samples: nrf9160: serial LTE modem sample #1099

junqingzou · 2019-09-05T09:51:37Z

The Serial LTE Modem (SLM) project is an enhancement to at_client sample
- support TCP/IP proprietary AT commands
- support GPS proprietary AT commands
- support communication to external MCU over UART

Like original at_client, all modem AT commands are still supported.
More proprietary AT commands will be added in the future.

NordicBuilder · 2019-09-05T09:54:18Z

Found the following issues, please fix and resubmit:

Gitlint issues

Commit 10ca9eeb07:
1: UC3 Title does not follow [subsystem]: [subject] (and should not start with literal subsys:): "This sample project exposes below nRF91 services to external MCU:"
2: B4 Second line is not empty: " .AT command"

rlubos · 2019-09-05T14:39:41Z

samples/nrf9160/serial_lte_modem/prj.conf

+CONFIG_SNTP=y
+
+# MQTT lib
+CONFIG_MQTT_SOCKET_LIB=y


Some early feedback - the upstream MQTT library (CONFIG_MQTT_LIB) is slightly updated copy of the downstream library (CONFIG_MQTT_SOCKET_LIB), and we plan to remove the downstream copy. Could you use the upstream version instead? You can check #391 to see what steps were needed to transition nRF Cloud library to use the upstream MQTT implementation.

The project used to use upstream MQTT and later adapted to downstream MQTT_SOCKET. No problem I'll change back to previous one, which need to update code accordingly, too.

Thanks and sorry for the confusion, I'm struggling with downstream MQTT copy removal for a way too long...

lemrey · 2019-09-05T15:11:47Z

Thank you! For your convenience you can check out the clang-format and use it with the .clang-format configuration file in this repository to automatically format your source files.

junqingzou · 2019-09-06T03:02:31Z

Beat the robot after 4 commits :) Will prepare a clean commit after review.
@rlubos Code changed back to use Zephyr MQTT. Please check.

junqingzou · 2019-09-06T03:07:23Z

@lemrey Thank you! It's a good practice to correct the code style following the guideline

junqingzou · 2019-09-06T04:37:09Z

CI/Jenkins/NRF — Job has Failed
NordicPlayground/fw-nrfconnect-zephyr is not ready for LwM2M service in this project.
Disable LwM2M service in prj.conf

rlubos

I didn't manage to go through all files yet, will finish on Monday. Looks pretty good so far!

samples/nrf9160/serial_lte_modem/CMakeLists.txt

samples/nrf9160/serial_lte_modem/README.rst

rlubos · 2019-09-06T13:26:51Z

samples/nrf9160/serial_lte_modem/README.rst

@@ -0,0 +1,125 @@
+.. _serial_lte_modem	:


@b-gent @carlescufi I'm not sure about the serial-lte-modem spec.docx file provided. It contains valuable information about the sample but is quite large, so I'm not sure if it could be easily converted to .rst format easily. Do we allow such supplement documentation in the repository?

samples/nrf9160/serial_lte_modem/prj.conf

samples/nrf9160/serial_lte_modem/src/lwm2m/th_lwm2m_service.c

rlubos · 2019-09-06T14:29:25Z

Also, the .nul.old file looks like it's not needed here.

rlubos

Still a few more files left, will be back to the review soon.

samples/nrf9160/serial_lte_modem/src/modem/th_modem_service.c

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

samples/nrf9160/serial_lte_modem/src/sntp/th_sntp_service.c

rlubos

Finally finished the review :)

It looks pretty good, left a few more comments to consider.

I have a question though. I understand that this is a fully functional application for nRF91, but to get the full image of the proposed solution, one would also need some kind of host application (possibly for nRF5) that would communicate over serial with the LTE modem sample. Do you have any plans to submit such a complementary sample as well (even a simple one)?

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

samples/nrf9160/serial_lte_modem/src/th_connect.h

rlubos · 2019-09-09T12:55:34Z

samples/nrf9160/serial_lte_modem/src/th_test.c

+
+	/* Test MQTT connection */
+#ifdef CONFIG_THIN_MQTT_SERVICE
+	/* th_mqtt_control(CMD_TYPE_MQTT_CONNECT, MQTT_CLIENT, */


Remove dead code, please.

This is left as reference code as we cannot test MQTT and LwM2M simultaneously.
I'll add comments for this.

In that case how about guarding it with
#if defined(CONFIG_THIN_MQTT_SERVICE) && !defined(CONFIG_THIN_LWM2M_SERVICE)?

