The code for adding/deleting fdb flow is repeated when
user-space does flow add/del and when we add/del from
the neigh update path - unify them to avoid the duplication.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
When replacing a tc flower rule, flower first requests to add the
new rule (new action), then deletes the old one.
But currently when asked to add a new tc flower flow, we append the
actions (and counters to it).
This can result in a fte with two flow counters or conflicting
actions (drop and encap action) which firmware complains/errs
about and isn't achieving what the user aimed for.
Instead, insert the flow using the new no-append flag which will add a
new HW rule, the old flow and rule will be deleted later by flower
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanmox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
If no-append flag is set, we will add a new FTE, instead of appending
the actions of the inserted rule when the same match already exists.
While here, move the has_flow_tag boolean indicator to be a flag too.
This patch doesn't change any functionality.
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanmox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
A chain is a group of priorities, so use the fdb parallel
sub namespaces to implement chains, and a flow table for each
priority in them.
Because these namespaces are parallel and in series to the slow path
fdb, the chains aren't connected to one another (but to the slow path),
and one must use a explicit goto action to reach a different chain.
Flow tables for the priorities will be created on demand and destroyed
once not used.
The Firmware has four pools of tables for sizes S/XS/M/L (4k, 64k, 1m, 4m).
We maintain ghost copies of the pools occupancy.
When a new table is to be created, we scan the pools from large to small
and find the 1st table size which can be now created. When a table is
destroyed, we update the relevant pool.
Multi chain/prio isn't enabled yet by this patch, for now all flows
will use the default chain 0, and prio 1.
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Be symmetric with the e-switch API to add rules which has a
specific function to add fwd rules which are used as part of
vport mirroring.
This patch doesn't change any functionality.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Towards supporting multi-chains and priorities, split the FDB fast path
to multiple namespaces (sub namespaces), each with multiple priorities.
This patch adds a new flow steering type, FS_TYPE_PRIO_CHAINS, which is
like current FS_TYPE_PRIO, but may contain only namespaces, and those
will be in parallel to one another in terms of managing of the flow
tables connections inside them. Meaning, while searching for the next
or previous flow table to connect for a new table inside such namespace
we skip the parallel namespaces in the same level under the
FS_TYPE_PRIO_CHAINS prio we originated from.
We use this new type for splitting the fast path prio into multiple
parallel namespaces, each containing normal prios.
The prios inside them (and their tables) will be connected to one
another, but not from one parallel namespace to another, instead the
last prio in each namespace will be connected to the next prio in
the containing FDB namespace, which is the slow path prio.
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Acked-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
If set, the firmware supports creating of flow tables with encap
enabled while VFs are configured, if we already created one
(restriction still applies on the first creation).
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Move to have clear separation on the code path to add nic vs e-switch
flows. While here we break the code that deals with adding offloaded
TC tool to few smaller stages, each on helper function.
Besides getting us simpler and readable code, these are pre-steps
for being able to have two HW flows serving one SW TC flow for some
e-switch use cases.
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Refactor the flow add utility functions to return err code instead of rule
pointers. This will allow for simpler logic when one tc rule is
duplicated to two HW rules in downstream patches.
Signed-off-by: Rabie Loulou <rabiel@mellanox.com>
Signed-off-by: Shahar Klein <shahark@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Currently, when a flow rule is created using the FS core layer, the caller
has to pass the entire flow counter object and not just the counter HW
handle (ID). This requires both the FS core and the caller to have
knowledge about the inner implementation of the FS layer flow counters
cache and limits the possible users.
Move to use the counter ID across the place when dealing with flows.
Doing this decoupling, now can we privatize the inner implementation
of the flow counters.
Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
There's no real reason for the e-switch logic to manage the creation of
counters for offloaded flows. The API already has the directive for the
caller to denote they want to attach a counter to the created flow.
As such, we go and move the management of flow counters to the mlx5e
tc offload logic. This also lets us remove an inelegant interface where
the FS layer had to provide a way to retrieve a counter from a flow rule.
Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
mlx5 updates for both net-next and rdma-next
* 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux: (21 commits)
net/mlx5: Expose DC scatter to CQE capability bit
net/mlx5: Update mlx5_ifc with DEVX UID bits
net/mlx5: Set uid as part of DCT commands
net/mlx5: Set uid as part of SRQ commands
net/mlx5: Set uid as part of SQ commands
net/mlx5: Set uid as part of RQ commands
net/mlx5: Set uid as part of QP commands
net/mlx5: Set uid as part of CQ commands
net/mlx5: Rename incorrect naming in IFC file
net/mlx5: Export packet reformat alloc/dealloc functions
net/mlx5: Pass a namespace for packet reformat ID allocation
net/mlx5: Expose new packet reformat capabilities
{net, RDMA}/mlx5: Rename encap to reformat packet
net/mlx5: Move header encap type to IFC header file
net/mlx5: Break encap/decap into two separated flow table creation flags
net/mlx5: Add support for more namespaces when allocating modify header
net/mlx5: Export modify header alloc/dealloc functions
net/mlx5: Add proper NIC TX steering flow tables support
net/mlx5: Cleanup flow namespace getter switch logic
net/mlx5: Add memic command opcode to command checker
...
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Eric reported that syzkaller triggered a splat in tcp_cleanup_ulp()
where assertion sock_owned_by_me() failed. This happened through
inet_csk_prepare_forced_close() first releasing the socket lock,
then calling into tcp_done(newsk) which is called after the
inet_csk_prepare_forced_close() and therefore without the socket
lock held. The sock_owned_by_me() assertion can generally be
removed as the only place where tcp_cleanup_ulp() is called from
now is out of inet_csk_destroy_sock() -> sk->sk_prot->destroy()
where socket is in dead state and unreachable. Therefore, add a
comment why the check is not needed instead.
Fixes: 8b9088f806 ("tcp, ulp: enforce sock_owned_by_me upon ulp init and cleanup")
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
dc_req_scat_data_cqe capability bit determines
if requester scatter to cqe is available for 64 bytes CQE over
DC transport type.
Signed-off-by: Yonatan Cohen <yonatanc@mellanox.com>
Reviewed-by: Guy Levi <guyle@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Yunsheng Lin says:
====================
Some cleanup and bugfix for desc filling
When retransmiting packets, skb_cow_head which is called in
hns3_set_tso may clone a new header. And driver will clear the
checksum of the header after doing DMA map, so HW will read the
old header whose L3 checksum is not cleared and calculate a
wrong L3 checksum.
Also When sending a big fragment using multiple buffer descriptor,
hns3 does one maping, but do multiple unmapping when tx is done,
which may cause unmapping problem.
This patchset does some cleanup before fixing the above problem.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
When sending a big fragment using multiple buffer descriptor,
hns3 does one maping, but do multiple unmapping when tx is done,
which may cause unmapping problem.
To fix it, this patch makes sure the value of desc_cb.length of
the non-first bd is zero. If desc_cb.length is zero, we do not
unmap the buffer.
Fixes: 76ad4f0ee7 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To keep symmetrical, this patch renames hns_nic_dma_unmap to
hns3_clear_desc.
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch unifies big tx fragment handling for tso and non-tso
case.
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To solve the L3 checksum error problem which happens when driver
does not clear L3 checksum, DMA map should be done after calling
skb_cow_head.
This patch moves DMA map into hns3_fill_desc to ensure that DMA
map is done after calling skb_cow_head.
Fixes: 76ad4f0ee7 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch removes hns3_fill_desc_tso in preparation for
fixing some desc filling bug, because for tso or non-tso
case, we will use the unified hns3_fill_desc.
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rahul Verma says:
====================
Align PTT and add various link modes.
This series aligns the ptt propagation as local ptt or global ptt.
