| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Upon a new UMR post, check if the WQE buffer contains
a previous UMR WQE. If so, modify the dynamic fields
instead of a whole WQE overwrite. This saves a memcpy.
In current setting, after 2 WQ cycles (12 UMR posts),
this will always be the case.
No degradation sensed.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
All UMR WQEs of an RQ share many common fields. We use
pre-initialized structures to save calculations in datapath.
One field (xlt_offset) was the only reason we saved a pre-initialized
copy per WQE index.
Here we remove its initialization (move its calculation to datapath),
and reduce the number of copies to one-per-RQ.
A very small datapath calculation is added, it occurs once per a MPWQE
(i.e. once every 256KB), but reduces memory consumption and gives
better cache utilization.
Performance testing:
Tested packet rate, no degradation sensed.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
When many packets reside on the same page, the bulking of
page_ref modifications reduces the total number of atomic
operations executed.
Besides the necessary 2 operations on page alloc/free, we
have the following extra ops per page:
- one on WQE allocation (bump refcnt to maximum possible),
- zero ops for SKBs,
- one on WQE free,
a constant of two operations in total, no matter how many
packets/SKBs actually populate the page.
Without this bulking, we have:
- no ops on WQE allocation or free,
- one op per SKB,
Comparing the two methods when PAGE_SIZE is 4K:
- As mentioned above, bulking method always executes 2 operations,
not more, but not less.
- In the default MTU configuration (1500, stride size is 2K),
the non-bulking method execute 2 ops as well.
- For larger MTUs with stride size of 4K, non-bulking method
executes only a single op.
- For XDP (stride size of 4K, no SKBs), non-bulking method
executes no ops at all!
Hence, to optimize the flows with linear SKB and XDP over Striding RQ,
we here remove the page_ref bulking method.
Performance testing:
ConnectX-5, Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz.
Single core packet rate (64 bytes).
Early drop in TC: no degradation.
XDP_DROP:
before: 14,270,188 pps
after: 20,503,603 pps, 43% improvement.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Add XDP support over Striding RQ.
Now that linear SKB is supported over Striding RQ,
we can support XDP by setting stride size to PAGE_SIZE
and headroom to XDP_PACKET_HEADROOM.
Upon a MPWQE free, do not release pages that are being
XDP xmit, they will be released upon completions.
Striding RQ is capable of a higher packet-rate than
conventional RQ.
A performance gain is expected for all cases that had
a HW packet-rate bottleneck. This is the case whenever
using many flows that distribute to many cores.
Performance testing:
ConnectX-5, 24 rings, default MTU.
CQE compression ON (to reduce completions BW in PCI).
XDP_DROP packet rate:
--------------------------------------------------
| pkt size | XDP rate | 100GbE linerate | pct% |
--------------------------------------------------
| 64byte | 126.2 Mpps | 148.0 Mpps | 85% |
| 128byte | 80.0 Mpps | 84.8 Mpps | 94% |
| 256byte | 42.7 Mpps | 42.7 Mpps | 100% |
| 512byte | 23.4 Mpps | 23.4 Mpps | 100% |
--------------------------------------------------
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Make the xdp_xmit indication available for Striding RQ
by taking it out of the type-specific union.
This refactor is a preparation for a downstream patch that
adds XDP support over Striding RQ.
In addition, use a bitmap instead of a boolean for possible
future flags.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Current Striding RQ HW feature utilizes the RX buffers so that
there is no wasted room between the strides. This maximises
the memory utilization.
This prevents the use of build_skb() (which requires headroom
and tailroom), and demands to memcpy the packets headers into
the skb linear part.
In this patch, whenever a set of conditions holds, we apply
an RQ configuration that allows combining the use of linear SKB
on top of a Striding RQ.
To use build_skb() with Striding RQ, the following must hold:
1. packet does not cross a page boundary.
2. there is enough headroom and tailroom surrounding the packet.
We can satisfy 1 and 2 by configuring:
stride size = MTU + headroom + tailoom.
This is possible only when:
a. (MTU - headroom - tailoom) does not exceed PAGE_SIZE.
b. HW LRO is turned off.
Using linear SKB has many advantages:
- Saves a memcpy of the headers.
- No page-boundary checks in datapath.
