do you have similar problems with a veth setup?
You could also try this:
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index b939b94e7e91..40ece9ab8b12 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -142,19 +142,19 @@ static void ip6_frag_expire(struct timer_list *t)
static struct frag_queue *
fq_find(struct net *net, __be32 id, const struct ipv6hdr *hdr, int iif)
{
- struct frag_v6_compare_key key = {
- .id = id,
- .saddr = hdr->saddr,
- .daddr = hdr->daddr,
- .user = IP6_DEFRAG_LOCAL_DELIVER,
- .iif = iif,
- };
+ struct frag_v6_compare_key key = {};
struct inet_frag_queue *q;
if (!(ipv6_addr_type(&hdr->daddr) & (IPV6_ADDR_MULTICAST |
IPV6_ADDR_LINKLOCAL)))
key.iif = 0;
+ key.id = id;
+ key.saddr = hdr->saddr;
+ key.daddr = hdr->daddr;
+ key.user = IP6_DEFRAG_LOCAL_DELIVER;
+ key.iif = iif;
+
q = inet_frag_find(&net->ipv6.frags, &key);
if (!q)
return NULL;
- Alex
^ permalink raw reply related
* Re: possible deadlock in bpf_tcp_close
From: Daniel Borkmann @ 2018-05-28 14:51 UTC (permalink / raw)
To: syzbot, ast, linux-kernel, netdev, syzkaller-bugs, john.fastabend
In-Reply-To: <000000000000ca8a42056d4530a5@google.com>
[ +John ]
On 05/28/2018 04:47 PM, syzbot wrote:
> 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=10fd82d7800000
> kernel config: https://syzkaller.appspot.com/x/.config?x=e4078980b886800c
> dashboard link: https://syzkaller.appspot.com/bug?extid=47ed903f50684f046b15
> compiler: gcc (GCC) 8.0.1 20180413 (experimental)
>
> Unfortunately, I don't have any reproducer for this crash yet.
>
> IMPORTANT: if you fix the bug, please add the following tag to the commit:
> Reported-by: syzbot+47ed903f50684f046b15@syzkaller.appspotmail.com
So this one does have [1] included in the tree. :-( John, can you take a look?
[1] https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git/commit/?id=c1659ee5bb2a9f2023f5065256d5fd742ed660d2
Thanks,
Daniel
> ======================================================
> WARNING: possible circular locking dependency detected
> 4.17.0-rc6+ #25 Not tainted
> ------------------------------------------------------
> syz-executor4/7489 is trying to acquire lock:
> (ptrval) (&htab->buckets[i].lock#2){+...}, at: bpf_tcp_close+0x822/0x10b0 kernel/bpf/sockmap.c:285
>
> but task is already holding lock:
> (ptrval) (clock-AF_INET6){++..}, at: bpf_tcp_close+0x241/0x10b0 kernel/bpf/sockmap.c:260
>
> which lock already depends on the new lock.
>
>
> the existing dependency chain (in reverse order) is:
>
> -> #1 (clock-AF_INET6){++..}:
> __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_delete_elem+0x7c6/0xaf0 kernel/bpf/sockmap.c:2338
> map_delete_elem+0x32e/0x4e0 kernel/bpf/syscall.c:815
> __do_sys_bpf kernel/bpf/syscall.c:2349 [inline]
> __se_sys_bpf kernel/bpf/syscall.c:2317 [inline]
> __x64_sys_bpf+0x342/0x510 kernel/bpf/syscall.c:2317
> do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
>
> -> #0 (&htab->buckets[i].lock#2){+...}:
> lock_acquire+0x1dc/0x520 kernel/locking/lockdep.c:3920
> __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
> get_signal+0x886/0x1960 kernel/signal.c:2482
> do_signal+0x98/0x2040 arch/x86/kernel/signal.c:810
> exit_to_usermode_loop+0x28a/0x310 arch/x86/entry/common.c:162
> prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline]
> syscall_return_slowpath arch/x86/entry/common.c:265 [inline]