Adds new transceiver modes, speed capabilities and board config,
which is utilized to display the enhanced link modes, media types
and speed. Enhances the link with detailed information.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Newly added link modes are required to be added
during setting link modes. If the new link mode
is not available during qed_set_link, it may cause
link getting down due to empty supported capability,
being passed to MFW, after setting autoneg off/on
with current/supported speed.
Signed-off-by: Rahul Verma <Rahul.Verma@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Set link mode after checking available "supported" link caps
of the port.
Signed-off-by: Rahul Verma <Rahul.Verma@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Added transceiver type, speed capability and board types
in HSI, are utilizing to display the accurate link
information in ethtool.
Signed-off-by: Rahul Verma <Rahul.Verma@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Added transceiver modes with different speed and media type,
speed capability and supported board types in HSI, which
will be utilizing to display correct specification of link
modes and speed type.
Signed-off-by: Rahul Verma <Rahul.Verma@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Align the use of local PTT to propagate through the qed_mcp* API's.
Global ptt should not be used.
Register access should be done through layers. Register address is
mapped into a PTT, PF translation table. Several interface functions
require a PTT to direct read/write into register. There is a pool of
PTT maintained, and several PTT are used simultaneously to access
device registers in different flows. Same PTT should not be used in
flows that can run concurrently.
To avoid running out of PTT resources, too many PTT should not be
acquired without releasing them. Every PF has a global PTT, which is
used throughout the life of PF, in most important flows for register
access. Generic functions acquire the PTT locally and release after
the use. This patch aligns the use of Global PTT and Local PTT
accordingly.
Signed-off-by: Rahul Verma <rahul.verma@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes the following sparse warning:
drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils_fw2x.c:282:5: warning:
symbol 'aq_fw2x_update_stats' was not declared. Should it be static?
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern says:
====================
net: Kernel side filtering for route dumps
Implement kernel side filtering of route dumps by protocol (e.g., which
routing daemon installed the route), route type (e.g., unicast), table
id and nexthop device.
iproute2 has been doing this filtering in userspace for years; pushing
the filters to the kernel side reduces the amount of data the kernel
sends and reduces wasted cycles on both sides processing unwanted data.
These initial options provide a huge improvement for efficiently
examining routes on large scale systems.
v2
- better handling of requests for a specific table. Rather than walking
the hash of all tables, lookup the specific table and dump it
- refactor mr_rtm_dumproute moving the loop over the table into a
helper that can be invoked directly
- add hook to return NLM_F_DUMP_FILTERED in DONE message to ensure
it is returned even when the dump returns nothing
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Unlike IPv6, IPv4 does not have routes marked with RTF_PREFIX_RT. If the
flag is set in the dump request, just return.
In the process of this change, move the CLONE check to use the new
filter flags.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similar to IPv4, IPv6 fib no longer contains cloned routes. If a user
requests a route dump for only cloned entries, no sense walking the FIB
and returning everything.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update the dump request parsing in MPLS for the non-INET case to
enable kernel side filtering. If INET is disabled the only filters
that make sense for MPLS are protocol and nexthop device.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update parsing of route dump request to enable kernel side filtering.
Allow filtering results by protocol (e.g., which routing daemon installed
the route), route type (e.g., unicast), table id and nexthop device. These
amount to the low hanging fruit, yet a huge improvement, for dumping
routes.
ip_valid_fib_dump_req is called with RTNL held, so __dev_get_by_index can
be used to look up the device index without taking a reference. From
there filter->dev is only used during dump loops with the lock still held.
Set NLM_F_DUMP_FILTERED in the answer_flags so the user knows the results
have been filtered should no entries be returned.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement kernel side filtering of routes by egress device index and
table id. If the table id is given in the filter, lookup table and
call mr_table_dump directly for it.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move per-table loops from mr_rtm_dumproute to mr_table_dump and export
mr_table_dump for dumps by specific table id.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement kernel side filtering of routes by egress device index and
protocol. MPLS uses only a single table and route type.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement kernel side filtering of routes by table id, egress device
index, protocol, and route type. If the table id is given in the filter,
lookup the table and call fib6_dump_table directly for it.