- No filler CQEs.
- Significantly smaller CQ.
- SKB data continuously resides in linear part, and not split to
small amount (linear part) and large amount (fragment).
This saves datapath cycles in driver and improves utilization
of SKB fragments in GRO.
- The fragments of a resulting GRO SKB follow the IP forwarding
assumption of equal-size fragments.
Some implementation details:
HW writes the packets to the beginning of a stride,
i.e. does not keep headroom. To overcome this we make sure we can
extend backwards and use the last bytes of stride i-1.
Extra care is needed for stride 0 as it has no preceding stride.
We make sure headroom bytes are available by shifting the buffer
pointer passed to HW by headroom bytes.
This configuration now becomes default, whenever capable.
Of course, this implies turning LRO off.
Performance testing:
ConnectX-5, single core, single RX ring, default MTU.
UDP packet rate, early drop in TC layer:
--------------------------------------------
| pkt size | before | after | ratio |
--------------------------------------------
| 1500byte | 4.65 Mpps | 5.96 Mpps | 1.28x |
| 500byte | 5.23 Mpps | 5.97 Mpps | 1.14x |
| 64byte | 5.94 Mpps | 5.96 Mpps | 1.00x |
--------------------------------------------
TCP streams: ~20% gain
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
When modifying the page mapping of a HW memory region
(via a UMR post), post the new values inlined in WQE,
instead of using a data pointer.
This is a micro-optimization, inline UMR WQEs of different
rings scale better in HW.
In addition, this obsoletes a few control flows and helps
delete ~50 LOC.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Do not busy-wait a pending UMR completion. Under high HW load,
busy-waiting a delayed completion would fully utilize the CPU core
and mistakenly indicate a SW bottleneck.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Gets the process of a UMR WQE post in one function,
in preparation for a downstream patch that inlines
the WQE data.
No functional change here.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
In Striding RQ, each WQE serves multiple packets
(hence called Multi-Packet WQE, MPWQE).
The size of a MPWQE is constant (currently 256KB).
Upon a ringparam set operation, we calculate the number of
MPWQEs per RQ. For this, first it is needed to determine the
number of packets that can reside within a single MPWQE.
In this patch we use the actual MTU size instead of ETH_DATA_LEN
for this calculation.
This implies that a change in MTU might require a change
in Striding RQ ring size.
In addition, this obsoletes some WQEs-to-packets translation
functions and helps delete ~60 LOC.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Knowing the MTU is required for RQ creation flow.
By our design, channels creation flow is totally isolated
from priv/netdev, and can be completed with access to
channels params and mdev.
Adding the MTU to the channels params helps preserving that.
In addition, we save it in RQ to make its access faster in
datapath checks.
Signed-off-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Fix spelling mistake in debug message text.
"dettaching" -> "detaching"
Signed-off-by: Talat Batheesh <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
With ConnectX-4, we expect the force teardown to fail in case that
DC was enabled, therefore change the message from error to warning.
Signed-off-by: Alaa Hleihel <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
1. This function is not used anywhere in mlx5 driver
2. It has a memcpy statement that makes no sense and produces build
warning with gcc8
drivers/net/ethernet/mellanox/mlx5/core/transobj.c: In function 'mlx5_core_query_xsrq':
drivers/net/ethernet/mellanox/mlx5/core/transobj.c:347:3: error: 'memcpy' source argument is the same as destination [-Werror=restrict]
Fixes: 01949d0109ee ("net/mlx5_core: Enable XRCs and SRQs when using ISSI > 0")
Reported-by: Arnd Bergmann <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
Instead of looking for the EQ of the CQ, remove that redundant code and
use the eq pointer stored in the cq struct.
Signed-off-by: Saeed Mahameed <[email protected]>
|
|
When using the -i feature to generate random ID numbers for test
cases in tdc, the function that writes the JSON to file doesn't
add a newline character to the end of the file, so we have to
add our own.
Signed-off-by: Lucas Bates <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
This commit has fix for RX traffic issues when we stress test the driver
with continuous ifconfig up/down under very high traffic conditions.
