In attach_default_qdiscs(), if a dev has multiple queues and queue 0 fails
to attach qdisc because there is no memory in attach_one_default_qdisc().
Then dev->qdisc will be noop_qdisc by default. But the other queues may be
able to successfully attach to default qdisc.
In this case, the fallback to noqueue process will be triggered. If the
original attached qdisc is not released and a new one is directly
attached, this will cause netdevice reference leaks.
The following is the bug log:
veth0: default qdisc (fq_codel) fail, fallback to noqueue
unregister_netdevice: waiting for veth0 to become free. Usage count = 32
leaked reference.
qdisc_alloc+0x12e/0x210
qdisc_create_dflt+0x62/0x140
attach_one_default_qdisc.constprop.41+0x44/0x70
dev_activate+0x128/0x290
__dev_open+0x12a/0x190
__dev_change_flags+0x1a2/0x1f0
dev_change_flags+0x23/0x60
do_setlink+0x332/0x1150
__rtnl_newlink+0x52f/0x8e0
rtnl_newlink+0x43/0x70
rtnetlink_rcv_msg+0x140/0x3b0
netlink_rcv_skb+0x50/0x100
netlink_unicast+0x1bb/0x290
netlink_sendmsg+0x37c/0x4e0
sock_sendmsg+0x5f/0x70
____sys_sendmsg+0x208/0x280
Fix this bug by clearing any non-noop qdiscs that may have been assigned
before trying to re-attach.
Fixes: bf6dba76d2 ("net: sched: fallback to qdisc noqueue if default qdisc setup fail")
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Link: https://lore.kernel.org/r/20220826090055.24424-1-wanghai38@huawei.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The memory allocated by using kzallloc_node and kcalloc has been cleared.
Therefore, the structure members of the new qdisc are 0. So there's no
need to explicitly assign a value of 0.
Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While reading weight_p, it can be changed concurrently. Thus, we need
to add READ_ONCE() to its reader.
Also, dev_[rt]x_weight can be read/written at the same time. So, we
need to use READ_ONCE() and WRITE_ONCE() for its access. Moreover, to
use the same weight_p while changing dev_[rt]x_weight, we add a mutex
in proc_do_dev_weight().
Fixes: 3d48b53fb2 ("net: dev_weight: TX/RX orthogonality")
Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that the bonding driver keeps track of the last TX time of ARP and
NS probes, we effectively revert the following commits:
32d3e51a82 ("net_sched: use macvlan real dev trans_start in dev_trans_start()")
07ce76aa9b ("net_sched: make dev_trans_start return vlan's real dev trans_start")
Note that the approach of continuing to hack at this function would not
get us very far, hence the desire to take a different approach. DSA is
also a virtual device that uses NETIF_F_LLTX, but there, many uppers
share the same lower (DSA master, i.e. the physical host port of a
switch). By making dev_trans_start() on a DSA interface return the
dev_trans_start() of the master, we effectively assume that all other
DSA interfaces are silent, otherwise this corrupts the validity of the
probe timestamp data from the bonding driver's perspective.
Furthermore, the hacks didn't take into consideration the fact that the
lower interface of @dev may not have been physical either. For example,
VLAN over VLAN, or DSA with 2 masters in a LAG.
And even furthermore, there are NETIF_F_LLTX devices which are not
stacked, like veth. The hack here would not work with those, because it
would not have to provide the bonding driver something to chew at all.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Netdev reference helpers have a dev_ prefix for historic
reasons. Renaming the old helpers would be too much churn
but we can rename the tracking ones which are relatively
recent and should be the default for new code.
Rename:
dev_hold_track() -> netdev_hold()
dev_put_track() -> netdev_put()
dev_replace_track() -> netdev_ref_replace()
Link: https://lore.kernel.org/r/20220608043955.919359-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
For some unknown reason qdisc_reset() is using
a convoluted way of freeing two lists of skbs.
Use __skb_queue_purge() instead.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20220414011004.2378350-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
commit 56b765b79e ("htb: improved accuracy at high rates") broke
"overhead X", "linklayer atm" and "mpu X" attributes.
"overhead X" and "linklayer atm" have already been fixed. This restores
the "mpu X" handling, as might be used by DOCSIS or Ethernet shaping:
tc class add ... htb rate X overhead 4 mpu 64
The code being fixed is used by htb, tbf and act_police. Cake has its
own mpu handling. qdisc_calculate_pkt_len still uses the size table
containing values adjusted for mpu by user space.
iproute2 tc has always passed mpu into the kernel via a tc_ratespec
structure, but the kernel never directly acted on it, merely stored it
so that it could be read back by `tc class show`.
