Now that the u-blox driver can let cmd_handler know that it hasn't
received enough data, let's disable the hex mode for readying
binary data.
On the SARA-R4 this mode limits the receiving MTU to 512 which is
unacceptable.
Signed-off-by: Michael Scott <mike@foundries.io>
Some modem commands can determine if they have enough data pulled from
the modem to continue or not. Let's allow those functions to return
EAGAIN which means there's more data needed from the modem.
Signed-off-by: Michael Scott <mike@foundries.io>
Modem commands are setup with arg_count to denote the number of params
to parse before returning. Let's honor that setting and return once
we've parse them.
This fixes a run-on bug where the parser would find as many of the
parameters via the supplied delimiter as it could.
Signed-off-by: Michael Scott <mike@foundries.io>
There is a blue LED used to display connectivity for the SparkFun
SARA-R4 shield. This LED is connected to the modem via GPIO 16
which is the default used by SARA-R4 driver when
CONFIG_MODEM_UBLOX_SARA_R4_NET_STATUS is enabled.
Signed-off-by: Michael Scott <mike@foundries.io>
The original Ublox SARA-R4 driver was written for the net_context
APIs. As a result many of the return values are not POSIX standard.
Let's go through all of the socket offload functions and make sure
we return standard values (0 or -1) and set errno where appropriate.
Signed-off-by: Michael Scott <mike@foundries.io>
SARA-R4 has a max read limit of 512 when using binary hex formatting.
Otherwise the max read limit is 1024.
Signed-off-by: Michael Scott <mike@foundries.io>
New socket functions were introduced to hide some of the modem_socket
internals:
- modem_socket_next_packet_size()
- modem_socket_wait_data()
- modem_socket_data_ready()
Let's use them.
Signed-off-by: Michael Scott <mike@foundries.io>
Let's use sock->is_connected to check whether we need to clean up
the modem's socket. It may have already been closed via URC.
Signed-off-by: Michael Scott <mike@foundries.io>
SARA-R4 AT command manual states +USOCO has a max timeout of 120s.
Let's use that instead of the default 10s command timeout.
Signed-off-by: Michael Scott <mike@foundries.io>
The SARA R412M requires manual activation of the PDP context. It also
reports RSSI instead of RSRP when on 2G. An off-by-one in the RSRP
calculation was also fixed in this commit.
Signed-off-by: Göran Weinholt <goran.weinholt@endian.se>
The u-blox SARA modems have the ability to output the network status
on a GPIO pin. This can be used to light up a LED when the modem is
connected to the network.
Signed-off-by: Göran Weinholt <goran.weinholt@endian.se>
The U2 modem will sometimes not power off even if we have given it the
signal to do so. We now signal it continually until the modem
indicates that power is off with VINT=0.
Signed-off-by: Göran Weinholt <goran.weinholt@endian.se>
This fixes two problems in modem_reset():
- mdata.net_iface is used without checking that it is valid.
Now, we validate mdata_net_iface before usage, and give a warning
if it was not set.
- if the modem does not perform network registration within the
given time of 20s, in the previous implementation the driver
gave up.
Now, the driver disables RF for a second, enables it again and
gives the modem another 20s to register. This has been seen
to help in roaming situations.
Tested on Sara R4.
Signed-off-by: Hans Wilmers <hans@wilmers.no>
- limit max. number of bytes when sending to socket
The number of bytes sent to a socket in one transaction
is limited to 512 in HEX mode (Sara-R4), and to 1024
otherwise. This corresponds to numbers given in the
manual for ublox cellular modems.
- report number of bytes actually sent, as reported by modem
After writing data to a socket, we now return the number of
bytes actually written, as reported by the modem.
Signed-off-by: Hans Wilmers <hans@wilmers.no>
In order to still process 128 bytes at a time, let's add 1 to the
length of match_buf which will be used to store a NUL char when
the match_buf is at max size (size - 1).
Otherwise we're processing 127 bytes at a time which is inefficient
compared to the data buffer sizes in net_buf.
Signed-off-by: Michael Scott <mike@foundries.io>
SARA-R4 modem supports offloaded DNS via AT+UDNSRN command.
Let's implement it.
