Resetting old qdisc on dev_queue->qdisc_sleeping in
dev_qdisc_reset() is redundant, because this qdisc,
even if not same with dev_queue->qdisc, is reset via
qdisc_put() right after calling dev_graft_qdisc() when
hitting refcnt 0.
This is very easy to observe with qdisc_reset() tracepoint
and stack traces.
Reported-by: Václav Zindulka <vaclav.zindulka@tlapnet.cz>
Tested-by: Václav Zindulka <vaclav.zindulka@tlapnet.cz>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Except for sch_mq and sch_mqprio, each dev queue points to the
same root qdisc, so when we reset the dev queues with
netdev_for_each_tx_queue() we end up resetting the same instance
of the root qdisc for multiple times.
Avoid this by checking the __QDISC_STATE_DEACTIVATED bit in
each iteration, so for sch_mq/sch_mqprio, we still reset all
of them like before, for the rest, we only reset it once.
Reported-by: Václav Zindulka <vaclav.zindulka@tlapnet.cz>
Tested-by: Václav Zindulka <vaclav.zindulka@tlapnet.cz>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
qdisc_destroy() calls ops->reset() and cleans up qdisc->gso_skb
and qdisc->skb_bad_txq, these are nearly same with qdisc_reset(),
so just call it directly, and cosolidate the code for the next
patch.
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch reverts the folowing commits:
commit 064ff66e2b
"bonding: add missing netdev_update_lockdep_key()"
commit 53d374979e
"net: avoid updating qdisc_xmit_lock_key in netdev_update_lockdep_key()"
commit 1f26c0d3d2
"net: fix kernel-doc warning in <linux/netdevice.h>"
commit ab92d68fc2
"net: core: add generic lockdep keys"
but keeps the addr_list_lock_key because we still lock
addr_list_lock nestedly on stack devices, unlikely xmit_lock
this is safe because we don't take addr_list_lock on any fast
path.
Reported-and-tested-by: syzbot+aaa6fa4949cc5d9b7b25@syzkaller.appspotmail.com
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently if the default qdisc setup/init fails, the device ends up with
qdisc "noop", which causes all TX packets to get dropped.
With the introduction of sysctl net/core/default_qdisc it is possible
to change the default qdisc to be more advanced, which opens for the
possibility that Qdisc_ops->init() can fail.
This patch detect these kind of failures, and choose to fallback to
qdisc "noqueue", which is so simple that its init call will not fail.
This allows the interface to continue functioning.
V2:
As this also captures memory failures, which are transient, the
device is not kept in IFF_NO_QUEUE state. This allows the net_device
to retry to default qdisc assignment.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On ingress and cls_act qdiscs init, save the block on ingress
mini_Qdisc and and pass it on to ingress classification, so it
can be used for the looking up a specified chain index.
Co-developed-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The only slightly tricky merge conflict was the netdevsim because the
mutex locking fix overlapped a lot of driver reload reorganization.
The rest were (relatively) trivial in nature.
Signed-off-by: David S. Miller <davem@davemloft.net>
There is networking hardware that isn't based on Ethernet for layers 1 and 2.
For example CAN.
CAN is a multi-master serial bus standard for connecting Electronic Control
Units [ECUs] also known as nodes. A frame on the CAN bus carries up to 8 bytes
of payload. Frame corruption is detected by a CRC. However frame loss due to
corruption is possible, but a quite unusual phenomenon.
While fq_codel works great for TCP/IP, it doesn't for CAN. There are a lot of
legacy protocols on top of CAN, which are not build with flow control or high
CAN frame drop rates in mind.
When using fq_codel, as soon as the queue reaches a certain delay based length,
skbs from the head of the queue are silently dropped. Silently meaning that the
user space using a send() or similar syscall doesn't get an error. However
TCP's flow control algorithm will detect dropped packages and adjust the
bandwidth accordingly.
When using fq_codel and sending raw frames over CAN, which is the common use
case, the user space thinks the package has been sent without problems, because
send() returned without an error. pfifo_fast will drop skbs, if the queue
length exceeds the maximum. But with this scheduler the skbs at the tail are
dropped, an error (-ENOBUFS) is propagated to user space. So that the user
space can slow down the package generation.
On distributions, where fq_codel is made default via CONFIG_DEFAULT_NET_SCH
during compile time, or set default during runtime with sysctl
net.core.default_qdisc (see [1]), we get a bad user experience. In my test case
with pfifo_fast, I can transfer thousands of million CAN frames without a frame
drop. On the other hand with fq_codel there is more then one lost CAN frame per
thousand frames.
