Add initial LoRaWAN support #23664

Mani-Sadhasivam · 2020-03-22T05:47:48Z

Hello,

This PR adds initial LoRaWAN support to Zephyr by using the loramac-node library from Semtech. This LoRaWAN support is added as a subsystem and sits on top of existing LoRa so that the underlying radio drivers get reused.

With the current support, a node can configure and send data to the network server via a gateway and also receive Acks in predefined window slots. A sample application is also included in this PR which demonstrates the CLASS-A functionality.

This PR has been verified using 96Boards Wistrio board and a SAMD21 board with SX1276 LoRa modem. Also, this has been tested in IN865 and EU868 region wherein EU868 the duty cycle restrictions are more strict so the sample application may not work correctly as it sends data in 5s interval.

TODO:

Add ABP join support
Add CLASS-B and C support with sample application/documentation
Improve LoRaWAN support by detecting Tx notification for both confirmed and unconfirmed messages.
Add Socket support with a suitable compression/fragmentation method. For this, I've started libschc project.
Add tests for LoRa and LoRaWAN

This LoRaWAN project is a collaborative work and following developers deserve explicit acknowledgments for their commendable contributions:

Kwon Tae Young - https://github.com/KwonTae-young
Kuba Sanak - https://github.com/KubaFYI
Andreas Sandberg - https://github.com/andysan

For more discussions other than this PR, please join #zephyr-lorawan slack channel.

Thanks,
Mani

zephyrbot · 2020-03-22T05:57:34Z

All checks are passing now.

checkpatch (informational only, not a failure)

-:769: WARNING:LONG_LINE_COMMENT: line over 80 characters
#769: FILE: subsys/lorawan/lorawan.c:188:
+	-ECONNRESET,		/* LORAMAC_EVENT_INFO_STATUS_DOWNLINK_REPEATED */

-:770: WARNING:LONG_LINE_COMMENT: line over 80 characters
#770: FILE: subsys/lorawan/lorawan.c:189:
+	-EMSGSIZE,		/* LORAMAC_EVENT_INFO_STATUS_TX_DR_PAYLOAD_SIZE_ERROR */

-:771: WARNING:LONG_LINE_COMMENT: line over 80 characters
#771: FILE: subsys/lorawan/lorawan.c:190:
+	-ECONNRESET,		/* LORAMAC_EVENT_INFO_STATUS_DOWNLINK_TOO_MANY_FRAMES_LOSS */

-:795: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'McpsConfirm', this function's name, in a string
#795: FILE: subsys/lorawan/lorawan.c:214:
+	LOG_DBG("Received McpsConfirm (for McpsRequest %d)",

-:811: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'McpsIndication', this function's name, in a string
#811: FILE: subsys/lorawan/lorawan.c:230:
+	LOG_DBG("Received McpsIndication %d", mcpsIndication->McpsIndication);

-:814: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'McpsIndication', this function's name, in a string
#814: FILE: subsys/lorawan/lorawan.c:233:
+		LOG_ERR("McpsIndication failed : %s",

-:836: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'MlmeConfirm', this function's name, in a string
#836: FILE: subsys/lorawan/lorawan.c:255:
+	LOG_DBG("Received MlmeConfirm (for MlmeRequest %d)",

-:840: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'MlmeConfirm', this function's name, in a string
#840: FILE: subsys/lorawan/lorawan.c:259:
+		LOG_ERR("MlmeConfirm failed : %s",

-:865: WARNING:EMBEDDED_FUNCTION_NAME: Prefer using '"%s...", __func__' to using 'MlmeIndication', this function's name, in a string
#865: FILE: subsys/lorawan/lorawan.c:284:
+	LOG_DBG("Received MlmeIndication %d", mlmeIndication->MlmeIndication);

- total: 0 errors, 9 warnings, 980 lines checked

NOTE: For some of the reported defects, checkpatch may be able to
      mechanically convert to the typical style using --fix or --fix-inplace.

Your patch has style problems, please review.

NOTE: Ignored message types: AVOID_EXTERNS BRACES CONFIG_EXPERIMENTAL CONST_STRUCT DATE_TIME FILE_PATH_CHANGES MINMAX NETWORKING_BLOCK_COMMENT_STYLE PRINTK_WITHOUT_KERN_LEVEL SPDX_LICENSE_TAG SPLIT_STRING VOLATILE

NOTE: If any of the errors are false positives, please report
      them to the maintainers.