NOTE! On SARA-R4 a new firmware *IS* required to support this feature:
L0.0.00.00.05.08 [Apr 17 2019 19:34:02]
Signed-off-by: Michael Scott <mike@foundries.io>
Managing the interface up / down events from driver code (as opposed
to L2 layers) has been an issue for quite some time. There are
race conditions which result in data aborts (referencing NULL
iface or NULL iface->if_dev) or ends up breaking ASSERT checks for
non-NULL iface.
Let's remove the handling for now and come up with something better
when a user actually needs it.
Signed-off-by: Michael Scott <mike@foundries.io>
In offload_recvfrom(), if we end up calling modem_cmd_send(),
then we should always wait for the results. We've already
checked for MSG_DONTWAIT and made sure that there is data ready.
We just need to get it from the modem buffer at this point.
Signed-off-by: Michael Scott <mike@foundries.io>
Users may be waiting for data via socket recv() function. We
notify them that data is ready when URC is received from modem.
Once unblocked, we read the data from the modem buffer which
is handled via on_cmd_sockread_common(). At this point, we
don't need to unblock waiting users again.
Signed-off-by: Michael Scott <mike@foundries.io>
The hex_to_binary() function is incorrectly assuming all buffers
passed to recv() will have an additinal byte for storing a NUL
terminating char. This should be optional as MQTT library uses
exact sized buffers for parsing socket data.
Signed-off-by: Michael Scott <mike@foundries.io>
The send_socket_data() is incorrectly sending 0 as successful result.
The socket APIs require the number of bytes sent to be returned.
Signed-off-by: Michael Scott <mike@foundries.io>
For parsing purposes we need to add a NUL to the end of match_buf.
When there is no CR/LF in the incoming rx bufs then we fill match_buf
to its max size. This ended up with an off by one error which was
overflowing match_buf into the following data.
To account for this, let's fill the buffer to size - 1 so that we
leave room for the NUL at the end and stop corrupting data.
Signed-off-by: Michael Scott <mike@foundries.io>
- rename buf_len to match_len for clarity
- pass in modem_cmd_handler_data instead of buffer pointer directly
- buffer adjustments for the command length are kept inside
parse_params()
Signed-off-by: Michael Scott <mike@foundries.io>
Let's add a field which the drivers can use to keep track of whether
they are connected or not. This will normally be enabled / disabled
in the socket connect and URC for socket close notify.
Signed-off-by: Michael Scott <mike@foundries.io>
Let's hide the internals of sock->packet_sizes[] by adding a function
which returns the size of the next waiting packet.
Signed-off-by: Michael Scott <mike@foundries.io>
Let's hide the internals of the modem_socket's sem_data_ready and
poll handling with 2 new functions:
- modem_socket_wait_data: take a semaphore and wait for data
- modem_socket_data_ready: give back the data ready semaphore and
unblock poll() users
Signed-off-by: Michael Scott <mike@foundries.io>
Add lock behavior for functions in modem_socket, to prevent race
conditions when performing socket data maintenance.
Signed-off-by: Michael Scott <mike@foundries.io>
Let's allow protocols to request portions of the current packet.
This is seen in the MQTT library when parsing headers for the type
and variable length of the packet.
This fixes basic parsing done in the MQTT library and probably others.
Signed-off-by: Michael Scott <mike@foundries.io>
Data transmission was paused when PHY update request control
PDU was enqueued in ULL. If there was pending data PDU in
ULL that was not enqueued towards LLL, this caused
transmission to stall.
Move the tx pause due to PHY update request/response being
enqueued to pre_tx_ack callback, this way all pending PDUs
in ULL is enqueued to LLL.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
The DT_FLASH_DEV_NAME define used by the sample was set to a define
symbol that was never generated. Change it to use
DT_INST_0_ATMEL_SAM0_NVMCTRL_LABEL on the sam0 family of SoCs that
had this issue.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
GitHub checks need to be uniquely named, and 'Documentation' conflicts
with the ci-tool name. Rename this while we are running both so the
names don't conflict.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
gperf is only used when CONFIG_USERSPACE is enabled. Users that use
arch's that don't support CONFIG_USERSPACE, or happen to not never
enable CONFIG_USERSPACE should not be artificially required to install
gperf.
