Fix in the do_ecb() function
In real HW this function busy waits for the ECB to be done.
In simulation with the POSIX arch a WFE was added to avoid that
infinite loop.
But this function is called with the ECB interrupts disabled.
In normal builds other interrupts will awake the CPU very soon
after entering into that WFE, and the problem was not discovered.
But, in controller only builds, during some conformance tests,
the loop will hang as no other interrupts are coming.
=> Replace the WFE() with a k_busy_wait (only for simulation)
Signed-off-by: Alberto Escolar Piedras <alpi@oticon.com>
Fix channel map update procedure implementation's handling
of different transaction collision by not asserting but
disconnecting the connection due to invalid behavior by
peer implementation.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Explicitly track the connection update related ticker stop
and start to avoid asserting due to ticker update being done
at the same time for compensating the clock drift.
The compensation related ticker update failure in this case
can be safely ignored as new anchor point is used anyway
at the instant of the connection update.
Fixes#8796
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
This adds common option to disable support for LE Data Length Update
procedure in controller and host.
This will reduce flash usage by compiling out le_data_len_change
event handler that will never be called if controller has been
compiled with BT_CTLR_DATA_LENGTH option disabled.
Signed-off-by: Mariusz Skamra <mariusz.skamra@codecoup.pl>
This adds common option to disable support for PHY Update
procedure in controller and host.
This will reduce flash usage by compiling out le_phy_update_complete
event handler that will never be called if controller has been
compiled with BT_CTLR_PHY option disabled.
Signed-off-by: Mariusz Skamra <mariusz.skamra@codecoup.pl>
Updated controller implementation to disallow disabling
initiator state using scan disable. But allow disabling an
already disabled scan state. Also, disallow enabling scan
state while in initiator state.
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Updated the controller implementation to not feature
exchange if already done once either by local or remote peer
device in an active connection session.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
The channels assigned to the controller are reordered so that the ones
previously used and now available for other purposes have continuous
numbers. When the controller can take advantage of the pre-programmed
PPI channels (when TIMER0 is used as the event timer), the now free
channels are 0-4, when it cannot, it is the channel 0.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Several PPI channels in nRF5 family SoCs are pre-programmed with fixed
settings. A few of them can be used in the bluetooth controller instead
of the freely programmable ones that are used currently. This commit
makes such replacements where possible so that the universal channels
can be left available for other purposes.
This commit also removes macros used previously in calls to functions
enabling and disabling particular PPI channels (as it is sufficient
to use the BIT macro to set bits corresponding to the channel numbers)
to prevent such problems like the one introduces by commit
9d1ca9c390 (channel 18 changed to 17
but the related macro definition not updated).
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Add support for exchanging both minimum and maximum
connection interval values in Connection Parameter Request
Procedure implementation.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
This is a pure refactoring of the k32src_wait function. It used the
following rules when refactoring:
Don't use the preprocessor when unprocessed C language suffices.
Don't undefine macro's.
Avoid global variables when possible.
Use consistent names for similiar things (hf_clock, lf_clock).
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
k_poll_signal was being used by both, struct and function. Besides
this being extremely error prone it is also a MISRA-C violation.
Changing the function to contain a verb, since it performs an action
and the struct will be a noun. This pattern must be formalized and
followed and across the project.
MISRA-C rules 5.7 and 5.9
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Fix an integer overflow in the scheduling implementation
that calculates whether resources required for next radio
event be retained.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
This commit optimizes access to RNG driver by taking advantage
of the data structures layout. As result, number of calls to RNG
driver is reduced.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
The entropy_nrf_get_entropy_isr(), which is specific to this driver,
is in fact equivalent of generic entropy_get_entropy_isr(..., 0).
This commit removes the entropy_nrf_get_entropy_isr() function
and replaces its usage by call to generic entropy API.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
The bluetooth controller has been using the flag '-Ofast' to keep
within a real-time limit. There are two problems with this; firstly,
when a project should be optimized for size it is standard to use -O2,
not -Ofast.
Secondly, optimization flags have been deemed to be non-portable, so
instead of directly using "-Ofast" we should use the intent-macro
OPTIMIZE_FOR_SPEED_FLAG to ensure toolchain portability.
Testing has shown that we are still within the real-time limit when
changing from -Ofast to -O2. -Ofast is about 1us and 1% faster, but
increases the code size by 13kB (5% of the available flash on a
nRF51).
Since the slowdown is comparatively small compared to the code size
increase we have decided to use -O2 in place of -Ofast.
Other optimization combinations were also measured and their results
can be seen below:
-Ofast in BLE Controller and in #pragma in entropy driver.
Memory region Used Size Region Size %age Used
FLASH: 138920 B 256 KB 52.99%
[bt] [INF] encode_control: l: 6, 6, 7; t: 75, 45, 132.
-O2 in BLE Controller and in #pragma in entropy driver.
Memory region Used Size Region Size %age Used
FLASH: 125840 B 256 KB 48.00%
[bt] [INF] encode_control: l: 6, 6, 7; t: 75, 51, 133
-O3 in BLE Controller and in #pragma in entropy driver.