Rather, tc would generate length-to-time tables that included the mpu
(and linklayer) in their construction, and the kernel used those tables.
Since v3.7, the tables were no longer used. Along with "mpu", this also
broke "overhead" and "linklayer" which were fixed in 01cb71d2d4
("net_sched: restore "overhead xxx" handling", v3.10) and 8a8e3d84b1
("net_sched: restore "linklayer atm" handling", v3.11).
"overhead" was fixed by simply restoring use of tc_ratespec::overhead -
this had originally been used by the kernel but was initially omitted
from the new non-table-based calculations.
"linklayer" had been handled in the table like "mpu", but the mode was
not originally passed in tc_ratespec. The new implementation was made to
handle it by getting new versions of tc to pass the mode in an extended
tc_ratespec, and for older versions of tc the table contents were analysed
at load time to deduce linklayer.
As "mpu" has always been given to the kernel in tc_ratespec,
accompanying the mpu-based table, we can restore system functionality
with no userspace change by making the kernel act on the tc_ratespec
value.
Fixes: 56b765b79e ("htb: improved accuracy at high rates")
Signed-off-by: Kevin Bracey <kevin@bracey.fi>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: Vimalkumar <j.vimal@gmail.com>
Link: https://lore.kernel.org/r/20220112170210.1014351-1-kevin@bracey.fi
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a netdevice_tracker inside struct net_device, to track
the self reference when a device has an active watchdog timer.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There is no reason for stopping all TX queues from dev_watchdog()
Not only this stops feeding the NIC, it also migrates all qdiscs
to be serviced on the cpu calling netif_tx_unlock(), causing
a potential latency artifact.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These are not fast path, there is no point in inlining them.
Also provide netif_freeze_queues()/netif_unfreeze_queues()
so that we can use them from dev_watchdog() in the following patch.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In following patches, dev_watchdog() will no longer stop all queues.
It will read queue->trans_start locklessly.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tx_timeout_show() assumed dev_watchdog() would stop all
the queues, to fetch queue->trans_timeout under protection
of the queue->_xmit_lock.
As we want to no longer disrupt transmits, we use an
atomic_long_t instead.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: david decotigny <david.decotigny@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The only valid values for a miniq pointer are NULL or a pointer to
miniq1 or miniq2, so testing for miniq_old != &miniq1 is functionally
equivalent to testing that it is NULL or equal to &miniq2.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Seth Forshee <sforshee@digitalocean.com>
Link: https://lore.kernel.org/r/20211026183721.137930-1-seth@forshee.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently rcu_barrier() is used to ensure that no readers of the
inactive mini_Qdisc buffer remain before it is reused. This waits for
any pending RCU callbacks to complete, when all that is actually
required is to wait for one RCU grace period to elapse after the buffer
was made inactive. This means that using rcu_barrier() may result in
unnecessary waits.
To improve this, store the current RCU state when a buffer is made
inactive and use poll_state_synchronize_rcu() to check whether a full
grace period has elapsed before reusing it. If a full grace period has
not elapsed, wait for a grace period to elapse, and in the non-RT case
use synchronize_rcu_expedited() to hasten it.
Since this approach eliminates the RCU callback it is no longer
necessary to synchronize_rcu() in the tp_head==NULL case. However, the
RCU state should still be saved for the previously active buffer.
Before this change I would typically see mini_qdisc_pair_swap() take
tens of milliseconds to complete. After this change it typcially
finishes in less than 1 ms, and often it takes just a few microseconds.
Thanks to Paul for walking me through the options for improving this.
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Signed-off-by: Seth Forshee <sforshee@digitalocean.com>
Link: https://lore.kernel.org/r/20211026130700.121189-1-seth@forshee.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The Qdisc::running sequence counter has two uses:
1. Reliably reading qdisc's tc statistics while the qdisc is running
(a seqcount read/retry loop at gnet_stats_add_basic()).
2. As a flag, indicating whether the qdisc in question is running
(without any retry loops).