As pointed out fq_codel is not suited for CAN hardware, so this patch changes
attach_one_default_qdisc() to use pfifo_fast for "ARPHRD_CAN" network devices.
During transition of a netdev from down to up state the default queuing
discipline is attached by attach_default_qdiscs() with the help of
attach_one_default_qdisc(). This patch modifies attach_one_default_qdisc() to
attach the pfifo_fast (pfifo_fast_ops) if the network device type is
"ARPHRD_CAN".
[1] https://github.com/systemd/systemd/issues/9194
Suggested-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Vincent Prince <vincent.prince.fr@gmail.com>
Acked-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is networking hardware that isn't based on Ethernet for layers 1 and 2.
For example CAN.
CAN is a multi-master serial bus standard for connecting Electronic Control
Units [ECUs] also known as nodes. A frame on the CAN bus carries up to 8 bytes
of payload. Frame corruption is detected by a CRC. However frame loss due to
corruption is possible, but a quite unusual phenomenon.
While fq_codel works great for TCP/IP, it doesn't for CAN. There are a lot of
legacy protocols on top of CAN, which are not build with flow control or high
CAN frame drop rates in mind.
When using fq_codel, as soon as the queue reaches a certain delay based length,
skbs from the head of the queue are silently dropped. Silently meaning that the
user space using a send() or similar syscall doesn't get an error. However
TCP's flow control algorithm will detect dropped packages and adjust the
bandwidth accordingly.
When using fq_codel and sending raw frames over CAN, which is the common use
case, the user space thinks the package has been sent without problems, because
send() returned without an error. pfifo_fast will drop skbs, if the queue
length exceeds the maximum. But with this scheduler the skbs at the tail are
dropped, an error (-ENOBUFS) is propagated to user space. So that the user
space can slow down the package generation.
On distributions, where fq_codel is made default via CONFIG_DEFAULT_NET_SCH
during compile time, or set default during runtime with sysctl
net.core.default_qdisc (see [1]), we get a bad user experience. In my test case
with pfifo_fast, I can transfer thousands of million CAN frames without a frame
drop. On the other hand with fq_codel there is more then one lost CAN frame per
thousand frames.
As pointed out fq_codel is not suited for CAN hardware, so this patch changes
attach_one_default_qdisc() to use pfifo_fast for "ARPHRD_CAN" network devices.
During transition of a netdev from down to up state the default queuing
discipline is attached by attach_default_qdiscs() with the help of
attach_one_default_qdisc(). This patch modifies attach_one_default_qdisc() to
attach the pfifo_fast (pfifo_fast_ops) if the network device type is
"ARPHRD_CAN".
[1] https://github.com/systemd/systemd/issues/9194
Signed-off-by: Vincent Prince <vincent.prince.fr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some interface types could be nested.
(VLAN, BONDING, TEAM, MACSEC, MACVLAN, IPVLAN, VIRT_WIFI, VXLAN, etc..)
These interface types should set lockdep class because, without lockdep
class key, lockdep always warn about unexisting circular locking.
In the current code, these interfaces have their own lockdep class keys and
these manage itself. So that there are so many duplicate code around the
/driver/net and /net/.
This patch adds new generic lockdep keys and some helper functions for it.
This patch does below changes.
a) Add lockdep class keys in struct net_device
- qdisc_running, xmit, addr_list, qdisc_busylock
- these keys are used as dynamic lockdep key.
b) When net_device is being allocated, lockdep keys are registered.
- alloc_netdev_mqs()
c) When net_device is being free'd llockdep keys are unregistered.
- free_netdev()
d) Add generic lockdep key helper function
- netdev_register_lockdep_key()
- netdev_unregister_lockdep_key()
- netdev_update_lockdep_key()
e) Remove unnecessary generic lockdep macro and functions
f) Remove unnecessary lockdep code of each interfaces.
After this patch, each interface modules don't need to maintain
their lockdep keys.
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With threaded interrupts enabled, the interrupt thread runs as SCHED_RR
with priority 50. If a user application with a higher priority preempts
the interrupt thread and tries to shutdown the network interface then it
will loop forever. The kernel will spin in the loop waiting for the
device to become idle and the scheduler will never consider the
interrupt thread because its priority is lower.
Avoid the problem by sleeping for a jiffy giving other tasks,
including the interrupt thread, a chance to run and make progress.