Reason for the issue is that, in existing liquidio_stop function NAPI is
disabled even before actual FW/HW interface is brought down via
send_rx_ctrl_cmd(lio, 0). Between time frame of NAPI disable and actual
interface down in firmware, firmware continuously enqueues rx traffic to
host. When interrupt happens for new packets, host irq handler fails in
scheduling NAPI as the NAPI is already disabled.
After "ifconfig <iface> up", Host re-enables NAPI but cannot schedule it
until it receives another Rx interrupt. Host never receives Rx interrupt as
it never cleared the Rx interrupt it received during interface down
operation. NIC Rx interrupt gets cleared only when Host processes queue and
clears the queue counts. Above anomaly leads to other issues like packet
overflow in FW/HW queues, backpressure.
Fix:
This commit fixes this issue by disabling NAPI only after informing
firmware to stop queueing packets to host via send_rx_ctrl_cmd(lio, 0).
send_rx_ctrl_cmd is not visible in the patch as it is already there in the
code. The DOWN command also waits for any pending packets to be processed
by NAPI so that the deadlock will not occur.
Signed-off-by: Raghu Vatsavayi <[email protected]>
Acked-by: Derek Chickles <[email protected]>
Signed-off-by: Felix Manlunas <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/sschmidt/wpan-next
Stefan Schmidt says:
====================
pull-request: ieee802154-next 2018-03-29
An update from ieee802154 for *net-next*
Colin fixed a unused variable in the new mcr20a driver.
Harry fixed an unitialised data read in the debugfs interface of the
ca8210 driver.
If there are any issues or you think these are to late for -rc1 (both can also
go into -rc2 as they are simple fixes) let me know.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
Signed-off-by: Roman Mashak <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The freescale.com address will no longer be available.
Signed-off-by: Claudiu Manoil <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Add a new compatible string for the RZ/G1C (R8A77470) SoC.
Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Fabrizio Castro <[email protected]>
Acked-by: Sergei Shtylyov <[email protected]>
Reviewed-by: Geert Uytterhoeven <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Jose Abreu says:
====================
Fix TX Timeout and implement Safety Features
Fix the TX Timeout handler to correctly reconfigure the whole system and
start implementing features for DWMAC5 cores, specifically the Safety
Features.
Changes since v1:
- Display error stats in ethtool
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
This adds initial suport for DWMAC5 and implements the Automotive Safety
Package which is available from core version 5.10.
The Automotive Safety Pacakge (also called Safety Features) offers us
with error protection in the core by implementing ECC Protection in
memories, on-chip data path parity protection, FSM parity and timeout
protection and Application/CSR interface timeout protection.
In case of an uncorrectable error we call stmmac_global_err() and
reconfigure the whole core.
Signed-off-by: Jose Abreu <[email protected]>
Cc: David S. Miller <[email protected]>
Cc: Joao Pinto <[email protected]>
Cc: Giuseppe Cavallaro <[email protected]>
Cc: Alexandre Torgue <[email protected]>
Cc: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Currently TX Timeout handler does not behaves as expected and leads to
an unrecoverable state. Rework current implementation of TX Timeout
handling to actually perform a complete reset of the driver state and IP.
We use deferred work to init a task which will be responsible for
resetting the system.
Signed-off-by: Jose Abreu <[email protected]>
Cc: David S. Miller <[email protected]>
Cc: Joao Pinto <[email protected]>
Cc: Giuseppe Cavallaro <[email protected]>
Cc: Alexandre Torgue <[email protected]>
Cc: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The style of the rx/tx queue's *_coal member assignment is:
static void foo_coal_set(...)
{
set the coal in hw;
update queue's foo_coal member; [1]
}
In other place, we call foo_coal_set(pp, queue->foo_coal), so the above [1]
is duplicated and could be removed.
Signed-off-by: Jisheng Zhang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Lorenzo Bianconi says:
====================
do not allow adding routes if disable_ipv6 is enabled
Do not allow userspace to add static ipv6 routes if disable_ipv6 is enabled.
