In order that host can continue a connection event by
enqueuing packets, the number of completed packets event
needs to be generated as soon it arrives on air. Hence, the
Controller now calls the radio_event_callback on every radio
Rx complete ISR.
The callback executes at the Radio ISR priority, take care
to only do as little as necessary in the callback, say, just
a semaphore give to wake up the hci_driver's thread.
Change-id: If48afd0f1390d450bc1e7ec66f1c9fd45208d9a4
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
For the host to be able to block on the recv_thread while waiting for a
number of completed packets HCI event, those are pulled from the radio
in a new, higher priority thread that also schedules the execution of
the lower priority event/ACL data recv_thread.
Change-id: I9d356bd297d0504cb16a032fb5fe5530693546e2
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Since callers of bt_recv() have so far anyway been required to know in
which context to call it (based on e.g. bt_hci_evt_is_prio) it's
cleaner to have two separate APIs: bt_recv and bt_recv_prio.
Change-Id: Icd0d9aed9c51ffd2def31432c4ffcc16a9f13ccd
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Advertising reports are the only HCI events which we can drop if we
are low on buffers. Allocate them therefore with K_NO_WAIT rather than
K_FOREVER.
Change-Id: I0b7c92647f9be54b8746da837037725f8161a452
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Since the controller already uses a thread for receiving data it makes
sense to enable the BLUETOOTH_RECV_IS_RX_THREAD option.
Change-Id: I927b20c1a0afaea8d000df28cc220a69ae817d59
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The controller uses K_FOREVER for the allocations so there's no
benefit in trying to use specialized pools (which exist to try to
guarantee availability of buffers for critical data).
Change-Id: I130f2c44a2f28af1284290e6a0b17dcba438862f
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This commit introduces a very light-weight and lock-less
scheduling of the Controller's deferred function calls
called the Mayfly.
Earlier work implementation used in the Controller had an
O(n) to schedule a function in a linked list that used IRQ
lock during modification of the linked list in the
Controller's ISR executions.
Mayfly is a compile time configurable matrix of queues
where an execution context-safe queue exists between two
execution context, one being the caller and the second
being the callee. Callee(s) are run in a software interrupt
but can also be run in an OS thread.
There are minor clean ups too in this commit related to
folder structure.
Change-id: I5ff44dcee6679d2f5ce9e8437d98d6c868782f3d
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Enable connection encryption setup in 3 connection
intervals. Peripheral will respond to Encryption Request
with Encryption Response in the same connection interval,
and also, will respond with Start Encryption Response PDU in
the 3rd connection interval, hence completing encryption
setup in 3 connection intervals. Encrypted data would be
transmitted as fast as in 3rd connection interval from the
connection establishment. Maximum CPU time in Radio ISR will
increase if this feature is selected.
Change-id: I16f2863fc2aaed624826505739519d4de1ac44c5
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Add Kconfig configuration to select Tx buffer size, this
value will be returned in the HCI LE Read Buffer Size
command response.
This configuration will allow lower Tx RAM usage when
larger PDU sizes are desired in Rx direction only.
Change-Id: I5106a448d78a0754c4b0f6fa61fd5dcacd87a67c
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Add Kconfig configuration and conditional compilation of
RSSI measurement during a connection.
Change-id: I5a2f23f76a7cbbd9569d53d31899a472bf39d3ee
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Add Kconfig configuration to conditionally enable Data
Length Update procedure support in the Controller.
This will save CPU time, flash and RAM, if this feature is
not desired.
Change-id: I4515c0c7cf9aeb333a289397ae3c9bac04a08e4e
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Add additional conditional compilation of code not needed
when LE Ping is disabled in Kconfig.
Change-id: Idab40b1371488e06a6f2009fb553f7655b7b86f3
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
The following readings were obtained after running the peripheral and
central_hr apps in qemu combined with the controller (hci_uart) on nRF51
and nRF52:
Main Stack 380
Idle Stack: 68
ISR stack: 532
Controller RX Stack: 388
HCI TX Stack: 516
The numbers set in this change provide a safety margin from the ones
measured empirically.
Change-id: Ice7ad7f081502e0ea1accf856a7937c0bf0783b2
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Add Kconfig configuration to conditionally enable LE Ping
feature in the Controller.
This will save CPU time, flash and RAM, if this feature is
not desired.
Change-id: I5fbbdbe8f45ac01c9b0d7b11e002a0d1db4d272e
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Move code that can be executed after radio tx/rx packet ptr
has been assigned in the radio h/w.
Change-id: I9c5a34ee6bb74c1265d7871bcdf93894e3e7b190
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Added code to profile radio ISR execution time and generate
an event, which can be used to populate a vendor HCI event.
This event and associated code is disabled, subsequent
patch will add kconfig to enable it.
Change-id: Ic3fa3e0f4e36829a22a25ffee039949eaae561a7
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Use BT_WARN instead of BT_ASSERT to display unknown rsp in
HCI layer.