For the first usage, the Qdisc::running sequence counter write section,
qdisc_run_begin() => qdisc_run_end(), covers a much wider area than what
is actually needed: the raw qdisc's bstats update. A u64_stats sync
point was thus introduced (in previous commits) inside the bstats
structure itself. A local u64_stats write section is then started and
stopped for the bstats updates.
Use that u64_stats sync point mechanism for the bstats read/retry loop
at gnet_stats_add_basic().
For the second qdisc->running usage, a __QDISC_STATE_RUNNING bit flag,
accessed with atomic bitops, is sufficient. Using a bit flag instead of
a sequence counter at qdisc_run_begin/end() and qdisc_is_running() leads
to the SMP barriers implicitly added through raw_read_seqcount() and
write_seqcount_begin/end() getting removed. All call sites have been
surveyed though, and no required ordering was identified.
Now that the qdisc->running sequence counter is no longer used, remove
it.
Note, using u64_stats implies no sequence counter protection for 64-bit
architectures. This can lead to the qdisc tc statistics "packets" vs.
"bytes" values getting out of sync on rare occasions. The individual
values will still be valid.
Signed-off-by: Ahmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The only factor differentiating per-CPU bstats data type (struct
gnet_stats_basic_cpu) from the packed non-per-CPU one (struct
gnet_stats_basic_packed) was a u64_stats sync point inside the former.
The two data types are now equivalent: earlier commits added a u64_stats
sync point to the latter.
Combine both data types into "struct gnet_stats_basic_sync". This
eliminates redundancy and simplifies the bstats read/write APIs.
Use u64_stats_t for bstats "packets" and "bytes" data types. On 64-bit
architectures, u64_stats sync points do not use sequence counter
protection.
Signed-off-by: Ahmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The not-per-CPU variant of qdisc tc (traffic control) statistics,
Qdisc::gnet_stats_basic_packed bstats, is protected with Qdisc::running
sequence counter.
This sequence counter is used for reliably protecting bstats reads from
parallel writes. Meanwhile, the seqcount's write section covers a much
wider area than bstats update: qdisc_run_begin() => qdisc_run_end().
That read/write section asymmetry can lead to needless retries of the
read section. To prepare for removing the Qdisc::running sequence
counter altogether, introduce a u64_stats sync point inside bstats
instead.
Modify _bstats_update() to start/end the bstats u64_stats write
section.
For bisectability, and finer commits granularity, the bstats read
section is still protected with a Qdisc::running read/retry loop and
qdisc_run_begin/end() still starts/ends that seqcount write section.
Once all call sites are modified to use _bstats_update(), the
Qdisc::running seqcount will be removed and bstats read/retry loop will
be modified to utilize the internal u64_stats sync point.
Note, using u64_stats implies no sequence counter protection for 64-bit
architectures. This can lead to the statistics "packets" vs. "bytes"
values getting out of sync on rare occasions. The individual values will
still be valid.
[bigeasy: Minor commit message edits, init all gnet_stats_basic_packed.]
Signed-off-by: Ahmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The code for handling active queue changes is identical
between mq and mqprio, reuse it.
Suggested-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
mq / mqprio make the default child qdiscs visible. They only do
so for the qdiscs which are within real_num_tx_queues when the
device is registered. Depending on order of calls in the driver,
or if user space changes config via ethtool -L the number of
qdiscs visible under tc qdisc show will differ from the number
of queues. This is confusing to users and potentially to system
configuration scripts which try to make sure qdiscs have the
right parameters.
Add a new Qdisc_ops callback and make relevant qdiscs TTRT.
Note that this uncovers the "shortcut" created by
commit 1f27cde313 ("net: sched: use pfifo_fast for non real queues")
The default child qdiscs beyond initial real_num_tx are always
pfifo_fast, no matter what the sysfs setting is. Fixing this
gets a little tricky because we'd need to keep a reference
on whatever the default qdisc was at the time of creation.
In practice this is likely an non-issue the qdiscs likely have
to be configured to non-default settings, so whatever user space
is doing such configuration can replace the pfifos... now that
it will see them.
Reported-by: Matthew Massey <matthewmassey@fb.com>
Reviewed-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
According to comment in qdisc_alloc(), sch->seqlock's lockdep
class key should be set to qdisc_tx_busylock, due to possible
type error, sch->busylock's lockdep class key is set to
qdisc_tx_busylock, which is duplicated because sch->busylock's
lockdep class key is already set in qdisc_alloc().
So fix it by replacing sch->busylock with sch->seqlock.