Move the existing route flags check for prefix only routes to the new
filter.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement kernel side filtering of routes by table id, egress device index,
protocol and route type. If the table id is given in the filter, lookup the
table and call fib_table_dump directly for it.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add struct fib_dump_filter for options on limiting which routes are
returned in a dump request. The current list is table id, protocol,
route type, rtm_flags and nexthop device index. struct net is needed
to lookup the net_device from the index.
Declare the filter for each route dump handler and plumb the new
arguments from dump handlers to ip_valid_fib_dump_req.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With dump filtering we need a way to ensure the NLM_F_DUMP_FILTERED
flag is set on a message back to the user if the data returned is
influenced by some input attributes. Normally this can be done as
messages are added to the skb, but if the filter results in no data
being returned, the user could be confused as to why.
This patch adds answer_flags to the netlink_callback allowing dump
handlers to set the NLM_F_DUMP_FILTERED at a minimum in the
NLMSG_DONE message ensuring the flag gets back to the user.
The netlink_callback space is initialized to 0 via a memset in
__netlink_dump_start, so init of the new answer_flags is covered.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf-next 2018-10-16
The following pull-request contains BPF updates for your *net-next* tree.
The main changes are:
1) Convert BPF sockmap and kTLS to both use a new sk_msg API and enable
sk_msg BPF integration for the latter, from Daniel and John.
2) Enable BPF syscall side to indicate for maps that they do not support
a map lookup operation as opposed to just missing key, from Prashant.
3) Add bpftool map create command which after map creation pins the
map into bpf fs for further processing, from Jakub.
4) Add bpftool support for attaching programs to maps allowing sock_map
and sock_hash to be used from bpftool, from John.
5) Improve syscall BPF map update/delete path for map-in-map types to
wait a RCU grace period for pending references to complete, from Daniel.
6) Couple of follow-up fixes for the BPF socket lookup to get it
enabled also when IPv6 is compiled as a module, from Joe.
7) Fix a generic-XDP bug to handle the case when the Ethernet header
was mangled and thus update skb's protocol and data, from Jesper.
8) Add a missing BTF header length check between header copies from
user space, from Wenwen.
9) Minor fixups in libbpf to use __u32 instead u32 types and include
proper perf_event.h uapi header instead of perf internal one, from Yonghong.
10) Allow to pass user-defined flags through EXTRA_CFLAGS and EXTRA_LDFLAGS
to bpftool's build, from Jiri.
11) BPF kselftest tweaks to add LWTUNNEL to config fragment and to install
with_addr.sh script from flow dissector selftest, from Anders.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
After commit 9f2959b6b5 ("net: phy: improve handling delayed work")
the sync parameter isn't needed any longer in phy_start_aneg_priv().
This allows to merge phy_start_aneg() and phy_start_aneg_priv().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The VF device's serial number is saved as a string in PCI slot's
kobj name, not the slot->number. This patch corrects the netvsc
driver, so the VF device can be successfully paired with synthetic
NIC.
Fixes: 00d7ddba11 ("hv_netvsc: pair VF based on serial number")
Reported-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet says:
====================
tcp: second round for EDT conversion
First round of EDT patches left TCP stack in a non optimal state.
- High speed flows suffered from loss of performance, addressed
by the first patch of this series.
- Second patch brings pacing to the current state of networking,
since we now reach ~100 Gbit on a single TCP flow.
- Third patch implements a mitigation for scheduling delays,
like the one we did in sch_fq in the past.
- Fourth patch removes one special case in sch_fq for ACK packets.
- Fifth patch removes a serious perfomance cost for TCP internal
pacing. We should setup the high resolution timer only if
really needed.
- Sixth patch fixes a typo in BBR.
- Last patch is one minor change in cdg congestion control.
Neal Cardwell also has a patch series fixing BBR after
EDT adoption.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
We store in tcp socket a cache of most recent high resolution
clock, there is no need to call local_clock() again, since
this cache is good enough.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There was a typo in this parameter name.