Update disable_ipv6 documentation according to that change
Changes since v1:
- added an extack message telling the user that IPv6 is disabled on the nexthop
device
- rebased on-top of net-next
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
Clarify that when disable_ipv6 is enabled even the ipv6 routes
are deleted for the selected interface and from now it will not
be possible to add addresses/routes to that interface
Signed-off-by: Lorenzo Bianconi <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Do not allow setting ipv6 routes from userspace if disable_ipv6 has been
enabled. The issue can be triggered using the following reproducer:
- sysctl net.ipv6.conf.all.disable_ipv6=1
- ip -6 route add a:b:c:d::/64 dev em1
- ip -6 route show
a:b:c:d::/64 dev em1 metric 1024 pref medium
Fix it checking disable_ipv6 value in ip6_route_info_create routine
Signed-off-by: Lorenzo Bianconi <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Pablo Neira Ayuso says:
====================
Netfilter/IPVS updates for net-next
The following patchset contains Netfilter/IPVS updates for your net-next
tree. This batch comes with more input sanitization for xtables to
address bug reports from fuzzers, preparation works to the flowtable
infrastructure and assorted updates. In no particular order, they are:
1) Make sure userspace provides a valid standard target verdict, from
Florian Westphal.
2) Sanitize error target size, also from Florian.
3) Validate that last rule in basechain matches underflow/policy since
userspace assumes this when decoding the ruleset blob that comes
from the kernel, from Florian.
4) Consolidate hook entry checks through xt_check_table_hooks(),
patch from Florian.
5) Cap ruleset allocations at 512 mbytes, 134217728 rules and reject
very large compat offset arrays, so we have a reasonable upper limit
and fuzzers don't exercise the oom-killer. Patches from Florian.
6) Several WARN_ON checks on xtables mutex helper, from Florian.
7) xt_rateest now has a hashtable per net, from Cong Wang.
8) Consolidate counter allocation in xt_counters_alloc(), from Florian.
9) Earlier xt_table_unlock() call in {ip,ip6,arp,eb}tables, patch
from Xin Long.
10) Set FLOW_OFFLOAD_DIR_* to IP_CT_DIR_* definitions, patch from
Felix Fietkau.
11) Consolidate code through flow_offload_fill_dir(), also from Felix.
12) Inline ip6_dst_mtu_forward() just like ip_dst_mtu_maybe_forward()
to remove a dependency with flowtable and ipv6.ko, from Felix.
13) Cache mtu size in flow_offload_tuple object, this is safe for
forwarding as f87c10a8aa1e describes, from Felix.
14) Rename nf_flow_table.c to nf_flow_table_core.o, to simplify too
modular infrastructure, from Felix.
15) Add rt0, rt2 and rt4 IPv6 routing extension support, patch from
Ahmed Abdelsalam.
16) Remove unused parameter in nf_conncount_count(), from Yi-Hung Wei.
17) Support for counting only to nf_conncount infrastructure, patch
from Yi-Hung Wei.
18) Add strict NFT_CT_{SRC_IP,DST_IP,SRC_IP6,DST_IP6} key datatypes
to nft_ct.
19) Use boolean as return value from ipt_ah and from IPVS too, patch
from Gustavo A. R. Silva.
20) Remove useless parameters in nfnl_acct_overquota() and
nf_conntrack_broadcast_help(), from Taehee Yoo.
21) Use ipv6_addr_is_multicast() from xt_cluster, also from Taehee Yoo.
22) Statify nf_tables_obj_lookup_byhandle, patch from Fengguang Wu.
23) Fix typo in xt_limit, from Geert Uytterhoeven.
24) Do no use VLAs in Netfilter code, again from Gustavo.
25) Use ADD_COUNTER from ebtables, from Taehee Yoo.
26) Bitshift support for CONNMARK and MARK targets, from Jack Ma.
27) Use pr_*() and add pr_fmt(), from Arushi Singhal.
28) Add synproxy support to ctnetlink.
29) ICMP type and IGMP matching support for ebtables, patches from
Matthias Schiffer.
30) Support for the revision infrastructure to ebtables, from
Bernie Harris.
31) String match support for ebtables, also from Bernie.
32) Documentation for the new flowtable infrastructure.
33) Use generic comparison functions in ebt_stp, from Joe Perches.
34) Demodularize filter chains in nftables.
35) Register conntrack hooks in case nftables NAT chain is added.
36) Merge assignments with return in a couple of spots in the
Netfilter codebase, also from Arushi.
37) Document that xtables percpu counters are stored in the same
memory area, from Ben Hutchings.