In the original thread it has been suggested to use wait_event() and
properly waiting for the state to occur. DaveM explained that this would
require to add expensive checks in the fast paths of packet processing.
Link: https://lkml.kernel.org/r/1393976987-23555-1-git-send-email-mkl@pengutronix.de
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
[bigeasy: Rewrite commit message, add comment, use
schedule_timeout_uninterruptible()]
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The introduction of this schedule point was done in commit
2ba2506ca7 ("[NET]: Add preemption point in qdisc_run")
at a time the loop was not bounded.
Then later in commit d5b8aa1d24 ("net_sched: fix dequeuer fairness")
we added a limit on the number of packets.
Now is the time to remove the schedule point, since the default
limit of 64 packets matches the number of packets a typical NAPI
poll can process in a row.
This solves a latency problem for most TCP receivers under moderate load :
1) host receives a packet.
NET_RX_SOFTIRQ is raised by NIC hard IRQ handler
2) __do_softirq() does its first loop, handling NET_RX_SOFTIRQ
and calling the driver napi->loop() function
3) TCP stores the skb in socket receive queue:
4) TCP calls sk->sk_data_ready() and wakeups a user thread
waiting for EPOLLIN (as a result, need_resched() might now be true)
5) TCP cooks an ACK and sends it.
6) qdisc_run() processes one packet from qdisc, and sees need_resched(),
this raises NET_TX_SOFTIRQ (even if there are no more packets in
the qdisc)
Then we go back to the __do_softirq() in 2), and we see that new
softirqs were raised. Since need_resched() is true, we end up waking
ksoftirqd in this path :
if (pending) {
if (time_before(jiffies, end) && !need_resched() &&
--max_restart)
goto restart;
wakeup_softirqd();
}
So we have many wakeups of ksoftirqd kernel threads,
and more calls to qdisc_run() with associated lock overhead.
Note that another way to solve the issue would be to change TCP
to first send the ACK packet, then signal the EPOLLIN,
but this changes P99 latencies, as sending the ACK packet
can add a long delay.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When tcf_block_get() fails in sfb_init(), q->qdisc is still a NULL
pointer which leads to a crash in sfb_destroy(). Similar for
sch_dsmark.
Instead of fixing each separately, Linus suggested to just accept
NULL pointer in qdisc_put(), which would make callers easier.
(For sch_dsmark, the bug probably exists long before commit
6529eaba33f0.)
Fixes: 6529eaba33 ("net: sched: introduce tcf block infractructure")
Reported-by: syzbot+d5870a903591faaca4ae@syzkaller.appspotmail.com
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Whenever MQ is not used on a multiqueue device, we experience
serious reordering problems. Bisection found the cited
commit.
The issue can be described this way :
- A single qdisc hierarchy is shared by all transmit queues.
(eg : tc qdisc replace dev eth0 root fq_codel)
- When/if try_bulk_dequeue_skb_slow() dequeues a packet targetting
a different transmit queue than the one used to build a packet train,
we stop building the current list and save the 'bad' skb (P1) in a
special queue. (bad_txq)
- When dequeue_skb() calls qdisc_dequeue_skb_bad_txq() and finds this
skb (P1), it checks if the associated transmit queues is still in frozen
state. If the queue is still blocked (by BQL or NIC tx ring full),
we leave the skb in bad_txq and return NULL.
- dequeue_skb() calls q->dequeue() to get another packet (P2)
The other packet can target the problematic queue (that we found
in frozen state for the bad_txq packet), but another cpu just ran
TX completion and made room in the txq that is now ready to accept
new packets.
- Packet P2 is sent while P1 is still held in bad_txq, P1 might be sent
at next round. In practice P2 is the lead of a big packet train
(P2,P3,P4 ...) filling the BQL budget and delaying P1 by many packets :/
To solve this problem, we have to block the dequeue process as long
as the first packet in bad_txq can not be sent. Reordering issues
disappear and no side effects have been seen.
Fixes: a53851e2c3 ("net: sched: explicit locking in gso_cpu fallback")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that 'TCQ_F_CPUSTATS' bit can be cleared, depending on the value of
'TCQ_F_NOLOCK' bit in the parent qdisc, we need to be sure that per-cpu
counters are present when 'reset()' is called for pfifo_fast qdiscs.
Otherwise, the following script:
# tc q a dev lo handle 1: root htb default 100
# tc c a dev lo parent 1: classid 1:100 htb \
> rate 95Mbit ceil 100Mbit burst 64k
[...]