Memory region Used Size Region Size %age Used
FLASH: 138920 B 256 KB 52.99%
[bt] [INF] encode_control: l: 6, 6, 7; t: 75, 50, 132.
No extra CFLAGS for BLE and entropy driver (pragmas removed), using
CONFIG_SIZE_OPTIMIZATIONS=y
Memory region Used Size Region Size %age Used
FLASH: 120124 B 256 KB 45.82%
[bt] [ERR] isr_rx_conn: assert: '!radio_is_ready()' failed
No extra CFLAGS for BLE and entropy driver (pragmas removed), using
CONFIG_SPEED_OPTIMIZATIONS=y
Memory region Used Size Region Size %age Used
FLASH: 138004 B 256 KB 52.64%
[bt] [INF] encode_control: l: 6, 6, 7; t: 61, 51, 130.
NB: RAM usage differences were insignificant.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
Now that log processing happens in a separate thread, the
BT_STACK_EXTRA macro is not needed (since there's no significant
overhead), and therefore the BT_STACK macros become unnecessary as
well.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Added implementation to avoid retransmitting NACK-ed Tx PDU,
to save on current consumption in retrying to transmit in
case peer device has no free buffer to receive the PDU.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
To allow the BLE stack to be used both in the real nRF platforms
and simulated ones, change the used macros in the code to the
COMPATIBLE ones.
Signed-off-by: Alberto Escolar Piedras <alpi@oticon.com>
Fix broken master role RSSI measurement. Since the original
contribution clean up into Zephyr, the radio shorts that was
set for measuring the RSSI for master role has been broken,
as it was cleared by the radio switching code further in the
Tx ISR.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Fix connection failed to be established regression
introduced by the commit 350c569aba ("Bluetooth:
controller: Avoid offseting to lldata").
As the Rx-ed PDU buffer is re-used to construct the
connection complete message towards HCI, the fields in the
Rx-ed PDU need to be backup for future use in the control
path. Here the channel selection bit is backup now.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Fix Connection Parameter Request Procedure implementation
to respond with sent interval_min and interval_max so that
certain peer devices dont reject the response as Invalid LL
Parameters.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactored the fix in the commit 685da02354 ("Bluetooth:
controller: Fix advertising random delay resolution calc")
to apply modulo in tick units before adding a tick.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Calling bt_recv in the Bluetooth host Tx thread by the
controller implementation caused deadlock in combined host
controller builds when HCI LE Create Connection Cancel
generated the HCI LE Connection Complete or HCI LE Enhanced
Connection Complete events.
Controller's HCI implementation has been updated to place
the generated event into Rx FIFO to avoid the deadlock.
Relates to commit a59f544fb4 ("bluetooth: controller:
Handle non-priority events correctly")
Relates to #10314.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
In commit d5836195d7 ("Bluetooth: controller: Increase advertising
random delay resolution"), the resolution of random_delay was
increased from 8-bit to 16-bit. Due to this switch the result
of HAL_TICKER_US_TO_TICKS() can now be a 0, which causes the following
crash:
***** Kernel OOPS! *****
Current thread ID = 0x200043f0
Faulting instruction address = 0x17914
Fatal fault in ISR! Spinning...
Let's make sure we don't pass a 0 to ticker_update() by increasing
the result of HAL_TICKER_US_TO_TICKS() by 1.
Signed-off-by: Michael Scott <mike@foundries.io>
Fix the default Tx buffers to 3. While the first Tx-ed
buffer generates the HCI Number of Completed Packets Event,
the controller needs to have 2 additional Tx buffers queued
so that the second Tx PDU has the More Data (MD) bit set so
as to have the connection event to continue to transmit any
additional Tx buffers that the Host will enqueue in the same
connection event.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Increase the resolution of advertising random delay from
1 ms unit to 1 us units. The controller scheduling will
floor it to nearest 32KHz clock unit on nRF5 series.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Under GNU C, sizeof(void) = 1. This commit merely makes it explicit u8.
Pointer arithmetics over void types is:
* A GNU C extension
* Not supported by Clang
* Illegal across all ISO C standards
See also: https://gcc.gnu.org/onlinedocs/gcc/Pointer-Arith.html
Signed-off-by: Mark Ruvald Pedersen <mped@oticon.com>
In order to avoid the build system complications that come from
including a timestamp, remove it by default from the version string in
HCI Vendor Extensions. Users can still include a unique identifier, be
it timestamp or not, using the CONFIG_BT_CTLR_HCI_VS_BUILD_INFO Kconfig
option.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
The Data Length Update implementation reused the flags used
by Encryption Procedure which caused invalid Encryption
Procedure sequence under conditions where Data Length Update
Procedure collide with Encryption Setup initiated by the
peer central device.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
There is an unnamed choice for the BT link layer selection.
Giving the choice a name would allow multiple declarations of the
option and the ability to select out-of-tree LL's.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
This allows to provide public address for controller without using
VS HCI command from host. Useful for controller only builds or
combined builds that are not using VS HCI commands.