This change makes gperf optional by moving it's presence-check to it's
usage.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
The callee-saved registers have been separated out and will not
be saved/restored if exception debugging is shut off.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The syscall exception frame was stored on the CPU struct during
syscall execution, but that's not right. System calls might "feel
like" exceptions, but they're actually perfectly normal kernel mode
code and can be preempted and migrated between CPUs at any time.
Put the field on the thread struct.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The code underneath z_fatal_error() (which is usually run in an
exception context, but is not required to be) was running with
interrupts enabled, which is a little surprising.
The only bug present currently is that the CPU ID extracted for
logging is subject to a race (i.e. it's possible but very unlikely
that such a handler might migrate to another CPU after the error is
flagged and log the wrong CPU ID), but in general users with custom
error handlers are likely to be surprised when their dying threads
gets preempted by other code before they can abort.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This test was whiteboxing the _current_cpu pointer to extract the CPU
ID. That's actually racy: the thread can be preempted and migrated to
another CPU between the _current_cpu expression and the read of the ID
field. Do it right.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The context switch implementation forgot to save the current flag
state of the old thread, so on resume the flags would be restored to
whatever value they had at the last interrupt preemption or thread
initialization. In practice this guaranteed that the interrupt enable
bit would always be wrong, becuase obviously new threads and preempted
ones have interrupts enabled, while arch_switch() is always called
with them masked. This opened up a race between exit from
arch_switch() and the final exit path in z_swap().
The other state bits weren't relevant -- the oddball ones aren't used
by Zephyr, and as arch_switch() on this architecture is a function
call the compiler would have spilled the (caller-save) comparison
result flags anyway.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Use of the _current_cpu pointer cannot be done safely in a preemptible
context. If a thread is preempted and migrates to another CPU, the
old CPU record will be wrong.
Add a validation assert to the expression that catches incorrect
usages, and fix up the spots where it was wrong (most important being
a few uses of _current outside of locks, and the arch_is_in_isr()
implementation).
Note that the resulting _current expression now requires locking and
is going to be somewhat slower. Longer term it's going to be better
to augment the arch API to allow SMP architectures to implement a
faster "get current thread pointer" action than this default.
Note also that this change means that "_current" is no longer
expressible as an lvalue (long ago, it was just a static variable), so
the places where it gets assigned now assign to _current_cpu->current
instead.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Coverity believes that a field is null when it isn't. Duplicate the
assert from 20 lines up in hopes it learns better.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
I think people might be reading differences into 'if' and 'depends on'
that aren't there, like maybe 'if' being needed to "hide" a symbol,
while 'depends on' just adds a dependency.
There are no differences between 'if' and 'depends on'. 'if' is just a
shorthand for 'depends on'. They work the same when it comes to creating
implicit menus too.
The way symbols get "hidden" is through their dependencies not being
satisfied ('if'/'depends on' get copied up as a dependency on the
prompt).
Since 'if' and 'depends on' are the same, an 'if' with just a single
symbol in it can be replaced with a 'depends on'. IMO, it's best to
avoid 'if' there as a style choice too, because it confuses people into
thinking there's deep Kconfig magic going on that requires 'if'.
Going for 'depends on' can also remove some nested 'if's, which
generates nicer symbol information and docs, because nested 'if's really
are so simple/dumb that they just add the dependencies from both 'if's
to all symbols within.
Replace a bunch of single-symbol 'if's with 'depends on' to despam the
Kconfig files a bit and make it clearer how things work. Also do some
other minor related dependency refactoring.
The replacement isn't complete. Will fix up the rest later. Splitting it
a bit to make it more manageable.
(Everything above is true for choices, menus, and comments as well.)
Detected by tweaking the Kconfiglib parsing code. It's impossible to
detect after parsing, because 'if' turns into 'depends on'.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
When running in qemu we utilize fdt-single_arch-zcu102-arm.dtb to
configure qemu. The current dtb requests just under 2G of memory in the
simulation, but we don't really need this and it impacts running on low
memory VMs.
So remove DDR ddr_bank2 and reduce ddr_bank1_3 from 1G - 256K to
128M - 256K in size.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Add sample for AES Galois/Counter Mode (GCM) of operation with a MACsec
GCM-AES test vector.
Also improve existing code by declaring expected ciphertext arrays as
constant.
Signed-off-by: Markus Fuchs <markus.fuchs@de.sauter-bc.com>