38) Revert mark_source_chains() sanity checks that break existing
rulesets, from Florian Westphal.
39) Use is_zero_ether_addr() in the ipset codebase, from Joe Perches.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
'Close-race-between-un-register_netdevice_notifier-and-pernet_operations'
Kirill Tkhai says:
====================
Close race between {un, }register_netdevice_notifier and pernet_operations
the problem is {,un}register_netdevice_notifier() do not take
pernet_ops_rwsem, and they don't see network namespaces, being
initialized in setup_net() and cleanup_net(), since at this
time net is not hashed to net_namespace_list.
This may lead to imbalance, when a notifier is called at time of
setup_net()/net is alive, but it's not called at time of cleanup_net(),
for the devices, hashed to the net, and vise versa. See (3/3) for
the scheme of imbalance.
This patchset fixes the problem by acquiring pernet_ops_rwsem
at the time of {,un}register_netdevice_notifier() (3/3).
(1-2/3) are preparations in xfrm and netfilter subsystems.
The problem was introduced a long ago, but backporting won't be easy,
since every previous kernel version may have changes in netdevice
notifiers, and they all need review and testing. Otherwise, there
may be more pernet_operations, which register or unregister
netdevice notifiers, and that leads to deadlock (which is was fixed
in 1-2/3). This patchset is for net-next.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
setup_net()/cleanup_net()
{un,}register_netdevice_notifier() iterate over all net namespaces
hashed to net_namespace_list. But pernet_operations register and
unregister netdevices in unhashed net namespace, and they are not
seen for netdevice notifiers. This results in asymmetry:
1)Race with register_netdevice_notifier()
pernet_operations::init(net) ...
register_netdevice() ...
call_netdevice_notifiers() ...
... nb is not called ...
... register_netdevice_notifier(nb) -> net skipped
... ...
list_add_tail(&net->list, ..) ...
Then, userspace stops using net, and it's destructed:
pernet_operations::exit(net)
unregister_netdevice()
call_netdevice_notifiers()
... nb is called ...
This always happens with net::loopback_dev, but it may be not the only device.
2)Race with unregister_netdevice_notifier()
pernet_operations::init(net)
register_netdevice()
call_netdevice_notifiers()
... nb is called ...
Then, userspace stops using net, and it's destructed:
list_del_rcu(&net->list) ...
pernet_operations::exit(net) unregister_netdevice_notifier(nb) -> net skipped
dev_change_net_namespace() ...
call_netdevice_notifiers()
... nb is not called ...
unregister_netdevice()
call_netdevice_notifiers()
... nb is not called ...
This race is more danger, since dev_change_net_namespace() moves real
network devices, which use not trivial netdevice notifiers, and if this
will happen, the system will be left in unpredictable state.
The patch closes the race. During the testing I found two places,
where register_netdevice_notifier() is called from pernet init/exit
methods (which led to deadlock) and fixed them (see previous patches).
The review moved me to one more unusual registration place:
raw_init() (can driver). It may be a reason of problems,
if someone creates in-kernel CAN_RAW sockets, since they
will be destroyed in exit method and raw_release()
will call unregister_netdevice_notifier(). But grep over
kernel tree does not show, someone creates such sockets
from kernel space.
Theoretically, there can be more places like this, and which are
hidden from review, but we found them on the first bumping there
(since there is no a race, it will be 100% reproducible).
Signed-off-by: Kirill Tkhai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Register netdevice notifier for every iptable entry
is not good, since this breaks modularity, and
the hidden synchronization is based on rtnl_lock().
This patch reworks the synchronization via new lock,
while the rest of logic remains as it was before.
This is required for the next patch.
Tested via:
while :; do
unshare -n iptables -t mangle -A OUTPUT -j TEE --gateway 1.1.1.2 --oif lo;
done
Signed-off-by: Kirill Tkhai <[email protected]>
Acked-by: Pablo Neira Ayuso <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Currently, driver registers it from pernet_operations::init method,
and this breaks modularity, because initialization of net namespace
and netdevice notifiers are orthogonal actions. We don't have
per-namespace netdevice notifiers; all of them are global for all
devices in all namespaces.
