In commit 37112105 (net: QDISC_STATE_RUNNING dont need atomic bit
ops) I moved QDISC_STATE_RUNNING flag to __state container, located in
the cache line containing qdisc lock and often dirtied fields.

I now move TCQ_F_THROTTLED bit too, so that we let first cache line read
mostly, and shared by all cpus. This should speedup HTB/CBQ for example.

Not using test_bit()/__clear_bit()/__test_and_set_bit allows to use an
"unsigned int" for __state container, reducing by 8 bytes Qdisc size.

Introduce helpers to hide implementation details.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
CC: Jesper Dangaard Brouer <hawk@diku.dk>
CC: Jarek Poplawski <jarkao2@gmail.com>
CC: Jamal Hadi Salim <hadi@cyberus.ca>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

HTB takes into account skb is segmented in stats updates.
Generalize this to all schedulers.

They should use qdisc_bstats_update() helper instead of manipulating
bstats.bytes and bstats.packets

Add bstats_update() helper too for classes that use
gnet_stats_basic_packed fields.

Note : Right now, TCQ_F_CAN_BYPASS shortcurt can be taken only if no
stab is setup on qdisc.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pawel reported a panic related to handling shared skbs in ixgbe
incorrectly. So we need to revert my previous patch to work around
this bug. Instead of reverting the patch completely, I just revert
the essential lines, so we can add the previous optimization
back more easily in future.

    commit 3511c913
    Author: Changli Gao <xiaosuo@gmail.com>
    Date:   Sat Oct 16 13:04:08 2010 +0000

        net_sched: remove the unused parameter of qdisc_create_dflt()
Reported-by: NPawel Staszewski <pstaszewski@itcare.pl>
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
Acked-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add dev_close_many and dev_deactivate_many to factorize another
sync-rcu operation on the netdevice unregister path.

$ modprobe dummy numdummies=10000
$ ip link set dev dummy* up
$ time rmmod dummy

Without the patch           With the patch

real    0m 24.63s           real    0m 5.15s
user    0m 0.00s            user    0m 0.00s
sys     0m 6.05s            sys     0m 5.14s
Signed-off-by: NOctavian Purdila <opurdila@ixiacom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The first parameter dev isn't in use in qdisc_create_dflt().
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
Acked-by: NJamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Change "return (EXPR);" to "return EXPR;"

return is not a function, parentheses are not required.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Reducing real_num_queues needs to flush the qdisc otherwise
skbs with queue_mappings greater then real_num_tx_queues can
be sent to the underlying driver.

The flow for this is,

dev_queue_xmit()
	dev_pick_tx()
		skb_tx_hash()  => hash using real_num_tx_queues
		skb_set_queue_mapping()
	...
	qdisc_enqueue_root() => enqueue skb on txq from hash
...
dev->real_num_tx_queues -= n
...
sch_direct_xmit()
	dev_hard_start_xmit()
		ndo_start_xmit(skb,dev) => skb queue set with old hash

skbs are enqueued on the qdisc with skb->queue_mapping set
0 < queue_mappings < real_num_tx_queues.  When the driver
decreases real_num_tx_queues skb's may be dequeued from the
qdisc with a queue_mapping greater then real_num_tx_queues.

This fixes a case in ixgbe where this was occurring with DCB
and FCoE. Because the driver is using queue_mapping to map
skbs to tx descriptor rings we can potentially map skbs to
rings that no longer exist.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Tested-by: NRoss Brattain <ross.b.brattain@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When calling qdisc_reset() the qdisc lock needs to be held.  In
this case there is at least one driver i4l which is using this
without holding the lock.  Add the locking here.
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

don't clone skb when skb isn't shared

When the tcf_action is TC_ACT_STOLEN, and the skb isn't shared, we don't need
to clone a new skb. As the skb will be freed after this function returns, we
can use it freely once we get a reference to it.
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
----
 include/net/sch_generic.h |   11 +++++++++--
 net/sched/act_mirred.c    |    6 +++---
 2 files changed, 12 insertions(+), 5 deletions(-)
Signed-off-by: NJamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When many cpus compete for sending frames on a given qdisc, the qdisc
spinlock suffers from very high contention.

The cpu owning __QDISC_STATE_RUNNING bit has same priority to acquire
the lock, and cannot dequeue packets fast enough, since it must wait for
this lock for each dequeued packet.