Tip: The bot edits this comment instead of posting a new one, so you can check the comment's history to see earlier messages.

Mani-Sadhasivam · 2020-03-22T11:12:37Z

@SebastianBoe I'm not sure what would be the best way to handle the library issue here. The loramac-node library was added by drivers/lora before, but for the lorawan support, subsys/lorawan gets processed first before drivers/lora so I moved the library definition to subsys/lorawan in this PR. By doing this, the lorawan-node library now depends on CONFIG_LORAWAN symbol. But the plain LoRa sample applications don't enable it in prj.conf.

Also, I'm not very keen to enable CONFIG_LORAWAN for the LoRa based applications since LoRaMAC is not at all used. What would you suggest?

jukkar · 2020-03-22T13:15:54Z

drivers/lora/sx1276.c

@@ -42,30 +48,25 @@ static struct sx1276_dio sx1276_dios[] =
 #define SX1276_MAX_DIO ARRAY_SIZE(sx1276_dios)

 struct sx1276_data {


The dev_data should probably be static.

jukkar · 2020-03-22T13:16:21Z

drivers/lora/sx1276.c

+#define BOARD_TCXO_WAKEUP_TIME	5
+
+/* TODO: Use Non-volatile memory for backup */
+static volatile u32_t backup_reg[2] = { 0, 0 };


Shouldn't we put these variables into dev_data?

There is no benefit of doing so... but will move it

jukkar · 2020-03-22T13:19:29Z

include/lorawan.h

@@ -0,0 +1,113 @@
+/*


We are trying to get rid of various includes from include/ so this is not good place for lorawan.h. You could place the include file into include/net/ directory as there are other radio API includes there already.

Okay. We had this discussion for previous LoRa support also. But we settled with include/drivers/lora.h. In this case, since we don't have include/subsys I think it can live under include/net

@Mani-Sadhasivam If lorawan.h is moving to /net, should the sample application fall under samples/net?

jukkar · 2020-03-22T13:21:14Z

samples/lorawan/class_a/prj.conf

+CONFIG_LOG=y
+CONFIG_SPI=y
+CONFIG_GPIO=y
+CONFIG_NEWLIB_LIBC=y


Why do we need newlib?

Some functions used by the stack are not present in the minimal libc.

@jukkar The reason is provided by @alexanderwachter

jukkar · 2020-03-22T13:24:46Z

samples/lorawan/class_a/src/main.c

+#include <logging/log.h>
+LOG_MODULE_REGISTER(lorawan_class_a);
+
+char data[MAX_DATA_LEN] = {'h', 'e', 'l', 'l', 'o', 'w', 'o', 'r', 'l', 'd'};


Could this be done other way around, so just set the data variable (without size) and if you need its length, the use sizeof(), like this:

char data[] = {'h', 'e', 'l', 'l', 'o', 'w', 'o', 'r', 'l', 'd'}; #define MAX_DATA_LEN sizeof(data)

jukkar · 2020-03-22T13:26:00Z

samples/lorawan/class_a/src/main.c

+	LOG_INF("Sending data...");
+	while (1) {
+		ret = lorawan_send(2, LORAWAN_DEFAULT_DATARATE, data,
+				   MAX_DATA_LEN, true, 1);


As the data length is only used here, you could just write this sizeof(data) and avoid the MAX_DATA_LEN define.

jukkar · 2020-03-22T13:28:52Z

subsys/lorawan/Kconfig

+# SPDX-License-Identifier: Apache-2.0
+
+menuconfig LORAWAN
+	bool "LoRaWAN"


I would recommened that you add [EXPERIMENTAL] until the API is a bit more stable. This way it is easier to change things in the API if needed.

jukkar · 2020-03-22T13:31:35Z

subsys/lorawan/lorawan.c

+static volatile bool send_status = false;
+
+/* MAC status strings */
+const char *to_status_str[] = {


Converting the status value to string, would be less error prone if you do the conversion directly in a function, so something like this:

const char *status2str(int status) { switch (status) { case LORAMAC_STATUS_OK: return "OK"; .... }

jukkar · 2020-03-22T13:31:45Z

subsys/lorawan/lorawan.c

+};
+
+/* MAC event info status strings */
+const char *to_event_info_status_str[] = {


I'm sorry @Mani-Sadhasivam , I'm not quite able to understand what the problem is.