Signed-off-by: Kirill Tkhai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Mike Looijmans says:
====================
of_net: Implement of_get_nvmem_mac_address helper
Posted this as a small set now, with an (optional) second patch that shows
how the changes work and what I've used to test the code on a Topic Miami board.
I've taken the liberty to add appropriate "Acked" and "Review" tags.
v4: Replaced "6" with ETH_ALEN
v3: Add patch that implements mac in nvmem for the Cadence MACB controller
Remove the integrated of_get_mac_address call
v2: Use of_nvmem_cell_get to avoid needing the assiciated device
Use void* instead of char*
Add devicetree binding doc
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
Call of_get_nvmem_mac_address() to fetch the MAC address from an nvmem
cell, if one is provided in the device tree. This allows the address to
be stored in an I2C EEPROM device for example.
Signed-off-by: Mike Looijmans <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
It's common practice to store MAC addresses for network interfaces into
nvmem devices. However the code to actually do this in the kernel lacks,
so this patch adds of_get_nvmem_mac_address() for drivers to obtain the
address from an nvmem cell provider.
This is particulary useful on devices where the ethernet interface cannot
be configured by the bootloader, for example because it's in an FPGA.
Signed-off-by: Mike Looijmans <[email protected]>
Reviewed-by: Florian Fainelli <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Jakub Kicinski says:
====================
nfp: flower: handle MTU changes
This set improves MTU handling for flower offload. The max MTU is
correctly capped and physical port MTU is communicated to the FW
(and indirectly HW).
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
Trigger a port mod message to request an MTU change on the NIC when any
physical port representor is assigned a new MTU value. The driver waits
10 msec for an ack that the FW has set the MTU. If no ack is received the
request is rejected and an appropriate warning flagged.
Rather than maintain an MTU queue per repr, one is maintained per app.
Because the MTU ndo is protected by the rtnl lock, there can never be
contention here. Portmod messages from the NIC are also protected by
rtnl so we first check if the portmod is an ack and, if so, handle outside
rtnl and the cmsg work queue.
Acks are detected by the marking of a bit in a portmod response. They are
then verfied by checking the port number and MTU value expected by the
app. If the expected MTU is 0 then no acks are currently expected.
Also, ensure that the packet headroom reserved by the flower firmware is
considered when accepting an MTU change on any repr.
Signed-off-by: John Hurley <[email protected]>
Reviewed-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Rename the 'change_mtu' app callback to 'check_mtu'. This is called
whenever an MTU change is requested on a netdev. It can reject the
change but is not responsible for implementing it.
Introduce a new 'repr_change_mtu' app callback that is hit when the MTU
of a repr is to be changed. This is responsible for performing the MTU
change and verifying it.
Signed-off-by: John Hurley <[email protected]>
Reviewed-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Florian Fainelli says:
====================
phylink: API changes
This patch series contains two API changes to PHYLINK which will later be used
by DSA to migrate to PHYLINK. Because these are API changes that impact other
outstanding work (e.g: MVPP2) I would rather get them included sooner to minimize
conflicts.
Thank you!
Changes in v2:
- added missing documentation to mac_link_{up,down} that the interface
must be configured in mac_config()
- added Russell's, Andrew's and my tags
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
Provide a pointer to the SFP bus in struct net_device, so that the
ethtool module EEPROM methods can access the SFP directly, rather
than needing every user to provide a hook for it.
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: Russell King <[email protected]>
Signed-off-by: Florian Fainelli <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: Russell King <[email protected]>
Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
In preparation for having DSA transition entirely to PHYLINK, we need to pass a
PHY interface type to the mac_link_{up,down} callbacks because we may have to
make decisions on that (e.g: turn on/off RGMII interfaces etc.). We do not pass
an entire phylink_link_state because not all parameters (pause, duplex etc.) are
defined when the link is down, only link and interface are.
Update mvneta accordingly since it currently implements phylink_mac_ops.
Acked-by: Russell King <[email protected]>
Signed-off-by: Florian Fainelli <[email protected]>
Acked-by: Russell King <[email protected]>
Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Shrikrishna Khare would no longer maintain the vmxnet3 driver. Taking
over the role of vmxnet3 maintainer.