One solution to this problem is to force all cpus spinning on a second
lock before trying to get the main lock, when/if they see
__QDISC_STATE_RUNNING already set.

The owning cpu then compete with at most one other cpu for the main
lock, allowing for higher dequeueing rate.

Based on a previous patch from Alexander Duyck. I added the heuristic to
avoid the atomic in fast path, and put the new lock far away from the
cache line used by the dequeue worker. Also try to release the busylock
lock as late as possible.

Tests with following script gave a boost from ~50.000 pps to ~600.000
pps on a dual quad core machine (E5450 @3.00GHz), tg3 driver.
(A single netperf flow can reach ~800.000 pps on this platform)

for j in `seq 0 3`; do
  for i in `seq 0 7`; do
    netperf -H 192.168.0.1 -t UDP_STREAM -l 60 -N -T $i -- -m 6 &
  done
done
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Acked-by: NAlexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

__QDISC_STATE_RUNNING is always changed while qdisc lock is held.

We can avoid two atomic operations in xmit path, if we move this bit in
a new __state container.

Location of this __state container is carefully chosen so that fast path
only dirties one qdisc cache line.

THROTTLED bit could later be moved into this __state location too, to
avoid dirtying first qdisc cache line.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Define three helpers to manipulate QDISC_STATE_RUNNIG flag, that a
second patch will move on another location.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

One of my test machine got a deadlock during "tc" sessions,
adding/deleting classes & filters, using traffic estimators.

After some analysis, I believe we have a potential use after free case
in est_timer() :

spin_lock(e->stats_lock); << HERE >>
read_lock(&est_lock);
if (e->bstats == NULL)   << TEST >>
	goto skip;

Test is done a bit late, because after estimator is killed, and before
rcu grace period elapsed, we might already have freed/reuse memory where
e->stats_locks points to (some qdisc->q.lock)

A possible fix is to respect a rcu grace period at Qdisc dismantle time.

On 64bit, sizeof(struct Qdisc) is exactly 192 bytes. Adding 16 bytes to
it (for struct rcu_head) is a problem because it might change
performance, given QDISC_ALIGNTO is 32 bytes.

This is why I also change QDISC_ALIGNTO to 64 bytes, to satisfy most
current alignment requirements.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This adds an additional queuing strategy, called pfifo_head_drop,
to remove the oldest skb in the case of an overflow within the queue -
the head element - instead of the last skb (tail). To remove the oldest
skb in congested situations is useful for sensor network environments
where newer packets reflect the superior information.
Reviewed-by: NFlorian Westphal <fw@strlen.de>
Acked-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NHagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This cleanup patch puts struct/union/enum opening braces,
in first line to ease grep games.

struct something
{

becomes :

struct something {
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After the recent mq change there is the new select_queue qdisc class
method used in tc_modify_qdisc, but it works OK only for direct child
qdiscs of mq qdisc. Grandchildren always get the first tx queue, which
would give wrong qdisc_root etc. results (e.g. for sch_htb as child of
sch_prio). This patch fixes it by using parent's dev_queue for such
grandchildren qdiscs. The select_queue method's return type is changed
BTW.

With feedback from: Patrick McHardy <kaber@trash.net>
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When new child qdiscs are attached to the mq qdisc, they are actually
attached as root qdiscs to the device queues. The lock selection for
new estimators incorrectly picks the root lock of the existing and
to be replaced qdisc, which results in a use-after-free once the old
qdisc has been destroyed.

Mark mq qdisc instances with a new flag and treat qdiscs attached to
mq as children similar to regular root qdiscs.

Additionally prevent estimators from being attached to the mq qdisc
itself since it only updates its byte and packet counters during dumps.
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds a classful dummy scheduler which can be used as root qdisc
for multiqueue devices and exposes each device queue as a child class.

This allows to address queues individually and graft them similar to regular
classes. Additionally it presents an accumulated view of the statistics of
all real root qdiscs in the dummy root.

Two new callbacks are added to the qdisc_ops and qdisc_class_ops:

- cl_ops->select_queue selects the tx queue number for new child classes.

- qdisc_ops->attach() overrides root qdisc device grafting to attach
  non-shared qdiscs to the queues.
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It will be used in a following patch by the multiqueue qdisc.
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In 5e140dfc "net: reorder struct Qdisc
for better SMP performance" the definition of struct gnet_stats_basic
changed incompatibly, as copies of this struct are shipped to
userland via netlink.

Restoring old behavior is not welcome, for performance reason.

