The nanostack hal's critical section uses a mutex for mutual exclusion,
which is nice for many use cases. But when one needs to use the critical
section from interrupts, the RTX will have a assertion failure and panic.
Add a configurable for mbed_lib, which can be used to enable a alternative
version of critical section, which uses the underlying OS primitives, which
disables the interrupts.
Note: the default behavior is not changed, one needs to override the
"nanostack-hal.critical-section-usable-from-interrupt" to have "true".
Reason for this change is that there is a need for sending events using
nanostack event queue from interrupt context, eg. from a socket callback.
Enables intrinsics for bit operations such as ctz, popc, and le32
conversion. Can be disabled to help debug toolchain issues. Has
proven to be surprisingly useful.
Much like the logging functions, assertions can be enabled, disabled
and forced with the littlefs.enable_assert config option. Integrates
with the new LFS_ASSERT macro.
The thread flag signaling mechanism is problematic if a separate event
loop thread is not used. The problem is, that one needs to know the
thread id of the loop dispathcer, and that knowledge can be derived
only from the caller thread of the initialization function.
Remove the magic connection from caller thread to dispatcher thread
by using event flags instead of thread flags on signaling. As the
event flags require the massive amount 20 bytes of RAM whereas
thread flags requires none, keep the code behind flag.
The separate eventloop thread may not be necessary on all uses, as one
can use the existing main thread for event dispatching. Add a
conditional nanostack-hal.event-loop-dispatch-from-application, which
disables the thread creation.
Note: the ns_hal_init must be ran from the same thread which will be
used to execute the event loop later.
Before this patch, when calling lfs_mkdir or lfs_file_open with root
as the target, littlefs wouldn't find the path properly and happily
run into undefined behaviour.
The fix is to populate a directory entry for root in the lfs_dir_find
function. As an added plus, this allowed several special cases around
root to be completely dropped.
This patch changes the way data is put in the TX_DATA_PORT register when
sending packet over Ethernet.
When this driver is compiled with release compilation profile
(space optimization compiler options) with Arm compiler version 5,
the line:
SMSC9220->TX_DATA_PORT = *pktptr;
generates the assembly instruction to get the pktptr pointed value:
LDM r2!, {r3}
with pktptr = r2
However, the code does not prevent the pktptr value from being unaligned
(to a 32 bits boundary) in that zone and the LDM instruction causes a
HardFault if this is the case. When the compiler option is not activated
(debug and develop compilation profiles), the compiler generates LDR
instruction instead which does not cause a HardFault.
The ARM v7-M states page B3-601: "Unaligned load-store multiples and
word or halfword exclusive accesses always fault."
To face that problem, we check if the data pointer is aligned or not. If
it is, we apply the same algorithm than before. If not, a local variable
is created and we copy in it, byte per byte, the contents at the
unaligned pointer. However, it will impact performances adding 8
instructions (one LD and one ST for each copied byte).
Change-Id: I11f6e82ce5521960d2ecf499f718f76fec29c0b0
Signed-off-by: Hugues de Valon <hugues.devalon@arm.com>
This patch fixes a memory bug. `eth_arch_enetif_init` method call
would attempt to write to un-initialized area of memory.
Change-Id: I9881de71d58fa14db609fe3e24617a210b896471
Signed-off-by: Galanakis, Minos <minos.galanakis@arm.com>
These tests are build around gtest and gmock and solely run on a host; cmake is
used as a build system:
- cd features/FEATURE_BLE/tests
- mkdir build
- cd build
- cmake ..
- make
- ./gatt-client-tests
These tests are build around gtest and gmock and solely run on a host; cmake is
used as a build system:
- cd features/FEATURE_BLE/tests
- mkdir build
- cd build
- cmake ..
- make
- ./gatt-client-tests
LoRaMacChannelPlan class provides APIs which are not usable for
PHY layer implementations who do not support custom channel plans.
So we had some code in APIs which was explicitely using magic numbers
for the channel mask. Although it turned out to be not a bug as a layer
down we were checking for custom channel support. However, we now
check for custom channel support before going deep into PHY layer that will
make the code run faster and we have done some cosmetics to the code for
readability.
Channel mask is manipulated with inline methods
The features/mbedtls/targets/TARGET_STM/* files include constant needed
for the error codes returned from the MD functions.
The features/mbedtls/targets/hash_wrappers.c provides thin redirection
layer for the hardware accelerated MD implementations that rely on the
old API.
The TESTS/mbedtls/multi/main.cpp has been changed to use the new API
as its build environment does not rely on the translation unit
containing the necessary wrappers.
Regression after ChanFS update: Due to parameter changes in the f_mkfs
function, the option to use a separate block for MBR (FDISK) was turned
back on. This should be off as it conflicts with an explicit MBR when
using the MBRBlockDevice.
The PHY layer had a lot of duplicated code in various geographic regions.
In this commit we have tried to concentrate all common functionaliy into
one single class which LoRaPHY that provides three kind of methods:
i) Non virtual base methods which are there for upper layer use, e.g.,
providing access to driver or generic PHY layer functionality which
needs to be exposed to upper layers.
ii) Virtual methods (no hard limit on implementation) that can be overriden
in derived classes. Some PHY implementations will need that as they may
come with very peculiar channel schemes, e.g., dynamic channel schemes
in US bands.
iii) Protected methods which are only available for the derived PHYs
We have adopted a mechanism for the dervied PHYs to announce their differenmtiating
parameters in their constructors by filling up a data structure known as lora_phy_params_t
which exists at base level. Access modifier for this data structure is protected so it can only be
used by the base or derived classes, i.e., no exposure to upper layers.
For extra functionality and differentiating controls, a derived PHY can override any virual method as necessary.
In addition to that we have adopted the Mbed-OS style guide and have changed data structures and code to reflect that.
Some data structures are removed.
* Algorithm to get alternate DR is modified. Current scheme, works as multiples of 6 as EU and EU like PHYs
provide 6 datarates. We make sure that we try a datarate at least once. If nuber of join retries is a multiple
of 6, we may try multiple times on each data rate.
* Most of the PHYs with dynamic channel plans, always override the above mentioned algorithm as the rules governing
this algorithm do not hild in their case.
* band_t data structure is enhanced with lower band frequency and higher band frequency. That enables us to validate
frequency based upon the band and hence we can have a single function for all PHYs to validate frequency.