> do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
>
> other info that might help us debug this:
>
> Possible unsafe locking scenario:
>
> CPU0 CPU1
> ---- ----
> lock(clock-AF_INET6);
> lock(&htab->buckets[i].lock#2);
> lock(clock-AF_INET6);
> lock(&htab->buckets[i].lock#2);
>
> *** DEADLOCK ***
>
> 2 locks held by syz-executor4/7489:
> #0: (ptrval) (rcu_read_lock){....}, at: bpf_tcp_close+0x0/0x10b0 kernel/bpf/sockmap.c:2106
> #1: (ptrval) (clock-AF_INET6){++..}, at: bpf_tcp_close+0x241/0x10b0 kernel/bpf/sockmap.c:260
>
> stack backtrace:
> CPU: 1 PID: 7489 Comm: syz-executor4 Not tainted 4.17.0-rc6+ #25
> 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_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_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
> get_signal+0x886/0x1960 kernel/signal.c:2482
> do_signal+0x98/0x2040 arch/x86/kernel/signal.c:810
> exit_to_usermode_loop+0x28a/0x310 arch/x86/entry/common.c:162
> prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline]
> syscall_return_slowpath arch/x86/entry/common.c:265 [inline]
> do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x455a09
> RSP: 002b:00007f95715e8ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca
> RAX: fffffffffffffe00 RBX: 000000000072c028 RCX: 0000000000455a09
> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000072c028
> RBP: 000000000072c028 R08: 0000000000000000 R09: 000000000072c000
> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
> R13: 00007fffdce9240f R14: 00007f95715e99c0 R15: 0000000000000002
> cgroup: cgroup2: unknown option "cgroup2"
> cgroup: cgroup2: unknown option "cgroup2"
> syz-executor4 uses obsolete (PF_INET,SOCK_PACKET)
> FAULT_INJECTION: forcing a failure.
> name failslab, interval 1, probability 0, space 0, times 1
> CPU: 0 PID: 7912 Comm: syz-executor0 Not tainted 4.17.0-rc6+ #25
> 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
> fail_dump lib/fault-inject.c:51 [inline]
> should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
> __should_failslab+0x124/0x180 mm/failslab.c:32
> should_failslab+0x9/0x14 mm/slab_common.c:1522
> slab_pre_alloc_hook mm/slab.h:423 [inline]
> slab_alloc_node mm/slab.c:3299 [inline]
> kmem_cache_alloc_node+0x272/0x780 mm/slab.c:3642
> __alloc_skb+0x111/0x780 net/core/skbuff.c:193
> alloc_skb include/linux/skbuff.h:989 [inline]
> alloc_skb_with_frags+0x137/0x760 net/core/skbuff.c:5266
> sock_alloc_send_pskb+0x87a/0xae0 net/core/sock.c:2095
> unix_dgram_sendmsg+0x4f9/0x1730 net/unix/af_unix.c:1672
> unix_seqpacket_sendmsg+0x115/0x18f net/unix/af_unix.c:2053
> sock_sendmsg_nosec net/socket.c:629 [inline]
> sock_sendmsg+0xd5/0x120 net/socket.c:639
> ___sys_sendmsg+0x805/0x940 net/socket.c:2117
> __sys_sendmsg+0x115/0x270 net/socket.c:2155
> __do_sys_sendmsg net/socket.c:2164 [inline]
> __se_sys_sendmsg net/socket.c:2162 [inline]
> __x64_sys_sendmsg+0x78/0xb0 net/socket.c:2162
> do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x455a09
> RSP: 002b:00007f85737fdc68 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
> RAX: ffffffffffffffda RBX: 00007f85737fe6d4 RCX: 0000000000455a09
> RDX: 0000000000000000 RSI: 00000000200013c0 RDI: 0000000000000013
> RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000015
> R13: 000000000000057f R14: 00000000006fc488 R15: 0000000000000000
> FAULT_INJECTION: forcing a failure.