Fix is to use a private structure for kernel, and
teach gnet_stats_copy_basic() to convert from kernel to user land,
using legacy structure (struct gnet_stats_basic)

Based on a report and initial patch from Michael Spang.
Reported-by: NMichael Spang <mspang@csclub.uwaterloo.ca>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

dev_queue_xmit enqueue's a skb and calls qdisc_run which
dequeue's the skb and xmits it. In most cases, the skb that
is enqueue'd is the same one that is dequeue'd (unless the
queue gets stopped or multiple cpu's write to the same queue
and ends in a race with qdisc_run). For default qdiscs, we
can remove the redundant enqueue/dequeue and simply xmit the
skb since the default qdisc is work-conserving.

The patch uses a new flag - TCQ_F_CAN_BYPASS to identify the
default fast queue. The controversial part of the patch is
incrementing qlen when a skb is requeued - this is to avoid
checks like the second line below:

+  } else if ((q->flags & TCQ_F_CAN_BYPASS) && !qdisc_qlen(q) &&
>>         !q->gso_skb &&
+          !test_and_set_bit(__QDISC_STATE_RUNNING, &q->state)) {

Results of a 2 hour testing for multiple netperf sessions (1,
2, 4, 8, 12 sessions on a 4 cpu system-X). The BW numbers are
aggregate Mb/s across iterations tested with this version on
System-X boxes with Chelsio 10gbps cards:

----------------------------------
Size |  ORG BW          NEW BW   |
----------------------------------
128K |  156964          159381   |
256K |  158650          162042   |
----------------------------------

Changes from ver1:

1. Move sch_direct_xmit declaration from sch_generic.h to
   pkt_sched.h
2. Update qdisc basic statistics for direct xmit path.
3. Set qlen to zero in qdisc_reset.
4. Changed some function names to more meaningful ones.
Signed-off-by: NKrishna Kumar <krkumar2@in.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

dev_queue_xmit() needs to dirty fields "state", "q", "bstats" and "qstats"

On x86_64 arch, they currently span three cache lines, involving more
cache line ping pongs than necessary, making longer holding of queue spinlock.

We can reduce this to one cache line, by grouping all read-mostly fields
at the beginning of structure. (Or should I say, all highly modified fields
at the end :) )

Before patch :

offsetof(struct Qdisc, state)=0x38
offsetof(struct Qdisc, q)=0x48
offsetof(struct Qdisc, bstats)=0x80
offsetof(struct Qdisc, qstats)=0x90
sizeof(struct Qdisc)=0xc8

After patch :

offsetof(struct Qdisc, state)=0x80
offsetof(struct Qdisc, q)=0x88
offsetof(struct Qdisc, bstats)=0xa0
offsetof(struct Qdisc, qstats)=0xac
sizeof(struct Qdisc)=0xc0
Signed-off-by: NEric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Patrick McHardy <kaber@trash.net> suggested:
> How about making this flag and the warning message (in a out-of-line
> function) globally available? Other qdiscs (f.i. HFSC) can't deal with
> inner non-work-conserving qdiscs as well.

This patch uses qdisc->flags field of "suspected" child qdisc.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After implementing qdisc->ops->peek() and changing sch_netem into
classless qdisc there are no more qdisc->ops->requeue() users. This
patch removes this method with its wrappers (qdisc_requeue()), and
also unused qdisc->requeue structure. There are a few minor fixes of
warnings (htb_enqueue()) and comments btw.

The idea to kill ->requeue() and a similar patch were first developed
by David S. Miller.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A packet dequeued and stored as gso_skb in qdisc_peek_dequeued() should
be seen as part of the queue for sch->q.qlen queries until it's really
dequeued with qdisc_dequeue_peeked(), so qlen needs additional updating
in these functions. (Updating qstats.backlog shouldn't matter here.)
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds qdisc_peek_dequeued() wrapper to emulate peek method
with qdisc->dequeue() and storing "peeked" skb in qdisc->gso_skb until
dequeuing. This is mainly for compatibility reasons not to break some
strange configs because peeking is expected for non-work-conserving
parent qdiscs to query work-conserving child qdiscs.

This implementation requires using qdisc_dequeue_peeked() wrapper
instead of directly calling qdisc->dequeue() for all qdiscs ever
querried with qdisc->ops->peek() or qdisc_peek_dequeued().
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

From: Patrick McHardy <kaber@trash.net>

Just as a demonstration how easy adding a peek operation to the
work-conserving qdiscs actually is. It doesn't need to keep or change
any internal state in many cases thanks to the guarantee that the
packet will either be dequeued or, if another packet arrives, the
upper qdisc will immediately ->peek again to reevaluate the state.