* In some PHYs, there were some extra channel masks were defined which were not being used. Hence removed.
* EIRP table corrected in some PHYs based upon spec.
* PHY functions in response to Mac commands are renamed to reflect what they exactly do.
for example accept_rx_param_setup_req() because that's what they do. they can either accept
the mac command or reject it.# Please enter the commit message for your changes.
This commit also introduces API change for disconnect(). disconnect() will
now return LORAWAN_STATUS_DEVICE_OFF for successfull disconnect.
* LoRaWANStack::handle_tx() can be called with NULL buffer when length is 0.
This commit fixes the case where user has provided NULL buffer and length
is > max_possible_size.
handle_tx() now always returns LORAWAN_STATUS_PARAMETER_INVALID if given
buffer is NULL pointer and length > 0.
General error checking is added and some asserts are added for events.
In order to reset LoRaMac's state in disconnect, we need to implement
an API which can be used to cancel all outstanding requests and reset
LoRaMac's internal state to idle.
This commit introduces LoRaMac::disconnect() which can be used for
this purpose.
Instead of initiating own timer objects we can use EventQueue::call_in() method
as we already have handle to EventQueue object.
Also setting timeout and starting timer has been combined to TimerStart method.
We now save roughly 500 bytes by removing storage of default
parameters in the loramac_params_t data structure. We use Mib to
get default values from PHY whenever needed instead.
loramac_sys_arams_t now contains only the runtime values set during operation
whenever defaults are needed we directly query the PHY layer or via Mib as the
need maybe.
Baseline is changed to use a single set of data structures that simplifies the
code in the LoRaWANStack and Mac layer. We are now following certian rules for naming
data structures.
- All structures visible outside their domain are prefixed as 'lorawan_'
- All mac structures are prefixed as 'loramac_'
- All subsystem or module strucutures carry their name in prefix, like 'mcps_'
PHY layer still have legacy camel case data structures which will be entertained
later while we will be simplifying PHY layer.
Test cases are also updated with the new data structure naming conventions.
One major difference from the previous baseline is the removal of static buffer
from mcps indication. And we do not copy data from stack buffer to rx_msg buffer.
This saves at least 512 bytes.
It may look like now that if we have received something but the user have not read
from the buffer, then the buffer will be overwritten and we will lose previous frame.
Yes, we will. But the same will happen even if we would have copied the buffer into rx_msg
because then the rx_msg gets overwritten. So we decide to abandon copying the buffer at
multiple locations. We inform the user about reception, if the user doesn't read and
the data gets overwritten, then so be it.
Channel planning was distributed over LoRaWANStack and LoRaMac previously.
We now centralize it by allocating the service to its own class.
Thus making the workflow consistent, i.e.,
Request for channel plan = Application->Interface->stack->Mac->ChannelPlan
Major change apart from adding the channel plan subsystem are the API changes
in LoRaMac class.
As a part of MAC layer breakdown into independent subsystems, we have
introduced a class for MIB service.
Major changes from the baseline are:
- making OpenRxWindow() public in LoRaMac.cpp
- Moving various data structures to central protocol data structure
MAC layer will services will be broken down into independent
subsystems. This is the first of those efforts. We have introduced
LoRaMacMlme class that handles everything related to MLME subsystem or
subservice. To accomodate subsystems we have grouped all protocol level
variables into one big data structure. A pointer to that data structure
will be passed around the subsystems in order to regulate the entire
system.
LoRaMac::Send() and LoRaMac::SetTxContWave*() APIs are made public as
they are needed to be accessed by the subsystems.
Ticker objects embeded in TimerEvent_t data structure were getting constructed
even for the non LORAWAN builds. And that's what was bloating the builds.
We now lazy initialize them using Mbed-OS utility clas SingletonPtr.
A central data structure has been created that carries all the protocol level
variables for the Mac layer. This is important as we are going to break down
MAC services into subsystems and we will pass around common data using that data structure.
Time handler class had a c style callback attached to it which
had been hampering us to be fully object oriented.
That particular callback is changed to Mbed Callback which is attatched
to a specific object hence allowing us to be fully object oriented.
LoRaWANTimer is now called as LoRaWANTimeHandler class as this class handles both
current time and timer functionalities.
Some refactoring on how LoRa objects are created was needed:
- LoRaWANTimeHandler object is created by LoRaWANStack and shares with LoRaMac and PHY.
- LoRaPHY object is now member of LoRaWANStack class instead of static variable in source file.
MAC layer is now a class rather than being a blob.
In addition to that Mac commands are now being handled in
a seperate subsystem (a class of its own). In future we
will do the same with othe sublayers of MAC like MLME, MCPS etc.
The drive behind this exercise is to make MAC and supporting layers
into an object oriented system.
Major bug fixes include:
- last join time inclusion in band parameters
- disabling rx2 window if we missed the slot already
- MLME uplink schdule hook
- nbRep according to spec
- maintaining datarate after successful joining
- suppressing MLME requests if MAC is in TX_DELAYED state
- Uplink dwell time verification
Some missing features are implemented. Details are as follows.
Support for LinkCheckRequet:
An application API is added, add_link_check_request() to delegate a
request for Link Check Request MAC command.
* Application provides a callback function that needs to be called on reception of
link check response.
* Mac command is piggybacked with data frames.
This API makes the sticky MAC command stick with the application payloads until/unless
the application un-sticks the said mac command using remove_link_check_request() API.
Handling fPending bit:
If in the Downlink, we get the fPending bit set in fctrl octet,
we attempt to send an empty message back to Network Server to
open additional Receive windows. This operation is independent
of the application. An RX_DONE event is queued bedore generating
the said empty message. Specification does not mention what can be the
type of that empty message. We have decided it to be of CONFIRMED
type as it gives us an added benefit of retries if the corresponding
RX slots are missed.
Radio event callbacks as Mbed callbacks:
radio_events_t structure has been carrying C-style callbacks which was
inherited from the legacy code. These callbacks has now been changed to
Mbed Callbacks that makes sure that we can be object oriented from now
on.
Setting up user defined data rate was found broken maybe because of
some rebase issue. Code has been setting always the default data rate
and ignoring used defined values.
We had a lot of static objects which would get constructed and hence
pull in some of the LoRaWAN code into the builds for other technologies.