# tc f a dev lo parent 1: protocol arp basic classid 1:100
[...]
# tc q a dev lo parent 1:100 handle 100: pfifo_fast
[...]
# tc q d dev lo root
can generate the following splat:
Unable to handle kernel paging request at virtual address dfff2c01bd148000
Mem abort info:
ESR = 0x96000004
Exception class = DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000004
CM = 0, WnR = 0
[dfff2c01bd148000] address between user and kernel address ranges
Internal error: Oops: 96000004 [#1] SMP
[...]
pstate: 80000005 (Nzcv daif -PAN -UAO)
pc : pfifo_fast_reset+0x280/0x4d8
lr : pfifo_fast_reset+0x21c/0x4d8
sp : ffff800d09676fa0
x29: ffff800d09676fa0 x28: ffff200012ee22e4
x27: dfff200000000000 x26: 0000000000000000
x25: ffff800ca0799958 x24: ffff1001940f332b
x23: 0000000000000007 x22: ffff200012ee1ab8
x21: 0000600de8a40000 x20: 0000000000000000
x19: ffff800ca0799900 x18: 0000000000000000
x17: 0000000000000002 x16: 0000000000000000
x15: 0000000000000000 x14: 0000000000000000
x13: 0000000000000000 x12: ffff1001b922e6e2
x11: 1ffff001b922e6e1 x10: 0000000000000000
x9 : 1ffff001b922e6e1 x8 : dfff200000000000
x7 : 0000000000000000 x6 : 0000000000000000
x5 : 1fffe400025dc45c x4 : 1fffe400025dc357
x3 : 00000c01bd148000 x2 : 0000600de8a40000
x1 : 0000000000000007 x0 : 0000600de8a40004
Call trace:
pfifo_fast_reset+0x280/0x4d8
qdisc_reset+0x6c/0x370
htb_reset+0x150/0x3b8 [sch_htb]
qdisc_reset+0x6c/0x370
dev_deactivate_queue.constprop.5+0xe0/0x1a8
dev_deactivate_many+0xd8/0x908
dev_deactivate+0xe4/0x190
qdisc_graft+0x88c/0xbd0
tc_get_qdisc+0x418/0x8a8
rtnetlink_rcv_msg+0x3a8/0xa78
netlink_rcv_skb+0x18c/0x328
rtnetlink_rcv+0x28/0x38
netlink_unicast+0x3c4/0x538
netlink_sendmsg+0x538/0x9a0
sock_sendmsg+0xac/0xf8
___sys_sendmsg+0x53c/0x658
__sys_sendmsg+0xc8/0x140
__arm64_sys_sendmsg+0x74/0xa8
el0_svc_handler+0x164/0x468
el0_svc+0x10/0x14
Code: 910012a0 92400801 d343fc03 11000c21 (38fb6863)
Fix this by testing the value of 'TCQ_F_CPUSTATS' bit in 'qdisc->flags',
before dereferencing 'qdisc->cpu_qstats'.
Changes since v1:
- coding style improvements, thanks to Stefano Brivio
Fixes: 8a53e616de ("net: sched: when clearing NOLOCK, clear TCQ_F_CPUSTATS, too")
CC: Paolo Abeni <pabeni@redhat.com>
Reported-by: Li Shuang <shuali@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Based on 1 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license as published by
the free software foundation either version 2 of the license or at
your option any later version
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-or-later
has been chosen to replace the boilerplate/reference in 3029 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190527070032.746973796@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Although devlink health report does a nice job on reporting TX
timeout and other NIC errors, unfortunately it requires drivers
to support it but currently only mlx5 has implemented it.
Before other drivers could catch up, it is useful to have a
generic tracepoint to monitor this kind of TX timeout. We have
been suffering TX timeout with different drivers, we plan to
start to monitor it with rasdaemon which just needs a new tracepoint.
Sample output:
ksoftirqd/1-16 [001] ..s2 144.043173: net_dev_xmit_timeout: dev=ens3 driver=e1000 queue=0
Cc: Eran Ben Elisha <eranbe@mellanox.com>
Cc: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This revert commit 46b1c18f9d ("net: sched: put back q.qlen into
a single location").
After the previous patch, when a NOLOCK qdisc is enslaved to a
locking qdisc it switches to global stats accounting. As a consequence,
when a classful qdisc accesses directly a child qdisc's qlen, such
qdisc is not doing per CPU accounting and qlen value is consistent.
