Network interface: Add status callback register #5457

ghost · 2017-11-08T13:20:55Z

Description

This adds connection status callback registering, and connection status reporting to NetworkInterface, and implementation of these to CellularInterface, mesh and ethernet interfaces

Status

READY

Migrations

YES

In NetworkInterface new APIs:
-attach() for connection status callback registering
-get_connection_status()
-set_blocking()

In PPPCellularInterface:
-connection_status_cb() removed

Related PRs

Todos

Documentation needs to be updated

Deploy notes

Steps to test or reproduce

0xc0170 · 2017-11-08T13:46:25Z

The destination branch: feature-status-callbacks . Is that correct?

Would @hasnainvirk review this, anyone else should be asked for review for this feature?

IN DEVELOPMENT

Let us know once this be ready .

ghost · 2017-11-08T13:56:21Z

@0xc0170 Yes, the destination branch is feature-status-callbacks.

Possible list of reviewers:
@hasnainvirk, @SeppoTakalo, @kjbracey-arm, @mikaleppanen, @mikter, @AnttiKauppila

SeppoTakalo · 2017-11-08T13:59:17Z

@TeemuKultala
Please add reviewers here:

0xc0170 · 2017-11-08T14:12:29Z

@SeppoTakalo Note, there are currently some problems with access for reviewers (I can not assign most of the names mentioned here), we are investigating

hasnainvirk

Looks okay to me for going into feature-status-callbacks. I would suggest taking all NetworkInterface users in confidence. This is going to effect them mildly.

ghost · 2017-11-09T10:43:56Z

@SeppoTakalo, @0xc0170 I cannot modify the Reviewers list

SeppoTakalo · 2017-11-09T13:32:31Z

@kjbracey-arm Please review.

theotherjimmy · 2017-11-13T16:37:03Z

@TeemuKultala what does "Status cb pr1" mean? Could you update your PR title?

ghost · 2017-11-15T07:23:35Z

@theotherjimmy "Status cb pr1" means "status callback pull request 1". I try to find a better name for this or for the next pull request

0xc0170 · 2017-11-16T13:19:34Z

features/netsocket/NetworkInterface.h

+     *
+     *  @param status_cb The callback for status changes
+     */
+    virtual void register_status_callback(mbed::Callback<void(ConnectionStatusType)> status_cb);


should this follow attach() for attaching a callback?

0xc0170 · 2017-11-16T13:21:27Z

features/netsocket/NetworkInterface.h

@@ -31,6 +32,11 @@ class NetworkStack;
 */
 class NetworkInterface {
 public:
+
+    enum ConnectionStatusType {
+        down, local_up, global_up, undefined


This should start uppercase Down, etc, each of them on the new line, good if with doxygen description for each status type.

0xc0170 · 2017-11-16T13:24:37Z

features/netsocket/NetworkInterface.h

+     *
+     *  @return         The connection status according to ConnectionStatusType
+     */
+    virtual ConnectionStatusType get_connection_status();


why is this not connection_status_t ? same as nsapi_error_t on the line 152

0xc0170 · 2017-11-16T13:27:23Z

Implementations needs to be still added to mesh and ethernet interfaces

This should be part of this patch, thus this needs some work (not yet complete) ?

ghost · 2017-11-16T13:38:54Z

@0xc0170 I have now ethernet and mesh changes available, and also fixes for other things you mentioned, should I make a new pull request for those things, or add them to this same pull request?

0xc0170 · 2017-11-16T13:55:15Z

You can add them here, feel free to push updates anytime

ghost · 2017-11-20T09:39:12Z

@0xc0170 Ethernet and mesh changes added here now, so should the status change back to needs: review, or how to proceed?

0xc0170 · 2017-11-20T13:52:54Z

features/FEATURE_LWIP/lwip-interface/EthernetInterface.h

+     *
+     *  @param status_cb The callback for status changes
+     */
+    virtual void register_status_callback(mbed::Callback<void(connection_status_t, int)> status_cb);


Still not answered, is this regular way how to register a callback? many API in mbed has attach, shall this be also aligned?

What about how to register callback? I would like to have an alignment if possible or an argument why this is register_status_callback_name

0xc0170 · 2017-11-20T13:54:50Z

features/netsocket/nsapi_types.h

@@ -55,6 +55,14 @@ enum nsapi_error {
    NSAPI_ERROR_CONNECTION_TIMEOUT  = -3017,     /*!< connection timed out */
 };

+ typedef enum connection_status {
+    DISCONNECTED, 


this should follow the rest of enumeration values (add prefix NSAPI), plus more specification (NSAPI_CONNECTION_STATUS or at least NSAPI_STATUS). See line 55 and others

ghost · 2017-11-21T08:43:57Z

@0xc0170 register_status_callback renamed to attach, enum changed too

0xc0170 · 2017-11-21T10:32:50Z

@TeemuKultala There are lot of reviewers assigned but only two reviewed so far.
I restarted jenkins PR , if it fails , please review the results

@kjbracey-arm @hasnainvirk Review please

Update: attach now , I see. Thanks

kjbracey

Looks like this is going to work out okay; main issue is that it doesn't seem to have picked up on the main point of the review with @AnttiKauppila and @lauri-piikivi about the callback API, which is to make the "connection status change" just one possible event on the attach, so the new state becomes a subreason of that.