Such objects have been now lazily initialized using utility class
SingletonPtr.
Current implementation uses high resolution timers to calculate elapsed time.
This prevents for example deep sleep completely and causes unnecessary timer
events.
This commit changes implamentation to use EventQueue::tick() to get elapsed time.
All compliance test related codes are now inside LORAWAN_COMPLIANCE_TEST
build flag. This will reduce memory usage in when compliance test codes
are not needed.
TxNextPacketTimer callback was being used for testing only (compliance testing to be precise).
Now there are independent methods and direct calls to automatic timers for the
compliance testing so there is no particular need for this timer anymore.
Application should be able to add some optional callbacks if it needs to.
Ofcourse there is a penalty of 8-12 bytes per callback, but there can be
certain needs of the application that needs to be met for example setting
up a link check request etc.
We have introduced a structure that contains callbacks for the application use.
- 'events' callback is mandatory, user must assign it. Because this callback brings
state changes for the application. We cannot segregate this into individual handlers
because of RAM penalty.
- Other calbacks (none of them are implemented yet are optional).
Example of using the API is provided with doxygen
Receive API should return the length of data written to the user buffer
as the Posix APIs suggest rather than sending the pending length of data
back.
That has actually been a typo mistake which actually wnt in even with doicumentation :)
The EventQueue thread in LoRaMac.cpp is disbanded and the LoRaWAN
protocol is redesigned to store a pointer for an application
provided EventQueue. It means that now the stack runs in the
same thread as application. Application provided EventQueue is used
to defer ISRs from radio driver and timer callbacks as well as the
application events are queued to the same event loop.
This class is the doorway for the user application into the
Mbed-OS implementation of LoRaWAN protocol. It implements LoRaWANBase
and hence would work with any stack implementation underneath, ensuring
seemless portability for applications.
It takes a pre-constructed object of LoRaRadio and delegates it in the
downward direction. Before calling connect() user must call initialize() function
in order to initialize stack and mac layers.
connect() APIs can be used to either provide relevent keys and connection method at
runtime or compile time (using Mbed config system).
enable_adaptive_datarate() and disable_adaptive_datarate() are used to turn on/off
automatic rate control. Otherwisem set_datarate() could be used to set a particular
data rate on the current channel.
set_confirmed_msg_retries() is valid only for CONFIRMED messages. It means that the stack will
retry for a given number of times before timing out.
set_channel_plan() and get_channel_plan() are used to set or get a particular channel plan.
These APIs are particularly useful in case of ABP (activation by personalization). Because
in case of OTAA(over the air activation), by default the stack takes in a CF List (carrier frequency list)
sent by the base station in conjunction with Network server. This list overwrites all user configured
channels or channel plan. set_channel_plan() can be used to set a single channel as well by setting the
parameter for number of channels to 1.
remove_channel_plan() or remove_channel() are used to remove a currently active channel plan or a specific
channel.
send() and receive() APIs follow posix design except the socket descriptor is replaced with port number here.
lora_event_callback() API is used to set a callback function from the application side which is used by the stack
to inform user of particular events like CONNECTED, DISCONNECTED, CRYPTO_FAILURE, TX_TIMEOUT etc.
LoRaWANStack class is our controller layer on top of our
current MAC and PHY layer. It provides services to an implementation
of LoRaWANBase class.
It is a singleton class owing to the fact that the mac layer underneath
is not a class object. Instead, it uses the MAC via setting mib, mlme, mcps
requests and getting responses back from the mac layer using confirmations and
indications.
In essense this class is a special handle for
mac layer underneath which is predominantly reference design based.
In future we may refactor the LoRaMac.cpp code to make it object oriented
and cleaner.
At one end, it binds the application selected radio driver with the PHY layer
and at the other end it provides services to upper layers handling the mac via
well defined APIs.
For proper selection of a PHY layer, user must use Mbed config system.
For this purpose an mbed_lib.json is provided which can be overriden by the
user defined mbed_app.json. By default the EU868 band is selected as a PHY layer.
User must set relevant keys for the selected connection mechanism.
The actual mac algorithms are being used as it is in the reference
implementation.
We introduce an internal class that starts a thread and constructs an event queue
to handle deffered calls from interrupt context for RTOS. The code base is
compatible with Mbed-OS 2 as well.
GetPhyEventHandlers() API provides mac callback funtions for PHY layer,
which are in turn delegated to radio driver from the PHY layer.
LoRaMacInitialization() is augmented with LoRaPHY parameter which let's
the MAC layer know which particular PHY layer is in use.
LoRaMacSetTxTimer() and LoRaMacStopTxTimer() are used when duty cycle is
off for testing purpose or to support custom application timers.
If the duty cycle is off, mac and phy layer work togather to figure
out the next possible transmission time.
LoRaMacCrypto APIs are provided which provide seemless integration of
mbedTLS into mac layer for cryptography. User application is supposed to
provide proper mbedTLS configuration file.
All other APIs are retained as it is.
LoRaPHY is the abstract class for the LoRa PHY layer which governs
the LoRaRadio and provides some common functionality to all regional
implementations.
We support 10 regions and every region comes loaded with default parameters.
These parameters can be changed by the Mac layer or explicitely by the stack
controller layer using APIs provided. This layer in essence detaches Mac completely
from PHY and provides more modular approach to the entire system.
Apart from class structure, the internal functionality is directly deduced from
semtech reference implementation that's why most of the internal data structures are
used on 'as is' basis.
In addition to that, the PHY layer provides APIs to control the LoRaRadio layer, i.e.,
the lora radio driver, ensuring that the radio is accessed from a single entry point.
A seperate data structure file is added which is common to PHY layers only.
All network interfaces for LoRaWAN protocol must implement this
class. In order to be compatible with Mbed-OS applications, any
implementation of this class must use the data structures and
Mbed-OS timers provided.
lorawan_data_structures may look repetitive but this is essential
as we have a plan to use a reference implementation for LoRaWAN mac
layer from Semtech. Some of the data structures provide seemless
transition from semtech implementation (as MAC layer) to the Mbed-OS
control layers above.
features/lorawan/lorastack is the placeholder for future items like mac and
phy layers. system/ will contain all the common bits.
All Mbed-OS drivers for LoRa radio devices must implement
this pure virtual class in order to be compliant with Mbed-OS
applications.