Thanks for the advice, applied.

rlubos · 2019-09-09T13:03:14Z

samples/nrf9160/serial_lte_modem/src/th_util.c

+#include <net/socket.h>
+
+/* Partly copy/paste from zephyr/subsys/net/ip/utils.c
+ * Cannot be used when BSD library selects NET_RAW_MODE


That's no longer the case, you should be able to use Zephyr implementation now (inet_pton).

Tested inet_pton(), not working

I need to investigate this, we use inet_ntop in several samples, both should work.

Ok, so I've tested it a bit and it seems that inet_pton converts the address string correctly. The only difference is that the inet_pton returns 1 on success (don't ask me why, IMO it's very confusing, but it's standardized here).

So the return value would need some extra attention here (return 0 if inet_pton returns 1, and for instance -EINVAL otherwise, IMO there's no need for more sophisticated error handling).

Actually I got 1 yesterday and thought it fails :)
Now I changed it as you suggested and tested OK.

samples/nrf9160/serial_lte_modem/src/main.c

junqingzou

@rlubos Thanks a lot for all the comments.
The nRF52 sample clients could be found here

rlubos · 2019-09-17T12:57:41Z

@carlescufi @b-gent Still need to settle about the user guide committed in *.docx format. Is this acceptable in NCS?

rlubos · 2019-09-17T12:59:12Z

@carlescufi Shall we link the repo with sample clients in the documentation (see #1099 (review)). The repo contains only Keil projects.

rlubos · 2019-09-17T13:02:06Z

@junqingzou Thank you for updating the PR. There are still a few details to determine, I'd also ask you to squash your commits before we can merge.

b-gent · 2019-09-17T13:41:22Z

Please wait with merging until I can check the docs (tomorrow)

rlubos · 2019-09-17T13:43:42Z

@b-gent Sure thing, take your time :)

carlescufi · 2019-09-17T15:39:28Z

@carlescufi @b-gent Still need to settle about the user guide committed in *.docx format. Is this acceptable in NCS?

No, I do not think so. This is a question for @ru-fu and @barsok

carlescufi · 2019-09-17T15:40:07Z

@carlescufi Shall we link the repo with sample clients in the documentation (see #1099 (review)). The repo contains only Keil projects.

Yes, definitely, we should. Even though they are not NCS clients at least they can serve as a starting point for users.

b-gent · 2019-09-17T20:09:44Z

@carlescufi @b-gent Still need to settle about the user guide committed in *.docx format. Is this acceptable in NCS?

I do not think so either, we cannot require our customers to have MS Word available to open a .docx file.
A potential solution is to generate PDF out of it and link to it from the RST file, but I don't recall having a PDF as part of our docset so not sure how that would work. Your thoughts @ru-fu ?

carlescufi · 2019-09-17T20:11:16Z

@carlescufi @b-gent Still need to settle about the user guide committed in *.docx format. Is this acceptable in NCS?

I do not think so either, we cannot require our customers to have MS Word available to open a .docx file.
A potential solution is to generate PDF out of it and link to it from the RST file, but I don't recall having a PDF as part of our docset so not sure how that would work. Your thoughts @ru-fu ?

I think we should ask the author, @junqingzou to just convert this to .rst or do it ourselves.

junqingzou · 2019-09-18T00:43:03Z

@carlescufi The spec doc is too much to be converted to .rst. Another option could be to put it with nRF52 sample clients, maintained here.

@rlubos Please DO NOT merge this for now. PMT is asking for raw AT-based serial modem, for which I have another project. This protocol-based project is mostly for exposing the app protocols. There are still things left to be clarified by PMT before decision of what to be in NCS. PS I'll rebase this project to latest master with a clean submit later.

ru-fu · 2019-09-18T06:54:14Z

The NCS documentation is in RST, so all documentation related to the samples must be converted to RST.
I'm not sure without looking into it how closely related this spec/user guide is. If it is general information that is used by the sample, it could be provided as a DevZone blog post for example. But from what @rlubos is saying, it sounds like it is required to understand what the sample is doing, and then it should be part of the sample documentation.

