After reading a frame, `read()` checks the status of the RX FIFO:
* If an overflow is detected, the FIFO gets flushed
* If there are more frames in the FIFO, the `cc2538_rf_process` will get polled again in order to read out the next frame.
#1550 changed `read()`, which now performs the above check for non-poll mode, but it then flushes the FIFO unconditionally. Therefore, if there are two or more frames in the FIFO, they will get flushed before the `cc2538_rf_process` has had a chance to read them out. This results in missed frames.
Reproducing this is trivial: Build a CC2538 sniffer and see how it will never show you .15.4 ACK frames. ACK reception completes while `read()` is still streaming the previous captured frame to the host. Upon completion, the FIFO will get flushed and the ACK will get lost.
This pull proposes removing the unconditional flush and reverting to the original logic for non-TSCH operation.
Pinging @thomas-ha here for input regarding poll mode.
This pull sits on top of #1778
Depending on the use case and on the timings,
aes_auth_crypt_check_status() sometimes never reported an available
result, leading to a deadlock of any protothread waiting for this event,
and to a WDT reset if a protothread was polling it.
This was caused by aes_auth_crypt_start() clearing the result available
interrupt after operations that may rightfully trigger it, leading to a
missed interrupt.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Introduce FLASH_CONF_FW_ADDR and FLASH_CONF_FW_SIZE in order to make it
possible to place the firmware anywhere, regardless of Coffee, and
without having to write a custom linker script. Also, handle the default
values properly in order to fix the link breakage reported by
Arthur Fabre <arthur@arthurfabre.com> with COFFEE_CONF_CUSTOM_PORT.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Add functions providing the SoC revision, SRAM size, and enabled
hardware features, as well as a function printing SoC information.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Add functions providing the last reset cause, one as an integer (ID),
and one as a string.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
The conditional definitions in project-conf.h depending on
CONTIKI_TARGET_<TARGET_NAME> were ignored at link time, which broke the
linker script if it used these definitions, so the flashed applications
could crash or malfunction.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
The block that controls the `.upload` target is unnecessarily replicated in multiple sub-board Makefiles. This was originally done because the SmartRF and the Launchpad can be programmed with the c2538-bsl script, whereas the sensortag cannot.
This commit moves the `cc2538-bsl` / `.upload` target logic to the top level cpu Makefile (`cpu/cc26xx-cc13xx/Makefile.cc26xx-cc13xx`). Board makefiles simply set the make variable `BOARD_SUPPORTS_BSL` to 1 to signal that they can be programmed by the BSL script. If `BOARD_SUPPORTS_BSL` is not equal to 1, trying to use the `.upload` target will return an error message.
For example:
```
$ make BOARD=sensortag/cc2650 cc26xx-demo.upload
using saved target 'srf06-cc26xx'
This board cannot be programmed with the ROM bootloader and therefore does not support the .upload target.
```
This patch adds a simple non-driver protection domain sample to serve
as an example for defining other non-driver protection domains. It
simply performs a ping-pong test of protection domain switching
latency during boot, including optional accesses to a private metadata
region, and prints out the results.
This patch extends the protection domain framework with a third plugin
that is a hybrid of the previous two. The hardware task switching
mechanism has a strictly-defined format for TSS data structures that
causes more space to be consumed than would otherwise be required.
This patch defines a smaller data structure that is allocated for each
protection domain, only requiring 32 bytes instead of 128 bytes. It
uses the same multi-segment memory layout as the TSS-based plugin and
leaves paging disabled. However, it uses a similar mechanism as the
paging plugin to perform system call dispatches and returns.
For additional information, please refer to cpu/x86/mm/README.md.
This patch extends the protection domain framework with an additional
plugin to use Task-State Segment (TSS) structures to offload much of
the work of switching protection domains to the CPU. This can save
space compared to paging, since paging requires two 4KiB page tables
and one 32-byte page table plus one whole-system TSS and an additional
32-byte data structure for each protection domain, whereas the
approach implemented by this patch just requires a 128-byte data
structure for each protection domain. Only a small number of
protection domains will typically be used, so
n * 128 < 8328 + (n * 32).
For additional information, please refer to cpu/x86/mm/README.md.
GCC 6 is introducing named address spaces for the FS and GS segments
[1]. LLVM Clang also provides address spaces for the FS and GS
segments [2]. This patch also adds support to the multi-segment X86
memory management subsystem for using these features instead of inline
assembly blocks, which enables type checking to detect some address
space mismatches.