This class comes loaded with all necessary data structures.
The implementations of this class can come out of tree.
Rather than tracking all in-flight blocks blocks during a lookahead,
littlefs uses an ack scheme to mark the first allocated block that
hasn't reached the disk yet. littlefs assumes all blocks since the
last ack are bad or in-flight, and uses this to know when it's out
of storage.
However, these unacked allocations were still being populated in the
lookahead buffer. If the whole block device fits in the lookahead
buffer, _and_ littlefs managed to scan around the whole storage while
an unacked block was still in-flight, it would assume the block was
free and misallocate it.
The fix is to only fill the lookahead buffer up to the last ack.
The internal free structure was restructured to simplify the runtime
calculation of lookahead size.
We currently don't have a mechanism for selecting tests based on the
available ram/heap, so the best solution right now is to disable these
tests specifically for this target.
Documentation states that nsapi_dns_query_multiple returns the number of
addresses found on success - it was returning 0.
Overloads using SocketAddress are relying on the return value, meaning
those calls didn't work at all.
Fixes#5921.
- New API to control whether coap itself sends a next GET(block2) request or not
- Fixes error IOTCLT-2203 mbed-coap does not handle PUT or POST if they indicate a smaller block size preference
Only clear the USB read buffer when endpointRead is called. This
allows data to be read with endpointReadResult without also allowing
USB to transfer more data. Instead additional data is transferred explicitly
with a call to endpointRead.
If a SETUP packet arrives shortly after an IN then the packets will be
processed in the wrong order - SETUP first then IN. This causes
the subsequent control transfer to fail. Fix this problem by processing
IN packets before processing SETUP packets.
Make the following improvements and fixes:
1.
Update the Kinetis USB driver so that endpointReadResult only reads the
result of the last read and does not trigger a new read. Instead
move the code to trigger new reads into endpointRead.
2.
Fix the race condition in controlIn caused by a call to
EP0read() followed immediately by EP0readStage(). This is done by
setting up to read the next setup packet (ignoring the status stage)
in endpointReadResult rather than in EP0readStage. This makes the
function EP0readStage unnecissary.
3.
Remove the Kinetis workaround in controlOut in USBDevice.cpp since
point 2 fixes this bug. For more info on this see the PR which
added this workaround - https://github.com/ARMmbed/mbed-os/pull/414
The ISTAT register is write 1 to clear. Because of this ORing this
register with itself fill clear all bits that are set. This patch
updates the code to use plain assignment so only desired
bit is cleared.
Attach callbacks with the assignment operator rather than with the
deprecated attach function. This fixes deprecation warnings.
This patch also adds the ability to attach a Callback directly.
The local static initialization in USB was causing multiple problems:
-Configurable descriptor values are set only once
-USB descriptor initialization causes a trap in debug builds since
this is acquiring a mutex in an interrupt handler
-Extra ram used since all descriptors are in RAM
This patch fixes these problems by making fixed descriptors
static const so they are stored in flash and never need to be
initialized and by making descriptors that do change a member
of the class so they are always initialized when requested
rather than once though lazy static local initialization.
This PR implement https://tools.ietf.org/html/rfc7959#section-2.5
"The error code 4.13 (Request Entity Too Large) can be returned at any
time by a server that does not currently have the resources to store
blocks for a block-wise request payload transfer that it would intend
to implement in an atomic fashion. (Note that a 4.13 response to a
request that does not employ Block1 is a hint for the client to try
sending Block1, and a 4.13 response with a smaller SZX in its Block1
Option than requested is a hint to try a smaller SZX.)
Add support for recently introduced NSAPI generic socket options to
control group membership. Previously applications using Nanostack would
have had to use Nanostack's native socket options.
* replace invalid using statement by typedef
* replace SecurityIOCapabilities_t by io_capability_t in send_pairing_response
* fix failure type in cancel_pairing
Such processing should be handled at the upper layer.
Initiating encryption of a given connection should be made with the start
encryption command. This command accepts the parameters Rand, EDIV and
LTK.
A request of encryption request generate an LTK request event that is
forwarded to the upper layer which then reply with the LTK presents in the
Security DB or reject the request.
Convert most functions of SecurityManager into virtual pure functions.
Protect access to the get_event_handler function and remove function set_app_event_handler.
deepikabhavnani did the hard work in tracking this issue down. Block
addresses are not cast to the correct type until after multiplying to
convert to byte addresses. This results in an overflow when the storage
is larger than 4 GB.
FATFilesystem declares sector count and size as uint32_t and block
device class arguments are addr and size which is uint64_t
While passing arguments to program/read/write API's of block device,
multiplication of uint32_t*uint32_t was not typecasted properly to
uint64_t which resulted in MSB truncation.
Eg. If block 0x800000 is accessed with block size 0x200, addr to be
passed (0x800000*0x200)0x100000000, but actual address passed was 0x0
which resulted in over-writting the root directory, and hence corrupted
filesystem
In the open call, the LFS_O_TRUNC flag was correctly zeroing the file, but
it wasn't actually writing the change out to disk. This went unnoticed because
in the cases where the truncate was followed by a file write, the
updated contents would be written out correctly.
Marking the file as dirty if the file isn't already truncated fixes the
problem with the least impact. Also added better test cases around
truncating files.
I added the function declarations of Ethernet functions that have a WEAK attribute. Although several Ethernet functions was called in rza1_emac.c, GR-LYCHEE don't have Ethernert feature. But there may be case that GR-LYCHEE uses LWIP feature.
In this case, since GR-LYCHEE will occur the build error, I addressed the error by defining the functions with a WEAK attribute. For reason of WEAK attribute, there is no influence in GR-PEACH and VK_RZ_A1H that have Ethernet feature.
A COMMON folder allows code reuse across different test cases. This
avoids code duplication or code enterying the application space.
The COMMON folder is uppercase to match naming conventions in Mbed OS.
I modified the debug message when using LWIP in RZ/A1 related mbed boards.
In eth_arch_enetif_init(), sys_thread_new() was called and task name is appeared as debug information, but task name for debug was a mistake.
For LWIP communication speedup in RZ_A1 related, I changed the below macro value and added the definition processing in RZ/A1 related header file(lwipopts_conf.h). For this reason, those macros are overrode by RZ/A1 related values, not default values.