Btw, it sounds like this sample covers a lot of features, so should it be an application instead of a sample? (Samples should only show one single feature, while applications show a full use case.)

rlubos · 2019-09-23T09:08:40Z

samples/nrf9160/serial_lte_modem/README.rst

+   * nRF9160 (UART2)
+
+* UART configuration
+   * Hardware flow control:	enabled


Tabulation slipped in (before enabled)

rlubos · 2019-09-23T09:11:49Z

samples/nrf9160/serial_lte_modem/src/main.c

+#define INVALID_DESCRIPTOR	-1
+
+#define AT_BUF_SIZE		CONFIG_INTER_CONNECT_UART_BUF_SIZE
+#define THREAD_STACK_SIZE	1048


I suppose 1024 was meant here? Or is this finetuned to 1048?

You could use KB(x) from zephyr/include/sys/util.h here, if 1048 is a mistake.
Otherwise a short comment could clarify that 1048 is correct.

It's a typo and updated to KB(1), Thanks.

rlubos · 2019-09-23T09:13:37Z

samples/nrf9160/serial_lte_modem/src/main.c

+		}
+
+		/* Read AT socket in non-blocking mode. */
+		r_bytes = recv(at_socket_fd, at_buf, AT_BUF_SIZE - 4,


Why AT_BUF_SIZE - 4 and not simply AT_BUF_SIZE?

4 is the reserved size for protocol overhead (STX, TYPE, LENGTH, BCC). Actually it should be 5, plus the Notification type as the buffer is reserved for receive.

rlubos · 2019-09-23T09:20:11Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.h

+ * @param param   Parameter data for command.
+ * @param length  Length of parameter data.
+ */
+void th_tcpip_control(u8_t cmd, const u8_t *param, uint8_t length);


uint8_t length -> u8_t length

rlubos · 2019-09-23T09:21:21Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+#include "th_connect.h"
+#include "th_util.h"
+
+#define TCPIP_BUFFER_SIZE	(CONFIG_INTER_CONNECT_UART_BUF_SIZE - 4)


4 is the reserved size for protocol overhead (STX, TYPE, LENGTH, BCC). Actually it should be 5, plus the Notification type as the buffer is reserved for receive.

rlubos · 2019-09-23T09:37:53Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+		ret = recv(client.sock, rx_buffer,
+			TCPIP_BUFFER_SIZE,
+			0);
+		if (ret < 0) {


We should also handle a situation when peer disconnects (ret == 0). In that case we should no longer use the socket - it should be closed and client.connected set to false.

I'd actually handle receive errors (ret < 0) in the same way, as they typically render the socket unusable.

Updated. The close of socket would be in DISCONNECT

rlubos · 2019-09-23T09:38:32Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+	int ret;
+
+	memcpy(&port, param, 2);
+	address_use_ip = *(param+2) & ADDRESS_TYPE_MASK;


Spaces around + here and below.

rlubos · 2019-09-23T09:40:21Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+	memcpy(&port, param, 2);
+	address_use_ip = *(param+2) & ADDRESS_TYPE_MASK;
+	address_length = *(param+2) & ADDRESS_LENGTH_MASK;
+	memcpy(address, param + 3, address_length);


Why's the address copied in UDP case, and not in TCP? I'd unify the behavior unless there's a reason to do it differently.

TCP has address handled in CONNECT. UDP has no CONNECT so has to handle address for every sendto and recvfrom.

rlubos · 2019-09-23T09:46:34Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+			   sizeof(struct sockaddr));
+		if (ret < 0) {
+			LOG_ERR("sendto() failed: %d", -errno);
+			return -errno;


We should close the socket here, otherwise, we leak a socket. Or break instead of return.

rlubos · 2019-09-23T09:50:20Z

samples/nrf9160/serial_lte_modem/src/tcpip/th_tcpip_service.c

+
+	default:
+		err = -EINVAL;
+		LOG_ERR("unknown %d", cmd);


"unknown command %d"? Lone "unknown" might be confusing when seen in logs.

lemrey

It looks good! I left many comments because it's large and there's many files involved but there's nothing critical except the handling of out of memory errors, which needs to be fixed imo.

lemrey · 2019-09-24T08:00:16Z

samples/nrf9160/serial_lte_modem/src/th_connect.c

+
+#include "th_connect.h"
+
+#define CONFIG_UART_0_NAME	 "UART_0"


Don't define macros prefixed with CONFIG_, they might be mistaken for Kconfig-generated macros.

lemrey · 2019-09-24T08:54:15Z

samples/nrf9160/serial_lte_modem/src/th_connect.c

+		LOG_ERR("Missing STX");
+		return;
+	}
+#if CONFIG_INTER_CONNECT_BCC


Prefer IS_ENABLED when possible, to make sure this code will keep being compile-able; it will be optimized out in the end.

