With this commit it is possible to add priority to sent or received
network packets. So user is able to send or receive higher priority
packets faster than lower level packets.
The traffic class support is activated by CONFIG_NET_TC_COUNT option.
The TC support uses work queues to separate the traffic. The
priority of the work queue thread specifies the ordering of the
network traffic. Each work queue thread handles traffic to one specific
work queue. Note that you should not enable traffic classes unless
you really need them by your application. Each TC thread needs
stack so this feature requires more memory.
It is possible to disable transmit traffic class support and keep the
receive traffic class support, or vice versa. If both RX and TX traffic
classes are enabled, then both will use the same number of queues
defined by CONFIG_NET_TC_COUNT option.
Fixes#6588
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Instead of always allocating both IPv6 and IPv4 address information
to every network interface, allow more fine grained address
configuration. So it is possible to have IPv6 or IPv4 only network
interfaces.
This commit introduces two new config options:
CONFIG_NET_IF_MAX_IPV4_COUNT and CONFIG_NET_IF_MAX_IPV6_COUNT
which tell how many IP address information structs are allocated
statically. At runtime when network interface is setup, it is then
possible to attach this IP address info struct to a specific
network interface. This can save considerable amount of memory
as the IP address information struct can be quite large (depends
on how many IP addresses user configures in the system).
Note that the value of CONFIG_NET_IF_MAX_IPV4_COUNT and
CONFIG_NET_IF_MAX_IPV6_COUNT should reflect the estimated number of
network interfaces in the system. So if if CONFIG_NET_IF_MAX_IPV6_COUNT
is set to 1 and there are two network interfaces that need IPv6
addresses, then the system will not be able to setup IPv6 addresses to
the second network interface in this case. This scenario might be
just fine if the second network interface is IPv4 only. The net_if.c
will print a warning during startup if mismatch about the counts and
the actual number of network interface is detected.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Move IP address settings from net_if to separate structs.
This is needed for VLAN support.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Create infrastructure that allows ethernet device driver to tell
if it supports network packet checksum offloading. This applies only
to IPv4, UDP or TCP checksums. The driver can enable/disable checksum
offloading separately for Tx and Rx network packets.
If the device (ethernet in this case) can calculate the network
packet checksum for IPv4, UDP or TCP, then do not calculate the
corresponding checksum by the stack itself.
Fixes#2987
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This introduces net_if_carrier_down so the L2 driver can inform when it
has lost connectivity so all packets shall be flushed and the interface
should be put down.
Fixes#5317
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Default network interface is the first interface if there are multiple
interfaces exist. But this creates an issue when different parts of
the network subsystem wants to choose particular default interface.
At-least with this Kconfig option user can select default network
interface.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
When CONFIG_NET_RPL enabled, do not run DAD (duplicate address
detection) for global addresses. In RPL mesh network global addresses
for nodes are determined by prefix from Border Router. DAD is not
necessary in this case.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
The meaning of this address type is the same as NET_ADDR_MANUAL,
but with a provision that DHCP can override such an address.
It's intended for the usecase when there's a default static
configuration for when DHCP is not available, but DHCP should
override it.
Before going to add another address type, there was an attempt
to repurpose TENTATIVE address state, but it doesn't work as
expected, as indeed, all existing address types/states already
have clearly semantics, and it makes sense to just another
address type to avoid confusion and unexpected behavior.
Fixes: #5696
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
The offset of the IP header in a received packet depends on the L2
header size. For Ethernet this is 14 bytes which puts the u32 IPv4
addresses on a non-u32 byte boundary. This causes chips that don't
support unaligned access (like the Cortex-M0) to fault.
The fixes in this patch are enough to ping the board and run the
http_server sample.
Signed-off-by: Michael Hope <mlhx@google.com>
https://tools.ietf.org/html/rfc4862#section-5.4.2 :
"""
Before sending a Neighbor Solicitation, an interface MUST join the
all-nodes multicast address and the solicited-node multicast address
of the tentative address.
"""
So, joining should happen before sending DAD packets, and it should
happen for each unicast address added. This is achieved by joining
from net_if_ipv6_addr_add() call. Note that we already leave
solicited-node group from net_if_ipv6_addr_rm(). In particular, we
leave it if DAD fails (as that function is called in this case).
