Skip to content

L
linux
Switch branch/tag
  1. 11 Apr, 2019 2 commits
    • Leandro Dorileo's avatar
      net/sched: cbs: fix port_rate miscalculation · e0a7683d
      Leandro Dorileo authored 
      The Credit Based Shaper heavily depends on link speed to calculate
the scheduling credits, we can't properly calculate the credits if the
device has failed to report the link speed.

In that case we can't dequeue packets assuming a wrong port rate that will
result into an inconsistent credit distribution.

This patch makes sure we fail to dequeue case:

1) __ethtool_get_link_ksettings() reports error or 2) the ethernet driver
failed to set the ksettings' speed value (setting link speed to
SPEED_UNKNOWN).

Additionally we properly re calculate the port rate whenever the link speed
is changed.

Fixes: 3d0bd028 ("net/sched: Add support for HW offloading for CBS")
Signed-off-by: default avatarLeandro Dorileo <leandro.maciel.dorileo@intel.com>
Reviewed-by: default avatarVedang Patel <vedang.patel@intel.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      e0a7683d
    • Leandro Dorileo's avatar
      net/sched: taprio: fix picos_per_byte miscalculation · 7b9eba7b
      Leandro Dorileo authored 
      The Time Aware Priority Scheduler is heavily dependent to link speed,
it relies on it to calculate transmission bytes per cycle, we can't
properly calculate the so called budget if the device has failed
to report the link speed.

In that case we can't dequeue packets assuming a wrong budget.
This patch makes sure we fail to dequeue case:

1) __ethtool_get_link_ksettings() reports error or 2) the ethernet
driver failed to set the ksettings' speed value (setting link speed
to SPEED_UNKNOWN).

Additionally we re calculate the budget whenever the link speed is
changed.

Fixes: 5a781ccb ("tc: Add support for configuring the taprio scheduler")
Signed-off-by: default avatarLeandro Dorileo <leandro.maciel.dorileo@intel.com>
Reviewed-by: default avatarVedang Patel <vedang.patel@intel.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      7b9eba7b
  3. 10 Apr, 2019 8 commits
    • Jakub Kicinski's avatar
      net: strparser: fix comment · 93e21254
      Jakub Kicinski authored 
      Fix comment.
Signed-off-by: default avatarJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      93e21254
    • David Ahern's avatar
      ipv4: Handle RTA_GATEWAY set to 0 · d73f80f9
      David Ahern authored 
      Govindarajulu reported a regression with Network Manager which sends an
RTA_GATEWAY attribute with the address set to 0. Fixup the handling of
RTA_GATEWAY to only set fc_gw_family if the gateway address is actually
set.

Fixes: f35b794b ("ipv4: Prepare fib_config for IPv6 gateway")
Reported-by: default avatarGovindarajulu Varadarajan <govind.varadar@gmail.com>
Signed-off-by: default avatarDavid Ahern <dsahern@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      d73f80f9
    • David S. Miller's avatar
      Merge branch 'net-sched-move-back-qlen-to-per-CPU-accounting' · 44b9b6ca
      David S. Miller authored 
      Paolo Abeni says:

====================
net: sched: move back qlen to per CPU accounting

The commit 46b1c18f ("net: sched: put back q.qlen into a single location")
introduced some measurable regression in the contended scenarios for
lock qdisc.

As Eric suggested we could replace q.qlen access with calls to qdisc_is_empty()
in the datapath and revert the above commit. The TC subsystem updates
qdisc->is_empty in a somewhat loose way: notably 'is_empty' is set only when
the qdisc dequeue() calls return a NULL ptr. That is, the invocation after
the last packet is dequeued.

The above is good enough for BYPASS implementation - the only downside is that
we end up avoiding the optimization for a very small time-frame - but will
break hard things when internal structures consistency for classful qdisc
relies on child qdisc_is_empty().

A more strict 'is_empty' update adds a relevant complexity to its life-cycle, so
this series takes a different approach: we allow lockless qdisc to switch from
per CPU accounting to global stats accounting when the NOLOCK bit is cleared.
Since most pieces of infrastructure are already in place, this requires very
little changes to the pfifo_fast qdisc, and any later NOLOCK qdisc can hook
there with little effort - no need to maintain two different implementations.

The first 2 patches removes direct qlen access from non core TC code, the 3rd
and 4th patches place and use the infrastructure to allow stats account
switching and the 5th patch is the actual revert.

 v1 -> v2:
  - fixed build issues
  - more descriptive commit message for patch 5/5
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      44b9b6ca
    • Paolo Abeni's avatar
      Revert: "net: sched: put back q.qlen into a single location" · 73eb628d
      Paolo Abeni authored 
      This revert commit 46b1c18f ("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
e5f0e8f8 ("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: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      73eb628d
    • Paolo Abeni's avatar
      net: sched: when clearing NOLOCK, clear TCQ_F_CPUSTATS, too · 8a53e616
      Paolo Abeni authored 
      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: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      8a53e616
    • Paolo Abeni's avatar
      net: sched: always do stats accounting according to TCQ_F_CPUSTATS · 9c01c9f1
      Paolo Abeni authored 
      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: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      9c01c9f1
    • Paolo Abeni's avatar
      net: sched: prefer qdisc_is_empty() over direct qlen access · 1f5e6fdd
      Paolo Abeni authored 
      When checking for root qdisc queue length, do not access directly q.qlen.
In the following patches we will move back qlen accounting to per CPU
values for NOLOCK qdiscs.

Instead, prefer the qdisc_is_empty() helper usage.
Signed-off-by: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      1f5e6fdd
    • Paolo Abeni's avatar
      net: caif: avoid using qdisc_qlen() · b0a231a2
      Paolo Abeni authored 
      Such helper does not cope correctly with NOLOCK qdiscs.
In the following patches we will move back qlen to per CPU
values for such qdiscs, so qdisc_qlen_sum() is not an option,
too.
Instead, use qlen only for lock qdiscs, and always set
flow off for NOLOCK qdiscs with a not empty tx queue.
Signed-off-by: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      b0a231a2
  5. 09 Apr, 2019 9 commits
  7. 08 Apr, 2019 21 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7