lemrey · 2019-09-24T08:54:46Z

samples/nrf9160/serial_lte_modem/src/th_connect.c

+	ARG_UNUSED(work);
+
+	if (data_handler_cb == NULL) {
+		LOG_ERR("Not itialized");


Typo: initialized

lemrey · 2019-09-24T08:55:51Z

samples/nrf9160/serial_lte_modem/src/th_connect.c

+	uart_dev = device_get_binding(uart_dev_name);
+	if (uart_dev == NULL) {
+		LOG_ERR("Cannot bind %s", uart_dev_name);
+		return -EINVAL;


Perhaps -EIO is a better choice here.

lemrey · 2019-09-24T08:56:37Z

samples/nrf9160/serial_lte_modem/src/th_connect.c

+	int err;
+	enum select_uart uart_id = CONFIG_INTER_CONNECT_UART;
+
+	if (data_handler == NULL) {


Consider using __ASSERT instead, if this function is not called directly by the user / part of the interface.

Here and other places where you check for initialization, see whether that can happen if the user misuses the interface or not. If it can't then it should be an __ASSERT.

lemrey · 2019-09-24T10:53:31Z

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

+
+#define STACKSIZE	1024
+#define MQTT_BUF_SIZE	(CONFIG_INTER_CONNECT_UART_BUF_SIZE - 4)
+#define PRIORITY	K_PRIO_PREEMPT(CONFIG_THIN_MQTT_THREAD_PRIO)


Why is this thread preemptive instead of cooperative? I ask because most (all?) threads in Zephyr are cooperative so I think we should have a good reason to let them be otherwise.

lemrey · 2019-09-24T10:55:52Z

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

+	if (c->transport.type == MQTT_TRANSPORT_NON_SECURE) {
+		fds.fd = c->transport.tcp.sock;
+	} else {
+#if defined(CONFIG_MQTT_LIB_TLS)


Is it possible to use IS_ENABLED here?

lemrey · 2019-09-24T10:57:16Z

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

+#ifdef CONFIG_THIN_MQTT_BROKER_USER_NAME
+	if (strlen(CONFIG_THIN_MQTT_BROKER_USER_NAME) > 0) {
+		client.user_name = k_malloc(sizeof(struct mqtt_utf8));
+		__ASSERT_NO_MSG(client.user_name != 0);


We shouldn't handle out of memory errors with __ASSERT, if it is a fatal error to be out of memory, we should allocate that memory statically. Otherwise, we should handle out of memory errors without crashing the application.

Please make sure to also fix this elsewhere.

As a side note, when comparing against pointers use NULL.

lemrey · 2019-09-24T10:58:04Z

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

+	if (strlen(CONFIG_THIN_MQTT_BROKER_USER_NAME) > 0) {
+		client.user_name = k_malloc(sizeof(struct mqtt_utf8));
+		__ASSERT_NO_MSG(client.user_name != 0);
+		client.user_name->utf8 = k_malloc(64);


lemrey · 2019-09-24T11:00:29Z

samples/nrf9160/serial_lte_modem/src/mqtt/th_mqtt_service.c

+
+/**@brief Function to unsubscribe to the configured topic
+ */
+int do_mqtt_unsubscribe(const u8_t *topic, u8_t length)


Why do you declare *topic as const and then down cast it on line 465?
If it needs indeed to be constant, down casting is dangerous 💣
If it doesn't need to be constant, you can remove the const and the cast on line 465.

NordicBuilder · 2019-10-01T05:38:40Z

All checks are passing now.

checkpatch (informational only, not a failure)

No additional types will be considered - file 'scripts/checkpatch/typedefsfile': No such file or directory
-:1300: WARNING:NEW_TYPEDEFS: do not add new typedefs
#1300: FILE: samples/nrf9160/serial_lte_modem/src/slm_at_host.h:23:
+typedef struct slm_at_cmd_list {

- total: 0 errors, 1 warnings, 2117 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 MULTISTATEMENT_MACRO_USE_DO_WHILE NETWORKING_BLOCK_COMMENT_STYLE PRINTK_WITHOUT_KERN_LEVEL 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.

rlubos

Thanks, that was a lot of new code. I've left some feedback in the comments.

rlubos · 2019-10-01T10:56:50Z

samples/nrf9160/serial_lte_modem/Kconfig

+
+config SLM_AT_MODE
+	bool "Serial LTE Modem by raw AT mode"
+  default 'y'


This one is still unaligned.