@SebastianBoe Till now, zephyr_library_named(loramac-node) is declared by drivers/lora/CMakeLists.txt but for LoRaWAN, we need to use library reference in subsys/lorawan. If we use set(ZEPHYR_CURRENT_LIBRARY loramac-node) in subsys/lorawan it doesn't work because this gets processed before drivers/lora and the library definition is missing.

To solve this issue, I moved the library definition to subsys/lorawan and it works well. But for the plain LoRa usecase where we don't need subsys/lorawan, this is creating issue because the loramac-node library definition is missing as subsys/lorawan is not enabled.

For me, the only way to fix this issue is by enabling CONFIG_LORAWAN for plain LoRa also but I don't favor it.

Can't you create more libraries? One for each entity; driver, subsystem, lorawan, lora ?

Then you wouldn't need to

set(ZEPHYR_CURRENT_LIBRARY loramac-node)

@SebastianBoe But the corresponding library will be used by the loramac-node module to include drivers: https://github.com/zephyrproject-rtos/loramac-node/blob/master/zephyr/CMakeLists.txt#L3

If we use different libraries, how will the loramac-node module add the drivers?

Couldn't it also create it's own library?

Sorry, I'm confused. If all 3 are different libraries, then how dirvers/lora and subsys/lorawan will use the drivers present in loramac-node? Am I missing something?

I am also confused. What is the problem here?

One system can use another by having it invoke functions in the other system.

Which functions are in which libraries is unimportant. Any library can invoke functions from any other library.

Here is the error after making different libraries: https://pastebin.ubuntu.com/p/NT9TBDkcNP/

Right, to solve this you would want to do

zephyr_include_directories(path/to/dir/of/LoRaMac.h)

this is assuming that the header files are fairly unique, if not you need to do something that is a bit more involved.

Used this as per Slack conversation: f3068ee#diff-4d3f1718e6c473abc54edf7fd4c5830d

jukkar · 2020-03-22T13:33:50Z

subsys/lorawan/lorawan.c

+static void McpsIndication(McpsIndication_t *mcpsIndication)
+{
+	if (mcpsIndication->Status != LORAMAC_EVENT_INFO_STATUS_OK) {
+		LOG_ERR("%s failed : %s", __func__,


Do not put __func__ to log prints. The logger functions can be configured to add them automatically.

alexanderwachter · 2020-03-22T15:31:08Z

drivers/lora/sx1276.c

-		LOG_ERR("Failed to read counter value (err %d)", err);
-		return 0;
-	}
+	return k_ms_to_ticks_ceil32(k_uptime_get_32());


Why don't you use k_cycle_get_32 directly here?

I tried it but the calculated delta between two points doesn't seem to be right. Seems like it takes accumulated count into account.

alexanderwachter · 2020-03-22T15:32:55Z

drivers/lora/sx1276.c

-	return ticks;
+u32_t RtcGetTimerElapsedTime(void)
+{
+	return (k_ms_to_ticks_ceil32(k_uptime_get_32()) - saved_time);


Use k_cycle_get_32 and store the saved_time in cycles.

alexanderwachter · 2020-03-22T15:33:56Z

drivers/lora/sx1276.c

+
+u32_t RtcSetTimerContext(void)
+{
+	saved_time = k_ms_to_ticks_ceil32(k_uptime_get_32());


I would recommend to use ticks for saved_time.

saved_time is already in ticks!

alexanderwachter · 2020-03-22T15:43:42Z

samples/lorawan/class_a/prj.conf

+CONFIG_LOG=y
+CONFIG_SPI=y
+CONFIG_GPIO=y
+CONFIG_NEWLIB_LIBC=y


Some functions used by the stack are not present in the minimal libc.

SebastianBoe · 2020-03-23T11:40:30Z

subsys/lorawan/lorawan.c

+#include <LoRaMac.h>
+
+#ifdef CONFIG_LORAMAC_REGION_AS923
+	#define CONFIG_LORAWAN_REGION LORAMAC_REGION_AS923


It's not OK to define macro's named CONFIG_*.