In the control path nobody uses directly qlen since commit
e5f0e8f8e4 ("net: sched: introduce and use qdisc tree flush/purge
helpers"), so we can remove the contented atomic ops from the
datapath.
v1 -> v2:
- complete the qdisc_qstats_atomic_qlen_dec() ->
qdisc_qstats_cpu_qlen_dec() replacement, fix build issue
- more descriptive commit message
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since stats updating is always consistent with TCQ_F_CPUSTATS flag,
we can disable it at qdisc creation time flipping such bit.
In my experiments, if the NOLOCK flag is cleared, per CPU stats
accounting does not give any measurable performance gain, but it
waste some memory.
Let's clear TCQ_F_CPUSTATS together with NOLOCK, when enslaving
a NOLOCK qdisc to 'lock' one.
Use stats update helper inside pfifo_fast, to cope correctly with
TCQ_F_CPUSTATS flag change.
As a side effect, q.qlen value for any child qdiscs is always
consistent for all lock classfull qdiscs.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The core sched implementation checks independently for NOLOCK flag
to acquire/release the root spin lock and for qdisc_is_percpu_stats()
to account per CPU values in many places.
This change update the last few places checking the TCQ_F_NOLOCK to
do per CPU stats accounting according to qdisc_is_percpu_stats()
value.
The above allows to clean dev_requeue_skb() implementation a bit
and makes stats update always consistent with a single flag.
v1 -> v2:
- do not move qdisc_is_empty definition, fix build issue
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The queue is marked not empty after acquiring the seqlock,
and it's up to the NOLOCK qdisc clearing such flag on dequeue.
Since the empty status lays on the same cache-line of the
seqlock, it's always hot on cache during the updates.
This makes the empty flag update a little bit loosy. Given
the lack of synchronization between enqueue and dequeue, this
is unavoidable.
v2 -> v3:
- qdisc_is_empty() has a const argument (Eric)
v1 -> v2:
- use really an 'empty' flag instead of 'not_empty', as
suggested by Eric
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the series fc8b81a598 ("Merge branch 'lockless-qdisc-series'")
John made the assumption that the data path had no need to read
the qdisc qlen (number of packets in the qdisc).
It is true when pfifo_fast is used as the root qdisc, or as direct MQ/MQPRIO
children.
But pfifo_fast can be used as leaf in class full qdiscs, and existing
logic needs to access the child qlen in an efficient way.
HTB breaks badly, since it uses cl->leaf.q->q.qlen in :
htb_activate() -> WARN_ON()
htb_dequeue_tree() to decide if a class can be htb_deactivated
when it has no more packets.
HFSC, DRR, CBQ, QFQ have similar issues, and some calls to
qdisc_tree_reduce_backlog() also read q.qlen directly.
Using qdisc_qlen_sum() (which iterates over all possible cpus)
in the data path is a non starter.
It seems we have to put back qlen in a central location,
at least for stable kernels.
For all qdisc but pfifo_fast, qlen is guarded by the qdisc lock,
so the existing q.qlen{++|--} are correct.
For 'lockless' qdisc (pfifo_fast so far), we need to use atomic_{inc|dec}()
because the spinlock might be not held (for example from
pfifo_fast_enqueue() and pfifo_fast_dequeue())
This patch adds atomic_qlen (in the same location than qlen)
and renames the following helpers, since we want to express
they can be used without qdisc lock, and that qlen is no longer percpu.
- qdisc_qstats_cpu_qlen_dec -> qdisc_qstats_atomic_qlen_dec()
- qdisc_qstats_cpu_qlen_inc -> qdisc_qstats_atomic_qlen_inc()
Later (net-next) we might revert this patch by tracking all these
qlen uses and replace them by a more efficient method (not having
to access a precise qlen, but an empty/non_empty status that might
be less expensive to maintain/track).
Another possibility is to have a legacy pfifo_fast version that would
be used when used a a child qdisc, since the parent qdisc needs
a spinlock anyway. But then, future lockless qdiscs would also
have the same problem.
Fixes: 7e66016f2c ("net: sched: helpers to sum qlen and qlen for per cpu logic")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
This pointer is RCU protected, so proper primitives should be used.
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The netfilter conflicts were rather simple overlapping
changes.
However, the cls_tcindex.c stuff was a bit more complex.
On the 'net' side, Cong is fixing several races and memory
leaks. Whilst on the 'net-next' side we have Vlad adding
the rtnl-ness support.