Sorry, Editor miss. Updated

rlubos · 2019-10-01T10:57:05Z

samples/nrf9160/serial_lte_modem/Kconfig

+#
+config SLM_TCPIP_AT_MODE
+	bool "Support TCP/IP by raw AT mode"
+  default 'y'


rlubos · 2019-10-01T10:57:21Z

samples/nrf9160/serial_lte_modem/Kconfig

+#
+config SLM_GPS_AT_MODE
+	bool "Support GPS by raw AT mode"
+  default 'y'


rlubos · 2019-10-01T11:21:34Z

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.h

+int slm_at_tcpip_parse(const u8_t *at_cmd, u8_t length);
+
+/**
+ * @brief Initialize GPS AT command parser.


GPS -> TCP/IP

rlubos · 2019-10-01T11:28:54Z

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.c

+
+/**@brief Socket operations. */
+enum slm_socket_operation {
+  AT_SOCKET_CLOSE,


Indent with tabulation, not spaces - here and in slm_tcpip_at_cmd_type.

rlubos · 2019-10-01T14:17:06Z

samples/nrf9160/serial_lte_modem/src/slm_at_gps.c

+
+/**@brief List of supported AT commands. */
+enum slm_gps_mode {
+  GPS_MODE_STANDALONE,


Tabs for indentation, here and below.

rlubos · 2019-10-01T14:26:22Z

samples/nrf9160/serial_lte_modem/src/slm_at_host.h

+ * @retval 0 If the operation was successful.
+ *           Otherwise, a (negative) error code is returned.
+ */
+int at_host_init(void);


We already have this API in NCS. If the sample needs to reimplement this functionality, please use different name (or just prefix with slm_ as other APIs here.

Updated to slm_at_host_init

rlubos · 2019-10-01T14:31:05Z

samples/nrf9160/serial_lte_modem/src/main.c

+		nrf_gpio_cfg_sense_set(CONFIG_SLM_MODEM_WAKEUP_PIN,
+			NRF_GPIO_PIN_SENSE_LOW);
+
+		/* The LTE modem also needs to be stopped by issuing a command*/


I think it would look nicer to have a single, multiline comment here:

/* * ... */

rlubos · 2019-10-01T14:42:12Z

samples/nrf9160/serial_lte_modem/client/at_client/main.c

@@ -0,0 +1,316 @@
+/**


I don't think we should introduce non-NCS samples into NCS codebase. @carlescufi Can you comment on that, was that agreed before?

@junqingzou I see that the sample was completely rewritten, and now supports AT-like commands for custom modules (gps/tcpip etc). We already have an at_client sample in NCS, are you familiar with it? Perhaps it would be possible to reuse it's codebase and prepare a NCS version of the client application?

Sorry for the misleading name "at_client", which is actually nRF52 project to work with the Serial LTE Modem. I'll remove it from this PR. Serial LTE Modem could work with either PC terminal or MCU code. "at_client" is a sample to run it in MCU code.

@rlubos Serial LTE Modem is copied from nrf\lib\at_host then add what PMT asked for. There might be more local AT commands to add based on customer requests.

rlubos · 2019-10-01T14:43:02Z

samples/nrf9160/serial_lte_modem/src/main.c

+#endif
+}
+
+#ifdef CONFIG_SLM_GPIO_WAKEUP


@carlescufi Could anyone more familiar with low-power modes in nRF have a look at this part?

@pizi-nordic and @anangl could you please take a look at this particular part?

rlubos

Thanks, I've noticed a few buffer overflows in this round that would need to be fixed.

Also, in order to finalize this PR, all of the checkpatch errors need to be fixed. While not critical, fixing warnings is also highly appreciated.

rlubos · 2019-10-03T12:37:30Z

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.c

+		client.ip_proto = IPPROTO_IP;
+		ret = -errno;
+	} else {
+		sprintf(buf, "#XSOCKET: %d, %d\r\n", client.sock, client.ip_proto);


This string is at minimum 16 bytes + null terminator, this will exceed the allocated buf size.

Like the gps code, use a global 64-byte buffer now.

rlubos · 2019-10-03T12:39:03Z

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.c

+		offset += ret;
+	}
+
+	sprintf(buf, "#XUDPSENDTO: %d\r\n", offset);


Same here, this string will be longer than 16 bytes.