Change-id: I63c792468d9c4768f69df73395ee026a03521704
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Peer controllers not supporting LE Ping feature will
respond with unknown response PDU, handle it internally in
the controller rather than exposing it to HCI layer.
Discovered that controller was passing the unknown rsp PDU
to HCI layer when LE Ping was send to Nexus 5.
Unknown response for slave feature request was forwarded to
HCI layer, during that implementation unknown rsp for LE
ping too was getting forwarded to HCI layer.
Change-id: I4396c482e5546d78239cf41d88728de996e48d7d
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Not all users are in an ISR context where we can't block, so give the
callers the freedom to choose if they want to block or not.
Jira: ZEP-1481
Change-Id: I19bd7e2df94c4eeb60886a17a78f872bd7bea887
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The execution time of k_sem_give() increased slightly recently, and
since it was being called from the radio ISR this had an impact in the
deadlines not being met.
This change moves the actual call to k_sem_give() to a lower-priority
job and thus out of the ISR to avoid its execution time affecting the
ISR's timing.
Change-id: I76c82df895c6daaffef52786b0c900ee15acb0aa
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
In preparation for the new Nordic MDK and nRF52840 IC support we switch
to using the SERIES config instead of the old "NRF52" macro that will be
deprecated in the new MDK.
JIRA: ZEP-1418
Change-Id: I563f025c9db9b7497116c5af23814d95c720f836
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
There are static initializer macros available for most kernel objects
which we should use whenever possible.
Change-Id: I496f4d05d26801eddd21fae53bdd4fcdc3246fe3
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Reduce the time limit for active clock startup and increase
the CPU usage time limit before hard real-time radio
transmit/receive.
Due to isr_wrapper overhead in ARM architectures, more time
was used before hard real-time radio transmission or
reception which was detected by controller implementation
and the controller aborted the radio transactions.
This commit permits more CPU utilization by the controller
before hard real-time radio transactions start.
Change-Id: Id976add80c70cabc753c43dfac6f6603588458d9
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
On CPUs like nRF51 which run at 16MHz, certain BLE control
procedure PDU processing take more CPU time than permitted
inside tIFS (150us). Current implementation of Data Length
Update procedure does not span over multiple connection
interval (unlike Encryption Setup, which is another control
procedure processing that would consume more CPU time)
hence taking more CPU time inside tIFS on nRF51.
During the radio ISR, the active clock and packet timer are
active and it is used to profile the CPU time taken which
is used to decide on whether there is sufficient time in
the current radio event to process the control packet.
This commit also fixes a potential bug that would cause
disconnection due to MIC failure on encrypted connections
that performed Data Length Update. Controller used to NACK
the request/response PDU if it was not in a state to resize
the receive buffers but did not reset the CCM counter. This
is now fixed by the change done to NACK control PDU based
on available CPU time in radio ISR.
Change-id: Id58322ad76a0dbc284738cdd9a7c0437c9e8c423
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Instead of having custom logic for determining the minimum of two
values, use the existing min() helper from misc/util.h.
Change-Id: I9809883d4a31126329373f293897dd49eb91e9ad
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When the Controller is configured to support less than 251
as the supported maximum data length, missing check caused
the Controller to incorrectly re-initialize its rx buffers
causing assert during the DLE procedure. This commit fixes
the procedure to correctly use the supported maximum length
if the peer requests a transmit length that exceeds the
supported receive length.
Change-id: I6ad7196e3db44b303ddf2ec06e0ae579bf2eb774
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Correct the local LE supported features bitmap to actually "or"
together all the bits that correspond to the set of features that
are implemented at this time.
Change-Id: I0c62ec566c775514250fcf062aeef6c9656719e3
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Implement the 2 HCI commands that provide ECC public key
and shared secret generation:
- LE Read Local P-256 Public Key
- LE Generate DHKey
Jira: ZEP-1246
Change-Id: I79388bfdb9f2e28b9377b4bb6ee2caca25f33f3e
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Implement the 3 missing HCI commands required to support
Data Length Extensions:
- LE Read Suggested Default Data Length
- LE Write Suggested Default Data Length
- LE Read Maximum Data Length
Note: Only octets are actually used at this time, not time.
Jira: ZEP-1246
Change-Id: Id76d8fedb5ecaf0001c8429cf22f9a3e2c910a44
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Fix Coverity, sizeof not portable, defect; by explicitly
using sizeof(void *).
suspicious_sizeof:
Passing argument mem_head of type void ** and argument 4U
/* sizeof (mem_head) */ to function memcpy is suspicious.
In this case, sizeof (void **) is equal to sizeof
(void *), but this is not a portable assumption.
Change-id: I4b4776466e16020876500feba0141985b8581017
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
External interrupts are indexed from value 16, wherein
0 to 15 are ARM cortex M exceptions. Fixed code in
_irq_is_priority_equal to fetch correct external
interrupt line ISR priority.