What I've decided to do, in order to resolve this, is revert the
conversion over to using a workqueue that Cong did, bringing us back
to pure RCU. I did it this way because I believe that either Cong's
races don't apply with have Vlad did things, or Cong will have to
implement the race fix slightly differently.
Signed-off-by: David S. Miller <davem@davemloft.net>
Extend tcf_chain with new filter_chain_lock mutex. Always lock the chain
when accessing filter_chain list, instead of relying on rtnl lock.
Dereference filter_chain with tcf_chain_dereference() lockdep macro to
verify that all users of chain_list have the lock taken.
Rearrange tp insert/remove code in tc_new_tfilter/tc_del_tfilter to execute
all necessary code while holding chain lock in order to prevent
invalidation of chain_info structure by potential concurrent change. This
also serializes calls to tcf_chain0_head_change(), which allows head change
callbacks to rely on filter_chain_lock for synchronization instead of rtnl
mutex.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Netlink has moved from bitmasks to group numbers long ago.
Signed-off-by: Jouke Witteveen <j.witteveen@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that call_rcu()'s callback is not invoked until after bh-disable
regions of code have completed (in addition to explicitly marked
RCU read-side critical sections), call_rcu() can be used in place
of call_rcu_bh(). Similarly, rcu_barrier() can be used in place o
frcu_barrier_bh(). This commit therefore makes these changes.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: <netdev@vger.kernel.org>
While noop_qdisc.gso_skb and noop_qdisc.skb_bad_txq are not used
in other places, it seems not correct to overwrite their fields
in dev_init_scheduler_queue().
noop_qdisc is essentially a shared and read-only object, even if
it is not marked as const because of some implementation detail.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes the following sparse warning:
net/sched/sch_generic.c:944:6: warning:
symbol 'qdisc_free_cb' was not declared. Should it be static?
Fixes: 3a7d0d07a3 ("net: sched: extend Qdisc with rcu")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, Qdisc API functions assume that users have rtnl lock taken. To
implement rtnl unlocked classifiers update interface, Qdisc API must be
extended with functions that do not require rtnl lock.
Extend Qdisc structure with rcu. Implement special version of put function
qdisc_put_unlocked() that is called without rtnl lock taken. This function
only takes rtnl lock if Qdisc reference counter reached zero and is
intended to be used as optimization.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Current implementation of qdisc_destroy() decrements Qdisc reference
counter and only actually destroy Qdisc if reference counter value reached
zero. Rename qdisc_destroy() to qdisc_put() in order for it to better
describe the way in which this function currently implemented and used.
Extract code that deallocates Qdisc into new private qdisc_destroy()
function. It is intended to be shared between regular qdisc_put() and its
unlocked version that is introduced in next patch in this series.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
An SKB is not on a list if skb->next is NULL.
Codify this convention into a helper function and use it
where we are dequeueing an SKB and need to mark it as such.
Signed-off-by: David S. Miller <davem@davemloft.net>
Checking netif_xmit_frozen_or_stopped() at the end of sch_direct_xmit()
is being bypassed. This is because "ret" from sch_direct_xmit() will be
either NETDEV_TX_OK or NETDEV_TX_BUSY, and only ret == NETDEV_TX_OK == 0
will reach the condition:
if (ret && netif_xmit_frozen_or_stopped(txq))
return false;
This patch cleans up the code by removing the whole condition.
For more discussion about this, please refer to
https://marc.info/?t=152727195700008
Signed-off-by: Song Liu <songliubraving@fb.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
After the previous patch, for NOLOCK qdiscs, q->seqlock is
always held when the dequeue() is invoked, we can drop
any additional locking to protect such operation.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
So that we can use lockdep on it.
The newly introduced sequence lock has the same scope of busylock,
so it shares the same lockdep annotation, but it's only used for
NOLOCK qdiscs.
With this changeset we acquire such lock in the control path around
flushing operation (qdisc reset), to allow more NOLOCK qdisc perf
improvement in the next patch.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently NOLOCK qdiscs pay a measurable overhead to atomically
manipulate the __QDISC_STATE_RUNNING. Such bit is flipped twice per
packet in the uncontended scenario with packet rate below the
line rate: on packed dequeue and on the next, failing dequeue attempt.
This changeset moves the bit manipulation into the qdisc_run_{begin,end}
helpers, so that the bit is now flipped only once per packet, with
measurable performance improvement in the uncontended scenario.