Like the gps part, use a global 64-byte buffer now.

rlubos · 2019-10-03T12:39:58Z

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.c

+	} else if (*(at_param) == '?') {
+		char buf[16];
+		if (client.sock != INVALID_SOCKET) {
+			sprintf(buf, "#XSOCKET: %d, %d\r\n", client.sock,


Also here, more than 16 bytes.

Like the gps part, use a global 64-byte buffer now.

rlubos · 2019-10-03T12:50:29Z

samples/nrf9160/serial_lte_modem/Kconfig

+
+config SLM_AT_MODE
+	bool "Serial LTE Modem by raw AT mode"
+  default 'y'


This one is still unaligned.

rlubos

Thanks, looks good from my side.

lemrey

Left a few comments.

samples/nrf9160/serial_lte_modem/Kconfig

samples/nrf9160/serial_lte_modem/prj.conf

lemrey · 2019-10-07T14:39:44Z

samples/nrf9160/serial_lte_modem/src/slm_at_gps.c

+static char buf[64];
+
+#define THREAD_STACK_SIZE	KB(1)
+#define THREAD_PRIORITY		K_PRIO_PREEMPT(7)


Why specifically 7, why preemptive?

No special consideration here. 7 means the same as main thread. What's the recommended priority for application threads? Also did not consider in depth on preemptive. Thought to just pick one between 0...NUM_PREEMPT_PRIORITIES(15)

The main thread default priority is not 7.

config MAIN_THREAD_PRIORITY int "Priority of initialization/main thread" default -2 if !PREEMPT_ENABLED default 0 help Priority at which the initialization thread runs, including the start of the main() function. main() can then change its priority if desired.

I suggest using a low priority: K_LOWEST_APPLICATION_THREAD_PRIO.

The project has "CONFIG_MAIN_THREAD_PRIORITY=7" in the prj.conf, which was copied from the MQTT sample long time ago. Anyway, I tried K_LOWEST_APPLICATION_THREAD_PRIO and confirmed it's good so I'll change to use it. Thanks a lot!

lemrey · 2019-10-07T14:42:25Z

samples/nrf9160/serial_lte_modem/src/slm_at_gps.c

+			LOG_ERR("Failed to stop GPS (err: %d)", -errno);
+			ret = -errno;
+		} else {
+			k_thread_abort(gps_thread_id);


side note: this will prevent the thread from running again until you reset..

Changed to use k_thread_suspend()/k_thread_resume().
Smoke test shows it works. Please review again.

samples/nrf9160/serial_lte_modem/src/slm_at_gps.c

samples/nrf9160/serial_lte_modem/src/slm_at_tcpip.c

samples/nrf9160/serial_lte_modem/src/slm_at_gps.c

samples/nrf9160/serial_lte_modem/prj.conf

samples/nrf9160/serial_lte_modem/nrf9160_pca10090ns.overlay

TerjeKWahl · 2019-10-16T07:48:48Z

samples/nrf9160/serial_lte_modem/README.rst

+******************
+
+* AT#XSOCKET=<op>,<type>
+* AT%XSOCKET?


Small typo: "AT%XSOCKET?" --> "AT#XSOCKET?"

Eagerly looking forward to this branch being finished and merged into master :-)

Thank you. Also missing "AT#XBIND" in README.rst. Will be added.

rlubos · 2019-10-21T09:22:01Z

@lemrey Your comments seem to be addressed, could you revisit?

rlubos · 2019-10-23T09:56:03Z

@b-gent @ru-fu This sample needs a doc review.

b-gent · 2019-10-23T12:11:25Z

@b-gent @ru-fu This sample needs a doc review.

Yes, it's on me. A lot was happening around this PR and I wanted it to 'cool down' a bit before a doc review. I'll review by EOD tomorrow.

b-gent · 2019-10-24T22:10:15Z

samples/nrf9160/serial_lte_modem/README.rst

+    #. nRF91 is woken up and sends a ``Ready\r\n`` message to the nRF52.
+    #. On receiving the message, nRF52 can proceed to issue AT commands.
+
+You can find a sample nRF52 client project in the :file:`client` folder, which can be put under "SDK_15.3.0/examples/peripheral" folder.


Please help me understand why we are suddenly talking about nRF5 SDK here?

