resolve lwip init twice issue #4444

Archcady · 2017-06-05T12:06:06Z

resolve wifi example test fail before lwip is init
We find for REALTEK_RTL8195AM target, the wifi example will hang at mbed_lwip_bringup().

0xc0170 · 2017-06-05T12:18:23Z

features/FEATURE_LWIP/lwip-interface/lwip_stack.c

@@ -469,6 +469,7 @@ nsapi_error_t mbed_lwip_emac_init(emac_interface_t *emac)
    eth_arch_enable_interrupts();
 #endif

+	netif_inited = true;


Tab used? please fix alignment

0xc0170 · 2017-06-05T12:25:54Z

As this change is in the generic, how was this tested? It touches wifi, what about the ethernet? any other platform?

This variable is set in mbed_lwip_bringup, isnt that correct place?

Archcady · 2017-06-06T07:56:58Z

We set this netif_inited here is because there's multiple place doing init of the lwip. The first time is by our RTWInterface constructor, the second time is during mbed_lwip_bringup in RTWInterface.connect().
That's why we submit this PR to discuss.

resolve wifi example test fail before lwip is init

0xc0170 · 2017-06-09T15:13:07Z

@geky @kjbracey-arm Can you please review?

tung7970 · 2017-06-09T20:04:52Z

RTWInterface calls mbed_lwip_init() in constructor, and calls mbed_lwip_bringup() in its connect() when a wifi connection is requested.

// Backwards compatibility with people using DEVICE_EMAC
nsapi_error_t mbed_lwip_init(emac_interface_t *emac)
{
    mbed_lwip_core_init();
    return mbed_lwip_emac_init(emac);
}

// Backwards compatibility with people using DEVICE_EMAC
nsapi_error_t mbed_lwip_bringup(bool dhcp, const char *ip, const char *netmask, const char *gw)
{
    return mbed_lwip_bringup_2(dhcp, false, ip, netmask, gw);
}

The problem is that mbed_lwip_init() calls mbed_lwip_emac_init() without setting netif_inited to true, so mbed_lwip_emac_init() ends up called twice.

Hope this helps.

0xc0170 · 2017-06-12T12:11:02Z

@geky @kjbracey-arm Can you please review?

bump

tung7970 · 2017-06-12T14:11:06Z

For the EMAC case, I think the proper place to set netif_inited is in mbed_lwip_init, iff mbed_lwip_emac_init() returns OK.

diff --git a/features/FEATURE_LWIP/lwip-interface/lwip_stack.c b/features/FEATURE_LWIP/lwip-interface/lwip_stack.c
index a1fa5848a..de5278bd7 100644
--- a/features/FEATURE_LWIP/lwip-interface/lwip_stack.c
+++ b/features/FEATURE_LWIP/lwip-interface/lwip_stack.c
@@ -468,8 +468,14 @@ nsapi_error_t mbed_lwip_emac_init(emac_interface_t *emac)
 // Backwards compatibility with people using DEVICE_EMAC
 nsapi_error_t mbed_lwip_init(emac_interface_t *emac)
 {
 +    nsapi_error_t ret;
 + 
      mbed_lwip_core_init();
 -    return mbed_lwip_emac_init(emac);
 +    ret = mbed_lwip_emac_init(emac);
 +    if (ret == NSAPI_ERROR_OK) {
 +        netif_inited = true;
 +    }
 +    return ret;
 }

0xc0170 · 2017-06-13T08:56:01Z

As I understand, this is still not completely fixed @tung7970 ?

tung7970 · 2017-06-13T09:34:47Z

@0xc0170 The problem is mbed_lwip_emac_int is called twice, once in mbed_lwip_init(), and the other in mbed_lwip_bringup_2(). The bug was introduced in the EMAC backward fix.

As long as netif_inited is correctly set to true after it's inited, wifi example should work properly. Either @Archcady's fix or the patch I posted should work.

kjbracey · 2017-06-13T09:41:22Z

I prefer @tung7970's version - but @Archcady's would be okay.

0xc0170 · 2017-06-14T09:11:29Z

/morph test-nightly

geky

Looks good to me

mbed-bot · 2017-06-14T18:06:57Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 550

Test failed!

sg- · 2017-06-15T16:27:38Z

/morph test-nightly

mbed-bot · 2017-06-15T23:18:19Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 561

Test failed!

adbridge · 2017-06-16T09:41:56Z

ARCH_PRO-GCC_ARM.features-feature_lwip-tests-mbedmicro-net-gethostbyname.DNS query

[1497558667.02][CONN][RXD] DNS: query "connector.mbed.com" => "(null)"
[1497558667.06][CONN][RXD] :49::FAIL: Expected 0 Was -3009

ARCH_PRO-GCC_ARM.features-feature_lwip-tests-mbedmicro-net-tcp_hello_world.TCP hello world

[1497559442.79][CONN][RXD] TCP client IP Address is 10.118.13.255
[1497559442.83][CONN][RXD] HTTP: Connection to developer.mbed.org:80
[1497559456.78][CONN][RXD] HTTP: ERROR
[1497559456.81][CONN][RXD] :88::FAIL: Expression Evaluated To FALSE

