Adds support for incoming and outgoing IPv4 fragmented packet support,
allowing a single packet that is too large to be sent to be split up
and sent successfully.
Signed-off-by: Jamie McCrae <spam@helper3000.net>
In order to avoid using multiple sources of truth for the platfom's
endianness, convert the in-tree code to use the (BIG|LITTLE)_ENDIAN
Kconfig variables exclusively, instead of the compiler's
__BYTE_ORDER__.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
In case loopback interface is enabled, it's most likely there will be
more than a one network interface if running on real hardware. Therefore
increase the default IPv4/IPv6 interface count to avoid the need to
increase it manually in every test suite.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Instead of doing a 1-complement addition for every 16-bit word, process
32-bit words and handle the 1-complement addition at the end.
This work is a colaboration with Diego Pino García. The algorithm is
inspired by:
https://blogs.igalia.com/dpino/2018/06/14/fast-checksum-computation/
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
Use a union to capture all big NET event structures with a default size
of 32bytes, this makes maintenance easier.
Signed-off-by: Krishna T <krishna.t@nordicsemi.no>
NAK message should be processed in the RENEWING state just like in
REBINDING, not ignored.
Additinonally, NAK processing should include the rejected IP address
removal from the nework interface, otherwise the IP address will remain
on the interface indefinitely (i. e. until removed manually), which
might disturb further operation.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
It could happen, that the requested IP address, stored in the dhcpv4
config structure in the network interface, could get overwritten with an
unexpected message, for example NAK from the DHCP server. In result, the
DHCPv4 module was no longer able to remove the requested address from
the network interface, as it simply no longer remembered what the
address was.
Fix this, by setting the requested address only when it's actually
provided by the DHCP server, i. e. when handling the OFFER message from
the server. Accordingly, the requested address will be cleared when the
SELECTING stage is entered, where all of the cleanups should've already
be done.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The function is no longer used, as its functionality has been replaced
with net_if_carrier_off().
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Align virtual L2 with interface state handling update. Introduce
net_virtual_enable() function, which gets called whenever a network
interface is brought up (operational). This, combined with already
existing net_virtual_disable() function, can be used to update the
carrier state on the virtual interface, based on the underlying
interface status.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
When packet is passed to transmit at the net_if level, verify the
carrier state (NET_IF_LOWER_UP) instead of administrative state
(NET_IF_UP).
The administrative state is checked anyway at higher level (net_context)
so no need to verify it again. This will allow to still transmit control
packets for example when interface is in a dormant state.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Introduce a new interface state management scheme, according to
RFC 2863. This includes the following changes:
* Introduce a new interface flag: NET_IF_LOWER_UP, along with
corresponding helper functions. The flag should be set/cleared on an
interface by a network driver/L2 to signalize physical readiness of an
interface to transmit data (for example cable plugged in).
* Introduce a new interface flag: NET_IF_DORMANT, along with
corresponding helper functions. The flag should be set on an
iterface when the interface is not ready to transmit application data,
for example still not joined a Wi-Fi network.
* Introduce a new interface flag: NET_IF_RUNNING, indicating that
interface is ready to transmit application data.
* Update the meaning of the NET_IF_UP flag - it now singnalizes whether
an interface has been brought up/down by the application (admin
up/down).
* Introduce operational state of an interface, derived from above. It
reflects the internal interface state.
The meaning of net_if_is_up() function and NET_EVENT_IF_UP/DOWN events
remains unchanged to retain backward compability - they reflect the
interface readiness to transmit application data.
To verify the administrative up/down state, a new function
net_if_is_admin_up() has been introduced, along with
NET_EVENT_IF_ADMIN_UP/DOWN events.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Socket offloading was not taken into consideration when network
interface was put down. Fix this by introducing a helper function, to
verify whether an interface is offloaded or not.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Improve the parse_arg() helper function used with shell ping command,
by allowing to parse arguments provided in hex.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Add new option to the net shell ping command, which allows to specify
the DSCP/ECN values for an outgoing ping request. The option is
compatible with Linux ping command.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Add support for setting DSCP/ECN values for an outgoing ping request.
Additionally, copy DSCP/ECN values from an incoming ping request into
the ping response, like Linux does.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Add support for setting DSCP/ECN values for an outgoing ping request.