Sorry this is a left-over of old code, where a nRF52 project is kept inside this nRF91 project. It's removed based on review comment. I changed the line like this "A sample nRF52 client project is to be shared by other means."

b-gent · 2019-10-24T22:12:58Z

I added a doc review commit, please check the diff and let me know if my changes are correct. If yes, then squash the doc commit.

lemrey · 2019-10-25T14:08:04Z

samples/nrf9160/serial_lte_modem/src/main.c

+}
+
+/**@brief Irrecoverable BSD library error. */
+void bsd_irrecoverable_error_handler(uint32_t err)


We have now removed the bsd_irrecoverable_error_handler(), from bsdlib.
This should be removed.

Rebased to BSDlibv0.5.0 and have this function removed from main.c.
Thanks!

lemrey · 2019-10-25T14:12:14Z

/jenkins_workspace/workspace/connect-nrf_documentation_master/_build/nrf/rst/doc/nrf/samples/nrf9160/serial_lte_modem/README.rst:1: WARNING: malformed hyperlink target.
/jenkins_workspace/workspace/connect-nrf_documentation_master/_build/nrf/rst/doc/nrf/ug_dev_model.rst:43: WARNING: Indirect hyperlink target "github documentation"  refers to target "github duplicate", which does not exist.
/jenkins_workspace/workspace/connect-nrf_documentation_master/_build/nrf/rst/doc/nrf/ug_dev_model.rst:43: WARNING: Unknown target name: "github duplicate".
/jenkins_workspace/workspace/connect-nrf_documentation_master/_build/nrf/rst/doc/nrf/samples/nrf9160/serial_lte_modem/README.rst:33: WARNING: unknown option: CONFIG_SLM_CONNECT_UART_0
/jenkins_workspace/workspace/connect-nrf_documentation_master/_build/nrf/rst/doc/nrf/samples/nrf9160/serial_lte_modem/README.rst:49: WARNING: unknown option: CONFIG_SLM_CONNECT_UART_2

Please take a look at the CI failures.

b-gent · 2019-10-25T22:24:41Z

Please take a look at the CI failures.

I fixed (hopefully) the ones coming from this PR. Two of them are unrelated and come from a user guide.

b-gent · 2019-10-25T22:28:37Z

I added another commit that fixes some CI issues and adds a Testing section. Please have a look and squash.

My last comment is about the statement in line 140:
"A sample nRF52 client project is to be shared by other means."

It sounds super cryptic and I doubt it adds any value. Unless you have a good reason to keep it there, I'd suggest to remove.

Generally, this looks good to me as it is now.

ru-fu · 2019-10-28T07:39:31Z

Please take a look at the CI failures.

I fixed (hopefully) the ones coming from this PR. Two of them are unrelated and come from a user guide.

There seems to be a wrong merge in links.txt.

b-gent · 2019-10-28T11:15:39Z

There seems to be a wrong merge in links.txt.

That is right, check the file links.txt on master, around line 100:
https://github.com/NordicPlayground/fw-nrfconnect-nrf/blob/master/doc/nrf/links.txt

The entry
.. _`GitHub duplicate`: https://help.github.com/en/articles/duplicating-a-repository

is removed on this branch and causes doc CI to fail.

rlubos · 2019-10-28T15:55:12Z

Hi @junqingzou,
I've noticed today that the AT parameters parsing is broken in the sample, after the recent changes on master. I've looked into the problem briefly, and it seems that we might need to feed the parser with an entire command (not only the parameters) and tweak it slightly, to parse AT commands correctly (it seems that it's good at parsing responses/notifications, but not at the actual commands). Otherwise, the whole parameter list is treated as unquoted string, which is a valid parameter for several AT commands. I will have a better look at this tomorrow and try to propose a fix.

Lifting the review till we sort out the command parsing

junqingzou · 2019-10-30T01:08:52Z

@rlubos yes I confirmed it's related to the change in this commit)

rlubos · 2019-10-30T10:55:59Z

@junqingzou, I've proposed a fix for the AT parser in #1359. With this change, the at_cmd_parser is capable of parsing full AT commands (including custom AT# prefixed ones).

This + rlubos@7989d75 fixes the command parsing in the sample. Please let me know if you're fine with the changes, and feel free to squash it into your commit.

b-gent · 2019-11-05T10:34:53Z

I added a commit that should fix the doc CI , let's see if it works now.

