The DFLL on samd5x has a hardware bug that requires a special
re-enabling sequence when it is disabled and then re-enabled again.
When running the clock on-demand, the hardware handles the disabling
and re-enabling so that sequence does not get executed.
To reproduce, run `tests/periph_uart` on `same54-xpro`.
Without this patch the test will get seemingly stuck on `sleep_test()`.
(In fact it keeps running, but the DFLL has the wrong frequency so the
UART baudrate is wrong).
In this test, on `same54-xpro` only UART0 is sourced from DFLL.
So if the UART is disabled the DFLL will be turned off as well.
Switch from the on-chip LDO to the on-chip buck voltage regulator
when not fast internal oscillators are used.
On `saml21-xpro` with `examples/default` this gives
**before:** 750 µA
** after:** 385 µA
nrf52 includes include $(RIOTBOARD)/$(BOARD)/Makefile.dep to know
if `nordic_softdevice_ble` is used, this changes dependency
resolution sinnce -include $(APPDIR)/Makefile.board.dep should
be resolved before.
This can be removed once #9913 is if `nordic_softdevice` is
deprecated.
- Add FEATURES_REQUIRED_ANY to dependency-debug:
Now `make dependency-debug` by default also stores the contents of
`FEATURES_REQUIRED_ANY`.
- makefiles/features_check.inc.mk: Break long lines
- {tests/minimal,tests/unittests,bootloaders/riotboot}:
Disable auto_init_% in addition to auto_init.
This works around weird behavior due to the USEMODULE being recursively expended
in the first iteration of dependency resolution: Modules added to DEFAULT_MODULE
get automatically added to USEMODULE during the first run, but not for
subsequent. This should be iron out later on.
Goals:
- Untangle dependency resolution and feature checking for better maintainability
- Improve performance of "make info-boards-supported"
Changes:
- Makefile.dep
- Dropped handling of default modules and recursion
- Now only dependencies of the current set of used modules and pkgs are
added
==> External recursion is needed to catch transient dependencies
- Changed Makefile.features:
- Dropped checking of provided features
- Dropped populating FEATURES_USED with provided features that are required
or optional
- Dropped populating FEATURES_MISSING with required but not provided
features
- Dropped adding modules implementing used features to USE_MODULE
==> This now only populates FEATURES_PROVIDED, nothing more
- Added makefiles/features_check.inc.mk:
- This performs the population of FEATURES_USED and FEATURES_MISSING now
- Added makefiles/features_modules.inc.mk:
- This performs now the addition of modules implementing used features
- Added makefiles/dependency_resolution.inc.mk:
- This now performs the recursion required to catch transient dependencies
- Also the feature check is performed recursively to handle also required
and optional features of the transient dependencies
- DEFAULT_MODULES are added repeatedly to allow it to be extended based on
used features and modules
==> This allows modules to have optional dependencies, as these
dependencies can be blacklisted
- Use simply expanded variables instead of recursively expended variables
(`foo := $(bar)` instead `foo = $(bar)`) for internal variables during feature
resolution. This improves performance significantly for
`make info-boards-supported`.
- Reduce dependency resolution steps in `make info-boards-supported`
- Globally resolve dependencies without any features (including arch)
provided
==> This results in the common subset of feature requirements and modules
used
- But for individual boards additional modules might be used on top due
to architecture specific dependencies or optional features
- Boards not supporting this subset of commonly required features are not
supported, so no additional dependency resolution is needed for them
- For each board supporting the common set of requirements a complete
dependency resolution is still needed to also catch architecture specific
hacks
- But this resolution is seeded with the common set of dependencies to
speed this up
Previously the function attempted to block here and manually service
the ISR.
This lead to unexpected results, in particular messages queuing up in
the threads message queue.
The result was that the radio would not end up in the correct state.
E.g. sending SLEEP to both interfaces while a transmission was ongoing
would lead to the interfaces waking up again.