Additionally, copy DSCP/ECN values from an incoming ping request into
the ping response, like Linux does.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Add a new net_context option, which allows to set DSCP/ECN values on a
net context. Those values are then encoded into outgoing packet
IPv4/IPv6 header.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Add new fields to the net_pkt structure, representing IP-specific
Differentiated services code point (DSCP) and Explicit Congestion
Notification (ECN) values. Those values are encoded in legacy
Type of Service (IPv4) and Trafic Class (IPv6) header fields.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
This change implements part two of the program laid out in the TSCH RFC,
see #50336#issuecomment-1250250154 :
> Consolidate IEEE 802.15.4 options in net_pkt
This change improves decoupling of generic net core code from
IEEE 802.15.4 internals. It also simplifies IEEE 802.15.4
attribute cloning and thereby makes it easier to maintain and less
error prone (and probably even faster as individual bits are no longer
copied over separately).
This enables us to extend and design IEEE 802.15.4 L2 attributes inside
the package in isolation from the net core code which will no longer
have to be changed when introducing changes or additions to the flags.
This flexibility will be built upon in later change sets to model the
IEEE 802.15.4 attributes closer to the spec.
The solution is inspired by Linux's sk_buff->cb attribute which addresses
the same concern as the attribute introduced in this change set:
https://elixir.bootlin.com/linux/v6.0.1/source/include/linux/skbuff.h#L871
As the inline comment says: The cb attribute can be made a union or even a
uint8[something] in the future, if further L2s need a control block, too.
Right now such full indirection would make the code overly abstract, so
I chose to compromise with maintainability in mind.
Care has been taken to ensure that this changes does not introduce
additional padding into the net package. To maintain zero-padding, future
changes to the net packet struct will have to ensure that the
IEEE 802.15.4 struct is 4-byte aligned (iff the IEEE 802.15.4 struct
continues with max uint32_t scalar members) which is no deviation from
the previous implementation.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
The net packet structure contains pointers to link-layer source and
destination addresses. Usually, these structures do not point to
separately allocated memory but directly into the packet's data buffer.
In case of a deep package clone (which includes copying the buffer) the
copy of the ll addresses continued to point into the old package
(contrary to a rather misleading inline comment). This was proven by an
additional failing unit test assertion.
As the original package may be unreferenced while the cloned package is
still being accessed, the ll address pointers of the cloned package may
become invalid.
The fix consists of two parts:
* First it is determined whether a given ll address actually points into
the buffer and if so at which logical cursor offset it is located.
* If the address points into the package buffer then the cursor API is
used to determine the corresponding physical memory location in the
cloned package. The ll address of the cloned package is then patched
to point to the cloned buffer.
Additional assertions were introduced to the existing unit test to ensure
that the newly generated address points to the correct content of the
cloned package.
The solution is implemented in a generic way so that the previously
redundant implementations were consolidated into a single one. The code
includes a check that ensures that the ll address check and manipulation
will be skipped in case of shallow package copies.
The change also addresses problems related to the "overwrite" flag of the
package:
* Package cloning assumes the overwrite flag to be set. Otherwise it
will not work correctly. This was not ensured inside the clone method.
* Package cloning manipulates the overwrite flag of the cloned package
but does not reset it to represent the same state as the original
package.
The change introduces a fix and unit test assertions for both problems.
Fixes: #51265
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
This is a public API for the subsystem, can be called from app
context, unlocks the local k_mutex on one of its three exit paths, and
it's quite clear that nothing ever locks that mutex!
The code used to work because k_mutex simply returned an error if you
tried to unlock an unlocked object. Now zync will panic (when
CONFIG_ZYNC_VALIDATE=y) if you try that.
Signed-off-by: Andy Ross <andyross@google.com>
Without pkt_unref(), the shell command "net udp bind" will fail to
receive after a few packets ("net allocs" shows that memory is still
allocated).
Signed-off-by: Andreas Müller <andreas.mueller@husqvarnagroup.com>
IEEE 802.15.4 short address support is incomplete in several places.
This change improves short address support without claiming to fix
it everywhere. Future iterations will have to continue where this change
leaves off.
