RTT cached in the TCP metrics are valuable for the initial timeout
because SYN RTT usually does not account for serialization delays
on low BW path.
However using it to seed the RTT estimator maybe disruptive because
other components (e.g., pacing) require the smooth RTT to be obtained
from actual connection.
The solution is to use the higher cached RTT to set the first RTO
conservatively like tcp_rtt_estimator(), but avoid seeding the other
RTT estimator variables such as srtt. It is also a good idea to
keep RTO conservative to obtain the first RTT sample, and the
performance is insured by TCP loss probe if SYN RTT is available.
To keep the seeding formula consistent across SYN RTT and cached RTT,
the rttvar is twice the cached RTT instead of cached RTTVAR value. The
reason is because cached variation may be too small (near min RTO)
which defeats the purpose of being conservative on first RTO. However
the metrics still keep the RTT variations as they might be useful for
user applications (through ip).
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
kbuild bot reported following m68k build error :
net/sched/sch_fq.c: In function 'fq_dequeue':
>> net/sched/sch_fq.c:491:2: error: implicit declaration of function
'prefetch' [-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors
While we are fixing this, move this prefetch() call a bit earlier.
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
__GFP_ZERO is an uncommon flag and perhaps is better
not used. static inline dma_zalloc_coherent exists
so convert the uses of dma_alloc_coherent with __GFP_ZERO
to the more common kernel style with zalloc.
Remove memset from the static inline dma_zalloc_coherent
and add just one use of __GFP_ZERO instead.
Trivially reduces the size of the existing uses of
dma_zalloc_coherent.
Realign arguments as appropriate.
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- Uses perfect flow match (not stochastic hash like SFQ/FQ_codel)
- Uses the new_flow/old_flow separation from FQ_codel
- New flows get an initial credit allowing IW10 without added delay.
- Special FIFO queue for high prio packets (no need for PRIO + FQ)
- Uses a hash table of RB trees to locate the flows at enqueue() time
- Smart on demand gc (at enqueue() time, RB tree lookup evicts old
unused flows)
- Dynamic memory allocations.
- Designed to allow millions of concurrent flows per Qdisc.
- Small memory footprint : ~8K per Qdisc, and 104 bytes per flow.
- Single high resolution timer for throttled flows (if any).
- One RB tree to link throttled flows.
- Ability to have a max rate per flow. We might add a socket option
to add per socket limitation.
Attempts have been made to add TCP pacing in TCP stack, but this
seems to add complex code to an already complex stack.
TCP pacing is welcomed for flows having idle times, as the cwnd
permits TCP stack to queue a possibly large number of packets.
This removes the 'slow start after idle' choice, hitting badly
large BDP flows, and applications delivering chunks of data
as video streams.
Nicely spaced packets :
Here interface is 10Gbit, but flow bottleneck is ~20Mbit
cwin is big, yet FQ avoids the typical bursts generated by TCP
(as in netperf TCP_RR -- -r 100000,100000)
15:01:23.545279 IP A > B: . 78193:81089(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.545394 IP B > A: . ack 81089 win 3668 <nop,nop,timestamp 11597985 1115>
15:01:23.546488 IP A > B: . 81089:83985(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.546565 IP B > A: . ack 83985 win 3668 <nop,nop,timestamp 11597986 1115>
15:01:23.547713 IP A > B: . 83985:86881(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.547778 IP B > A: . ack 86881 win 3668 <nop,nop,timestamp 11597987 1115>
15:01:23.548911 IP A > B: . 86881:89777(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.548949 IP B > A: . ack 89777 win 3668 <nop,nop,timestamp 11597988 1115>
15:01:23.550116 IP A > B: . 89777:92673(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.550182 IP B > A: . ack 92673 win 3668 <nop,nop,timestamp 11597989 1115>
15:01:23.551333 IP A > B: . 92673:95569(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.551406 IP B > A: . ack 95569 win 3668 <nop,nop,timestamp 11597991 1115>
15:01:23.552539 IP A > B: . 95569:98465(2896) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.552576 IP B > A: . ack 98465 win 3668 <nop,nop,timestamp 11597992 1115>
15:01:23.553756 IP A > B: . 98465:99913(1448) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.554138 IP A > B: P 99913:100001(88) ack 65248 win 3125 <nop,nop,timestamp 1115 11597805>