In the below "features/FEATURE_LWIP" folders, same as Cortex-M targets, I changed the folder structure to combine files that can be shared as RZ/A1 related. And I renamed the folder name to "TARGET_RZ_A1XX" in order to make commonality explicit.
- "features/FEATURE_LWIP" folder
<before>
\features\FEATURE_LWIP\lwip-interface\lwip-eth\arch\TARGET_RZ_A1H
\features\FEATURE_LWIP\lwip-interface\lwip-eth\arch\TARGET_VK_RZ_A1H
<after>
\features\FEATURE_LWIP\lwip-interface\lwip-eth\arch\TARGET_RZ_A1XX
Call underlying HAL implementation to enter critical section/disable interrupts
before incrementing the global critical section counter.
Modify HAL implementations to track first entrances to the critical section and
only update the saved state on first enter.
Take SHA1 for example, without the fix, SHA H/W is not stopped in either case:
(1) ctx->total == 0 in mbedtls_sha1_hw_finish()
(2) mbedtls_sha1_hw_finish() is not called by upper layer
For example, even though MBEDTLS_SHA512_C is disabled (via #undef MBEDTLS_SHA512_C),
mbedtls_sha512_context is still necessary due to referenced in sha512.h.
Currently, trng_api.c is located in targets/ and AES/DES/SHA alter. are located in mbedtls/.
They have shared crypto code.
If they could locate at same location e.g. mbedtls/, the shared crypto code placement would be more reasonable.
Add counter to track crypto init among crypto sub-modules. It includes:
1. Enable crypto clock
2. Enable crypto interrupt
As counter gets zero, crypto clock is disabled to save power.
1. aes.h/des.h/sha1.h/sha256.h/sha512.h includes config.h before aes_alt.h/des_alt.h/sha1_alt.h/sha256_alt.h/sha512_alt.h.
2. aes_alt.h/des_alt.h/sha1_alt.h/sha256_alt.h/sha512_alt.h should not be included in any other location.
3. Just include aes.h/des.h/sha1.h/sha256.h/sha512.h in aes_alt.c/des_alt.c/sha1_alt.c/sha256_alt.c/sha512_alt.c.
1. aes.h/des.h/sha1.h/sha256.h/sha512.h includes config.h before aes_alt.h/des_alt.h/sha1_alt.h/sha256_alt.h/sha512_alt.h.
2. aes_alt.h/des_alt.h/sha1_alt.h/sha256_alt.h/sha512_alt.h should not be included in any other location.
3. Just include aes.h/des.h/sha1.h/sha256.h/sha512.h in aes_alt.c/des_alt.c/sha1_alt.c/sha256_alt.c/sha512_alt.c.
This was a small hole in the logic that handles initializing the
lookahead buffer. To imitate exhaustion (so the block allocator
will trigger a scan), the lookahead buffer is rewound a full
lookahead and set up to look like it is exhausted. However,
unlike normal allocation, this rewind was not kept aligned to
a multiple of the scan size, which is limited by both the
lookahead buffer and the total storage size.
This bug went unnoticed for so long because it only causes
problems when the block device is both:
1. Not aligned to the lookahead buffer (not a power of 2)
2. Smaller than the lookahead buffer
While this seems like a strange corner case for a block device,
this turned out to be very common for internal flash, especially
when a handleful of blocks are reserved for code.
This bug was a result of an annoying corner case around intermingling
signed and unsigned offsets. The boundary check that prevents seeking
a file to a position before the file was preventing valid seeks with
positive offsets.
This corner case is a bit more complicated than it looks because the
offset is signed, while the size of the file is unsigned. Simply
casting both to signed or unsigned offsets won't handle large files.
The API is as per posix standard, but does not provide stats for file/directory.
Stats buffer (block size, total block count, free block count) is filled for
entire mounted filesystem.
Merged from the existing TARGET_DISCO_F413ZH code (which has the same MCU)
and TARGET_NUCLEO_F412ZG code (which has the same Nucleo-144 board and
pinout).
Update the code to check all addresses for all interfaces. Move
the code from mbed_lwip_socket_bind() to a new function called
mbed_lwip_is_local_addr()
Return NSAPI_ERROR_PARAMETER when:
Binding to a non-local address
Socket listen() is called without calling bind() first
Socket accept() is called without calling listen() first
The K64F Ethernet driver installs an interrupt handler that sets thread
flags, and this could be called before the thread was initialised, so it
would use a NULL thread ID.
This triggers an RTX error-checking trap in debug builds, and could also
lead to other problems with received packets not being processed.
Adjusted so the RX interrupt handler does nothing if the thread isn't
initialised yet, and manually trigger a RX event flag after initialising
the thread in case any interrupts were ignored.
An alternative would have been to implement eth_arch_enable_interrupts,
but this mechanism is not present in the EMAC world - drivers will have
to start returning interrupts in their power up.
Fixes#5680
When specifying a full channel mask in .lib and setting the channel
to zero, device does proper channel scan.
For Thread, channel must be specified for router.
Sleepy devices don't respect the channel settings but do a full channel
scan at the start.
+ Modified the channel mask print to use hex.
ExhaustibleBlockDevice.h
- Fix typos for consistent spelling.
ObservingBlockDevice.h
- Fix typos for consistent spelling.
ReadOnlyBlockDevice.h
- Fix typos for consistent spelling.
README.md
- Fix typos, mostly for branding.
DESIGN.md
- Make minor changes for consistent spelling and precise language.
SPEC.md
- Make minor changes for consistent spelling and precise language.
README.md
- Make minor changes for consistent spelling and precise language.
Not all devices have enough heap to fit a simulated heap block device,
however using a simulated heap block device is preferred if available
(reduced flash wear, faster testing).
Added MBED_TEST_SIM_BLOCKDEVICE for tests that only need a simulated
block device (wear_leveling + resilience), and added support for targets
that are known to have enough heap.
GattServer::write on Nordic's targets use sd_ble_gatts_hvx to send an
handle value Notification or Indication; This function can fail if the
connection handle is invalid or if Updates are not enabled for this
connection.
This patch workaround those limitations.
Need to avoid a TX packet error from messing up the TX FIFO. Since this API will return busy if a packet is in transmission anyways, this is acceptable.
Unfortunately there are several issues colluding to force the timeout
this high.