K64F-GCC_ARM.features-feature_lwip-tests-mbedmicro-net-gethostbyname.DNS query

[1497564549.07][CONN][RXD] DNS: query "connector.mbed.com" => "(null)"
[1497564549.10][CONN][RXD] :49::FAIL: Expected 0 Was -3009

K64F-GCC_ARM.features-feature_lwip-tests-mbedmicro-net-tcp_hello_world.TCP hello world

[1497565521.38][CONN][RXD] TCP client IP Address is 10.118.13.255
[1497565521.43][CONN][RXD] HTTP: Connection to developer.mbed.org:80
[1497565534.86][CONN][RXD] HTTP: ERROR
[1497565534.90][CONN][RXD] :88::FAIL: Expression Evaluated To FALSE

So looks like all the failures were LWIP network failures when trying to communicate with developer.mbed.org !

0xc0170 · 2017-06-17T11:30:44Z

/morph test-nightly

mbed-bot · 2017-06-17T19:57:10Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 578

Test failed!

0xc0170 · 2017-06-19T08:02:30Z

Seems like network issue was back again the last weekend ? I restarted one job to confirm, waiting for a result
@geky @studavekar

adbridge · 2017-06-19T15:08:36Z

/morph test-nightly

mbed-bot · 2017-06-19T23:28:11Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 587

All builds and test passed!

0xc0170 requested review from geky and kjbracey June 5, 2017 12:17

0xc0170 reviewed Jun 5, 2017

View reviewed changes

0xc0170 added the needs: review label Jun 5, 2017

Archcady force-pushed the lwip_issue branch from 87ce384 to 2fb0e4b Compare June 6, 2017 03:22

Archcady added 2 commits June 7, 2017 12:04

resolve lwip init twice issue

ef1ddae

resolve wifi example test fail before lwip is init

replace tab with space

a812696

Archcady force-pushed the lwip_issue branch from 2fb0e4b to a812696 Compare June 7, 2017 04:08

Archcady mentioned this pull request Jun 13, 2017

[OOB 5.5.0] REALTEK_RTL8195AM cannot find any networks ARMmbed/mbed-os-example-wifi#35

Closed

A better approach to flip "netif_inited"

b1ca77b

0xc0170 added needs: CI release-version: 5.5.1 and removed needs: review labels Jun 13, 2017

0xc0170 mentioned this pull request Jun 14, 2017

[OOB 5.5.0] ublox_evk_odin_w2 never gets to main ARMmbed/mbed-os-example-wifi#36

Closed

andreaslarssonublox pushed a commit to u-blox/mbed-os that referenced this pull request Jun 14, 2017

resolve lwip init twice as done in ARMmbed#4444

f6f662e

kjbracey approved these changes Jun 14, 2017

View reviewed changes

geky approved these changes Jun 14, 2017

View reviewed changes

andreaslarssonublox mentioned this pull request Jun 15, 2017

Updated u-blox ODIN-W2 driver binaries to 2.1 rc1 #4567

Merged

adbridge mentioned this pull request Jun 16, 2017

Check Ethernet before including lwipopts_conf.h #4523

Merged

theotherjimmy merged commit c08864b into ARMmbed:master Jun 19, 2017

theotherjimmy added ready for merge and removed needs: CI ready for merge labels Jun 19, 2017

sarahmarshy mentioned this pull request Jun 28, 2017

Odin board fails to connect #4663

Closed

resolve lwip init twice issue #4444

resolve lwip init twice issue #4444

Uh oh!

Conversation

Archcady commented Jun 5, 2017

Uh oh!

0xc0170 Jun 5, 2017

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Jun 5, 2017

Uh oh!

Archcady commented Jun 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Jun 9, 2017

Uh oh!

tung7970 commented Jun 9, 2017 • edited by 0xc0170 Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Jun 12, 2017

Uh oh!

tung7970 commented Jun 12, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Jun 13, 2017

Uh oh!

tung7970 commented Jun 13, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kjbracey commented Jun 13, 2017

Uh oh!

0xc0170 commented Jun 14, 2017

Uh oh!

geky left a comment

Choose a reason for hiding this comment

Uh oh!

mbed-bot commented Jun 14, 2017

Result: FAILURE

Output

Uh oh!

sg- commented Jun 15, 2017

Uh oh!

mbed-bot commented Jun 15, 2017

Result: FAILURE

Output

Uh oh!

adbridge commented Jun 16, 2017

Uh oh!

0xc0170 commented Jun 17, 2017

Uh oh!

mbed-bot commented Jun 17, 2017

Result: FAILURE

Output

Uh oh!

0xc0170 commented Jun 19, 2017

Uh oh!

adbridge commented Jun 19, 2017

Uh oh!

mbed-bot commented Jun 19, 2017

Result: SUCCESS

Output

Uh oh!

Uh oh!

Archcady commented Jun 6, 2017 •

edited

Loading

tung7970 commented Jun 9, 2017 •

edited by 0xc0170

Loading

tung7970 commented Jun 12, 2017 •

edited

Loading

tung7970 commented Jun 13, 2017 •

edited

Loading