This is going to conflict thoroughly with #5558 - whatever this does is going to have to be reimplemented in line with its style. That needn't worry you excessively, but it may help concentrate the mind with regard to genericity.

kjbracey · 2017-11-23T08:26:49Z

features/FEATURE_LWIP/lwip-interface/EthernetInterface.cpp

 {
 }
+Callback<void(nsapi_connection_status_t, int)> EthernetInterface::_connection_status_cb = NULL;


These things mustn't be static. Nothing stopping you having multiple Ethernet interfaces.

kjbracey · 2017-11-23T08:31:45Z

features/FEATURE_LWIP/lwip-interface/EthernetInterface.h

+     *
+     *  @param status_cb The callback for status changes
+     */
+    virtual void attach(mbed::Callback<void(nsapi_connection_status_t, int)> status_cb);


The callback here isn't following the discussion we had in that 2017-11-08 meeting - that agreed generalising to be effectively (nsapi_event_t reason, uint32_t parameter), so for this initial event the "reason" would be "connection status change", and the parameter would be the nsapi_connection_status_t.

kjbracey · 2017-11-23T08:32:17Z

features/FEATURE_LWIP/lwip-interface/EthernetInterface.h

@@ -111,6 +123,12 @@ class EthernetInterface : public EthInterface
    char _ip_address[IPADDR_STRLEN_MAX];
    char _netmask[NSAPI_IPv4_SIZE];
    char _gateway[NSAPI_IPv4_SIZE];
+
+
+    static Callback<void(nsapi_connection_status_t, int)> _connection_status_cb;


As mentioned above - nothing should be static

kjbracey · 2017-11-23T08:38:08Z

features/FEATURE_LWIP/lwip-interface/lwip_stack.c

@@ -599,7 +617,12 @@ nsapi_error_t mbed_lwip_bringup_2(bool dhcp, bool ppp, const char *ip, const cha

    netif_set_default(&lwip_netif);
    netif_set_link_callback(&lwip_netif, mbed_lwip_netif_link_irq);
-    netif_set_status_callback(&lwip_netif, mbed_lwip_netif_status_irq);
+    if (status_cb) {
+        netif_set_status_callback(&lwip_netif, status_cb);


Don't you always need to call our callback to unblock the blocking connect? Oh, you've assuming the person doing the attach will call back in to our status callback.

No that doesn't make sense, and will cause complication with the stack generalisation (#5558). The callback to lwIP should always be mbed_lwip_netif_status_irq. That can call status_cb if set - you'll need to record its setting.

kjbracey · 2017-11-23T08:39:24Z

features/netsocket/NetworkInterface.cpp

+
+nsapi_error_t NetworkInterface::set_blocking(bool blocking)
+{
+    return NSAPI_ERROR_UNSUPPORTED;


If you're adding this, you should implement it for lwIP and Nanostack, which I don't think you are.

kjbracey · 2017-11-23T08:39:41Z

features/netsocket/NetworkInterface.h

+     *
+     *  @return         The connection status according to ConnectionStatusType
+     */
+    virtual nsapi_connection_status_t get_connection_status();


Should be const.

kjbracey · 2017-11-23T08:39:55Z

features/netsocket/NetworkInterface.h

+
+    /** Set blocking status of connect()
+     *
+     *  @param blocking true if connect is blocking


Note that the default is blocking.

kjbracey · 2017-11-23T08:40:47Z

features/netsocket/cellular/generic_modem_driver/PPPCellularInterface.cpp

@@ -263,9 +263,11 @@ PPPCellularInterface::PPPCellularInterface(FileHandle *fh, bool debug)
    _fh = fh;
    _debug_trace_on = debug;
    _stack = DEFAULT_STACK;
+    _connection_status_cb = NULL;
+    _ppp_cb = Callback<void(nsapi_connection_status_t, int)>(this, &PPPCellularInterface::ppp_status_cb);


I think you should be able to simplify this to callback(this, &PPPCellularInterface::ppp_status_cb). Although not sure why it's a variable at all? Just pass that straight to the bringup function.

kjbracey · 2017-11-23T08:41:50Z

features/netsocket/cellular/generic_modem_driver/PPPCellularInterface.cpp

@@ -398,7 +406,11 @@ bool PPPCellularInterface::nwk_registration(uint8_t nwk_type)

 bool PPPCellularInterface::is_connected()
 {
-    return dev_info.ppp_connection_up;
+    if (_connect_status == NSAPI_STATUS_DISCONNECTED) {


return __connect_status != NSAPI_STATUS_DISCONNECTED; perhaps?