(This is only slightly modified Patrick's patch.)
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add Qdisc_ops peek() method in order to replace requeuing.

Based on ideas and patches of Herbert Xu, Patrick McHardy and
David S. Miller.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jay Cliburn noticed and diagnosed a bug triggered in
dev_gso_skb_destructor() after last change from qdisc->gso_skb
to qdisc->requeue list. Since gso_segmented skbs can't be queued
to another list this patch brings back qdisc->gso_skb for them.
Reported-by: NJay Cliburn <jcliburn@gmail.com>
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The idea is that we can use this to get rid of
->requeue().
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use new qdisc_root_sleeping_lock() instead of qdisc_root_lock() as
sch_tree_lock() because this lock could be used while dev is
deactivated, but we never need to use this with noop_qdisc as a root.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

While passing a qdisc root lock to gen_new_estimator() and
gen_replace_estimator() dev could be deactivated or even before
grafting proper root qdisc as qdisc_sleeping (e.g. qdisc_create), so
using qdisc_root_lock() is not enough. This patch adds
qdisc_root_sleeping_lock() for this, plus additional checks, where
necessary.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

dev_deactivate() can skip rescheduling of a qdisc by qdisc_watchdog()
or other timer calling netif_schedule() after dev_queue_deactivate().
We prevent this checking aliveness before scheduling the timer. Since
during deactivation the root qdisc is available only as qdisc_sleeping
additional accessor qdisc_root_sleeping() is created.

With feedback from Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We can now kill them synchronously with all of the
previous dev_deactivate() cures.

This makes netdev destruction and shutdown saner as
the qdiscs hold references to the device.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This new state lets dev_deactivate() mark a qdisc as having been
deactivated.

dev_queue_xmit() and ing_filter() check for this bit and do not
try to process the qdisc if the bit is set.

dev_deactivate() polls the qdisc after setting the bit, waiting
for both __QDISC_STATE_RUNNING and __QDISC_STATE_SCHED to clear.

This isn't perfect yet, but subsequent changesets will make it so.
This part is just one piece of the puzzle.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Patrick McHardy <kaber@trash.net> noticed that it would be nice to
handle NET_XMIT_BYPASS by NET_XMIT_SUCCESS with an internal qdisc flag
__NET_XMIT_BYPASS and to remove the mapping from dev_queue_xmit().

David Miller <davem@davemloft.net> spotted a serious bug in the first
version of this patch.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Patrick McHardy <kaber@trash.net> noticed:
"The other problem that affects all qdiscs supporting actions is
TC_ACT_QUEUED/TC_ACT_STOLEN getting mapped to NET_XMIT_SUCCESS
even though the packet is not queued, corrupting upper qdiscs'
qlen counters."

and later explained:
"The reason why it translates it at all seems to be to not increase
the drops counter. Within a single qdisc this could be avoided by
other means easily, upper qdiscs would still increase the counter
when we return anything besides NET_XMIT_SUCCESS though.

This means we need a new NET_XMIT return value to indicate this to
the upper qdiscs. So I'd suggest to introduce NET_XMIT_STOLEN,
return that to upper qdiscs and translate it to NET_XMIT_SUCCESS
in dev_queue_xmit, similar to NET_XMIT_BYPASS."

David Miller <davem@davemloft.net> noticed:
"Maybe these NET_XMIT_* values being passed around should be a set of
bits. They could be composed of base meanings, combined with specific
attributes.

So you could say "NET_XMIT_DROP | __NET_XMIT_NO_DROP_COUNT"

The attributes get masked out by the top-level ->enqueue() caller,
such that the base meanings are the only thing that make their
way up into the stack. If it's only about communication within the
qdisc tree, let's simply code it that way."

This patch is trying to realize these ideas.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It is the only legal environment in which this can be
used.

Add some commentary explaining the situation.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The stab bits can't be referenced uniless the full
packet scheduler layer is enabled.

Reported by Stephen Rothwell.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add size table functions for qdiscs and calculate packet size in
qdisc_enqueue().

Based on patch by Patrick McHardy
 http://marc.info/?l=linux-netdev&m=115201979221729&w=2Signed-off-by: NJussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>