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
When running combined build on nRF5 with disabled VS command it is
possible to simply read static random address from FICR in host.
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
The return of memset is never checked. This patch explicitly ignore
the return to avoid MISRA-C violations.
The only directory excluded directory was ext/* since it contains
only imported code.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
In the BLE controller, NRF radio HAL, for the PPI registers used
for the SW TIFS.
To allow easier testing, replace direct use of registers
from Nordic's nrfx MDK with accesses via its HAL inlined functions
Signed-off-by: Alberto Escolar Piedras <alpi@oticon.com>
Replase magic numbers with HCI Error Code definitions in the
LE controller implementation.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Updated the implementation of Connection Update Procedure to
not assert when peer master violates the Bluetooth
Specification v5.0 Vol.6 Part B Section 5.3 Procedure
Collisions. Instead disconnect the link with reason
Different Transaction Collision (0x2A).
Certain phones in the market perform Connection Update
Procedure and do not correctly handle remote initiated
colliding PHY update procedures. They try to perform both
the transactions involving an instant simultaneously
violating the Bluetooth Specifications.
Implementation in Zephyr is updated to gracefully handle
the violating remote master device, and not fatally assert
in the local device.
Relates to commit 8b3fd6963c ("Bluetooth: controller: Fix
assert on different transaction collision")
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Updated the implementation of PHY update procedure to not
assert when peer master violates the Bluetooth Specification
v5.0 Vol.6 Part B Section 5.3 Procedure Collisions. Instead
disconnect the link with reason Different Transaction
Collision (0x2A).
Certain phones in the market perform Connection Update
Procedure and do not correctly handle remote initiated
colliding PHY update procedures. They try to perform both
the transactions involving an instant simultaneously
violating the Bluetooth Specifications.
Implementation in Zephyr is updated to gracefully handle
the violating remote master device, and not fatally assert
in the local device.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Up until now, Zephyr has patched Kconfig to use the last 'default' with
a satisfied condition, instead of the first one. I'm not sure why the
patch was added (it predates Kconfiglib), but I suspect it's related to
Kconfig.defconfig files.
There are at least three problems with the patch:
1. It's inconsistent with how Kconfig works in other projects, which
might confuse newcomers.
2. Due to oversights, earlier 'range' properties are still preferred,
as well as earlier 'default' properties on choices.
In addition to being inconsistent, this makes it impossible to
override 'range' properties and choice 'default' properties if the
base definition of the symbol/choice already has 'range'/'default'
properties.
I've seen errors caused by the inconsistency, and I suspect there
are more.
3. A fork of Kconfiglib that adds the patch needs to be maintained.
Get rid of the patch and go back to standard Kconfig behavior, as
follows:
1. Include the Kconfig.defconfig files first instead of last in
Kconfig.zephyr.
2. Include boards/Kconfig and arch/<arch>/Kconfig first instead of
last in arch/Kconfig.
3. Include arch/<arch>/soc/*/Kconfig first instead of last in
arch/<arch>/Kconfig.
4. Swap a few other 'source's to preserve behavior for some scattered
symbols with multiple definitions.
Swap 'source's in some no-op cases too, where it might match the
intent.
5. Reverse the defaults on symbol definitions that have more than one
default.
Skip defaults that are mutually exclusive, e.g. where each default
has an 'if <some board>' condition. They are already safe.
6. Remove the prefer-later-defaults patch from Kconfiglib.
Testing was done with a Python script that lists all Kconfig
symbols/choices with multiple defaults, along with a whitelist of fixed
symbols. The script also verifies that there are no "unreachable"
defaults hidden by defaults without conditions
As an additional test, zephyr/.config was generated before and after the
change for several samples and checked to be identical (after sorting).
This commit includes some default-related cleanups as well:
- Simplify some symbol definitions, e.g. where a default has 'if FOO'
when the symbol already has 'depends on FOO'.
- Remove some redundant 'default ""' for string symbols. This is the
implicit default.
Piggyback fixes for swapped ranges on BT_L2CAP_RX_MTU and
BT_L2CAP_TX_MTU (caused by confusing inconsistency).
Piggyback some fixes for style nits too, e.g. unindented help texts.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
After a dedicated header is allocated to the simulated SOC versions
there is no need anymore to have guards in these other files
Signed-off-by: Alberto Escolar Piedras <alpi@oticon.com>
To avoid issues with differences between the simulated and the real
SOC let's separate the simulated one into its own header
Signed-off-by: Alberto Escolar Piedras <alpi@oticon.com>
Added implementation to check and wait for stable 32KHz
clock source before starting connectable/directed
advertising state and initiating state.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Fixed the implementation to generate command status as per
Bluetooth v5.0 specification instead of the incorrect
command complete that was generated before.
Also, the unsupported features status will be generated
before the invalid parameters status.
Relates to commit 258c7ccff1 ("Bluetooth: controller: Fix
HCI LE Set PHY invalid behavior check")
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