cmonr · 2017-12-28T21:17:42Z

@TeemuKultala, please take another look. The UDP DTLS handshake test case appears to be failing.

ghost · 2018-01-02T06:09:09Z

#5720 should provide a fix for the UDP DTLS handshake issue

0xc0170 · 2018-01-02T16:54:22Z

/morph build

mbed-ci · 2018-01-02T17:17:08Z

Build : SUCCESS

Build number : 778
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/5457/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build

mbed-ci · 2018-01-02T18:15:17Z

Exporter Build : SUCCESS

Build number : 430
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/5457/

mbed-ci · 2018-01-02T19:22:49Z

Test : FAILURE

Build number : 613
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/5457/613

0xc0170 · 2018-01-03T10:10:21Z

/morph build

mbed-ci · 2018-01-03T10:35:58Z

Build : SUCCESS

Build number : 784
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/5457/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build

mbed-ci · 2018-01-03T11:20:52Z

Exporter Build : SUCCESS

Build number : 438
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/5457/

mbed-ci · 2018-01-03T13:02:32Z

Test : SUCCESS

Build number : 619
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/5457/619

This is the original content of feature-status-callbacks, reviewed in ARMmbed#5457

0xc0170 added the needs: review label Nov 8, 2017

hasnainvirk approved these changes Nov 8, 2017

View reviewed changes

adbridge requested review from SeppoTakalo, kjbracey, mikaleppanen, mikter and AnttiKauppila November 9, 2017 11:31

0xc0170 reviewed Nov 16, 2017

View reviewed changes

0xc0170 added needs: work and removed needs: review labels Nov 16, 2017

0xc0170 changed the title ~~Status cb pr1~~ Network interface: Add status callback register Nov 16, 2017

0xc0170 added needs: review and removed needs: work labels Nov 20, 2017

0xc0170 requested changes Nov 20, 2017

View reviewed changes

kjbracey suggested changes Nov 23, 2017

View reviewed changes

0xc0170 added the needs: work label Nov 23, 2017

Teemu Kultala added 16 commits January 3, 2018 08:22

Callback changes

5dde919

Unnecessary line breaks removed

b1633c6

default implementations into use

a20e02d

ethernet, mesh callbacks added

ace2320

local address info added

195a9da

Changes after review

16b7556

nsapi enum changes

079727b

doxygen change

ccc4e44

nonblocking connects added

e9f6b40

new state added to nsapi_connection_status_t

5356508

link callback change

d245f1f

bringup change

a8c0311

changes after review

7124dcf

types change

a572680

types change

fcac41c

header fix

2fcbf06

0xc0170 merged commit 469d74f into ARMmbed:feature-status-callbacks Jan 3, 2018

0xc0170 removed the needs: work label Jan 3, 2018

jarlamsa mentioned this pull request Feb 7, 2018

Status callbacks #6032

Merged

jarlamsa pushed a commit to jarlamsa/mbed-os that referenced this pull request Feb 9, 2018

NSAPI status callback changes

5a74826

This is the original content of feature-status-callbacks, reviewed in ARMmbed#5457

Network interface: Add status callback register #5457

Network interface: Add status callback register #5457

Conversation

ghost commented Nov 8, 2017 • edited by ghost Loading

Description

Status

Migrations

Related PRs

Todos

Deploy notes

Steps to test or reproduce

0xc0170 commented Nov 8, 2017

ghost commented Nov 8, 2017

SeppoTakalo commented Nov 8, 2017

0xc0170 commented Nov 8, 2017

hasnainvirk left a comment

Choose a reason for hiding this comment

ghost commented Nov 9, 2017

SeppoTakalo commented Nov 9, 2017

theotherjimmy commented Nov 13, 2017

ghost commented Nov 15, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

0xc0170 commented Nov 16, 2017

ghost commented Nov 16, 2017

0xc0170 commented Nov 16, 2017

ghost commented Nov 20, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ghost commented Nov 21, 2017

0xc0170 commented Nov 21, 2017 • edited Loading

kjbracey left a comment • 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

kjbracey Nov 23, 2017 • 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

cmonr commented Dec 28, 2017

ghost commented Jan 2, 2018

0xc0170 commented Jan 2, 2018

mbed-ci commented Jan 2, 2018

Triggering tests

mbed-ci commented Jan 2, 2018

mbed-ci commented Jan 2, 2018

0xc0170 commented Jan 3, 2018

mbed-ci commented Jan 3, 2018

Triggering tests

mbed-ci commented Jan 3, 2018

mbed-ci commented Jan 3, 2018

ghost commented Nov 8, 2017 •

edited by ghost

Loading

0xc0170 commented Nov 21, 2017 •

edited

Loading

kjbracey left a comment •

edited

Loading

kjbracey Nov 23, 2017 •

edited

Loading