In an effort to reduce the footprint of the LwM2M client, let's
lower the default # of observes handled by the client from 20 to
10.
This saves ~640 bytes of SRAM.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Remove "used" member from observe_node structure and replace by
checking the ctx for non-NULL value.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
The slist attr_list doesn't scale well when added to the LwM2M object,
object instance and resource instance structures. The goal of a
robust LwM2M client is to let the user create MANY object instances
and these will have many resource instances each. The amount of SRAM
taken up by the attr_lists will only increase over time, regardless
of the actual # of write attribute structures reserved via the
LWM2M_NUM_ATTR config setting.
Instead, let's remove the slist from these structures and add a
reference pointer to the lwm2m_attr structure. We can use this
reference to create the one to many relationship between the objects,
object instances and resource instances for a much smaller amount of
code and SRAM resources.
The sacrifice for these savings will be a larger # of iterations when
looking up assigned write attributes and matching them to their
references. However, due to the # of write attributes current being
handled, the # of iterations during this process is very manageable.
Example flash and SRAM savings when building for nrf52_blenano2:
Before patch:
Memory region Used Size Region Size %age Used
FLASH: 139532 B 512 KB 26.61%
SRAM: 36576 B 64 KB 55.81%
IDT_LIST: 148 B 2 KB 7.23%
After patch:
Memory region Used Size Region Size %age Used
FLASH: 139284 B 512 KB 26.57%
SRAM: 36000 B 64 KB 54.93%
IDT_LIST: 148 B 2 KB 7.23%
Summary: This patch saves ~248 bytes of flash and ~576 bytes of SRAM
for the typical configuration of LwM2M client in Zephyr.
NOTE: these values will vary by architecture.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
When any error is returned from update_attrs() in engine_add_observer()
an EINVAL is returned back to the caller. Let's return whatever error
code was generated in update_attrs() instead.
Also, add handling where previously errors were ignored.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
For code clarity, let's move notification_attrs structure to the top
of lwm2m_engine.c. While we're at it, we can re-order it's members
for memory alignment.
NOTE: This patch does not change the current flash or SRAM usage but
further additions to the notification_attrs structure could.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
The path member of the object instance and resource instance structures
can easily be removed to save several bytes per instance over the entire
LwM2M subsystem. So let's remove it.
Example savings when building for nrf52_blenano:
SRAM usage before patch: 37952 B
SRAM usage after patch: 36576 B
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Let's optimize the order of the following structures to account for
memory alignment:
lwm2m_engine_obj
lwm2m_engine_res_inst
lwm2m_output_context
lwm2m_output_context
Tested building for nrf52_blenano hardware:
SRAM usage before patch: 38240 B
SRAM usage after patch: 37952 B
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
There are valid use cases where the model layer must know the true
destination address. So far only the fact that it was one of the
addresses that the model subscribes to (its element's unicast
included) has been knowable.
Solve the issue by moving the destination address from the internal
net_rx context to the public bt_mesh_msg_ctx struct.
Fixes#7453
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
During the compression an empty settings entry was
copied to the scratch sector form the oldest sector in case
lack of newer entry version in fcb storage.
This doesn't make sense as empty entry and lack of entry has
similar meaning.
Signed-off-by: Andrzej Puzdrowski <andrzej.puzdrowski@nordicsemi.no>
sendto() is one and only caller of send_data(), a function of handful
of lines, and yet send_data() is located a hundred lines away from
it. Such "spaghetti functions" complicate review, debugging, and
refactoring of the IP stack.
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
It's highly unlikely that snprintk() will return a negative value, but
that's a possibility that will make the `pos` variable be set to a
value outside the boundaries of the statically allocated `buf` array.
Also clamp writes to ensure that the statically allocated buffer won't
be overwritten with a large token length.
Fixes#7070.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
This patch restricts the stack sentinel to only be allowed if the
USERSPACE configuration option is not set. The stack sentinel feature
is redundant if used in conjunction with the USERSPACE, due to the
protection mechanisms in place for stacks.
Signed-off-by: Andy Gross <andy.gross@linaro.org>
There are certain use cases where the application needs to be able to
explicitly set a specific identity address. This was previously
possible using the bt_storage API, however now that it's gone another
solution is needed.
This patch adds a ne bt_set_id_addr() API which the application can
use to set a specific identity address before calling bt_enable().