> name failslab, interval 1, probability 0, space 0, times 0
> CPU: 1 PID: 7947 Comm: syz-executor0 Not tainted 4.17.0-rc6+ #25
> 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
> fail_dump lib/fault-inject.c:51 [inline]
> should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
> __should_failslab+0x124/0x180 mm/failslab.c:32
> should_failslab+0x9/0x14 mm/slab_common.c:1522
> slab_pre_alloc_hook mm/slab.h:423 [inline]
> slab_alloc_node mm/slab.c:3299 [inline]
> kmem_cache_alloc_node_trace+0x26f/0x770 mm/slab.c:3661
> __do_kmalloc_node mm/slab.c:3681 [inline]
> __kmalloc_node_track_caller+0x33/0x70 mm/slab.c:3696
> __kmalloc_reserve.isra.39+0x3a/0xe0 net/core/skbuff.c:137
> __alloc_skb+0x14d/0x780 net/core/skbuff.c:205
> alloc_skb include/linux/skbuff.h:989 [inline]
> alloc_skb_with_frags+0x137/0x760 net/core/skbuff.c:5266
> sock_alloc_send_pskb+0x87a/0xae0 net/core/sock.c:2095
> unix_dgram_sendmsg+0x4f9/0x1730 net/unix/af_unix.c:1672
> unix_seqpacket_sendmsg+0x115/0x18f net/unix/af_unix.c:2053
> sock_sendmsg_nosec net/socket.c:629 [inline]
> sock_sendmsg+0xd5/0x120 net/socket.c:639
> ___sys_sendmsg+0x805/0x940 net/socket.c:2117
> __sys_sendmsg+0x115/0x270 net/socket.c:2155
> __do_sys_sendmsg net/socket.c:2164 [inline]
> __se_sys_sendmsg net/socket.c:2162 [inline]
> __x64_sys_sendmsg+0x78/0xb0 net/socket.c:2162
> do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x455a09
> RSP: 002b:00007f85737fdc68 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
> RAX: ffffffffffffffda RBX: 00007f85737fe6d4 RCX: 0000000000455a09
> RDX: 0000000000000000 RSI: 00000000200013c0 RDI: 0000000000000013
> RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000015
> R13: 000000000000057f R14: 00000000006fc488 R15: 0000000000000001
> FAULT_INJECTION: forcing a failure.
> name failslab, interval 1, probability 0, space 0, times 0
> CPU: 1 PID: 8051 Comm: syz-executor1 Not tainted 4.17.0-rc6+ #25
> 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
> fail_dump lib/fault-inject.c:51 [inline]
> should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
> __should_failslab+0x124/0x180 mm/failslab.c:32
> should_failslab+0x9/0x14 mm/slab_common.c:1522
> slab_pre_alloc_hook mm/slab.h:423 [inline]
> slab_alloc mm/slab.c:3378 [inline]
> __do_kmalloc mm/slab.c:3716 [inline]
> __kmalloc+0x2c8/0x760 mm/slab.c:3727
> kmalloc include/linux/slab.h:517 [inline]
> map_get_next_key+0x24a/0x640 kernel/bpf/syscall.c:863
> __do_sys_bpf kernel/bpf/syscall.c:2352 [inline]
> __se_sys_bpf kernel/bpf/syscall.c:2317 [inline]
> __x64_sys_bpf+0x357/0x510 kernel/bpf/syscall.c:2317
> do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x455a09
> RSP: 002b:00007fbd35c1ac68 EFLAGS: 00000246 ORIG_RAX: 0000000000000141
> RAX: ffffffffffffffda RBX: 00007fbd35c1b6d4 RCX: 0000000000455a09
> RDX: 000000000000002c RSI: 0000000020003000 RDI: 0000000000000004
> RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000014
> R13: 000000000000003d R14: 00000000006f4658 R15: 0000000000000000
> FAULT_INJECTION: forcing a failure.
> name fail_page_alloc, interval 1, probability 0, space 0, times 1
> CPU: 1 PID: 8104 Comm: syz-executor1 Not tainted 4.17.0-rc6+ #25
> 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
> fail_dump lib/fault-inject.c:51 [inline]
> should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
> should_fail_alloc_page mm/page_alloc.c:3060 [inline]
> prepare_alloc_pages mm/page_alloc.c:4319 [inline]
> __alloc_pages_nodemask+0x34e/0xd70 mm/page_alloc.c:4358
> alloc_pages_vma+0xdd/0x540 mm/mempolicy.c:2057
> wp_page_copy+0x24c/0x1440 mm/memory.c:2490
> do_wp_page+0x425/0x1990 mm/memory.c:2776
> handle_pte_fault mm/memory.c:3979 [inline]
> __handle_mm_fault+0x2996/0x4310 mm/memory.c:4087
>
>
> ---
> 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.