1. The tests push littlefs to the limits of how many errors it can
handle before failing. As a side effect this causes a massive amount
of debug/warn/error logging about the situation.
2. The logging can't be turned off for specific tests. Note: This might
change with the introduction of test-configs.
3. Logging is fixed to a baud rate of 9600. Previous testing was carried
out with a baud rate of 115200, which is the reason for the original
timeout.
Previously, send() was somewhat soft - it only ever made one send
call to the underlying stack, so it would typically take as much data
as would fit in the buffer, and only block if it was unable to write
anything.
This is not the intent of a POSIX socket/filehandle write. It should try
to send everything if blocking, and only send less if interrupted by a
signal:
- If the O_NONBLOCK flag is clear, write() shall block the calling
thread until the data can be accepted.
- If the O_NONBLOCK flag is set, write() shall not block the thread.
If some data can be written without blocking the thread, write()
shall write what it can and return the number of bytes written.
Otherwise, it shall return -1 and set errno to [EAGAIN].
This "send all" behaviour is of slightly limited usefulness in POSIX, as
you still usually have to worry about the interruption possibility:
- If write() is interrupted by a signal before it writes any data, it
shall return -1 with errno set to [EINTR].
- If write() is interrupted by a signal after it successfully writes
some data, it shall return the number of bytes written.
But as mbed OS does not have the possibility of signal interruption, if we
strengthen send to write everything, we can make applications' lives
easier - they can just do "send(large amount)" confident that it will
all go in one call (if no errors).
So, rework to make multiple sends to the underlying stack, blocking as
necessary, until all data is written.
This change does not apply to recv(), which is correct in only blocking until
some data is available:
- If O_NONBLOCK is set, read() shall return -1 and set errno to [EAGAIN].
- If O_NONBLOCK is clear, read() shall block the calling thread until some
data becomes available.
- The use of the O_NONBLOCK flag has no effect if there is some data
available.
The index field of FATFS_DIR does not encapsulate all the context
required to reposition the directory traversal. ChaN provides
f_rewinddir() but no directory seek, so rewind if necessary then step
through until the desired index is reached.
This patch includes:
* Documentation update.
* Code simplification and correctness; the value overloads based on the
size of the hrm counter has been removed in favor of runtime check which
is more correct.
* The control point characteristic has been removed since HeartRate value
byte does not support the accumulated energy expanded.
Generic implementation of the GAP class. It allows porters to have a working Gap
implementation by implementing the following abstraction layer primitives:
- pal::Gap: Adaptation for GAP related primitives.
- pal::EventQueue: simple interface to the inner event queue of the stack.
pal::SimpleEventQueue can also be used as an implementation.
- pal::GenericAccessService: Accessors to the Generic Access Service present in
the GATT server.
To help generic code, an interface of an event queue at the PAL level has been
added. Implementation can either rely on the event mechanism internal to the
stack or use the SimpleEventQueue implementation provided by this patch.
This interface expose the primitives needed to realize operations defined in
the GAP layer. Data types, event and function definitions follow closely HCI
commands and events defined in the Bluetooth specification.
This class and its components are not used by BLE API and never been used in the
last two years.
This patch deprecate all the components in GapEvent.h and exclude its definitions
from the documentation process.
Updated NanostackRfPhyEfr32 with a receive queue.
Cleaned up debug messages, re-added to non-threaded calls.
Removed debug print override
Removed tr_debug override
Removed normal-operation prints that could have timing implications if enabled
Removed dead NVIC code (and a couple of dead log outputs)
Variable length flag was lost during attribute settings,
so variable length GATT attributes should have been set
to the predefined maximum length.
This fixes issue #86.
Change-Id: Ia0cd236ecd903fdb9e62a21bffef57d1e63764b9
- The class has been documented with a detailed example and explanation of its
responsabilities.
- Members have been reordered and regrouped logically.
- Deprecated annotation has been added to relevant members.
- BLE::BLE, BLE::waitForEvent has been deprecated to match with their existing
description.
- Improve detailed description of the class.
- Add undefined params tparams and return documentation.
- Add pre and postcondition when valuable.
- Put the elements of the file in the ble.common group.
We currently set the lwIP pbuf pool size small - to 5 x 576-byte
buffers.
This is insufficient to hold a single DTLS handshake flight, so can
cause cloud client connections to fail. STM-based platforms are failing
handshake because of this. (K64F works because it doesn't use the pbuf
pool for reception, but lwIP does recommend drivers use the pbuf pool).
Not changing the default memory sizes here, as intended for a patch
release, but adding mbed configuration options to allow the numbers to
be adjusted for memory tuning in an application.
In a future minor revision, I would recommend increasing the default
PBUF_POOL_SIZE - we are well below lwIP's out-of-the-box default - and
offsetting by a reduction in MEM_SIZE for the drivers that don't use
PBUF_RAM.
The constructor doc incorrectly suggested that only short (16-bit)
UUIDs were accepted. The same doc also referred to properties
instead of characteristics. (And to "value length", which seemed to
be completely out of place in the context of the current code.)
Target of LPC1769 links to mbed LPC1768.
The PinNames.h has conditional compile for the pin names.
LWIP lpc17xx emac driver modified to allow LPC1769 target
This changes introduce a platform adaptation over ATT/GATT that can be implemented by porter.
Unlike the GattClient interface, the ATT/GATT adaptation is simple, follow closely the Bluetooth specification and won't change over time.
Implementation of the GattClient interface is realized by the class GenericGattClient which accept in input a pal::GattClient.
This change will also free design space once adopted by partners, addition to the GattClient interface won't require partner support.
From C++11 and beyond string literals must be seperated by space
so that they are recongizable as seperate tokens.
Context macro in PPPCellularInterface (CTX) has been causing issues
as it was not augmented with a space from a nearby string literal.
STM32F439xI-family MD5, SHA1 and SHA256 hardware acceleration
occasionally produces incorrect output (#5079).
Don't enable MD5, SHA1 and SHA256 HW acceleration on STM32F439xI-family
targets by default until issue #5079 is fixed.
Cellular example had build issues with IAR8, combination of define and string
as argument to send function resulted in above error. Typecasting to const char *
didn't help, hence replacing the define explicitly.