Fixes#7434
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Instead of manually iterating all app keys and net keys, use the
bt_mesh_subnet_del() helper on all subnets. This will also clear any
app keys, and ensures that persistent storage is cleared as well.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Both the local sequence number as well as the Replay Protection List
(RPL) are states that may potentially change very often. In order not
to wear out the flash with these updates it makes sense to try to
avoid too frequent writes.
For the local sequence number a simple solution is not to write the
number on every increment. This patch introduces a new Kconfig option
to define after how many increments the sequence number gets written.
When the stack gets initialized it automatically adds the configured
number to the last stored one, thereby guaranteeing that the node
starts off with a number that's larger than the last used one.
The RPL is more problematic, since in principle it needs to be updated
every single time that we receive and process a message. Especially
security sentitive nodes will want this stored immediately to flash.
To give some use-case dependent flexibility, this patch introduces a
new Kconfig option to specify a timeout after which the RPL gets
written to flash.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
These are needed both for bt_mesh_provision() as well as persistent
storage-based network creation.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This variable is both redundant as well as problematic when it comes
to adding persistent-storage-based provisioning information, which
will not come through main.c or the bt_mesh_provision() API. Just
remove it and use bt_mesh.valid which serves the same purpose in
practice.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add APIs for storing core network values, such as Net and App Keys, IV
Index, Sequence number, RPL, etc.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add initial skeleton for doing settings-based persistent storage for
the mesh network state. This patch only includes restoring some core
network state such as IV Index, Sequence number, Net Keys, App Keys
and the Replay Protection list. The remaining state, and actually
storing the state, is left for follow-up patches.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The sequence number is incremented from several different places in
the stack. The way it was done was potentially race condition prone,
and was also problematic from the perspective of updating the
sequence number in persistent storage. Create a dedicated helper for
incrementing the sequence number (solves the race) which can in later
patches be used to add the persistent storage support.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This for loop runs inside an "if (!sub)" branch, so explicitly setting
sub to NULL in the loop is redundant.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
These app key and net key (subnet) helpers will soon be needed to be
called from the persistent storage code.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add support for multiple instances of a file system by
making use of mount point as the disk volume name which
is used by the file system library while formatting or
mounting a disk.
Also moved out file system specific data structures from
public fs.h header and handled them in corresponding
file system interface files by introducing open files and
open directories concept which is already being used in
NFFS interface module. Now it is extended to FatFs as well.
Signed-off-by: Ramakrishna Pallala <ramakrishna.pallala@intel.com>
Add support for enabling multiple disk interfaces (Flash, RAM)
simultaneously in Zephyr by introducing a simple disk interface
framework where we can register multiple disks which would
interface with different storage devices. This would enable us
to have multiple instances of FATFS in Zephyr.
Add support for mass storage drive disk name which will be
used as an argument when calling the disk interface API's.
Enable multiple volumes support configuration in
ELM FAT library.
Signed-off-by: Ramakrishna Pallala <ramakrishna.pallala@intel.com>
Extend storage_dev type beyond 'struct device' by changing
the variable type to void pointer. This will be needed if we
have additional transactional layers b/w filesystem and the
actual backend device.
Signed-off-by: Ramakrishna Pallala <ramakrishna.pallala@intel.com>
Enable settings and increase the system workqueue size to deal with
the stack usage. This also makes it possible to test unpairing support
with the shell's 'clear' command.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
There's a bit of unnecessary space in the bt_keys struct. Re-design
some fields for a more compact format, which is particularly helpful
now that the struct gets stored as-is to flash through the settings
API.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Integrate the bt_keys submodule with bt_settings. Add a new
bt_keys_store() API to write keys to flash, and extend the existing
bt_keys_clear() to remove the keys from flash.
Along with this, add some helpers for genrating settings key values
containing a bluetooth address, as well as for decoding them to get
the binary bt_addr_le_t.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add a new linker section for a list of submodule settings handlers,
and iterate the list from the various settings callbacks.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The same functionality is now supported by the settings-based
solution, so remove bt_storage out of the way. There were stubs in
bt_storage to handle per-peer information (e.g. pairing keys) but this
was never actually implemented in full. The next step is to add this
support to the settings-based solution.
Leave the code for generating temporary IRK and identity address in
case BT_SETTINGS is not enabled. Also leave the code for using vendor
HCI to read the identity address, in which case the settings
implementation will not touch it.