Fixes: 96009c7d50 ("sched: replace __QDISC_STATE_RUNNING bit with a spin lock")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As MISSED and DRAINING state are used to indicate a non-empty
qdisc, qdisc->empty is not longer needed, so remove it.
Acked-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> # flexcan
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently pfifo_fast has both TCQ_F_CAN_BYPASS and TCQ_F_NOLOCK
flag set, but queue discipline by-pass does not work for lockless
qdisc because skb is always enqueued to qdisc even when the qdisc
is empty, see __dev_xmit_skb().
This patch calls sch_direct_xmit() to transmit the skb directly
to the driver for empty lockless qdisc, which aviod enqueuing
and dequeuing operation.
As qdisc->empty is not reliable to indicate a empty qdisc because
there is a time window between enqueuing and setting qdisc->empty.
So we use the MISSED state added in commit a90c57f2ce ("net:
sched: fix packet stuck problem for lockless qdisc"), which
indicate there is lock contention, suggesting that it is better
not to do the qdisc bypass in order to avoid packet out of order
problem.
In order to make MISSED state reliable to indicate a empty qdisc,
we need to ensure that testing and clearing of MISSED state is
within the protection of qdisc->seqlock, only setting MISSED state
can be done without the protection of qdisc->seqlock. A MISSED
state testing is added without the protection of qdisc->seqlock to
aviod doing unnecessary spin_trylock() for contention case.
As the enqueuing is not within the protection of qdisc->seqlock,
there is still a potential data race as mentioned by Jakub [1]:
thread1 thread2 thread3
qdisc_run_begin() # true
qdisc_run_begin(q)
set(MISSED)
pfifo_fast_dequeue
clear(MISSED)
# recheck the queue
qdisc_run_end()
enqueue skb1
qdisc empty # true
qdisc_run_begin() # true
sch_direct_xmit() # skb2
qdisc_run_begin()
set(MISSED)
When above happens, skb1 enqueued by thread2 is transmited after
skb2 is transmited by thread3 because MISSED state setting and
enqueuing is not under the qdisc->seqlock. If qdisc bypass is
disabled, skb1 has better chance to be transmited quicker than
skb2.
This patch does not take care of the above data race, because we
view this as similar as below:
Even at the same time CPU1 and CPU2 write the skb to two socket
which both heading to the same qdisc, there is no guarantee that
which skb will hit the qdisc first, because there is a lot of
factor like interrupt/softirq/cache miss/scheduling afffecting
that.
There are below cases that need special handling:
1. When MISSED state is cleared before another round of dequeuing
in pfifo_fast_dequeue(), and __qdisc_run() might not be able to
dequeue all skb in one round and call __netif_schedule(), which
might result in a non-empty qdisc without MISSED set. In order
to avoid this, the MISSED state is set for lockless qdisc and
__netif_schedule() will be called at the end of qdisc_run_end.
2. The MISSED state also need to be set for lockless qdisc instead
of calling __netif_schedule() directly when requeuing a skb for
a similar reason.
3. For netdev queue stopped case, the MISSED case need clearing
while the netdev queue is stopped, otherwise there may be
unnecessary __netif_schedule() calling. So a new DRAINING state
is added to indicate this case, which also indicate a non-empty
qdisc.
4. As there is already netif_xmit_frozen_or_stopped() checking in
dequeue_skb() and sch_direct_xmit(), which are both within the
protection of qdisc->seqlock, but the same checking in
__dev_xmit_skb() is without the protection, which might cause
empty indication of a lockless qdisc to be not reliable. So
remove the checking in __dev_xmit_skb(), and the checking in
the protection of qdisc->seqlock seems enough to avoid the cpu
consumption problem for netdev queue stopped case.
1. https://lkml.org/lkml/2021/5/29/215
Acked-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> # flexcan
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mlx5 devices were observed generating MLX5_PORT_CHANGE_SUBTYPE_ACTIVE
events without an intervening MLX5_PORT_CHANGE_SUBTYPE_DOWN. This
breaks link flap detection based on Linux carrier state transition
count as netif_carrier_on() does nothing if carrier is already on.
Make sure we count such events.
netif_carrier_event() increments the counters and fires the linkwatch
events. The latter is not necessary for the use case but seems like
the right thing to do.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
The netdev qeueue might be stopped when byte queue limit has
reached or tx hw ring is full, net_tx_action() may still be
rescheduled if STATE_MISSED is set, which consumes unnecessary
cpu without dequeuing and transmiting any skb because the
netdev queue is stopped, see qdisc_run_end().