CONFIG_ is a reserved namespace for Kconfig.

SebastianBoe · 2020-03-23T11:50:03Z

I'm sorry @Mani-Sadhasivam , I'm not quite able to understand what the problem is.

SebastianBoe · 2020-03-25T14:08:52Z

drivers/lora/CMakeLists.txt

@@ -1,5 +1,9 @@
 # SPDX-License-Identifier: Apache-2.0

-zephyr_library_named(loramac-node)
+if(TARGET loramac-node)


An explanation in a comment above is in order. E.g. something like:

sx1276.c is added to the loramac-node library when it exists because it needs the same include directories as the loramac-node has.

When it does not exist a library of the same name is created because ...

stale

carlescufi · 2020-10-05T15:40:38Z

@jukkar @andysan @tejlmand @alexanderwachter @KubaFYI @KwonTae-young @chrta could you please take another look at this PR? It's been open for a while and should be ready now.

chrta

I will approve once you address the comments from @tejlmand from May

drivers/lora/hal_common.c

tejlmand · 2020-10-06T09:15:26Z

Code still uses:

include($ENV{ZEPHYR_BASE}/cmake/app/boilerplate.cmake NO_POLICY_SCOPE)

#23664 (comment)

and the alignment comment has also not been addressed.

Mani-Sadhasivam · 2020-10-06T14:23:45Z

@chrta @tejlmand Addressed review comments. Sorry I missed those!

carlescufi · 2020-10-07T12:49:20Z

@jukkar @andysan @alexanderwachter @KubaFYI @KwonTae-young any additional comments before we merge?

jukkar

LGTM

Add LoRaWAN API for Zephyr. Only config, join and send methods are implemented. Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

Add initial support for LoRaWAN based on Semtech's loramac-node library. Current implementation only supports OTAA config and sending data to LoRaWAN server like ThingsNetwork. While at it, this commit also moves the "loramac-node" library definition from drivers/lora to subsys/lorawan. This is required because, subsys/lorawan gets processed before drivers/lora and that creates issue while building. Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

andysan

LGTM

subsys/lorawan/lorawan.c

samples/lorawan/class_a/src/main.c

KwonTae-young · 2020-10-08T01:41:03Z

LGTM

This sample application shows how to configure an end node in Class-A mode and to send data to network server via gateway. Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

SystemMaxRxError is used to negotiate overall timing error for Rx in the loramac-node library. Hence, add support for configuring this parameter from Kconfig. Signed-off-by: Kuba Sanak <contact@kuba.fyi> Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

For consistency with other parts of Zephyr, the public APIs available in lorawan subsystem now returns error codes from the set defined in errno.h. Signed-off-by: Kuba Sanak <contact@kuba.fyi> [mani: reworked the code and commit a bit for upstream] Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

Add CODEOWNERS entry for LoRaWAN Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

The error handling code currently has a couple of issues: * It relies on ordered lists and upstream not changing any constants. * Converted messages are not stored in constant memory which means that log_strdup is needed whenever they are printed. This change also factors out error handling to a separate file, lw_priv.{c,h}, to facilitate reuse in a future secure element and state storage implementation. Signed-off-by: Andreas Sandberg <andreas@sandberg.pp.se>

Mani-Sadhasivam requested review from jukkar, tbursztyka, alexanderwachter, carlescufi and galak March 22, 2020 05:47

Mani-Sadhasivam requested review from MaureenHelm and nashif as code owners March 22, 2020 05:47

zephyrbot added area: API Changes to public APIs area: Samples Samples area: Modules labels Mar 22, 2020

Mani-Sadhasivam force-pushed the lorawan-upstream branch from fb75a90 to 872e94d Compare March 22, 2020 06:03

Mani-Sadhasivam added the area: LoRa label Mar 22, 2020

jukkar previously requested changes Mar 22, 2020

View reviewed changes

alexanderwachter previously requested changes Mar 22, 2020

View reviewed changes

SebastianBoe reviewed Mar 23, 2020

View reviewed changes

Mani-Sadhasivam force-pushed the lorawan-upstream branch from 872e94d to 934f2f1 Compare March 24, 2020 15:31

Mani-Sadhasivam requested a review from pfalcon as a code owner March 24, 2020 15:31