The purpose of this change was to:
* use the short address returned by association responses,
* automatically bind IEEE 802.15.4 datagram sockets to the short
address if available,
* use the short address in outgoing packages where applicable,
* improve validation of association/disassociation frames,
* model association more closely to the spec by tying it to the
existence of a short address in the MAC PIB thereby removing
redundancy in the PIB (which makes race conditions less probable),
* keep both, the short and extended addresses, of the coordinator.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
This is a follow up to commit bb86f8b967.
Also custom IEEE 802.15.4 based L2 implementations may need to use
packet handling without the IP layer. Add support for such cases.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Zephyr doesn't seem to allow on adding route to the device from
which it received Router Advertisement without SLLAO field set
(in particular device that doesn't include Link Layer address).
Changes done:
* Added creating new NBR without Link Layer address set on RA
reception (LL is added only if SLLAO option is set, otherwise
NBR doesn't have LL address, but exists in the table and has
valid IPv6 link local address).
* Removed two asserts preventing from adding route to the NBR
that doesn't have Link Layer address set.
Signed-off-by: Kamil Kasperczyk <kamil.kasperczyk@nordicsemi.no>
Since db11fcd174 [net/net_pkt: Fully
separate struct net_pkt from struct net_buf], net_pkt_compact() can not
fail anymore.
This commit acknowledges this and simplifies the API accordingly.
Signed-off-by: Reto Schneider <reto.schneider@husqvarnagroup.com>
Allocated, but undersized packets must not just be logged, but also
unreferenced before returning an error.
Signed-off-by: Reto Schneider <reto.schneider@husqvarnagroup.com>
In the function find_available_port() a port is randomly selected. Because
the random value is always >= 0x8000, it is redundant to check if it is
<= 1023 afterwards.
This commit removes the redundant check.
Signed-off-by: Christian Taedcke <christian.taedcke@lemonbeat.com>
If a L2 link has been established, then the DHCP is taking too long as
it has to go through its capped exponential backoff timers to trigger
discover (The DHCP starts immediately during init, this is itself wrong,
it should start on a link UP notification) that delays the DHCP for
few seconds to a minute.
And if we do stop and start DHCP then also it goes through the initial
delays (though configurable), which is also not ideal.
Add support for restarting DHCP without any delay, i.e., release and
send discover immediately.
This is also useful in case L2 switches to a different subnet, in this
case Zephyr doesn't restart DHCP automatically, this API can be used by
L2 apps/drivers to restart DHCP to get new subnet IP.
Signed-off-by: Krishna T <krishna.t@nordicsemi.no>
* reduced cyclomatic complexity
* group validation by family to make the validation easier to understand
and extend
* change preprocessor markup where possible to allow for complete code
elimination when features (esp. IP) are disabled
* renamed net_context_get/set_ip_proto() to net_context_get_proto()
While the latter is formally part of the public API and might therefore
have to be deprecated rather than renamed, it is considered internal API
by the net developers, see
https://github.com/zephyrproject-rtos/zephyr/pull/48751#discussion_r942402612
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
The net_core:process_data() and connection:net_conn_input() methods are
the central network packet reception pipeline which:
1) guide network packets through all network layers,
2) decode, validate and filter packages along the way and
3) distribute packages to connections/sockets on all layers.
This code seems to have grown complex and rather cluttered over time as
all protocols, layers and socket implementations meet there in one single
place.
The code also reveals its origin as a pure IP stack which makes it hard
to introduce non-IP protocols and their supporting socket infrastructure
in a modularized way.
For an outside contributor it seems almost impossible to add another
protocol, protocol layer, filter rule or socket implementation without
breaking things.
This change doesn't try to solve all issues at once. It focuses
exclusively on aspects that maintain backwards compatibility:
* Improve modularization and encapsulation on implementation level by
disentangling code that mixes up layers, protocols and socket
implementations.
* Make IP just one protocol among others by removing assymmetry in
protocol handling logic and introduce preprocessor markup so that
IP-specific code can be eliminated by the preprocessor if not needed.
* Use preprocessor markup to delineate hook points for future
modularization or expansion without introducing structural changes (as
this would almost certainly break the API).
* Reduce cyclomatic complexity, use positive rather than negative logic,
improve variable naming, replace if/elseif/else blocks with switches,
reduce variable span, introduce inline comments where code does not
speak for itself, etc. as much as possible to make the code overall
more human-friendly.