^ permalink raw reply
* possible deadlock in bpf_tcp_close
From: syzbot @ 2018-05-28 14:47 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=10fd82d7800000
kernel config: https://syzkaller.appspot.com/x/.config?x=e4078980b886800c
dashboard link: https://syzkaller.appspot.com/bug?extid=47ed903f50684f046b15
compiler: gcc (GCC) 8.0.1 20180413 (experimental)
Unfortunately, I don't have any reproducer for this crash yet.
IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+47ed903f50684f046b15@syzkaller.appspotmail.com
======================================================
WARNING: possible circular locking dependency detected
4.17.0-rc6+ #25 Not tainted
------------------------------------------------------
syz-executor4/7489 is trying to acquire lock:
(ptrval) (&htab->buckets[i].lock#2){+...}, at:
bpf_tcp_close+0x822/0x10b0 kernel/bpf/sockmap.c:285
but task is already holding lock:
(ptrval) (clock-AF_INET6){++..}, at: bpf_tcp_close+0x241/0x10b0
kernel/bpf/sockmap.c:260
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (clock-AF_INET6){++..}:
__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_delete_elem+0x7c6/0xaf0 kernel/bpf/sockmap.c:2338
map_delete_elem+0x32e/0x4e0 kernel/bpf/syscall.c:815
__do_sys_bpf kernel/bpf/syscall.c:2349 [inline]
__se_sys_bpf kernel/bpf/syscall.c:2317 [inline]
__x64_sys_bpf+0x342/0x510 kernel/bpf/syscall.c:2317
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
-> #0 (&htab->buckets[i].lock#2){+...}:
lock_acquire+0x1dc/0x520 kernel/locking/lockdep.c:3920
__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
get_signal+0x886/0x1960 kernel/signal.c:2482
do_signal+0x98/0x2040 arch/x86/kernel/signal.c:810
exit_to_usermode_loop+0x28a/0x310 arch/x86/entry/common.c:162
prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline]
syscall_return_slowpath arch/x86/entry/common.c:265 [inline]
do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290
entry_SYSCALL_64_after_hwframe+0x49/0xbe
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(clock-AF_INET6);
lock(&htab->buckets[i].lock#2);
lock(clock-AF_INET6);
lock(&htab->buckets[i].lock#2);
*** DEADLOCK ***
2 locks held by syz-executor4/7489:
#0: (ptrval) (rcu_read_lock){....}, at: bpf_tcp_close+0x0/0x10b0
kernel/bpf/sockmap.c:2106
#1: (ptrval) (clock-AF_INET6){++..}, at:
bpf_tcp_close+0x241/0x10b0 kernel/bpf/sockmap.c:260
stack backtrace:
CPU: 1 PID: 7489 Comm: syz-executor4 Not tainted 4.17.0-rc6+ #25
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_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_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
get_signal+0x886/0x1960 kernel/signal.c:2482
do_signal+0x98/0x2040 arch/x86/kernel/signal.c:810
exit_to_usermode_loop+0x28a/0x310 arch/x86/entry/common.c:162
prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline]
syscall_return_slowpath arch/x86/entry/common.c:265 [inline]
do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x455a09
RSP: 002b:00007f95715e8ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca
RAX: fffffffffffffe00 RBX: 000000000072c028 RCX: 0000000000455a09
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000072c028
RBP: 000000000072c028 R08: 0000000000000000 R09: 000000000072c000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007fffdce9240f R14: 00007f95715e99c0 R15: 0000000000000002
cgroup: cgroup2: unknown option "cgroup2"
cgroup: cgroup2: unknown option "cgroup2"
syz-executor4 uses obsolete (PF_INET,SOCK_PACKET)
FAULT_INJECTION: forcing a failure.