Mani-Sadhasivam force-pushed the lorawan-upstream branch 2 times, most recently from 2a25fee to 324e427 Compare March 25, 2020 13:59

SebastianBoe reviewed Mar 25, 2020

View reviewed changes

Mani-Sadhasivam force-pushed the lorawan-upstream branch from 324e427 to 0f3f546 Compare March 25, 2020 14:26

Mani-Sadhasivam requested review from SebastianBoe, jukkar and alexanderwachter March 25, 2020 14:26

Mani-Sadhasivam force-pushed the lorawan-upstream branch from 0f3f546 to 768a688 Compare March 25, 2020 14:36

SebastianBoe requested review from tejlmand and removed request for SebastianBoe March 31, 2020 13:23

Mani-Sadhasivam removed area: Networking has-conflicts Issue/PR has conflicts with another issue/PR labels Oct 5, 2020

chrta reviewed Oct 6, 2020

View reviewed changes

drivers/lora/hal_common.c Show resolved Hide resolved

Mani-Sadhasivam force-pushed the lorawan-upstream branch from ff3f110 to 39ac1a4 Compare October 6, 2020 14:22

chrta approved these changes Oct 7, 2020

View reviewed changes

tejlmand approved these changes Oct 7, 2020

View reviewed changes

jukkar approved these changes Oct 7, 2020

View reviewed changes

Mani-Sadhasivam added 2 commits October 7, 2020 19:35

include: Add LoRaWAN API

7d2a7be

Add LoRaWAN API for Zephyr. Only config, join and send methods are implemented. Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

Mani-Sadhasivam force-pushed the lorawan-upstream branch from 39ac1a4 to 2e22205 Compare October 7, 2020 14:06

andysan approved these changes Oct 7, 2020

View reviewed changes

mniestroj reviewed Oct 7, 2020

View reviewed changes

subsys/lorawan/lorawan.c Outdated Show resolved Hide resolved

samples/lorawan/class_a/src/main.c Outdated Show resolved Hide resolved

Mani-Sadhasivam and others added 5 commits October 8, 2020 13:03

samples: lorawan: Add Class-A LoRaWAN sample application

69cd892

This sample application shows how to configure an end node in Class-A mode and to send data to network server via gateway. Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

CODEOWNERS: Add entry for LoRaWAN

645f9ff

Add CODEOWNERS entry for LoRaWAN Signed-off-by: Manivannan Sadhasivam <mani@kernel.org>

Mani-Sadhasivam force-pushed the lorawan-upstream branch from 2e22205 to 264f9d9 Compare October 8, 2020 07:35

Mani-Sadhasivam requested a review from mniestroj October 8, 2020 07:35

mniestroj approved these changes Oct 8, 2020

View reviewed changes

carlescufi merged commit cfb6257 into zephyrproject-rtos:master Oct 8, 2020

mniestroj mentioned this pull request Oct 9, 2020

drivers: lora: run TimerIrqHandler() from system workqueue #27770

Closed

		@@ -42,30 +48,25 @@ static struct sx1276_dio sx1276_dios[] =
		#define SX1276_MAX_DIO ARRAY_SIZE(sx1276_dios)

		struct sx1276_data {

Add initial LoRaWAN support #23664

Add initial LoRaWAN support #23664

Conversation

Mani-Sadhasivam commented Mar 22, 2020 • edited Loading

zephyrbot commented Mar 22, 2020 • edited Loading

checkpatch (informational only, not a failure)

Mani-Sadhasivam commented Mar 22, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SebastianBoe Mar 25, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SebastianBoe commented Mar 23, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

carlescufi commented Oct 5, 2020 • edited Loading

chrta left a comment

Choose a reason for hiding this comment

tejlmand commented Oct 6, 2020

Mani-Sadhasivam commented Oct 6, 2020

carlescufi commented Oct 7, 2020

jukkar left a comment

Choose a reason for hiding this comment

andysan left a comment

Choose a reason for hiding this comment

KwonTae-young commented Oct 8, 2020

Mani-Sadhasivam commented Mar 22, 2020 •

edited

Loading

zephyrbot commented Mar 22, 2020 •

edited

Loading

SebastianBoe Mar 25, 2020 •

edited

Loading

carlescufi commented Oct 5, 2020 •

edited

Loading