This patch fixes it by checking the netdev queue state before
calling qdisc_run() and clearing STATE_MISSED if netdev queue is
stopped during qdisc_run(), the net_tx_action() is rescheduled
again when netdev qeueue is restarted, see netif_tx_wake_queue().
As there is time window between netif_xmit_frozen_or_stopped()
checking and STATE_MISSED clearing, between which STATE_MISSED
may set by net_tx_action() scheduled by netif_tx_wake_queue(),
so set the STATE_MISSED again if netdev queue is restarted.
Fixes: 6b3ba9146f ("net: sched: allow qdiscs to handle locking")
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently qdisc_run() checks the STATE_DEACTIVATED of lockless
qdisc before calling __qdisc_run(), which ultimately clear the
STATE_MISSED when all the skb is dequeued. If STATE_DEACTIVATED
is set before clearing STATE_MISSED, there may be rescheduling
of net_tx_action() at the end of qdisc_run_end(), see below:
CPU0(net_tx_atcion) CPU1(__dev_xmit_skb) CPU2(dev_deactivate)
. . .
. set STATE_MISSED .
. __netif_schedule() .
. . set STATE_DEACTIVATED
. . qdisc_reset()
. . .
.<--------------- . synchronize_net()
clear __QDISC_STATE_SCHED | . .
. | . .
. | . some_qdisc_is_busy()
. | . return *false*
. | . .
test STATE_DEACTIVATED | . .
__qdisc_run() *not* called | . .
. | . .
test STATE_MISS | . .
__netif_schedule()--------| . .
. . .
. . .
__qdisc_run() is not called by net_tx_atcion() in CPU0 because
CPU2 has set STATE_DEACTIVATED flag during dev_deactivate(), and
STATE_MISSED is only cleared in __qdisc_run(), __netif_schedule
is called at the end of qdisc_run_end(), causing tx action
rescheduling problem.
qdisc_run() called by net_tx_action() runs in the softirq context,
which should has the same semantic as the qdisc_run() called by
__dev_xmit_skb() protected by rcu_read_lock_bh(). And there is a
synchronize_net() between STATE_DEACTIVATED flag being set and
qdisc_reset()/some_qdisc_is_busy in dev_deactivate(), we can safely
bail out for the deactived lockless qdisc in net_tx_action(), and
qdisc_reset() will reset all skb not dequeued yet.
So add the rcu_read_lock() explicitly to protect the qdisc_run()
and do the STATE_DEACTIVATED checking in net_tx_action() before
calling qdisc_run_begin(). Another option is to do the checking in
the qdisc_run_end(), but it will add unnecessary overhead for
non-tx_action case, because __dev_queue_xmit() will not see qdisc
with STATE_DEACTIVATED after synchronize_net(), the qdisc with
STATE_DEACTIVATED can only be seen by net_tx_action() because of
__netif_schedule().
The STATE_DEACTIVATED checking in qdisc_run() is to avoid race
between net_tx_action() and qdisc_reset(), see:
commit d518d2ed86 ("net/sched: fix race between deactivation
and dequeue for NOLOCK qdisc"). As the bailout added above for
deactived lockless qdisc in net_tx_action() provides better
protection for the race without calling qdisc_run() at all, so
remove the STATE_DEACTIVATED checking in qdisc_run().
After qdisc_reset(), there is no skb in qdisc to be dequeued, so
clear the STATE_MISSED in dev_reset_queue() too.
Fixes: 6b3ba9146f ("net: sched: allow qdiscs to handle locking")
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
V8: Clearing STATE_MISSED before calling __netif_schedule() has
avoid the endless rescheduling problem, but there may still
be a unnecessary rescheduling, so adjust the commit log.
Signed-off-by: David S. Miller <davem@davemloft.net>
Lockless qdisc has below concurrent problem:
cpu0 cpu1
. .
q->enqueue .
. .
qdisc_run_begin() .
. .
dequeue_skb() .
. .
sch_direct_xmit() .
. .
. q->enqueue
. qdisc_run_begin()
. return and do nothing
. .
qdisc_run_end() .
cpu1 enqueue a skb without calling __qdisc_run() because cpu0
has not released the lock yet and spin_trylock() return false
for cpu1 in qdisc_run_begin(), and cpu0 do not see the skb
enqueued by cpu1 when calling dequeue_skb() because cpu1 may
enqueue the skb after cpu0 calling dequeue_skb() and before
cpu0 calling qdisc_run_end().