Signed-off-by: Ronak Doshi <[email protected]>
Signed-off-by: Shrikrishna Khare <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Florian Fainelli says:
====================
net: Broadcom drivers coalescing fixes
Following Tal's review of the adaptive RX/TX coalescing feature added to the
SYSTEMPORT and GENET driver a number of things showed up:
- adaptive TX coalescing is not actually a good idea with the current way
the estimator will program the ring, this results in a higher CPU load, NAPI
on TX already does a reasonably good job at maintaining the interrupt count low
- both SYSTEMPORT and GENET would suffer from the same issues while configuring
coalescing parameters where the values would just not be applied correctly
based on user settings, so we fix that too
Tal, thanks again for your feedback, I would appreciate if you could review that
the new behavior appears to be implemented correctly.
Thanks!
Changes in v2:
- added Tal's reviewed-by to the first patch
- split DIM initialization from coalescing parameters initialization
- avoid duplicating the same code in bcmgenet_set_coalesce() when configuring RX rings
- fixed the condition where default DIM parameters would be applied when
adaptive RX coalescing would be enabled, do this only if it was disabled before
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
There were a number of issues with setting the RX coalescing parameters:
- we would not be preserving values that would have been configured
across close/open calls, instead we would always reset to no timeout
and 1 interrupt per packet, this would also prevent DIM from setting its
default usec/pkts values
- when adaptive RX would be turned on, we woud not be fetching the
default parameters, we would stay with no timeout/1 packet per interrupt
until the estimator kicks in and changes that
- finally disabling adaptive RX coalescing while providing parameters
would not be honored, and we would stay with whatever DIM had previously
determined instead of the user requested parameters
Fixes: 9f4ca05827a2 ("net: bcmgenet: Add support for adaptive RX coalescing")
Signed-off-by: Florian Fainelli <[email protected]>
Reviewed-by: Tal Gilboa <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
There were a number of issues with setting the RX coalescing parameters:
- we would not be preserving values that would have been configured
across close/open calls, instead we would always reset to no timeout
and 1 interrupt per packet, this would also prevent DIM from setting its
default usec/pkts values
- when adaptive RX would be turned on, we woud not be fetching the
default parameters, we would stay with no timeout/1 packet per
interrupt until the estimator kicks in and changes that
- finally disabling adaptive RX coalescing while providing parameters
would not be honored, and we would stay with whatever DIM had
previously determined instead of the user requested parameters
Fixes: b6e0e875421e ("net: systemport: Implement adaptive interrupt coalescing")
Signed-off-by: Florian Fainelli <[email protected]>
Reviewed-by: Tal Gilboa <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Adaptive TX coalescing is not currently giving us any advantages and
ends up making the CPU spin more frequently until TX completion. Deny
and disable adaptive TX coalescing for now and rely on static
configuration, we can always add it back later.
Reviewed-by: Tal Gilboa <[email protected]>
Signed-off-by: Florian Fainelli <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
NETIF_F_HW_VLAN_[CS]TAG_FILTER features require more than just a bit
flip in dev->features in order to keep the driver in a consistent state.
These features notify the driver of each added/removed vlan, but toggling
of vlan-filter does not notify the driver accordingly for each of the
existing vlans.
This patch implements a similar solution to NETIF_F_RX_UDP_TUNNEL_PORT
behavior (which notifies the driver about UDP ports in the same manner
that vids are reported).
Each toggling of the features propagates to the 8021q module, which
iterates over the vlans and call add/kill ndo accordingly.
Signed-off-by: Gal Pressman <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Fix to return a negative error code from the hash filter init error
handling case instead of 0, as done elsewhere in this function.
Fixes: 5c31254e35a8 ("cxgb4: initialize hash-filter configuration")
Signed-off-by: Wei Yongjun <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
To make the test a bit clearer and to reduce object size a little.
Miscellanea:
o remove now unnecessary static const array
$ size ip_set_hash_mac.o*
text data bss dec hex filename
22822 4619 64 27505 6b71 ip_set_hash_mac.o.allyesconfig.new
22932 4683 64 27679 6c1f ip_set_hash_mac.o.allyesconfig.old
10443 1040 0 11483 2cdb ip_set_hash_mac.o.defconfig.new
10507 1040 0 11547 2d1b ip_set_hash_mac.o.defconfig.old
Signed-off-by: Joe Perches <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>
|