15:01:23.554204 IP B > A: . ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.554234 IP B > A: . 65248:68144(2896) ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.555620 IP B > A: . 68144:71040(2896) ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.557005 IP B > A: . 71040:73936(2896) ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.558390 IP B > A: . 73936:76832(2896) ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.559773 IP B > A: . 76832:79728(2896) ack 100001 win 3668 <nop,nop,timestamp 11597993 1115>
15:01:23.561158 IP B > A: . 79728:82624(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.562543 IP B > A: . 82624:85520(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.563928 IP B > A: . 85520:88416(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.565313 IP B > A: . 88416:91312(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.566698 IP B > A: . 91312:94208(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.568083 IP B > A: . 94208:97104(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.569467 IP B > A: . 97104:100000(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.570852 IP B > A: . 100000:102896(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.572237 IP B > A: . 102896:105792(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.573639 IP B > A: . 105792:108688(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.575024 IP B > A: . 108688:111584(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.576408 IP B > A: . 111584:114480(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
15:01:23.577793 IP B > A: . 114480:117376(2896) ack 100001 win 3668 <nop,nop,timestamp 11597994 1115>
TCP timestamps show that most packets from B were queued in the same ms
timeframe (TSval 1159799{3,4}), but FQ managed to send them right
in time to avoid a big burst.
In slow start or steady state, very few packets are throttled [1]
FQ gets a bunch of tunables as :
limit : max number of packets on whole Qdisc (default 10000)
flow_limit : max number of packets per flow (default 100)
quantum : the credit per RR round (default is 2 MTU)
initial_quantum : initial credit for new flows (default is 10 MTU)
maxrate : max per flow rate (default : unlimited)
buckets : number of RB trees (default : 1024) in hash table.
(consumes 8 bytes per bucket)
[no]pacing : disable/enable pacing (default is enable)
All of them can be changed on a live qdisc.
$ tc qd add dev eth0 root fq help
Usage: ... fq [ limit PACKETS ] [ flow_limit PACKETS ]
[ quantum BYTES ] [ initial_quantum BYTES ]
[ maxrate RATE ] [ buckets NUMBER ]
[ [no]pacing ]
$ tc -s -d qd
qdisc fq 8002: dev eth0 root refcnt 32 limit 10000p flow_limit 100p buckets 256 quantum 3028 initial_quantum 15140
Sent 216532416 bytes 148395 pkt (dropped 0, overlimits 0 requeues 14)
backlog 0b 0p requeues 14
511 flows, 511 inactive, 0 throttled
110 gc, 0 highprio, 0 retrans, 1143 throttled, 0 flows_plimit
[1] Except if initial srtt is overestimated, as if using
cached srtt in tcp metrics. We'll provide a fix for this issue.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update documentation to add fanout policies that are available.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of hard-coding reciprocal_divide function, use the inline
function from reciprocal_div.h.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We currently allow for different fanout scheduling policies in pf_packet
such as scheduling by skb's rxhash, round-robin, by cpu, and rollover.
Also allow for a random, equidistributed selection of the socket from the
fanout process group.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's no need to call ether_setup() in the driver since prior alloc_etherdev()
call already arranges for it.
Suggested-by: Denis Kirjanov <kda@linux-powerpc.org>
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Veaceslav Falico says:
====================
bonding: remove vlan special handling
v1: Per Jiri's advice, remove the exported netdev_upper struct to keep it
inside dev.c only, and instead implement a macro to iterate over the
list and return only net_device *.
v2: Jiri noted that we need to see every upper device, but not only the
first level. Modify the netdev_upper logic to include a list of lower
devices and for both upper/lower lists every device would see both its
first-level devices and every other devices that is lower/upper of it.