name failslab, interval 1, probability 0, space 0, times 1
CPU: 0 PID: 7912 Comm: syz-executor0 Not tainted 4.17.0-rc6+ #25
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
fail_dump lib/fault-inject.c:51 [inline]
should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
__should_failslab+0x124/0x180 mm/failslab.c:32
should_failslab+0x9/0x14 mm/slab_common.c:1522
slab_pre_alloc_hook mm/slab.h:423 [inline]
slab_alloc_node mm/slab.c:3299 [inline]
kmem_cache_alloc_node+0x272/0x780 mm/slab.c:3642
__alloc_skb+0x111/0x780 net/core/skbuff.c:193
alloc_skb include/linux/skbuff.h:989 [inline]
alloc_skb_with_frags+0x137/0x760 net/core/skbuff.c:5266
sock_alloc_send_pskb+0x87a/0xae0 net/core/sock.c:2095
unix_dgram_sendmsg+0x4f9/0x1730 net/unix/af_unix.c:1672
unix_seqpacket_sendmsg+0x115/0x18f net/unix/af_unix.c:2053
sock_sendmsg_nosec net/socket.c:629 [inline]
sock_sendmsg+0xd5/0x120 net/socket.c:639
___sys_sendmsg+0x805/0x940 net/socket.c:2117
__sys_sendmsg+0x115/0x270 net/socket.c:2155
__do_sys_sendmsg net/socket.c:2164 [inline]
__se_sys_sendmsg net/socket.c:2162 [inline]
__x64_sys_sendmsg+0x78/0xb0 net/socket.c:2162
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x455a09
RSP: 002b:00007f85737fdc68 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 00007f85737fe6d4 RCX: 0000000000455a09
RDX: 0000000000000000 RSI: 00000000200013c0 RDI: 0000000000000013
RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000015
R13: 000000000000057f R14: 00000000006fc488 R15: 0000000000000000
FAULT_INJECTION: forcing a failure.
name failslab, interval 1, probability 0, space 0, times 0
CPU: 1 PID: 7947 Comm: syz-executor0 Not tainted 4.17.0-rc6+ #25
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
fail_dump lib/fault-inject.c:51 [inline]
should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
__should_failslab+0x124/0x180 mm/failslab.c:32
should_failslab+0x9/0x14 mm/slab_common.c:1522
slab_pre_alloc_hook mm/slab.h:423 [inline]
slab_alloc_node mm/slab.c:3299 [inline]
kmem_cache_alloc_node_trace+0x26f/0x770 mm/slab.c:3661
__do_kmalloc_node mm/slab.c:3681 [inline]
__kmalloc_node_track_caller+0x33/0x70 mm/slab.c:3696
__kmalloc_reserve.isra.39+0x3a/0xe0 net/core/skbuff.c:137
__alloc_skb+0x14d/0x780 net/core/skbuff.c:205
alloc_skb include/linux/skbuff.h:989 [inline]
alloc_skb_with_frags+0x137/0x760 net/core/skbuff.c:5266
sock_alloc_send_pskb+0x87a/0xae0 net/core/sock.c:2095
unix_dgram_sendmsg+0x4f9/0x1730 net/unix/af_unix.c:1672
unix_seqpacket_sendmsg+0x115/0x18f net/unix/af_unix.c:2053
sock_sendmsg_nosec net/socket.c:629 [inline]
sock_sendmsg+0xd5/0x120 net/socket.c:639
___sys_sendmsg+0x805/0x940 net/socket.c:2117
__sys_sendmsg+0x115/0x270 net/socket.c:2155
__do_sys_sendmsg net/socket.c:2164 [inline]
__se_sys_sendmsg net/socket.c:2162 [inline]
__x64_sys_sendmsg+0x78/0xb0 net/socket.c:2162
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x455a09
RSP: 002b:00007f85737fdc68 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 00007f85737fe6d4 RCX: 0000000000455a09
RDX: 0000000000000000 RSI: 00000000200013c0 RDI: 0000000000000013
RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000015
R13: 000000000000057f R14: 00000000006fc488 R15: 0000000000000001
FAULT_INJECTION: forcing a failure.