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When TCP implements its own pacing (when no fq packet scheduler is used),
it is arming high resolution timer after a packet is sent.
But in many cases (like TCP_RR kind of workloads), this high resolution
timer expires before the application attempts to write the following
packet. This overhead also happens when the flow is ACK clocked and
cwnd limited instead of being limited by the pacing rate.
This leads to extra overhead (high number of IRQ)
Now tcp_wstamp_ns is reserved for the pacing timer only
(after commit "tcp: do not change tcp_wstamp_ns in tcp_mstamp_refresh"),
we can setup the timer only when a packet is about to be sent,
and if tcp_wstamp_ns is in the future.
This leads to a ~10% performance increase in TCP_RR workloads.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the new EDT model, sch_fq no longer has to special
case TCP pure acks, since their skb->tstamp will allow them
being sent without pacing delay.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit fefa569a9d ("net_sched: sch_fq: account for schedule/timers
drifts") we added a mitigation for scheduling jitter in fq packet scheduler.
This patch does the same in TCP stack, now it is using EDT model.
Note that this mitigation is valid for both external (fq packet scheduler)
or internal TCP pacing.
This uses the same strategy than the above commit, allowing
a time credit of half the packet currently sent.
Consider following case :
An skb is sent, after an idle period of 300 usec.
The air-time (skb->len/pacing_rate) is 500 usec
Instead of setting the pacing timer to now+500 usec,
it will use now+min(500/2, 300) -> now+250usec
This is like having a token bucket with a depth of half
an skb.
Tested:
tc qdisc replace dev eth0 root pfifo_fast
Before
netperf -P0 -H remote -- -q 1000000000 # 8000Mbit
540000 262144 262144 10.00 7710.43
After :
netperf -P0 -H remote -- -q 1000000000 # 8000 Mbit
540000 262144 262144 10.00 7999.75 # Much closer to 8000Mbit target
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sk_pacing_rate has beed introduced as a u32 field in 2013,
effectively limiting per flow pacing to 34Gbit.
We believe it is time to allow TCP to pace high speed flows
on 64bit hosts, as we now can reach 100Gbit on one TCP flow.
This patch adds no cost for 32bit kernels.
The tcpi_pacing_rate and tcpi_max_pacing_rate were already
exported as 64bit, so iproute2/ss command require no changes.
Unfortunately the SO_MAX_PACING_RATE socket option will stay
32bit and we will need to add a new option to let applications
control high pacing rates.
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 1787144 10.246.9.76:49992 10.246.9.77:36741
timer:(on,003ms,0) ino:91863 sk:2 <->
skmem:(r0,rb540000,t66440,tb2363904,f605944,w1822984,o0,bl0,d0)
ts sack bbr wscale:8,8 rto:201 rtt:0.057/0.006 mss:1448
rcvmss:536 advmss:1448
cwnd:138 ssthresh:178 bytes_acked:256699822585 segs_out:177279177
segs_in:3916318 data_segs_out:177279175
bbr:(bw:31276.8Mbps,mrtt:0,pacing_gain:1.25,cwnd_gain:2)
send 28045.5Mbps lastrcv:73333
pacing_rate 38705.0Mbps delivery_rate 22997.6Mbps
busy:73333ms unacked:135 retrans:0/157 rcv_space:14480
notsent:2085120 minrtt:0.013
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In EDT design, I made the mistake of using tcp_wstamp_ns
to store the last tcp_clock_ns() sample and to store the
pacing virtual timer.
This causes major regressions at high speed flows.
Introduce tcp_clock_cache to store last tcp_clock_ns().
This is needed because some arches have slow high-resolution
kernel time service.
tcp_wstamp_ns is only updated when a packet is sent.
Note that we can remove tcp_mstamp in the future since
tcp_mstamp is essentially tcp_clock_cache/1000, so the
apparent socket size increase is temporary.
Fixes: 9799ccb0e9 ("tcp: add tcp_wstamp_ns socket field")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Or Gerlitz