Also, convert some annoying spamming warnings to pr_debug in
bond_arp_send_all.
v3: move renaming part completely to patch 1 (did I forget to git add
before commiting?) and address Jiri's input about comments/style of
patch 2.
v4: as Vlad found spotted, bond_arp_send_all() won't work in a config where
we have a device with ip on top of our upper vlan. It fails to send
packets because we don't tag the packet, while the device on top of
vlan will emit tagged packets through this vlan. Fix this by first
searching for all upper vlans, and for each vlan - for the devs on top
of it. If we find the dev - then tag the packet with the underling's
vlan_id, otherwise just search the old way - for all devices on top of
bonding. Also, move the version changes under "---" so they won't get
into the commit message, if/when applied.
The aim of this patchset is to remove bondings' own vlan handling as much
as possible and replace it with the netdev upper device functionality.
The upper device functionality is extended to include also lower devices
and to have, for each device, a full view of every lower and upper device,
but not only the first-level ones. This might permit in the future to
avoid, for any grouping/teaming/upper/lower devices, to maintain its own
lists of slaves/vlans/etc.
This is achieved by adding a helper function to upper dev list handling -
netdev_upper_get_next_dev(dev, iter), which returns the next device after
the list_head **iter, and sets *iter to the next list_head *. This patchset
also adds netdev_for_each_upper_dev(dev, upper, iter), which iterates
through the whole dev->upper_dev_list, setting upper to the net_device.
The only special treatment of vlans remains in rlb code.
This patchset solves several issues with bonding, simplifies it overall,
RCUify further and exports upper list functions for any other users which
might also want to get rid of its own vlan_lists or slaves.
I'm testing it continuously currently, no issues found, will update on
anything.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
They're simply annoying and will spam dmesg constantly if we hit them, so
convert to pr_debug so that we still can access them in case of debugging.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently there are no real users of vlan_list/current_alb_vlan, only the
helpers which maintain them, so remove them.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, if there are vlans on top of bond, alb_send_learning_packets()
will never send LPs from the bond itself (i.e. untagged), which might leave
untagged clients unupdated.
Also, the 'circular vlan' logic (i.e. update only MAX_LP_BURST vlans at a
time, and save the last vlan for the next update) is really suboptimal - in
case of lots of vlans it will take a lot of time to update every vlan. It
is also never called in any hot path and sends only a few small packets -
thus the optimization by itself is useless.
So remove the whole current_alb_vlan/MAX_LP_BURST logic from
alb_send_learning_packets(). Instead, we'll first send a packet untagged
and then traverse the upper dev list, sending a tagged packet for each vlan
found. Also, remove the MAX_LP_BURST define - we already don't need it.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Create alb_send_lp_vid(), which will handle the skb/lp creation, vlan
tagging and sending, and use it in alb_send_learning_packets().
This way all the logic remains in alb_send_learning_packets(), which
becomes a lot more cleaner and easier to understand.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We always hold the rtnl_lock() in __bond_release_one(), so use
vlan_uses_dev() instead of bond_vlan_used().
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, bond_has_this_ip() is aware only of vlan upper devices, and thus
will return false if the address is associated with the upper bridge or any
other device, and thus will break the arp logic.
Fix this by using the upper device list. For every upper device we verify
if the address associated with it is our address, and if yes - return true.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, bond_arp_send_all() is aware only of vlans, which breaks
configurations like bond <- bridge (or any other 'upper' device) with IP
(which is quite a common scenario for virt setups).
To fix this we convert the bond_arp_send_all() to first verify if the rt
device is the bond itself, and if not - to go through its list of upper
vlans and their respectiv upper devices (if the vlan's upper device matches
- tag the packet), if still not found - go through all of our upper list
devices to see if any of them match the route device for the target. If the
match is a vlan device - we also save its vlan_id and tag it in
bond_arp_send().