name failslab, interval 1, probability 0, space 0, times 0
CPU: 1 PID: 8051 Comm: syz-executor1 Not tainted 4.17.0-rc6+ #25
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
fail_dump lib/fault-inject.c:51 [inline]
should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
__should_failslab+0x124/0x180 mm/failslab.c:32
should_failslab+0x9/0x14 mm/slab_common.c:1522
slab_pre_alloc_hook mm/slab.h:423 [inline]
slab_alloc mm/slab.c:3378 [inline]
__do_kmalloc mm/slab.c:3716 [inline]
__kmalloc+0x2c8/0x760 mm/slab.c:3727
kmalloc include/linux/slab.h:517 [inline]
map_get_next_key+0x24a/0x640 kernel/bpf/syscall.c:863
__do_sys_bpf kernel/bpf/syscall.c:2352 [inline]
__se_sys_bpf kernel/bpf/syscall.c:2317 [inline]
__x64_sys_bpf+0x357/0x510 kernel/bpf/syscall.c:2317
do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x455a09
RSP: 002b:00007fbd35c1ac68 EFLAGS: 00000246 ORIG_RAX: 0000000000000141
RAX: ffffffffffffffda RBX: 00007fbd35c1b6d4 RCX: 0000000000455a09
RDX: 000000000000002c RSI: 0000000020003000 RDI: 0000000000000004
RBP: 000000000072bea0 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000014
R13: 000000000000003d R14: 00000000006f4658 R15: 0000000000000000
FAULT_INJECTION: forcing a failure.
name fail_page_alloc, interval 1, probability 0, space 0, times 1
CPU: 1 PID: 8104 Comm: syz-executor1 Not tainted 4.17.0-rc6+ #25
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
fail_dump lib/fault-inject.c:51 [inline]
should_fail.cold.4+0xa/0x1a lib/fault-inject.c:149
should_fail_alloc_page mm/page_alloc.c:3060 [inline]
prepare_alloc_pages mm/page_alloc.c:4319 [inline]
__alloc_pages_nodemask+0x34e/0xd70 mm/page_alloc.c:4358
alloc_pages_vma+0xdd/0x540 mm/mempolicy.c:2057
wp_page_copy+0x24c/0x1440 mm/memory.c:2490
do_wp_page+0x425/0x1990 mm/memory.c:2776
handle_pte_fault mm/memory.c:3979 [inline]
__handle_mm_fault+0x2996/0x4310 mm/memory.c:4087
---
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.
^ permalink raw reply
* Re: [PATCH v3 07/11] net: sched: implement reference counted action release
From: Jiri Pirko @ 2018-05-28 14:30 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-8-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:25PM CEST, vladbu@mellanox.com wrote:
>Implement helper delete function that uses new action ops 'delete', instead
>of destroying action directly. This is required so act API could delete
>actions by index, without holding any references to action that is being
>deleted.
>
>Implement function __tcf_action_put() that releases reference to action and
>frees it, if necessary. Refactor action deletion code to use new put
>function and not to rely on rtnl lock. Remove rtnl lock assertions that are
>no longer needed.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* Re: [PATCH 1/5] ptp: rework gianfar_ptp as QorIQ common PTP driver
From: Richard Cochran @ 2018-05-28 14:29 UTC (permalink / raw)
To: Yangbo Lu; +Cc: netdev, devicetree, linux-kernel, claudiu.manoil, Rob Herring
In-Reply-To: <20180525044038.37756-1-yangbo.lu@nxp.com>
On Fri, May 25, 2018 at 12:40:34PM +0800, Yangbo Lu wrote:
> gianfar_ptp was the PTP clock driver for 1588 timer
> module of Freescale QorIQ eTSEC (Enhanced Three-Speed
> Ethernet Controllers) platforms. Actually QorIQ DPAA
> (Data Path Acceleration Architecture) platforms is
> also using the same 1588 timer module in hardware.
>
> This patch is to rework gianfar_ptp as QorIQ common
> PTP driver to support both DPAA and eTSEC. Moved
> gianfar_ptp.c to drivers/ptp/, renamed it as
> ptp_qoriq.c, and renamed many variables. There were
> not any function changes.
>
> Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
For the series:
Acked-by: Richard Cochran <richardcochran@gmail.com>
^ permalink raw reply
* Re: [PATCH v3 08/11] net: sched: don't release reference on action overwrite
From: Jiri Pirko @ 2018-05-28 14:27 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-9-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:26PM CEST, vladbu@mellanox.com wrote:
>Return from action init function with reference to action taken,
>even when overwriting existing action.
>
>Action init API initializes its fourth argument (pointer to pointer to tc
>action) to either existing action with same index or newly created action.