The Serial LTE Modem (SLM) project is an enhancement to at_client sample - support TCP/IP proprietary AT commands - support GPS proprietary AT commands - support communication to external MCU over UART Like original at_client, all modem AT commands are still supported. More proprietary AT commands could be added in the future. Signed-off-by: Jun Qing Zou <jun.qing.zou@nordicsemi.no>

lemrey

LGTM

rlubos

LGTM, tested the sample and it works fine 👍

die4lie · 2020-05-22T09:02:55Z

Hello there. I still can't understand why with this example with frequent AT command sending nrf9160 bricks. AT client example works normally, but serial lte modem sample bricks very easily, just try to send few AT commands with less than 100 ms delay...

ru-fu · 2020-05-25T07:28:12Z

Hi - questions like this will get lost here. Please ask at https://devzone.nordicsemi.com/ and our support team can help you.

junqingzou requested review from lemrey and rlubos as code owners September 5, 2019 09:51

rlubos reviewed Sep 5, 2019

View reviewed changes

rlubos reviewed Sep 6, 2019

View reviewed changes

rlubos reviewed Sep 9, 2019

View reviewed changes

junqingzou commented Sep 17, 2019

View reviewed changes

rlubos requested a review from b-gent September 17, 2019 12:55

rlubos reviewed Sep 23, 2019

View reviewed changes

lemrey reviewed Sep 24, 2019

View reviewed changes

rlubos reviewed Oct 1, 2019

View reviewed changes

rlubos previously approved these changes Oct 3, 2019

View reviewed changes

rlubos previously approved these changes Oct 7, 2019

View reviewed changes

lemrey reviewed Oct 7, 2019

View reviewed changes

TerjeKWahl reviewed Oct 16, 2019

View reviewed changes

junqingzou requested a review from ru-fu as a code owner October 24, 2019 22:08

b-gent reviewed Oct 24, 2019

View reviewed changes

lemrey reviewed Oct 25, 2019

View reviewed changes

b-gent approved these changes Oct 25, 2019

View reviewed changes

rlubos mentioned this pull request Oct 30, 2019

at_cmd_parser: parse AT commands #1359

Merged

lemrey approved these changes Nov 5, 2019

View reviewed changes

rlubos approved these changes Nov 5, 2019

View reviewed changes

rlubos merged commit 5ff8db8 into nrfconnect:master Nov 5, 2019

pkchan mentioned this pull request Nov 6, 2019

drivers: at_cmd: corrected warning in set_notification_handler(). #1424

Merged

junqingzou deleted the serial_lte_modem branch November 6, 2019 10:36

samples: nrf9160: serial LTE modem sample #1099

samples: nrf9160: serial LTE modem sample #1099

Conversation

junqingzou commented Sep 5, 2019 • edited Loading

NordicBuilder commented Sep 5, 2019 • edited Loading

Gitlint issues

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lemrey commented Sep 5, 2019 • edited Loading

junqingzou commented Sep 6, 2019 • edited Loading

junqingzou commented Sep 6, 2019

junqingzou commented Sep 6, 2019

rlubos left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rlubos commented Sep 6, 2019

rlubos left a comment

Choose a reason for hiding this comment

rlubos left a 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

junqingzou Sep 17, 2019 • 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

junqingzou left a comment

Choose a reason for hiding this comment

rlubos commented Sep 17, 2019

rlubos commented Sep 17, 2019 • edited Loading

rlubos commented Sep 17, 2019

b-gent commented Sep 17, 2019

rlubos commented Sep 17, 2019

carlescufi commented Sep 17, 2019

carlescufi commented Sep 17, 2019

b-gent commented Sep 17, 2019

carlescufi commented Sep 17, 2019

junqingzou commented Sep 18, 2019

ru-fu commented Sep 18, 2019

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

junqingzou Sep 24, 2019 • 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

lemrey left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lemrey Sep 24, 2019 • 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

NordicBuilder commented Oct 1, 2019 • edited Loading

junqingzou commented Sep 5, 2019 •

edited

Loading

NordicBuilder commented Sep 5, 2019 •

edited

Loading

lemrey commented Sep 5, 2019 •

edited

Loading

junqingzou commented Sep 6, 2019 •

edited

Loading

junqingzou Sep 17, 2019 •

edited

Loading

rlubos commented Sep 17, 2019 •

edited

Loading

junqingzou Sep 24, 2019 •

edited

Loading

lemrey Sep 24, 2019 •

edited

Loading

NordicBuilder commented Oct 1, 2019 •

edited

Loading