Lockless qdisc has below another concurrent problem when
tx_action is involved:
cpu0(serving tx_action) cpu1 cpu2
. . .
. q->enqueue .
. qdisc_run_begin() .
. dequeue_skb() .
. . q->enqueue
. . .
. sch_direct_xmit() .
. . qdisc_run_begin()
. . return and do nothing
. . .
clear __QDISC_STATE_SCHED . .
qdisc_run_begin() . .
return and do nothing . .
. . .
. qdisc_run_end() .
This patch fixes the above data race by:
1. If the first spin_trylock() return false and STATE_MISSED is
not set, set STATE_MISSED and retry another spin_trylock() in
case other CPU may not see STATE_MISSED after it releases the
lock.
2. reschedule if STATE_MISSED is set after the lock is released
at the end of qdisc_run_end().
For tx_action case, STATE_MISSED is also set when cpu1 is at the
end if qdisc_run_end(), so tx_action will be rescheduled again
to dequeue the skb enqueued by cpu2.
Clear STATE_MISSED before retrying a dequeuing when dequeuing
returns NULL in order to reduce the overhead of the second
spin_trylock() and __netif_schedule() calling.
Also clear the STATE_MISSED before calling __netif_schedule()
at the end of qdisc_run_end() to avoid doing another round of
dequeuing in the pfifo_fast_dequeue().
The performance impact of this patch, tested using pktgen and
dummy netdev with pfifo_fast qdisc attached:
threads without+this_patch with+this_patch delta
1 2.61Mpps 2.60Mpps -0.3%
2 3.97Mpps 3.82Mpps -3.7%
4 5.62Mpps 5.59Mpps -0.5%
8 2.78Mpps 2.77Mpps -0.3%
16 2.22Mpps 2.22Mpps -0.0%
Fixes: 6b3ba9146f ("net: sched: allow qdiscs to handle locking")
Acked-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow a policer action to enforce a rate-limit based on packets-per-second,
configurable using a packet-per-second rate and burst parameters.
e.g.
tc filter add dev tap1 parent ffff: u32 match \
u32 0 0 police pkts_rate 3000 pkts_burst 1000
Testing was unable to uncover a performance impact of this change on
existing features.
Signed-off-by: Baowen Zheng <baowen.zheng@corigine.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: Louis Peens <louis.peens@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
kmalloc() of sufficiently big portion of memory is cache-aligned
in regular conditions. If some debugging options are used,
there is no reason qdisc structures would need 64-byte alignment
if most other kernel structures are not aligned.
This get rid of QDISC_ALIGN and QDISC_ALIGNTO.
Addition of privdata field will help implementing
the reverse of qdisc_priv() and documents where
the private data is.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Allen Pais <allen.lkml@gmail.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently there is concurrent reset and enqueue operation for the
same lockless qdisc when there is no lock to synchronize the
q->enqueue() in __dev_xmit_skb() with the qdisc reset operation in
qdisc_deactivate() called by dev_deactivate_queue(), which may cause
out-of-bounds access for priv->ring[] in hns3 driver if user has
requested a smaller queue num when __dev_xmit_skb() still enqueue a
skb with a larger queue_mapping after the corresponding qdisc is
reset, and call hns3_nic_net_xmit() with that skb later.
Reused the existing synchronize_net() in dev_deactivate_many() to
make sure skb with larger queue_mapping enqueued to old qdisc(which
is saved in dev_queue->qdisc_sleeping) will always be reset when
dev_reset_queue() is called.
Fixes: 6b3ba9146f ("net: sched: allow qdiscs to handle locking")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since the quiesce/activate rework, __netdev_watchdog_up() is directly
called in the ucc_geth driver.
Unfortunately, this function is not available for modules and thus
ucc_geth cannot be built as a module anymore. Fix it by exporting
__netdev_watchdog_up().
Since the commit introducing the regression was backported to stable
branches, this one should ideally be as well.
Fixes: 79dde73cf9 ("net/ethernet/freescale: rework quiesce/activate for ucc_geth")
Signed-off-by: Valentin Longchamp <valentin@longchamp.me>
Signed-off-by: David S. Miller <davem@davemloft.net>
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>