Also, clean the function a bit to be more readable.
CC: Vlad Yasevich <vyasevic@redhat.com>
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Convert bond_vlan_used() to traverse the upper device list to see if we
have any vlans above us. It's protected by rcu, and in case we are holding
rtnl_lock we should call vlan_uses_dev() instead - it's faster.
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new macro netdev_for_each_upper_dev_rcu(dev, upper, iter) iterates
through the dev->upper_dev_list starting from the first element, using
the netdev_upper_get_next_dev_rcu(dev, &iter).
Must be called under RCU read lock.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This function returns the next dev in the dev->upper_dev_list after the
struct list_head **iter position, and updates *iter accordingly. Returns
NULL if there are no devices left.
Caller must hold RCU read lock.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We already don't need it cause we see every upper/lower device in the list
already.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds lower_dev_list list_head to net_device, which is the same
as upper_dev_list, only for lower devices, and begins to use it in the same
way as the upper list.
It also changes the way the whole adjacent device lists work - now they
contain *all* of upper/lower devices, not only the first level. The first
level devices are distinguished by the bool neighbour field in
netdev_adjacent, also added by this patch.
There are cases when a device can be added several times to the adjacent
list, the simplest would be:
/---- eth0.10 ---\
eth0- --- bond0
\---- eth0.20 ---/
where both bond0 and eth0 'see' each other in the adjacent lists two times.
To avoid duplication of netdev_adjacent structures ref_nr is being kept as
the number of times the device was added to the list.
The 'full view' is achieved by adding, on link creation, all of the
upper_dev's upper_dev_list devices as upper devices to all of the
lower_dev's lower_dev_list devices (and to the lower_dev itself), and vice
versa. On unlink they are removed using the same logic.
I've tested it with thousands vlans/bonds/bridges, everything works ok and
no observable lags even on a huge number of interfaces.
Memory footprint for 128 devices interconnected with each other via both
upper and lower (which is impossible, but for the comparison) lists would be:
128*128*2*sizeof(netdev_adjacent) = 1.5MB
but in the real world we usualy have at most several devices with slaves
and a lot of vlans, so the footprint will be much lower.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rename the structure to reflect the upcoming addition of lower_dev_list.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jeff Kirsher says:
====================
This series contains updates to ixgbe.
Jacob provides a fix for 82599 devices where it can potentially keep link
lights up when the adapter has gone down.
Mark provides a fix to resolve the possible use of uninitialized memory
by checking the return value on EEPROM reads.
Don provides 2 patches, one to fix a issue where we were traversing the
Tx ring with the value of IXGBE_NUM_RX_QUEUES which currently happens
to have the correct value but this is misleading. A change later, could
easily make this no longer correct so when traversing the Tx ring, use
netdev->num_tx_queues. His second patch does some minor clean ups of log
messages.
Emil provides the remaining ixgbe patches. First he fixes the link test
where forcing the laser before the link check can lead to inconsistent
results because it does not guarantee that the link will be negotiated
correctly. Then he initializes the message buffer array to 0 in order
to avoid using random numbers from the memory as a MAC address for the
VF. Emil also fixes the read loop for the I2C data to account for the
offset for SFP+ modules. Lastly, Emil provides several patches to add
support for QSFP modules where 1Gbps support is added as well as support
for older QSFP active direct attach cables which pre-date SFF-8436 v3.6.
v2: Fixed patch 4 description and added blank line based on feedback from
Sergei Shtylyov
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of using a custom 'FEC_NAPI_WEIGHT', just use the generic
'NAPI_POLL_WEIGHT' definition instead.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a comment related to RFC4960 explaning why we do not check for initial
TSN, and while at it, remove yoda notation checks and clean up code from
checks of mandatory conditions. That's probably just really minor, but makes
reviewing easier.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After hearing many people over past years complaining against TSO being
bursty or even buggy, we are proud to present automatic sizing of TSO
packets.