Fixes#5282.
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
The reachable calculation for ND is using fractions combined with
integers and getting rounded to very small results (1ms or 0ms).
Let's split up the fraction into it's numerator and denominator
and perform the math in a better way to get the correct results.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
No need to do any IPv6 neighbor checks if the packet is routed back
to us by loopback driver.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
There was no handler functions for adding, removing and looking up
IPv4 multicast addresses in the network interface.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Because a next networking API call will lead to a crash. Given that
logging can be easily disabled (disabled by default so far!), don't
be shy and call by the name (i.e. error).
Jira: ZEP-2105
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
Create support for registering a callback that will be called
if the device leaves or joins IPv6 multicast group.
Jira: ZEP-1673
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the expire send timer expires, then it sends the packet.
If that happens, then we must not try to send the same packet
again if we receive ACK etc. which can cause re-sends to happen.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Make sure that the sent flag is only set after we have really
sent the packet and the driver has verified that.
If the net_pkt_set_sent() is called while still in tcp.c, then
depending on how fast the device is, it might happen that the
retry timer expires before the packet is actually sent. This was
seen in frdm-k64f with ethernet and various debug prints activated.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Check the return value of k_poll() as instructed by Coverity.
If debugging is enabled we print error if this happens, otherwise
this error is ignored. In our case the return value should always
be 0.
Coverity-CID: 173652
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the network interface is down, then IPv6 DAD (Duplicate
Address Detection) cannot be done if new IPv6 address is added
to the network interface. This can happen in Bluetooth, where
network interface is taken up only after there is a BT connection.
The DAD is delayed and done later after the network interface is up.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The address lifetime timer was cancelled always even if the address
timer was never installed.
Jira: ZEP-2397
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
From
http://pubs.opengroup.org/onlinepubs/7908799/xns/netinetin.h.html:
in_addr_t
An unsigned integral type of exactly 32 bits.
[] the in_addr structure [] includes at least the following member:
in_addr_t s_addr
In other words, POSIX requires s_addr to be a single integer value,
whereas Zephyr defines it as an array, and then access as s_addr[0]
everywhere. Fix that by following POSIX definition, which helps to
port existing apps to Zephyr.
Jira: ZEP-2264
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
net_if_get_default() was documented as returning "Default interface
or NULL if no interfaces are configured.", but actually didn't
return NULL in the latter case. Instead, it effectively returned
a pointer to random area of memory, shared with other system
structures, so calling functions like net_if_ipv4_set_netmask(),
etc. could trash unrelated memory.
Jira: ZEP-2105
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
And print a warning if that happens.
This will also avoid to raise an ASSERT on net_if_tx_thread()'s k_poll
as this one will be called with no events to work with.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Convert code to use u{8,16,32,64}_t and s{8,16,32,64}_t instead of C99
integer types.
Jira: ZEP-2051
Change-Id: I4ec03eb2183d59ef86ea2c20d956e5d272656837
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
- net_pkt becomes a stand-alone structure with network packet meta
information.
- network packet data is still managed through net_buf, mostly named
'frag'.
- net_pkt memory management is done through k_mem_slab
- function got introduced or relevantly renamed to target eithe net_pkt
or net_buf fragments.
- net_buf's sent_list ends up in net_pkt now, and thus helps to save
memory when TCP is enabled.
Change-Id: Ibd5c17df4f75891dec79db723a4c9fc704eb843d
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
There have been long lasting confusion between net_buf and net_nbuf.
While the first is actually a buffer, the second one is not. It's a
network buffer descriptor. More precisely it provides meta data about a
network packet, and holds the chain of buffer fragments made of net_buf.
Thus renaming net_nbuf to net_pkt and all names around it as well
(function, Kconfig option, ..).
Though net_pkt if the new name, it still inherit its logic from net_buf.
'
This patch is the first of a serie that will separate completely net_pkt
from net_buf.