Commit f602c936 incorrectly started defining LWIP_DEBUG as either 0 or
1.
lwipopts.h locally was changed to use #if, but all existing users
continued to use #ifdef. Therefore defining it to 0 was accidentally
enabling it for quite a few users.
This would have pulled in some unwanted lwIP code, and affected some
drivers, eg k64f_emac.c allocating an extra 2K of stack.
Correct lwipopts.h to either define it as 1 or leave it undefined, and
change the #if tests back to #ifdef, so all are consistent.
When building with VisualGDB the compilation fails due to the lack of required include files Dir.h and File.h.
Compiler report:
1>------ Build started: Project: exported-mbed-STM32F429-nucleo, Configuration: Debug VisualGDB ------
1> FileSystem.cpp
1> mbed-os/features/filesystem/FileSystem.cpp: In instantiation of 'class Managed<mbed::File>':
1> mbed-os\features\filesystem\FileSystem.cpp(141,19): note : required from here
1>mbed-os\features\filesystem\FileSystem.cpp(130,7): error : invalid use of incomplete type 'class mbed::File'
1> class Managed : public F {
1> ^~~~~~~
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(39,7): note : forward declaration of 'class mbed::File'
1> class File;
1> ^~~~
1> mbed-os/features/filesystem/FileSystem.cpp: In member function 'virtual int mbed::FileSystem::open(mbed::FileHandle**, const char*, int)':
1>mbed-os\features\filesystem\FileSystem.cpp(141,19): error : cannot convert 'Managed<mbed::File>*' to 'mbed::File*' in initialization
1> File *f = new Managed<File>;
1> ^~~~~~~~~~~~~
1>mbed-os\features\filesystem\FileSystem.cpp(142,16): error : invalid use of incomplete type 'class mbed::File'
1> int err = f->open(this, path, flags);
1> ^~
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(39,7): note : forward declaration of 'class mbed::File'
1> class File;
1> ^~~~
1>mbed-os\features\filesystem\FileSystem.cpp(144,16): warning : possible problem detected in invocation of delete operator: [-Wdelete-incomplete]
1> delete f;
1> ^
1>mbed-os\features\filesystem\FileSystem.cpp(141,11): warning : 'f' has incomplete type
1> File *f = new Managed<File>;
1> ^
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(39,7): note : forward declaration of 'class mbed::File'
1> class File;
1> ^~~~
1> mbed-os\features\filesystem\FileSystem.cpp(144,16): note : neither the destructor nor the class-specific operator delete will be called, even if they are declared when the class is defined
1> delete f;
1> ^
1>mbed-os\features\filesystem\FileSystem.cpp(148,13): error : cannot convert 'mbed::File*' to 'mbed::FileHandle*' in assignment
1> *file = f;
1> ^
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(39,7): note : class type 'mbed::File' is incomplete
1> class File;
1> ^~~~
1> mbed-os/features/filesystem/FileSystem.cpp: In instantiation of 'class Managed<mbed::Dir>':
1> mbed-os\features\filesystem\FileSystem.cpp(153,18): note : required from here
1>mbed-os\features\filesystem\FileSystem.cpp(130,7): error : invalid use of incomplete type 'class mbed::Dir'
1> class Managed : public F {
1> ^~~~~~~
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(38,7): note : forward declaration of 'class mbed::Dir'
1> class Dir;
1> ^~~
1> mbed-os/features/filesystem/FileSystem.cpp: In member function 'virtual int mbed::FileSystem::open(mbed::DirHandle**, const char*)':
1>mbed-os\features\filesystem\FileSystem.cpp(153,18): error : cannot convert 'Managed<mbed::Dir>*' to 'mbed::Dir*' in initialization
1> Dir *d = new Managed<Dir>;
1> ^~~~~~~~~~~~
1>mbed-os\features\filesystem\FileSystem.cpp(154,16): error : invalid use of incomplete type 'class mbed::Dir'
1> int err = d->open(this, path);
1> ^~
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(38,7): note : forward declaration of 'class mbed::Dir'
1> class Dir;
1> ^~~
1>mbed-os\features\filesystem\FileSystem.cpp(156,16): warning : possible problem detected in invocation of delete operator: [-Wdelete-incomplete]
1> delete d;
1> ^
1>mbed-os\features\filesystem\FileSystem.cpp(153,10): warning : 'd' has incomplete type
1> Dir *d = new Managed<Dir>;
1> ^
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(38,7): note : forward declaration of 'class mbed::Dir'
1> class Dir;
1> ^~~
1> mbed-os\features\filesystem\FileSystem.cpp(156,16): note : neither the destructor nor the class-specific operator delete will be called, even if they are declared when the class is defined
1> delete d;
1> ^
1>mbed-os\features\filesystem\FileSystem.cpp(160,12): error : cannot convert 'mbed::Dir*' to 'mbed::DirHandle*' in assignment
1> *dir = d;
1> ^
1> mbed-os\features\filesystem\FileSystem.cpp(18):
1> mbed-os\features\filesystem\FileSystem.h(38,7): note : class type 'mbed::Dir' is incomplete
1> class Dir;
1> ^~~
1> mbed-os/features/filesystem/FileSystem.cpp: In instantiation of 'int Managed<F>::close() [with F = mbed::Dir]':
1> mbed-os\features\filesystem\FileSystem.cpp(162,1): note : required from here
1>mbed-os\features\filesystem\FileSystem.cpp(133,27): error : incomplete type 'mbed::Dir' used in nested name specifier
1> int err = F::close();
1> ~~~~~~~~^~
1> mbed-os/features/filesystem/FileSystem.cpp: In instantiation of 'int Managed<F>::close() [with F = mbed::File]':
1> mbed-os\features\filesystem\FileSystem.cpp(162,1): note : required from here
1>mbed-os\features\filesystem\FileSystem.cpp(133,27): error : incomplete type 'mbed::File' used in nested name specifier
========== Build: 0 succeeded, 1 failed, 0 up-to-date, 0 skipped ==========
If only buffer-full events and LF characters trigger the flush of
the send buffer then only line-based communication can be
implemented over the BLE UART Service. This patch extends the
service API by adding an explicit `flush` method to force sending
the buffer contents, thus enabling protocols with short (e.g.,
single character) messages.
