The newly introduced i2c_poll_transfer() is now used for register
read for PCAL9535A GPIO driver. The old write then read does not
work because every read or write ends with STOP. However,
the chip requires RESTART after writing the register address.
So the new generic transfer function is perfect for this.
Change-Id: I56d7ebe08f68cb04731c72138d60645ef124f65e
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
() For interrupt driven transfer, the driver already has
a generic transfer function, so this simply exposes it
for public consumption.
() The polling write is extended to support generic
transfer.
Change-Id: I657db8a6376282c723728b7a875813d389cf7d27
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Add an API to I2C for generic transfer. This is for those data
transactions which are more complex than simple read or write.
One example use of this is to read register data from a slave
device. The master has to write the address of the register
to slave, then start reading from slave for register data.
In this situation, simple read or write is no longer
sufficient, as for some slave devices, they require RESTART
when data transfer direction changes. Doing a simple write via
i2c_write() results in STOP at end of transfer, which results
in the slave not responding to read request as it is not
expected to send anything back.
Change-Id: I22f9c5ba2f7a7c48a33b9cb1aaf559e596a85d39
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
() Re-organize the steps involved in data transfer, according to
the datasheet's flowchart.
() Extract the common code for transfer initialization and put
it into its own function.
() i2c_write() and i2c_polling_write() are now using the common
data sending function.
Change-Id: Ieb90253ee10ddceb3b5d05b258e7fc6253d18729
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Polling in/out are basic UART functions that are supposed to
work on every UART controllers. Therefore, both poll_in and
poll_out are now mandatory. The API no longer checks if
they are defined before calling. This is to avoid the if
statements so to speed up things a bit.
Change-Id: I3bd1e7a2dfa5405699ebaec3eb34cfc0434b0073
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Both poll_in and poll_out are supposed to be implemented
in each driver, as these are basic function.
Change-Id: If1b6b11834f57934fe7e776bfcf818e088abdd0a
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Passing checkpatch check to use tabs instead of spaces for everything.
Change-Id: Ieffa0b0cc4b24598e7cbce7edf9ed640c8fb27dd
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
Moving the includes to include/tinycrypt. This will help make it
clear when looking at source files where each header originated from.
Change-Id: Ic79978da286f9c288868074a69262d89488925b8
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
XIP is currently implemented in Zephyr by copying only the RW
sections from flash into RAM. For these sections, if the VMA spacing
is different than the LMA spacing the sections in RAM will not be
at the right addresses. Warn about this situation.
Change-Id: I59a55fc27703103e37a7f0bbc34f236047621e0e
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
- Several recently added binary sections were not being taken
into consideration, unknown sections now trigger test case
failure
- Use same VMA/LMA terminology that objdump uses and show both
addresses
Change-Id: I641eef64aaed6612f62e5aa092f66baaa2797c70
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Bad things happen if an SW IRQ is asserted before it is initialized.
Change-Id: I4ad48480c4d7bb5fc14d778d2566d73086300fa1
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This commit breaks the sanitychecks
This reverts commit 93fa1053b21efe3d030ea060a31b80d6008d5bfd.
Change-Id: Ia52092e845c33234817a1cb27cd1f17064c65e7e
Signed-off-by: Anas Nashif <nashif@linux.intel.com>
Updated as per Allan's comments to correct use of Zephyr
kernel in lieu of codename variable
Change-Id: Iaf5f8531c5452b6d8fb13318f00050aa63d99200
Signed-off-by: L.S. Cook <leonax.cook@intel.com>
Editing language.rst file to clarify a few items
Also correcting a few things I missed earlier
Also fix title case without hyphen to be consistent
Change-Id: I463d1cdb436957a14af6ea40933f892a51b161aa
Signed-off-by: L.S. Cook <leonax.cook@intel.com>
Passing checkpatch check to use tabs instead of spaces for everything.
Change-Id: I2fdd507d48db83fda8d183043c5a4ab1b97148dc
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
Moving the includes to include/tinycrypt. This will help make it
clear when looking at source files where each header originated from.
Change-Id: I062cb3c6ecc5bc2ed2d28228d5926646b6b5f912
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
Fixes the tinycrypt library to use Zephyr checkpath acceptable line
lengths.
Change-Id: Ia7bef46a39fa029cda826ec5b2d7d42036d9bdfe
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
If the other tasks try to trigger a SW IRQ before MONITORTASK
has initilized it, the CPU generates an exception.
Change-Id: Id2745d728a7cf15132096585c073f85f8306cb38
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Needed to get simple uart driver working in qemu with
SLIP networking or qemu<->qemu network testing. Because
this is only needed in qemu the patch only touches ia32
platform.
This can conflict with Bluetooth in x86 and qemu so caveat emptor.
Change-Id: Iba20543f968c8fd37ee36747d9aa1ad3f782057d
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If bt_conn_security failed required security level should be set
back to current security. Otherwise it might not be possible to
try later on with lower security.