One part of the problem is that tcp_tso_should_defer() uses an heuristic
relying on upcoming ACKS instead of a timer, but more generally, having
big TSO packets makes little sense for low rates, as it tends to create
micro bursts on the network, and general consensus is to reduce the
buffering amount.
This patch introduces a per socket sk_pacing_rate, that approximates
the current sending rate, and allows us to size the TSO packets so
that we try to send one packet every ms.
This field could be set by other transports.
Patch has no impact for high speed flows, where having large TSO packets
makes sense to reach line rate.
For other flows, this helps better packet scheduling and ACK clocking.
This patch increases performance of TCP flows in lossy environments.
A new sysctl (tcp_min_tso_segs) is added, to specify the
minimal size of a TSO packet (default being 2).
A follow-up patch will provide a new packet scheduler (FQ), using
sk_pacing_rate as an input to perform optional per flow pacing.
This explains why we chose to set sk_pacing_rate to twice the current
rate, allowing 'slow start' ramp up.
sk_pacing_rate = 2 * cwnd * mss / srtt
v2: Neal Cardwell reported a suspect deferring of last two segments on
initial write of 10 MSS, I had to change tcp_tso_should_defer() to take
into account tp->xmit_size_goal_segs
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Van Jacobson <vanj@google.com>
Cc: Tom Herbert <therbert@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements RFC6980: Drop fragmented ndisc packets by
default. If a fragmented ndisc packet is received the user is informed
that it is possible to disable the check.
Cc: Fernando Gont <fernando@gont.com.ar>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix phy0 address to match the reg property defined in phy0 node.
Signed-off-by: Boris BREZILLON <b.brezillon@overkiz.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace misleading -1 (-EPERM) by a more appropriate return code (-ENXIO)
in macb_mii_probe function.
Save macb_mii_probe return before branching to err_out_unregister to avoid
erronous 0 return.
Signed-off-by: Boris BREZILLON <b.brezillon@overkiz.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some slave devices may have set a dev->needed_headroom value which is
different than the default one, most likely in order to prepend a
hardware descriptor in front of the Ethernet frame to send. Whenever a
new slave is added to a bridge, ensure that we update the
needed_headroom value accordingly to account for the slave
needed_headroom value.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reduce cacheline usage from 2 to 1 cacheline for sctp_globals structure. By
reordering elements, we can close gaps and simply achieve the following:
Current situation:
/* size: 80, cachelines: 2, members: 10 */
/* sum members: 57, holes: 4, sum holes: 16 */
/* padding: 7 */
/* last cacheline: 16 bytes */
Afterwards:
/* size: 64, cachelines: 1, members: 10 */
/* padding: 7 */
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use devm_ioremap_resource instead of of_iomap() and devm_kzalloc()
instead of kmalloc() to make cleanup paths simpler. This patch also
fixes the resource leak caused by missing corresponding iounamp()
of the of_iomap().
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for QSFP active direct attach (DA) cables which
pre-date SFF-8436 v3.6.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch makes sure that QSFP+ modules use the SFP+ code path for
setting up link.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch adds GB speed support for QSFP+ modules.
Autonegotiation is not supported with QSFP+. The user will have to set
the desired speed on both link partners using ethtool advertise setting.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes the read loop for the I2C data to account for the offset.
Also includes a whitespace cleanup and removes ret_val as it is not needed.
CC: Ben Hutchings <bhutchings@solarflare.com>
Reported-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Some minor log messages cleanup, changing the level one message is logged,
adding a bit of detail to another and put all the text on one line.
Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch initializes the msgbuf array to 0 in order to avoid using random
numbers from the memory as MAC address for the VF.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch is a partial reverse of:
commit dfcc4615f0
Author: Jacob Keller <jacob.e.keller@intel.com>
Date: Thu Nov 8 07:07:08 2012 +0000
ixgbe: ethtool ixgbe_diag_test cleanup
Specifically forcing the laser before the link check can lead to
inconsistent results because it does not guarantee that the link will be
negotiated correctly. Such is the case when dual speed SFP+ module is
connected to a gigabit link partner.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
We were transversing the tx_ring with IXGBE_NUM_RX_QUEUES. Now this define
happens to have the correct value but this is misleading and a change later
could easily make this no longer true. I updated it to netdev->num_tx_queues
like we use in ixgbe_get_strings().
Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes the possible use of uninitialized memory by checking the
return value on eeprom reads. These issues were identified by static
analysis. In many cases error messages will be produced so that corrupted
eeprom issues will be more visible.
Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes an issue with the 82599 adapter where it can potentially keep
link lights up when the adapter has gone down. The patch adds a function which
ensures link is disabled, and calls this function when the adapter transitions
to a down state.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Ben Hutchings says:
====================
1. Further cleanup and refactoring in preparation for EF10.
2. Remove ethtool stats that are always zero on Falcon boards.
3. Add an ethtool stat for merged TX completions.
4. Prepare to support merged RX completions.
5. Prepare to support more hwmon sensors.
6. Add support for new events that are generated by EF10 firmware.
7. Update MC reboot detection for EF10.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
- set the protocol field in the skb structure according to the encapsulated
payload
- make the gateway component send a uevent in case of "gw client mode"
de-selection
- increment version number
- minor code rearrangement
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.20 (GNU/Linux)
iQIcBAABCAAGBQJSHcUOAAoJEADl0hg6qKeOvKgQAJXYAreupV+M6lNOSYLpuHcF
kI8+TdhNqLb/zb9zQGEfm+iU847GDTcvQfnVnpKFtfIN7HZWNYl3oYePK5fJx7La
FQBWN3hgi2fliDQ2lzjth9bw6RSiANYGXNvaTHWsgwVr+Wfq1hG3UOulf6k01vNO
9+CGF9u2XHPTH/x33ZKyVshX70kBSsIGKJIkaGRnCed02WOw9PWCoN2Pg+vsRLFJ
jYu2JUO/QaQr0cCYHzfc+TAZorhbfyoHoYuM5CKq8vK+S2lVQQE9PXNq9TH2sVfi
PlFzdfdb+H0M9vmYySj+opddhCw/93Gat+zfJeE+GmqihsN5lcq69VgDHYke6B57
gwOvydB2xXDEHPQ2tTh93YYJTe7lO1Dy4dDidfxlirvpH9iWT+eZSHQbhce7OuIB
b+lv4xkiFCyJvu8TRMCx/W8qG+sgp1UPicSFaodmQDU5k/vL44EwxvFlEMFYdsWK
NZLsVS0+eiuhzkNwFJdaePHEd6ejUf7gTqHmugJsJUXPTwUoYVGuejhoPTNOCJKo
9vkSJ+Xt82XfTOyKKomrN9IisdxYuKv5Q4HzjvFHasjTxADBpYOCxy/JWa0yl9nv
YvyRGz1XoEik7FTP78EcpS2cDAQIW8aLF1LN+wbJiPyhCYUDJMlWxYIm3AA/Jy6v
gjQ7Ztfjb2tsz4D0IUdb
=jA/h
-----END PGP SIGNATURE-----
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
Included changes:
- set the protocol field in the skb structure according to the encapsulated
payload
- make the gateway component send a uevent in case of "gw client mode"
de-selection
- increment version number
- minor code rearrangement
Signed-off-by: David S. Miller <davem@davemloft.net>
This function checks the upper bound but it doesn't check for negative
numbers:
if (txq > QLCNIC_MAX_TX_RINGS) {
I've solved this by making "txq" a u32 type. I chose that because
->tx_count in the ethtool_channels struct is a __u32.