This also allows simplifying the qdisc teardown code path - since
qdisc_is_running() is now effective for each qdisc type - and avoid a
possible race between qdisc_run() and dev_deactivate_many(), as now
the some_qdisc_is_busy() can properly detect NOLOCK qdiscs being busy
dequeuing packets.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After the qdisc lock was dropped in pfifo_fast we allow multiple
enqueue threads and dequeue threads to run in parallel. On the
enqueue side the skb bit ooo_okay is used to ensure all related
skbs are enqueued in-order. On the dequeue side though there is
no similar logic. What we observe is with fewer queues than CPUs
it is possible to re-order packets when two instances of
__qdisc_run() are running in parallel. Each thread will dequeue
a skb and then whichever thread calls the ndo op first will
be sent on the wire. This doesn't typically happen because
qdisc_run() is usually triggered by the same core that did the
enqueue. However, drivers will trigger __netif_schedule()
when queues are transitioning from stopped to awake using the
netif_tx_wake_* APIs. When this happens netif_schedule() calls
qdisc_run() on the same CPU that did the netif_tx_wake_* which
is usually done in the interrupt completion context. This CPU
is selected with the irq affinity which is unrelated to the
enqueue operations.
To resolve this we add a RUNNING bit to the qdisc to ensure
only a single dequeue per qdisc is running. Enqueue and dequeue
operations can still run in parallel and also on multi queue
NICs we can still have a dequeue in-flight per qdisc, which
is typically per CPU.
Fixes: c5ad119fb6 ("net: sched: pfifo_fast use skb_array")
Reported-by: Jakob Unterwurzacher <jakob.unterwurzacher@theobroma-systems.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
syzbot reported one use-after-free in pfifo_fast_enqueue() [1]
Issue here is that we can not reuse skb after a successful skb_array_produce()
since another cpu might have consumed it already.
I believe a similar problem exists in try_bulk_dequeue_skb_slow()
in case we put an skb into qdisc_enqueue_skb_bad_txq() for lockless qdisc.
[1]
BUG: KASAN: use-after-free in qdisc_pkt_len include/net/sch_generic.h:610 [inline]
BUG: KASAN: use-after-free in qdisc_qstats_cpu_backlog_inc include/net/sch_generic.h:712 [inline]
BUG: KASAN: use-after-free in pfifo_fast_enqueue+0x4bc/0x5e0 net/sched/sch_generic.c:639
Read of size 4 at addr ffff8801cede37e8 by task syzkaller717588/5543
CPU: 1 PID: 5543 Comm: syzkaller717588 Not tainted 4.16.0-rc4+ #265
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:17 [inline]
dump_stack+0x194/0x24d lib/dump_stack.c:53
print_address_description+0x73/0x250 mm/kasan/report.c:256
kasan_report_error mm/kasan/report.c:354 [inline]
kasan_report+0x23c/0x360 mm/kasan/report.c:412
__asan_report_load4_noabort+0x14/0x20 mm/kasan/report.c:432
qdisc_pkt_len include/net/sch_generic.h:610 [inline]
qdisc_qstats_cpu_backlog_inc include/net/sch_generic.h:712 [inline]
pfifo_fast_enqueue+0x4bc/0x5e0 net/sched/sch_generic.c:639
__dev_xmit_skb net/core/dev.c:3216 [inline]
Fixes: c5ad119fb6 ("net: sched: pfifo_fast use skb_array")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot+ed43b6903ab968b16f54@syzkaller.appspotmail.com
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
pfifo_fast used to drop based on qdisc_dev(qdisc)->tx_queue_len,
so we have to resize skb array when we change tx_queue_len.
Other qdiscs which read tx_queue_len are fine because they
all save it to sch->limit or somewhere else in qdisc during init.
They don't have to implement this, it is nicer if they do so
that users don't have to re-configure qdisc after changing
tx_queue_len.
Cc: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduce a new qdisc ops ->change_tx_queue_len() so that
each qdisc could decide how to implement this if it wants.
Previously we simply read dev->tx_queue_len, after pfifo_fast
switches to skb array, we need this API to resize the skb array
when we change dev->tx_queue_len.
To avoid handling race conditions with TX BH, we need to
deactivate all TX queues before change the value and bring them
back after we are done, this also makes implementation easier.
Cc: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Expose the number of times the link has been going UP or DOWN, and
update the "carrier_changes" counter to be the sum of these two events.
