This adds restrictions for setting scan response data as specified by
Core spec:
* on legacy, only complete data can be set
* on legacy, data cannot be longer than 31 octets
* on ext, if non-scannable, data can only be discarded
* if ext enabled, only complete data can be set
* if ext enabled, data cannot be discarded
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
This allows to set scan response data using ext adv commands.
New AUX_SCAN_RSP can be created without analyzing previous PDU since
the only value that need to be copied from previous PDU is AdvA which
is placed at fixed location. TxPower is omitted since, if enabled, it
is already present in AUX_ADV_IND and does not need to be presed in
other PDUs. ADI support is to be added later.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
AdvA in aux is mandatory for any connectable and scannable set so make
sure it's always added.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
Connectable and scannable instances always have aux so we need to force
AuxPtr to be included in primary PDU and have aux allocated even if no
data are set.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
Connectable and scannable instances always have AuxPtr so we can return
error immediately if controller is configured with no aux.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
Prevent redundant collision resolves and potential incorrect and harmful
increment of lazy_current.
During collision scenarios with high CPU load, the ticker_worker may be
called a second time before the ticker_job gets to run. This will cause
ticker operations on the previosly expired node (expected), and in some
cases increment lazy_current, even though the node was not sceduled to
execute via the req/ack mechanism.
By moving the request check before collision resolve, CPU time is saved,
and lazy_current will not incorrectly be incremented if node is in
collision.
The problem may be seen as a connection suddenly not receiving packets,
or MIC error on master, because the event counter goes out of sync.
Signed-off-by: Morten Priess <mtpr@oticon.com>
Use a helper config BT_CTLR_CONN_RSSI_SUPPORT so that it is only
enabled when supported by the controller.
Signed-off-by: Rubin Gerritsen <rubin.gerritsen@nordicsemi.no>
Now that device_api attribute is unmodified at runtime, as well as all
the other attributes, it is possible to switch all device driver
instance to be constant.
A coccinelle rule is used for this:
@r_const_dev_1
disable optional_qualifier
@
@@
-struct device *
+const struct device *
@r_const_dev_2
disable optional_qualifier
@
@@
-struct device * const
+const struct device *
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
A build assert in dummy.c lists the following requirement:
[...] receive thread priority shall be higher than the Bluetooth
Host's Tx and the Controller's receive thread priority.
This is required in order to dispatch Number of Completed Packets
event before any new data arrives on a connection to the Host threads.
The drivers uses a priority that is equal to the Host TX thread,
and since they don't use the CONFIG define that is only available
to the controller then this BUILD_ASSERT will not catch the
requirement.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
Add the nRF52805-specific file with radio related defintions.
Use also specific configurations for certain PPI channels used by
the controller, as this SoC has limited number of programmable PPI
channels.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Change hard-coded (D)PPI channel bitfield values to the BIT() macro
calls with the corresponding channel base value supplied as parameter,
to make it possible to actually change this base value if needed.
Remove unused *_EXCLUDE macros.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
nRF5340 does have SWI peripheral, hence use it instead of
the single available EGU peripheral. Use of SWI will allow
controller's LLL, ULL_HIGH and ULL_LOW execution context to
be independently configured to different interrupt priority
levels.
When ULL_HIGH priority equals ULL_LOW priority, only SWI2
is used by controller. Otherwise, SWI3 is used for ULL_LOW.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Detect and handle AD data set race condition between thread
and ISR context.
Fixes#27637.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Add cpu_dsb() hal interface so that data synchronization
barrier be used in ARM Cortex M4 and M33 architecture
implementation.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Set the AD data double buffer first index as volatile
because it is modified in LLL ISR context.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactor out into function the filling of sync info
structure in the common extended advertising payload format
header.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactor out into function the filling of aux ptr structure
in the common extended advertising payload format header.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactor out into functions the calculation of common
extended advertising payload format header length.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Moved the assignment of PDU length out side of a conditional
so that code to calculate PDU length can be reused.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactor out sync info field population into a utility
function with set and clear interface to add or remove
the common extended advertising header format fields.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
It fixes:
../../../../include/toolchain/gcc.h:169:30: warning: statement will
never be executed [-Wswitch-unreachable]
169 | #define __fallthrough __attribute__((fallthrough))
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Fix conditional compilations that fail when combinations
of Broadcaster, Observer, Peripheral and/or Central are
selected to build an application.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
-Wimplicit-fallthrough=2 requires a fallthrough comment or a compiler
to tells gcc that this happens intentionally.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Return BT_HCI_ERR_INVALID_CONN_ID if an invalid handle is
given by the HCI; currently sometimes BT_HCI_ERR_CMD_DISALLOWED
is returned
Signed-off-by: Andries Kruithof <Andries.Kruithof@nordicsemi.no>
Core 5.2, Vol 4, Part E, section 5.3.1 clarifies that advertising set
handle is assigned by host when advertising set is created and this
happens only on LE Set Extended Advertising Parameters.
An advertising set handle is an arbitrary number within allowed range,
i.e. 0x00-0xEF and not 0..max_supported-1.
This patch adds option to enable advertising set handle mapping from
HCI range as defined by Core specification to zero-based handles used
by LL. If enabled, HCI handle will be remapped to LL handle for each
command, otherwise HCI handle will be used as an LL handle. The latter
effectively skips mapping logic and should be used with Zephyr host
which uses zero based indexes.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
There's not need to mark advertising set as created by either legacy
or advertising command since this is determined by current so-called
HCI advertising mode which determines whether host uses legacy or
extended advertising commands.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
As per Core 5.2, Part B, section 3.1.1 controller should not allow
mixing of legacy and extended advertising commands. Once 1st command
is received from host, we should only allow commands of the same type
and return an error on other commands.
If legacy advertising interface is selected by 1st command received,
we need to make sure that set with handle=0 is created since it's
used by all legacy commands.
When not using external host, we assume that only extended advertising
interface will be used thus all checks can be skipped (optimized and
compiled-out).
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
A lot of commands return CC event with status parameter only, especially
when error is returned. This patch adds a helper to create such event,
similar to cmd_status.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
Proper error code should be returned if advertising set with given
handle does not exist instead of generic "invalid parameters" one.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
AdvA in extended advertising PDUs is put at different offset than in
legacy PDUs thus it's better to let caller calculate and pass AdvA and
TxAdd explicitly. This way we won't need to check whether scan request
was on legacy or extended PDU every time.
Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@codecoup.pl>
When a connect indication contains a channel map of all zeros, the
adv->conn is left NULL'ed after return and next connect attempt leads
to a crash.
This was identified via the Sweyntooth test suite executing the script
re. issue 6.14 towards an Oticon target.
Signed-off-by: Erik Brockhoff <erbr@oticon.com>
It does not auto-initiate the DLE procedure if the default data length
parameters are not equal to the initial parameters.
Signed-off-by: Rubin Gerritsen <rubin.gerritsen@nordicsemi.no>
Fix incorrect use of LLL ticker user id in ULL_LOW execution
context.
Regression introduced in commit 23ef75f6d9 ("Bluetooth:
controller: split: Fix Low Latency Prepare Done").
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