Change-Id: I3ad7151f2cdff15147b9948bbdab516b8cc16ce2
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
This fixes the bug that affected pairing as peripheral.
With this patch pairing flag is cleared and timeout timer stops
if all keys has been distributed.
Change-Id: If7575197d5a42a0ca5bb158ea5607934e92abb81
Signed-off-by: Mariusz Skamra <mariusz.skamra@tieto.com>
Rename bt_conn_connected to notify_connected. This will avoid
confusion with BT_CONN_CONNECTED state name. Same goes with
bt_conn_disconnected.
Change-Id: I6e9d50016036b30345068ef2a6bb556c45d7dd06
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Delaying only connected callback resulted in races where security
or pairing callback could be called before bt_conn connected.
Currently only procedure that relies on read features is connection
parameters update and that should not conflict with other traffic.
Change-Id: I835535abbff21ea8d86b728f5681f90ef9fd337c
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Print current number of free MAC buf count when allocating
or freeing mbufs.
Change-Id: I3c434671b28ac4b4d8c8d968899e891391d849d9
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If mbuf could not be allocated, then do not try to free it.
Change-Id: Idf48f9f2eda260b8560a5388c920a9d9a3ef8f16
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The stack usage was not printed often enough when receiving
and sending 802.15.4 packets.
Change-Id: I61a4b6acdc10cb1681e2aeb63cb6af7a35834dff
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The TX stack usage data and statistics were only printed
if there was an error. This happened very seldom which was
not very helpful.
Change-Id: I059cf9ee3449e5b39d872ef6026df968d86d0209
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If IPv6 packet length is incorrect, then print the actual
length values for debugging aid.
Change-Id: I20d4d73efd1e0103f670e8e53ba6417a4795ab80
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If one has 802.15.4 network and tries to compile without
6lowpan, then print warning as that combination is not a
usable network setup. It can be used for testing but not
for real live setup.
Change-Id: I3b540686c8556ed0306723834ca38633aff63165
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the fragmentation header is missing, then do not discard
the message but pass it to upper layers as is. The fragmentation
header might be missing if the packet fits into 128 bytes.
Change-Id: I39d29e7ad80be5531262a0ac413d8137da0c64f5
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
No need to compile sicslowmac.c if we do not have 6lowpan.
Change-Id: I8624089f4b360fb12ef1c4a6d7104f0b190f048c
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This collects received statistics from layer 2 (MAC).
Change-Id: Ie869c1209bd06f0a5ccad5094b0cafe5fe74d787
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This collects byte counts from layer 2 (MAC).
Change-Id: I16c25f7ae8b9aa485e3c5f1573154cd2270cff93
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The statistics counts were not initialized so the numbers were
mostly bogus.
Change-Id: If4c33701fbdc4d4603d708c69950a8a18f8eaeaf
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
There is no need to define keys types if SMP is disabled.
Change-Id: If6b653906df4623f440b258afd574722498109d0
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Gather central and peripheral related code togerther so it can be
placed under single blocks of ifdefs.
Change-Id: Ia5fe45b86173b0e647b582bf2cd96fc7c9cb7feb
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Gather client related code togerther so it can be placed under
single blocks of ifdefs.
Change-Id: I6f2f5a06c0308b885d01fc6af300b162baec0823
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Gather SMP related code together so that it can be placed under
single blocks of ifdefs.
Change-Id: I3c2e71243b694e9984ae8e88025a6ead7969e020
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
This fix a regression that cause identity address to be passed to
LE Create Connection HCI command. This was affecting connecting to
iPhone. Regression was introduced in beaf30fda7de1476c742a
(Bluetooth: Simplify check_pending_conn function).
check_pending_conn now takes two addresses: ID and real address.
ID is used to find connection object and real is used in LE Create
Connection HCI command. If no IRK is present both addresses have
same value. This is to avoid double lookup of ID address for every
advertising report.
Change-Id: I6b86a698087889c4e6c2a1cbc9a688a804cfeb84
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
This reduces number of idefs cuts and make code much easier to read.
In case of SMP not supported (or connection support since SMP depends
on it) find_id_addr is simplified and will most likely be inlined by
compiler resulting in no-op.
Change-Id: Ib16d1508c5664fcb58f619b090a5d62b2ecfe8c5
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
Added -e option to echo-client and it will not quit but just
send data forever to the echo-server.
Change-Id: I9085220844b5bc45b2b7fffc46fb1b9e93235307
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If we receive too long message in dummy radio driver,
then we need to discard the packet as we cannot handle
it anyway.
Change-Id: Ibed0ac5f1568dd791dec3e780c7f6c71914d02cd
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Made the test_15_4 application to run much faster when it is
running inside qemu (started with make qemu0).
Change-Id: I4455eca4f0426e8acc8b87d93eb80f4d179a1b48
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The 802.15.4 monitor application had issues with connecting
pipes together and making sure the first byte from qemu is
not lost. Before starting real data connection we send five
null bytes and the discard them in receiving side.
Change-Id: I376ae4f79aaa1ea742640999ec5b3d9c57a1279a
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>