While at it, also update the sysfs-class-net documentation to cover:
carrier_changes (3.15), carrier_up_count (4.16) and carrier_down_count
(4.16)
Signed-off-by: David Decotigny <decot@googlers.com>
[Florian:
* rebase
* add documentation
* merge carrier_changes with up/down counters]
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While working on fixing another bug, I ran into the following panic
on arm64 by simply attaching clsact qdisc, adding a filter and running
traffic on ingress to it:
[...]
[ 178.188591] Unable to handle kernel read from unreadable memory at virtual address 810fb501f000
[ 178.197314] Mem abort info:
[ 178.200121] ESR = 0x96000004
[ 178.203168] Exception class = DABT (current EL), IL = 32 bits
[ 178.209095] SET = 0, FnV = 0
[ 178.212157] EA = 0, S1PTW = 0
[ 178.215288] Data abort info:
[ 178.218175] ISV = 0, ISS = 0x00000004
[ 178.222019] CM = 0, WnR = 0
[ 178.224997] user pgtable: 4k pages, 48-bit VAs, pgd = 0000000023cb3f33
[ 178.231531] [0000810fb501f000] *pgd=0000000000000000
[ 178.236508] Internal error: Oops: 96000004 [#1] SMP
[...]
[ 178.311855] CPU: 73 PID: 2497 Comm: ping Tainted: G W 4.15.0-rc7+ #5
[ 178.319413] Hardware name: FOXCONN R2-1221R-A4/C2U4N_MB, BIOS G31FB18A 03/31/2017
[ 178.326887] pstate: 60400005 (nZCv daif +PAN -UAO)
[ 178.331685] pc : __netif_receive_skb_core+0x49c/0xac8
[ 178.336728] lr : __netif_receive_skb+0x28/0x78
[ 178.341161] sp : ffff00002344b750
[ 178.344465] x29: ffff00002344b750 x28: ffff810fbdfd0580
[ 178.349769] x27: 0000000000000000 x26: ffff000009378000
[...]
[ 178.418715] x1 : 0000000000000054 x0 : 0000000000000000
[ 178.424020] Process ping (pid: 2497, stack limit = 0x000000009f0a3ff4)
[ 178.430537] Call trace:
[ 178.432976] __netif_receive_skb_core+0x49c/0xac8
[ 178.437670] __netif_receive_skb+0x28/0x78
[ 178.441757] process_backlog+0x9c/0x160
[ 178.445584] net_rx_action+0x2f8/0x3f0
[...]
Reason is that sch_ingress and sch_clsact are doing mini_qdisc_pair_init()
which sets up miniq pointers to cpu_{b,q}stats from the underlying qdisc.
Problem is that this cannot work since they are actually set up right after
the qdisc ->init() callback in qdisc_create(), so first packet going into
sch_handle_ingress() tries to call mini_qdisc_bstats_cpu_update() and we
therefore panic.
In order to fix this, allocation of {b,q}stats needs to happen before we
call into ->init(). In net-next, there's already such option through commit
d59f5ffa59 ("net: sched: a dflt qdisc may be used with per cpu stats").
However, the bug needs to be fixed in net still for 4.15. Thus, include
these bits to reduce any merge churn and reuse the static_flags field to
set TCQ_F_CPUSTATS, and remove the allocation from qdisc_create() since
there is no other user left. Prashant Bhole ran into the same issue but
for net-next, thus adding him below as well as co-author. Same issue was
also reported by Sandipan Das when using bcc.
Fixes: 46209401f8 ("net: core: introduce mini_Qdisc and eliminate usage of tp->q for clsact fastpath")
Reference: https://lists.iovisor.org/pipermail/iovisor-dev/2018-January/001190.html
Reported-by: Sandipan Das <sandipan@linux.vnet.ibm.com>
Co-authored-by: Prashant Bhole <bhole_prashant_q7@lab.ntt.co.jp>
Co-authored-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
When dev_requeue_skb() is called with bulked skb list, only the
first skb of the list will be requeued to qdisc layer, and leak
the others without free them.
TCP is broken due to skb leak since no free skb will be considered
as still in the host queue and never be retransmitted. This happend
when dev_requeue_skb() called from qdisc_restart().
qdisc_restart
|-- dequeue_skb
|-- sch_direct_xmit()
|-- dev_requeue_skb() <-- skb may bluked
Fix dev_requeue_skb() to requeue the full bluked list. Also change
to use __skb_queue_tail() in __dev_requeue_skb() to avoid skb out
of order.
Fixes: a53851e2c3 ("net: sched: explicit locking in gso_cpu fallback")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cong Wang