This bug was added in aa4a1f7df7 ('qlcnic: Enable Tx queue changes using
ethtool for 82xx Series adapter.').
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Wei Liu says:
====================
xen-netback: switch to NAPI + kthread 1:1 model
This series implements NAPI + kthread 1:1 model for Xen netback.
This model
- provides better scheduling fairness among vifs
- is prerequisite for implementing multiqueue for Xen network driver
The second patch has the real meat:
- make use of NAPI to mitigate interrupt
- kthreads are not bound to CPUs any more, so that we can take
advantage of backend scheduler and trust it to do the right thing
Benchmark is done on a Dell T3400 workstation with 4 cores, running 4
DomUs. Netserver runs in Dom0. DomUs do netperf to Dom0 with
following command: /root/netperf -H Dom0 -fm -l120
IRQs are distributed to 4 cores by hand in the new model, while in the
old model vifs are automatically distributed to 4 kthreads.
* New model
%Cpu0 : 0.5 us, 20.3 sy, 0.0 ni, 28.9 id, 0.0 wa, 0.0 hi, 24.4 si, 25.9 st
%Cpu1 : 0.5 us, 17.8 sy, 0.0 ni, 28.8 id, 0.0 wa, 0.0 hi, 27.7 si, 25.1 st
%Cpu2 : 0.5 us, 18.8 sy, 0.0 ni, 30.7 id, 0.0 wa, 0.0 hi, 22.9 si, 27.1 st
%Cpu3 : 0.0 us, 20.1 sy, 0.0 ni, 30.4 id, 0.0 wa, 0.0 hi, 22.7 si, 26.8 st
Throughputs: 2027.89 2025.95 2018.57 2016.23 aggregated: 8088.64
* Old model
%Cpu0 : 0.5 us, 68.8 sy, 0.0 ni, 16.1 id, 0.5 wa, 0.0 hi, 2.8 si, 11.5 st
%Cpu1 : 0.4 us, 45.1 sy, 0.0 ni, 31.1 id, 0.4 wa, 0.0 hi, 2.1 si, 20.9 st
%Cpu2 : 0.9 us, 44.8 sy, 0.0 ni, 30.9 id, 0.0 wa, 0.0 hi, 1.3 si, 22.2 st
%Cpu3 : 0.8 us, 46.4 sy, 0.0 ni, 28.3 id, 1.3 wa, 0.0 hi, 2.1 si, 21.1 st
Throughputs: 1899.14 2280.43 1963.33 1893.47 aggregated: 8036.37
We can see that the impact is mainly on CPU usage. The new model moves
processing from kthread to NAPI (software interrupt).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
As we move to 1:1 model and melt xen_netbk and xenvif together, it would
be better to use single prefix for all functions in xen-netback.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements 1:1 model netback. NAPI and kthread are utilized
to do the weight-lifting job:
- NAPI is used for guest side TX (host side RX)
- kthread is used for guest side RX (host side TX)
Xenvif and xen_netbk are made into one structure to reduce code size.
This model provides better scheduling fairness among vifs. It is also
prerequisite for implementing multiqueue for Xen netback.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The data flow from DomU to DomU on the same host in current copying
scheme with tracking facility:
copy
DomU --------> Dom0 DomU
| ^
|____________________________|
copy
The page in Dom0 is a page with valid MFN. So we can always copy from
page Dom0, thus removing the need for a tracking facility.
copy copy
DomU --------> Dom0 -------> DomU
Simple iperf test shows no performance regression (obviously we copy
twice either way):
W/ tracking: ~5.3Gb/s
W/o tracking: ~5.4Gb/s
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Matt Wilson <msw@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Whenever the GW client mode is deselected, a DEL event has
to be sent in order to tell userspace that the current
gateway has been lost. Send the uevent on state change only
if a gateway was currently selected.
Reported-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Yuchung Cheng