With this patch the operation will just return -ERRNO so the caller can
try again.
To reproduce, try the attached patch for the `gnrc_networking` example:
On master you will find that the radio still consumes ~2.4mA after 'shutdown'.
(It is in fact in the state TRXOFF as it woke up again)
With this change the radio should consume less than 1µA (DEEP SLEEP).
diff --git a/examples/gnrc_networking/main.c b/examples/gnrc_networking/main.c
index 6301f4291d..93b96eb939 100644
--- a/examples/gnrc_networking/main.c
+++ b/examples/gnrc_networking/main.c
@@ -23,12 +23,47 @@
#include "shell.h"
#include "msg.h"
+#include "periph/pm.h"
+
+#include "net/netopt.h"
+#include "net/gnrc/netif.h"
+
#define MAIN_QUEUE_SIZE (8)
static msg_t _main_msg_queue[MAIN_QUEUE_SIZE];
extern int udp_cmd(int argc, char **argv);
+extern void send(char *addr_str, char *port_str, char *data, unsigned int num,
+ unsigned int delay);
+
+static int send_and_shutdown(int argc, char **argv)
+{
+ (void) argc;
+ (void) argv;
+
+ /* the address must not exist */
+ char addr[] = "fe80::2068:3123:59f5:d238%7";
+ char port[] = "1234";
+ char data[] = "Hello World!";
+
+ send(addr, port, data, 1, 0);
+
+ /* disable radio */
+ gnrc_netif_t* netif = NULL;
+ netopt_state_t state = NETOPT_STATE_SLEEP;
+ while ((netif = gnrc_netif_iter(netif))) {
+ /* retry while busy */
+ while (gnrc_netapi_set(netif->pid, NETOPT_STATE, 0, &state,
+ sizeof(netopt_state_t)) == -EBUSY);
+ }
+
+ pm_set(0);
+
+ return 0;
+}
+
static const shell_command_t shell_commands[] = {
+ { "shutdown", "turn off the radio & shut down", send_and_shutdown },
{ "udp", "send data over UDP and listen on UDP ports", udp_cmd },
{ NULL, NULL, NULL }
};
diff --git a/examples/gnrc_networking/udp.c b/examples/gnrc_networking/udp.c
index e8a559846e..cb80855b76 100644
--- a/examples/gnrc_networking/udp.c
+++ b/examples/gnrc_networking/udp.c
@@ -36,7 +36,7 @@ static gnrc_netreg_entry_t server = GNRC_NETREG_ENTRY_INIT_PID(GNRC_NETREG_DEMUX
KERNEL_PID_UNDEF);
-static void send(char *addr_str, char *port_str, char *data, unsigned int num,
+void send(char *addr_str, char *port_str, char *data, unsigned int num,
unsigned int delay)
{
gnrc_netif_t *netif = NULL;
Also adapt the defines to the documentation
- CPUs define up to 4 power modes (from zero, the lowest power mode,
to PM_NUM_MODES-1, the highest)
- >> there is an implicit extra idle mode (which has the number PM_NUM_MODES) <<
Previously on saml21 this would always generate pm_set(3) which is an illegal state.
Now pm_layered will correctly generate pm_set(2) for IDLE modes.
Idle power consumption dropped from 750µA to 368µA and wake-up from standby is also
possible. (Before it would just enter STANDBY again as the mode register was never
written with the illegal value.)
With lwIP we have a chunked UDP payload, so just providing the
stack-internal buffer is not possible. To be able to iterate over such
a chunked payload, this change allows the `sock_*_recv_buf()` functions
to use the internal buffer context as an iteration state.
As the internal buffer space can be released when the function would
return 0, `sock_recv_buf_free()` becomes unnecessary.
When a previously disabled DFLL gets enabled again, the frequency will
be incorrect. Follow the procedure outlined in the errata sheet, section 2.8.3
to work around the issue.
This fixes wake from standby.