Change-id: I9cfd411480e78dfc9635e72d14df9d667a9d8400
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Coverity analysis discovered NULL pointer being
dereferenced when passing a auto variable. The variable is
now correctly assigned with address of a valid default
value variable. As per design, the dereferencing will not
happen as the master role does not use the passed parameter
only slave role uses it to prepare the connection parameter
request PDU.
Change-id: I3f8519b23a83cb8c50c7fba81810eff7737ff74a
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Coverity analysis discovered that observer filter policy
field was 1 bit, whereas valid range for extended scanner
filter policy feature implemented in controller is 0 to 3.
Increase the bit field size from 1 to 2.
Change-Id: Id4b2e354961dfb3b45f72fa4e0ab18de7425bbb5
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Added implementation for HCI Reset Command. Implementation
gracefully disables any running advertiser, observer, and/
or connection roles, and it resets controller context members.
The HCI Reset Command is implemented in such a way that
driver instances shared with other sub-systems and
application is not disturbed and instance/references used
by Bluetooth Controller are gracefully returned back.
Jira: ZEP-1282
Change-id: Ifb9ae6807736b5ec2d9f346cf2a590322056bcee
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
During the initial integration of controller to Zephyr OS,
radio hardware access was abstracted out into hal/radio.c
file. Bug introduced in hal/radio.c has been fixed so that
whitelist feature works again.
Change-id: Ie5faf80b1a008ef326613548a5a28a4ba52e7ef7
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
The conditional defaults must come last in order to be properly
processed.
Change-Id: Id7a152ca1a1584935029e212d0dd8f37494d1cf4
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add implementation in HCI and Controller to support
HCI_LE_Remove_Device_From_White_List, as it is listed as
mandatory under BT Spec. v4.2, Part E, Section 3.19 LE
Controller Requirements.
Change-id: Icef88dffc85746f3cc7adb7fb692ae5578274ed2
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Switch from fiber_start() to k_thread_spawn() and from NANOKERNEL to
POST_KERNEL init level.
Change-Id: I34fb11cbe20216c8646ebacb07be304a67e3cd0a
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Use the unified kernel API k_yield() instead of fiber_yield().
Change-Id: I8f52031f52f7ac8783033a51751dc22decdfa59a
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Currently the ideal buffer counts for a controller-only build are 6 RX
buffers and 7 TX buffers: ATT_MTU of 158 bytes can be sent in one
connection interval of 6 tx/rx -es wherein connection interval is
7.5ms
Change-Id: I64b4620c5e8e7db8d7ed72fa1db82e266e121f27
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Restructure the Bluetooth options more logically.
- Both host and controller are now behind the same high level
CONFIG_BLUETOOTH.
- Selecting controller support disables other HCI driver selection, so
the controller isn't in the same list as HCI drivers any more.
- Under the top-level there's a "Custom stack" option, which when
enabled opens up the option of choosing CONFIG_NBLE.
There are various other cleanups and simplifications in this patch as
well, since splitting these up would have been fairly tricky while
making sure all test cases still build.
Change-Id: I5bb715cb9d20201cb8b72fbd149c8a09a4b2d7d2
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When building Zephyr in the controller-only configuration,
assertions that happen in the Link Layer code are not visible
to the Host which is running on another HCI and connected via
UART or USB to it. This patch allows the Controller code
to output the assertion line number when in such a configuration,
allowing the Host to view the event to help debugging.
The event format used is temporary and will be replaced by a
standardized Vendor Specific specification to come at a later
time.
Change-Id: I013ca6783a3fdedc47b171132919dd4798c66285
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
The right convention for help text is tab + 2 spaces.
Change-Id: I2722a8b33f5f74be110dc43fbcecc12841f0db84
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When a peer controller does not recognize one of the LL control
PDUs received, it will issue an LL_UNKNOWN_RSP PDU to let the
peer know that it does not recognize the request.
The controller now handles this incoming PDU and completes the
procedure by issuing the appropriate HCI event in the case of
an LL_FEATURE_REQ and _RSP pair.
Jira: ZEP-1220
Change-Id: I7c04a346441f04deee41198daa6309c11ae1b571
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Remove hardcoded use of crystal as 32KHz clock source and
20 ppm accuracy; and use the values from config for sleep
clock source and sleep clock accuracy value in Bluetooth
Controller.
Change-id: I1c0d53ecf8ad158153d5186a6680b5eb03d1641b
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
During implementation of alternate encryption procedure
usable in nRF51, commited in
c41d3edda8, re-encryption
procedure was broken. This commit fixes the issue, and
now re-encryption should work on both nRF51 and nRF52.
Change-id: Ia41200f42b1d46e1f3f35ff44b582d4ffcc5f4fa
Signed-off-by: Vinayak Chettimada <vinayak.kariappa.chettimada@nordicsemi.no>
Move controller code from drivers/controller to
subsys/bluetooth/controller.
Change-Id: I73f675188485aa3267507bad7647796e593a3da0
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