Introduce a new bt_unpair() API to replace the removed
bt_storage_clear(), since the latter was actually doing more than just
storage management: it was also handling runtime storage of pairing
information. Later, the bt_unpair() implementation will be extended to
clear settings-based pairing storage.
There is one feature that the bt shell module looses: the ability to
give a specific identity address to the "init" command as a parameter.
We might look later in the future if this is really needed, and add a
separate API for this.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Introduce a basic skeleton for peristent storage based on the settings
subsystem. Also enable support for this to the peripheral sample
application, so the new code gets exersized by CI. For now, the
implementation provides the same level support as the bt_storage API
ever did, i.e. for the identity address and the IRK.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This is bug-fix for issue #7311
FCB: CRC write size in append_finish doesn't honor
flash min write size
This patch changes write size to the minimum supported
write size.
Signed-off-by: Andrzej Puzdrowski <andrzej.puzdrowski@nordicsemi.no>
FS_FLASH_MAP_STORAGE keyword enables the storage partition,
but it was depend on flash_map module which is unused by
NFFS. This patch makes it independent thanks
to it is possible to enable the storage partition
without flash_map module.
FS_FLASH_MAP_STORAGE was renamed to
CONFIG_FS_FLASH_STORAGE_PARTITION
as it is independent for flash_map.
Signed-off-by: Andrzej Puzdrowski <andrzej.puzdrowski@nordicsemi.no>
The IEEE 802.1Q chapter I.3 contains a proper network packet
priority to traffic class mapping. The original mapping was
clearly incorrect.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
In the future, when bootstrap support is added, this config won't
be used. Let's remove it.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Now that the LWM2M_OP_* bits have been renumbered, we no longer need
a custom BIT() macro for the LwM2M code. Let's remove it and use
BIT() instead.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Remove unused OP flag LWM2M_OP_NONE and renumber the existing flags
so that the operations used in object permissions land in the lowest
bits, and extended operations come later.
We may eventually add more permission / data flags, so let's try and
keep them inside a 1 byte boundary (flags with bits 0 to 7).
NOTE: LWM2M_OP_DELETE is currently not checked as a permission but
it may be in the future so it is in the lower bits.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
When a data pointer or data length is not set, the read and write
handlers should return ENOENT to generate the correct LwM2M error
code (COAP_RESPONSE_CODE_NOT_FOUND).
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Before ever call of string_to_path(), the lwm2m_obj_path object
was being cleared via memset. Let's move the memset into
string_to_path() to remove the duplicate code.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Eliminate several similar code-blocks by replacing
engine_get_resource() with a more useful function called:
path_to_objs()
By supplying an lwm2m_obj_path object, it will find and set
the related values for:
struct lwm2m_engine_obj_inst
struct lwm2m_engine_obj_field
struct lwm2m_engine_res_inst
NOTE: NULLs can be supplied where the returned value is not
important.
Signed-off-by: Michael Scott <michael@opensourcefoundries.com>
Add functions that will return correct source IPv4 address
according to given destination address. This is done similar
way as for IPv6.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The code was updating net_rx->seq to make sure sdu_recv() gets the
right sequence number (seq_auth for segmented messages), however later
net_rx->seq was also used for enqueuing to the Friend Queue, causing
the queued messages to have the wrong value.
To fix this, don't update net_rx->seq, rather pass an explicit
sequence number value to sdu_recv(), which is just net_rx->seq for
unsegmented messages, and seq_auth for segmented messages.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The 'valid' member of struct bt_gatt_ccc_cfg was redundant, since
setting 'peer' to BT_ADDR_LE_ANY does the same job. What's worse, the
handling of 'valid' was also buggy in that some places looking for
valid CCC structs only matched the address, meaning it might yield a
positive match for invalid entries.
Fix these issues by removing the 'valid' struct member, and solely
using the 'peer' member to identify valid entries. Also simplify the
code by acknowledging that no CCC entry is essentially the same as the
value '0' written to CCC.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
With fragmentation disabled echo server responds to packet fragment with
ICMPv6 message Type: "Parameter Problem (4)", Code: "unrecognized Next
Header type encountered (1)". If a fragment with payload length 15 is
received in response sent by echo server IPv6 payload length and ICMPv6
checksum are wrong. This patch solves the issue by correcting payload
length.
Signed-off-by: Ruslan Mstoi <ruslan.mstoi@intel.com>