>In case of existing index(and bind argument is zero), init function returns
>without incrementing action reference counter. Caller of action init then
>proceeds working with action, without actually holding reference to it.
>This means that action could be deleted concurrently.
>
>Change action init behavior to always take reference to action before
>returning successfully, in order to protect from concurrent deletion.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* [PATCH net-next v16 1/8] sched: Add Common Applications Kept Enhanced (cake) qdisc
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake; +Cc: Pete Heist, Georgios Amanakis, Dave Taht
In-Reply-To: <152751762093.30935.15398977119837536494.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 v16 7/8] sch_cake: Add overhead compensation support to the rate shaper
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake
In-Reply-To: <152751762093.30935.15398977119837536494.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 5d9fdfd083c9..54cde99fc0ad 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
* [PATCH net-next v16 0/8] sched: Add Common Applications Kept Enhanced (cake) qdisc
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake
Cc: Georgios Amanakis, Pete Heist, Yuchung Cheng, Neal Cardwell,
Dave Taht, netfilter-devel
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
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 callback
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 | 6
include/uapi/linux/pkt_sched.h | 113 +
net/netfilter/core.c | 21
net/netfilter/nf_conntrack_core.c | 37
net/sched/Kconfig | 11
net/sched/Makefile | 1
net/sched/sch_cake.c | 2987 +++++++++++++++++++++++++++++++++++++
7 files changed, 3176 insertions(+)
create mode 100644 net/sched/sch_cake.c
^ permalink raw reply
* [PATCH net-next v16 2/8] sch_cake: Add ingress mode
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake
In-Reply-To: <152751762093.30935.15398977119837536494.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 v16 6/8] sch_cake: Add DiffServ handling
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake
In-Reply-To: <152751762093.30935.15398977119837536494.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 fecd9caac0cc..5d9fdfd083c9 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 v16 3/8] sch_cake: Add optional ACK filter
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake; +Cc: Yuchung Cheng, Neal Cardwell
In-Reply-To: <152751762093.30935.15398977119837536494.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 v16 5/8] sch_cake: Add NAT awareness to packet classifier
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake; +Cc: netfilter-devel
In-Reply-To: <152751762093.30935.15398977119837536494.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..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)
+ 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, "No conntrack support in kernel",
+ tb[TCA_CAKE_NAT]);
+ 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 v16 4/8] netfilter: Add nf_ct_get_tuple_skb callback
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake; +Cc: netfilter-devel
In-Reply-To: <152751762093.30935.15398977119837536494.stgit@alrua-kau>
This adds a callback to netfilter to extract a conntrack tuple from an skb
that works before the _nfct skb field has been initialised (e.g., in an
ingress qdisc). The tuple is copied to the caller to avoid issues with
reference counting.
The callback will return false when conntrack is not loaded, allowing it to
be accessed 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 | 6 ++++++
net/netfilter/core.c | 21 +++++++++++++++++++++
net/netfilter/nf_conntrack_core.c | 37 +++++++++++++++++++++++++++++++++++++
3 files changed, 64 insertions(+)
diff --git a/include/linux/netfilter.h b/include/linux/netfilter.h
index 85a1a0b32c66..7cbe7e9ce527 100644
--- a/include/linux/netfilter.h
+++ b/include/linux/netfilter.h
@@ -375,6 +375,12 @@ 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 *);
extern void (*nf_ct_destroy)(struct nf_conntrack *) __rcu;
+
+struct nf_conntrack_tuple;
+extern bool (*skb_ct_get_tuple)(struct nf_conntrack_tuple *,
+ const struct sk_buff *) __rcu;
+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) {}
#endif
diff --git a/net/netfilter/core.c b/net/netfilter/core.c
index 0f6b8172fb9a..520565198f0e 100644
--- a/net/netfilter/core.c
+++ b/net/netfilter/core.c
@@ -572,6 +572,27 @@ void nf_conntrack_destroy(struct nf_conntrack *nfct)