[1] https://gcc.gnu.org/onlinedocs/gcc/Named-Address-Spaces.html
[2] http://llvm.org/releases/3.3/tools/clang/docs/LanguageExtensions.html#target-specific-extensions
If the setup of socket 0 to 3 with 4+2+1+1KB is detected then the W5100 is _not_ initialized, otherwise it does set up socket 0 and 1 with 4KB each. Either way socket 0 is used - now with 4KB instead of 8KB as before.
This patch implements a simple, lightweight form of protection domains
using a pluggable framework. Currently, the following plugin is
available:
- Flat memory model with paging.
The overall goal of a protection domain implementation within this
framework is to define a set of resources that should be accessible to
each protection domain and to prevent that protection domain from
accessing other resources. The details of each implementation of
protection domains may differ substantially, but they should all be
guided by the principle of least privilege. However, that idealized
principle is balanced against the practical objectives of limiting the
number of relatively time-consuming context switches and minimizing
changes to existing code.
For additional information, please refer to cpu/x86/mm/README.md.
This patch also causes the C compiler to be used as the default linker
and assembler.
The UEFI GenFw program inserts headers ahead of the code in the UEFI
binary. The linker script adjusts the starting address of the .text
section to account for that. This prevents the symbols from being
perturbed. This patch accounts for a recent change in the size of the
headers added by the GenFw program.
This patch configures Isolated Memory Regions (IMRs) to block DMA to
code and data regions that do not contain any data that needs to be
DMA-accessible.
The Intel Quark X1000 SoC includes support for Isolated Memory Regions
(IMRs), which are specified using range registers and associated
control registers that are accessible via the message bus. This patch
adds a driver for accessing those registers.
The Intel Quark X1000 SoC includes a message bus that is accessible
via PCI configuration registers. It communicates to various SoC
components such as the Isolated Memory Region (IMR) registers and the
Remote Management Unit. This patch adds a driver for accessing the
message bus.
Bug fixes include:
- keep interrupts disabled during lpm_sleep() so that we don't
miss any interrupts we may have been expecting
- check that the pending etimer isn't already expired (and don't sleep
at all if it is)
- check that the about-to-be scheduled rtimer wakeup is neither too
far into the future nor too close into the future (or even in the
past) before actually setting the interrupt (should fix#1509); If
the time is out of bounds we use a default min or max value instead.
- Correctly handle LPM_MODE_MAX_SUPPORTED set to zero (and added a
macro for the zero value) so that sleeping can be disabled altogether
- If no etimer is set, we specify a wakeup time which is reasonably far
into the future instead of setting none at all (this will save on
power consumption whenever no etimers are set).
Also did a bit of refactoring in that some long functions were broken
into multiple functions.
This patch revises the I2C and GPIO initialization code to always be
run during platform boot rather than within each process that requires
it.
This patch also revises the gpio-output example to use a pin that is
set as an output by the default pinmux configuration. Previously, it
used a pin that was set as an output by the pinmux configuration that
is in effect when the OS does not change the pinmux configuration.
This patch permits interrupts to be generated by both the I2C and GPIO
controllers for simultaneously-executing applications. The controllers
share a single interrupt pin, INTC. Prior to this patch,
quarkX1000_gpio_init() routed INTA to PIRQC and IRQ 10 (due to an
incorrect assumption that INTA is connected to the GPIO controller),
and quarkX1000_i2c_init() routed INTC to PIRQC and IRQ 9. The I2C
controller initialization is a prerequisite for GPIO initialization,
so the final configuration was that INTA and INTC were both routed to
PIRQC and IRQ 10. Thus, only the GPIO ISR was being invoked, even if
the I2C controller was actually responsible for the interrupt.
This patch refactors the I2C and GPIO ISR setup and handler code so
that the shared portions are combined in
cpu/x86/drivers/legacy_pc/shared-isr.[ch]. The I2C and GPIO drivers
communicate their interrupt information to the shared component by
placing structures in a specific section of the binary.
When sending a command to the CC13xx/CC25xx RF core, we wait for command completion by checking the LSB of CMDSTA (correctly so). However, in doing so we also zero out the 3 CMDSTA return bytes. For some commands, those bytes contain useful information (e.g. an RSSI value) and are required by the caller.
This problem manifests itself e.g. in PROP mode `channel_clear()`, whereby the caller will always see an RSSI value of 0.
This pull therefore fixes the logic in `rf_core_send_cmd()` to check for command completion by blocking on the CMDSTA result byte without zeroing out the 3 return bytes.
Fixes#1465