Background: These are preparative steps for the introduction of IEEE
802.15.RAW sockets, DGRAM sockets and sockets bound to PAN IDs and device
addresses similar to what the Linux kernel does.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
Introducing additional socket implementations is rather involved right
now due to some more or less convoluted code that had grown over time.
This change introduces an additional configuration variable in preparation
for additional socket API drivers. The idea is to reduce redundant code
and make existing code more readable by better exposing its actual intent.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
The code contained several repeated composite IPv4/v6 and UDP/TCP
preprocessor statements that can be simplified by introducing a hidden
NET_IP preprocessor constant that captures what probably is actually
"meant" by this code.
While we were on it we also used the new constant to further isolate
IP-specific code from non-IP specific generics.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
As of today <zephyr/zephyr.h> is 100% equivalent to <zephyr/kernel.h>.
This patch proposes to then include <zephyr/kernel.h> instead of
<zephyr/zephyr.h> since it is more clear that you are including the
Kernel APIs and (probably) nothing else. <zephyr/zephyr.h> sounds like a
catch-all header that may be confusing. Most applications need to
include a bunch of other things to compile, e.g. driver headers or
subsystem headers like BT, logging, etc.
The idea of a catch-all header in Zephyr is probably not feasible
anyway. Reason is that Zephyr is not a library, like it could be for
example `libpython`. Zephyr provides many utilities nowadays: a kernel,
drivers, subsystems, etc and things will likely grow. A catch-all header
would be massive, difficult to keep up-to-date. It is also likely that
an application will only build a small subset. Note that subsystem-level
headers may use a catch-all approach to make things easier, though.
NOTE: This patch is **NOT** removing the header, just removing its usage
in-tree. I'd advocate for its deprecation (add a #warning on it), but I
understand many people will have concerns.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The `%i` format specifier is for `int` and is equivalent to
`%d` for printf formatting.
However, for `size_t`, the correct format specifier is `%zu`.
Fixes#49825
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
The reorder process in rather complicated. Add a runtime check to always
check afterwards if the reordering queue is still consistent. When this
test fails, discard the contents of the complete reorder buffer.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
When all the data of the new packet is already present in the
queue_recv_data, make sure the packet is discarded.
This commit adds a test line that reproduces the issue and implements a
fix.
Also in appending the packet, call the official net_buf_frag_add function
instead of updating the last net_buf pointer, since the net_pkt_remove_tail
called in between, might have removed the last net_buf.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
TCP is a streaming protocol, this means it can set the packet boundary
at an arbitrary location. Therefor a re-transmitted packet may have the
packet boundary at a different location from the original missing segment.
The reordering logic should be able to deal with this situation and
throw away the overlapping data. This fix clears the out of order queue
when the queued data is not relevant.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
TCP is a streaming protocol, this means it can set the packet boundary
at an arbitrary location. Therefor a re-transmitted packet may have the
packet boundary at a different location from the original missing segment.
The reordering logic should be able to deal with this situation and
throw away the overlapping data. Now also for data that needs to be added
to the existing queue
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
TCP is a streaming protocol, this means it can set the packet boundary
at an arbitrary location. Therefor a re-transmitted packet may have the
packet boundary at a different location from the original missing segment.
The reordering logic should be able to deal with this situation and
throw away the overlapping data.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
Instead of waiting for the retransmit timeout, retransmit as soon as
missing data is deduced based on a triple-duplicate ACK.
Increase the number of buffers in the testcase, to allow for at least 4
packets in flight to trigger the triple-duplicate ACK.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
When out of order data is received, send out a duplicate ACK to notify
the sender that there is data received out of sequence, so it can
retransmit the missing section.
Also avoid sending acknowlegdement to acknowlegdement messages that are
having an out of order sequence number.
Signed-off-by: Sjors Hettinga <s.a.hettinga@gmail.com>
This change makes the packet socket and ieee802154 l2 drivers aware of
AF_PACKET sockets, see https://github.com/linux-wpan/wpan-tools/tree/master/examples
for examples which inspired this change.
Signed-off-by: Florian Grandel <jerico.dev@gmail.com>
Networking statistics framework is used to define handler and the data
structure, Wi-Fi management layer implements the handler and also adds a
new offload API to get statistics from the Wi-Fi driver.
Signed-off-by: Krishna T <krishna.t@nordicsemi.no>