}
EXPORT_SYMBOL(nf_conntrack_destroy);
+bool (*skb_ct_get_tuple)(struct nf_conntrack_tuple *,
+ const struct sk_buff *) __rcu __read_mostly;
+EXPORT_SYMBOL(skb_ct_get_tuple);
+
+bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
+ const struct sk_buff *skb)
+{
+ bool (*get_tuple)(const struct sk_buff *, struct nf_conntrack_tuple *);
+ bool ret = false;
+
+ rcu_read_lock();
+ get_tuple = rcu_dereference(skb_ct_get_tuple);
+ if (!get_tuple)
+ goto out;
+ ret = get_tuple(dst_tuple, skb);
+out:
+ 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 41ff04ee2554..eee5c76f638c 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -1611,6 +1611,41 @@ static void nf_conntrack_attach(struct sk_buff *nskb, const struct sk_buff *skb)
nf_conntrack_get(skb_nfct(nskb));
}
+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),
@@ -1808,6 +1843,7 @@ void nf_conntrack_cleanup_start(void)
{
conntrack_gc_work.exiting = true;
RCU_INIT_POINTER(ip_ct_attach, NULL);
+ RCU_INIT_POINTER(skb_ct_get_tuple, NULL);
}
void nf_conntrack_cleanup_end(void)
@@ -2135,6 +2171,7 @@ void nf_conntrack_init_end(void)
/* For use by REJECT target */
RCU_INIT_POINTER(ip_ct_attach, nf_conntrack_attach);
RCU_INIT_POINTER(nf_ct_destroy, destroy_conntrack);
+ RCU_INIT_POINTER(skb_ct_get_tuple, nf_conntrack_get_tuple_skb);
}
/*
^ permalink raw reply related
* [PATCH net-next v16 8/8] sch_cake: Conditionally split GSO segments
From: Toke Høiland-Jørgensen @ 2018-05-28 14:27 UTC (permalink / raw)
To: netdev, cake
In-Reply-To: <152751762093.30935.15398977119837536494.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 54cde99fc0ad..2b9ec946250f 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
* Re: [PATCH v3 06/11] net: sched: add 'delete' function to action ops
From: Jiri Pirko @ 2018-05-28 14:21 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-7-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:24PM CEST, vladbu@mellanox.com wrote:
>Extend action ops with 'delete' function. Each action type to implements
>its own delete function that doesn't depend on rtnl lock.
>
>Implement delete function that is required to delete actions without
>holding rtnl lock. Use action API function that atomically deletes action
>only if it is still in action idr. This implementation prevents concurrent
>threads from deleting same action twice.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* Re: [PATCH v3 05/11] net: sched: implement action API that deletes action by index
From: Jiri Pirko @ 2018-05-28 14:21 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-6-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:23PM CEST, vladbu@mellanox.com wrote:
>Implement new action API function that atomically finds and deletes action
>from idr by index. Intended to be used by lockless actions that do not rely
>on rtnl lock.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* Re: [PATCH v3 04/11] net: sched: always take reference to action
From: Jiri Pirko @ 2018-05-28 14:19 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-5-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:22PM CEST, vladbu@mellanox.com wrote:
>Without rtnl lock protection it is no longer safe to use pointer to tc
>action without holding reference to it. (it can be destroyed concurrently)
>
>Remove unsafe action idr lookup function. Instead of it, implement safe tcf
>idr check function that atomically looks up action in idr and increments
>its reference and bind counters. Implement both action search and check
>using new safe function
>
>Reference taken by idr check is temporal and should not be accounted by
>userspace clients (both logically and to preserver current API behavior).
>Subtract temporal reference when dumping action to userspace using existing
>tca_get_fill function arguments.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ permalink raw reply
* Re: [PATCH v3 03/11] net: sched: implement unlocked action init API
From: Jiri Pirko @ 2018-05-28 14:17 UTC (permalink / raw)
To: Vlad Buslov; +Cc: netdev, jhs, xiyou.wangcong, davem, ast, daniel, kliteyn
In-Reply-To: <1527455849-22327-4-git-send-email-vladbu@mellanox.com>
Sun, May 27, 2018 at 11:17:21PM CEST, vladbu@mellanox.com wrote:
>Add additional 'rtnl_held' argument to act API init functions. It is
>required to implement actions that need to release rtnl lock before loading
>kernel module and reacquire if afterwards.
>
>Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
^ 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