Change-Id: Iecb32d2a0d8f4647692e5328e54b5c35454194cd
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Use UNALIGNED_GET and UNALIGNED_PUT throughout the networking stack to
access fields from the IP address structure. These structures can be
mapped directly to buffers and the macros are required for correct
unaligned memory access.
Jira: ZEP-2012
Change-Id: I55f9da7b143a22fa869d5d215c661de988cd9b91
Signed-off-by: Bogdan Davidoaia <bogdan.davidoaia@linaro.org>
This does not save any space but sets the fields in the
structs in more natural order. Move IPv6 related
fields in net_if into internal ipv6 struct so that all
IPv6 fields are located inside one struct. Same thing
is done for IPv4 fields which are now located inside
IPv4 internal struct in net_if.
There is no functionality changes by this commit.
Change-Id: I7d72ec0a28e2b88c79a4c294655d5ef6da6ccb25
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
We need a way to know when IPv6 address is successfully set
into network interface.
If IPv6 DAD (Duplicate Address Detection) succeeds or fails,
we send a management event for that. This can be used by
other components to detect when the network interface is in
usable state.
Change-Id: Ifb22415fe21f31f5dba4f55455d6e0f89b414d32
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Some of the IPv6 utility functions were missing const in
one of the parameters that are not modified by the corresponding
function.
Change-Id: Ic9fe53daac288570c14423fd9410dcf15d1c5cfa
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Router lifetime timer initialized in router_init but not submitted.
Change-Id: If5a77f413832db52eff99e7191f82d2a8fc8f081
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Do not call callback if destination ll address is not set. This
happens when dst is multicast or broadcast.
Change-Id: I34dda92799a987d9cff031dc97f4a01b94437561
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
There is no transition from tentative state etc... It should directly go
to preferred.
This is fixing ND processing when DAD is disabled: source address was
never set, as tentative state is not a valid in
is_proper_ipv6_address().
Change-Id: I6f0a0fdd99dd13d28849f9749f89ec21fdd370d8
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
If IPv6 router is added, then NET_EVENT_IPV6_ROUTER_ADD event
is generated. When router is removed, NET_EVENT_IPV6_ROUTER_DEL
is then generated. This is useful info if we want to have a generic
connectivity to outside of our local network.
Change-Id: Ia03958a071ceb998127894025c99ab72a8b648d5
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
k_delayed_work_init shall only be called once per work and not every
time before as that can trash the workqueue list if the work is pending.
Change-Id: Ib6398a6a843dd4d820529487ad76a9dc9d9e152e
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
If the earlier prefix timeout is not cancelled before setting
it again, the system will hang.
Change-Id: I6e271294c8e5d43e3ceae4780b5d1c26bab2b296
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Make sure that both the TX and RX threads are running before
we turn the network interface up.
Change-Id: Ie7e5938403e90e1ae76047c221cf9fd2f5a1167b
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
NET_IF_UP may change during the lifetime of k_pool which means we would
have to reconfigure everytime the flag changes but NET_IF_UP is already
checked during net_if_send_data thus it should never reach the queue in
the first place making this check unnecessary.
Jira: ZEP-1888
Change-Id: Iaa8471bee886a6f7e701a1dd243fb199def26589
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
No need to print error if we have already joined a pre-defined
multicast group.
Change-Id: I1237d5fbf0b63267ccb00432522513358c48d31d
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Wrong type modifier %zu in debug print, should be %d.
Change-Id: I82feca91d9e083a03263477569c552acc7c9767f
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Now that k_poll landed in the kernel, it's worth using it to save
memory and reduce the number of threads at runtime.
Such switch has been first done in bluetooth (see hci_core.c and conn.c
in subsys/bluetooth/host). Since network interfaces kind of follows the
same design for sending data, it was then easy to copy the same change as
in bluetooth.
Change-Id: I7f9734b88ac818284bbabaedc946b4765b905ebb
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Instead of magic constant (0) use the K_NO_WAIT when spawning
a thread.
Change-Id: Ib3a181fdfa246843d9c708592ccc754e54bc6005
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
As the networking cannot work without RX or TX threads, mark
them as essential.
Change-Id: Icb52c77fcdcef35d77c36e5ec855fed438a38407
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