Enable the compile-time option MBEDTLS_AES_ROM_TABLES in the mbed TLS
main config.h file in mbed OS. This option has the effect of labelling
the AES tables as 'const' so they are placed in ROM, which saves some
RAM space.
d65b6b0 Update unittests for nsdynmemlib API change (#71)
bc69b8b Disable CoAP duplicate message detection (#69)
ccb65f8 Change year 2017 to copyright (#68)
76490a7 Add option to join COAP multicast group (#67)
381d910 Register to multicast groups (#66)
dce323c Add transaction delete to CoAP service (#65)
feea33e Add option to select used socket interface (#63)
5a5c0f9 Merge pull request #62 from ARMmbed/coap_separation
0d26c00 Modifying file headers and Makefile to adapt from libcoap to mbed-coap
d323c3a Fixing unit tests based on new coap library
d1a3d25 Modifying Makefile and source file based on new coap library
git-subtree-dir: features/nanostack/FEATURE_NANOSTACK/coap-service
git-subtree-split: d65b6b0eb890be93f667debe8da83aa498021ccf
Merged lwip 2.0.2 stable path 1 from https://github.com/ARMmbed/lwip
Patch allows lwip to do autonomous address configuration for prefixes
that are not on-link.
ad7cf16 Add David's IPv6 improvements to CHANGELOG
af9d783 Fix (bogus) MSVC 2010 warning about uninitialized variable usage in ip6.c It's wrong because the variables are initialized during first loop iteration due to best_addr == NULL
68358d7 nd6: cull destination cache on router removal
7323fa1 nd6: some work on basic RFC 4861 compliance
10eb2ca ip6: improve source address selection
6c06ecd ip6/nd6: route using on-link prefixes, not addresses
9f1714d nd6: improve router selection
2486b41 netif: more ip6 state changes invoke status callback
519d809 nd6: fix Duplicate Address Detection
9f3c6dd nd6: check link status before sending packets
8c761a2 nd6: improve address autoconfiguration support
Some block devices (for example I2C EEPROM) can be erased at the byte
level. These aren't really block devices, but fall under the block
device API. For these devices, the fat filesystem needs to use a lower
bound on the block size. In this case we used 512 bytes is used since
it is already a standard.
This function provides a shortcut to reformatting a mounted filesystem.
Also, since this function is a virtual member of the FileSystem class,
the user does not need to know the underlying filesystem to reformat
the underlying storage.
Fixes errors reported in Github
- #4798
- Fixed Lwm2m blockwise data transfer (using Block1 option) with Leshan
- ARMmbed/mbed-client#512
- Fixed an ongoing blocktransfer (Block 1 option) interruption by out of order Block to return error 4.08 Request Entity Incomplete as specified in the coap specification.
- ARMmbed/mbed-client#511
- Fixed detected duplicate message duplications will not stop mbed-client
Some devices have RAM memory split into two sections.
This becames a problem for GCC based toolchains as they don't
support splitting .bss or .data sections into two memory parts.
When we run out of memory from .bss sections, allocating the stack
by malloc() allows it to be moved to .data section which might already
be in the second memory section. For example KW24D platform.
* Use _C flags at compile time in SHA to avoid compiling in unconfigured features
* Don't define ECP_SHORTWEIERSTRASS since it is part of the application's configuration
MAX32630FTHR gets BLE, uses timer peripheral 5 for HCI timing
libexactLE update for 610,620
Fixup hal_sleep and hal_deepsleep for 620
LEDn outputs now forced to open-drain
Warning #1300-D: inherits implicit virtual
- Adding the virtual keyword in the derived class prevents the warning
Warning #1-D: last line of file ends without a newline
- New line added at the end of file
Warning #997-D:
function "MeshInterfaceNanostack::initialize(NanostackPhy *)" is hidden
by "ThreadInterface::initialize" -- virtual function override intended?
- virtual keyword removed from "MeshInterfaceNanostack::initialize"
Warning #1300-D: inherits implicit virtual
- Adding the virtual keyword in the derived class prevents the warning
PPP is running close to the edge of its default thread stack size of 512 bytes. When it experiences an FCS error on the incoming data (for example. caused by character loss when the incoming serial rate is too high for it to process in time) it performs some additional work which overruns the thread's stack, hitting the OS "stack underflow" check. Increasing the PPP stack size to 768 bytes resolves this problem.
This is only an issue when multiple fatfss are used simultaneously.
Repeated use of a single fatfs instance (even with different storages)
do not show this issue.
Full support requires path prefixes being applied for every function
that takes a path. Note: this is only required filesystems after the
first mounted filesystem. The first filesystem has no penalty.
Antti Kauppila
Teppo Järvelin
Ari Parkkila
Anna Bridge
Deepak Venugopal
Cruz Monrreal
Donatien Garnier
paul-szczepanek-arm
Vincent Coubard
Christopher Haster
Amanda Butler
Bartek Szatkowski
Krzysztof Stachowiak
daid
Tero Jääskö
Kevin Bracey
David Saada
Rob Meades
Antti Yli-Tokola
Martin Kojtal
Teemu Kultala
Brendan McDonnell
Hugues de Valon
Galanakis, Minos
Veijo Pesonen
Hasnain Virk
Andrew Leech
Krzysztof Stachowiak
Kimmo Vaisanen
Jarno Lamsa
ccli8
Arto Kinnunen
andreas.larsson
Juha Ylinen
Kari Severinkangas
Wilfried Chauveau
Oren Cohen
adustm
bcostm
Jani Suonpera
Russ Butler
Russ Butler
Daniel Schuler
deepikabhavnani
Seppo Takalo
TomoYamanaka
Steven Cartmell
Tero Heinonen
Jimmy Brisson
gorazd
Adam Heinrich
Ashok Rao
Mika Leppänen
Ryan Kurte
Abdelhak Bougouffa
Steven Cooreman
Mahadevan Mahesh
jeromecoutant
Colin Hogben
Steven Cooreman
Laurent MEUNIER
Kevin Gilbert
Andrea Corrado
Gabor Kertesz
Chris Snow
Alexander Zilberkant
Kevin Gilbert
Akos Kiss
Juhani Puurula
Piotr Grygorczuk
Sarah Marsh
cyliangtw
Shrikant Tudavekar
Andres Amaya Garcia
Yuguo Zou
0x6d61726b
Jaeden Amero
Yogesh Pande
Adrien Chardon
Jesse Marroquin
Kari Haapalehto
Azim Khan