[WIZnet][W7500*] Support for mbed OS 5 #4226

khj098765 · 2017-04-26T00:05:34Z

Description

For support mbed OS 5 PR
Modifying the code for mbed ci shield test -> analogin_api.c, W7500x_adc.c, gpio_irq_api.c, W7500x_gpio.c
Modifying the code for fix bug -> W7500x_uart.c(register control driver problem), W7500x_uart.h, pinmap.c(pullup, pulldown problem)

Status

READY

Migrations

NO

Modifying the code for mbed ci shield test -> analogin_api.c, W7500x_adc.c, gpio_irq_api.c, W7500x_gpio.c Modifying the code for fix bug -> W7500x_uart.c(register control driver problem), W7500x_uart.h, pinmap.c(pullup, pulldown problem)

andrewc-arm · 2017-04-26T01:58:47Z

Hi,

I am the Partner Enablement Korea lead for WIZnet.
This PR is very important due to WIZnet/ARM co-marketing event which is due on May 1st.
WIZnet has done all the pretesting (Greentea + CI-test) and analysis already.
PE team and WIZnet have agreed on waiver of RTC and PWM test.
Except those, everything else works well.

Please let me know how we can expedite this process.
And due to the urgency, I will be co-leading this issue actively.

Thanks,

0xc0170 · 2017-04-26T08:09:38Z

targets/TARGET_WIZNET/TARGET_W7500x/W7500x_Peripheral_Library/W7500x_adc.c

@@ -42,6 +42,7 @@ void ADC_InterruptClear (void)
 void ADC_Init (void)
 {
 	// ADC_CLK on
+    ADC_PowerDownEnable(ENABLE);


please keep the indentation 4 spaces as it was (or this might be a tab ?)

Modifications completed

0xc0170 · 2017-04-26T08:11:55Z

targets/TARGET_WIZNET/TARGET_W7500x/pinmap.c

@@ -119,15 +119,31 @@ void pin_mode(PinName pin, PinMode pupd)

    switch(port_num)    {
        case PortA:
+            if(pupd != 0)


Use

if () { }

as the rest of this file

Modifications completed

0xc0170 · 2017-04-26T08:12:22Z

targets/TARGET_WIZNET/TARGET_W7500x/gpio_irq_api.c

@@ -110,10 +119,11 @@ int gpio_irq_init(gpio_irq_t *obj, PinName pin, gpio_irq_handler handler, uint32
    else
        obj->irq_n = PORT3_IRQn;

-    //obj->event = EDGE_FALL;


please remove any "dead" code like this, same o nthe line 109

Modifications completed

0xc0170 · 2017-04-26T08:12:42Z

targets/TARGET_WIZNET/TARGET_W7500x/gpio_irq_api.c

    port_generic_handler(GPIOA, 0);
+    test_tmp++;


test variable that should be removed?

Modifications completed

0xc0170 · 2017-04-26T08:14:23Z

targets/TARGET_WIZNET/TARGET_W7500x/analogin_api.c

@@ -58,11 +58,11 @@ void analogin_init(analogin_t *obj, PinName pin)
    obj->pin = pin;

    // The ADC initialization is done once
-    if (adc_inited == 0) {
+    //if (adc_inited == 0) {


second ADC pin will reinit entire ADC? What is this change fixing? What was wrong with the previous version?

For future reference, It would be beneficial if any logical change would have separate commit, like this one, would be explained via commit message why and what has changed.

Problems occurred when Two or more pins were used...
So we modified this code.

0xc0170 · 2017-04-26T08:16:57Z

WIZnet has done all the pretesting (Greentea + CI-test) and analysis already.

It would be good to share test results from mbed test run for these platforms that we are about to enable.

I left few comments, mostly style related.

Note, we are soon enough going to land new RTX2 to master.

- deleted dead code - deleted test code

khj098765 · 2017-04-26T23:57:58Z

modified this issues
we have already been shared to andrew about Greentea TEST & Ci TEST results.
have another problem?
Please tell us.

andrewc-arm · 2017-04-27T02:34:00Z

It would be good to share test results from mbed test run for these platforms that we are about to enable.

This is the test result log. WIZnet_W7500_test_logs.zip
Please note that I explicitly asked WIZnet to remove RTC and PWM test items since those will be waivered and added to the board's disclaimer.

0xc0170 · 2017-04-27T08:22:17Z

targets/TARGET_WIZNET/TARGET_W7500x/gpio_irq_api.c

@@ -119,7 +114,6 @@ int gpio_irq_init(gpio_irq_t *obj, PinName pin, gpio_irq_handler handler, uint32
    else
        obj->irq_n = PORT3_IRQn;

-    obj->event = EDGE_NONE;


why was this event removed? it's a style commit, but this is functional change? or dead code not used?

Sorry, It is a mistake. I put it back.

0xc0170 · 2017-04-27T08:25:12Z

targets/TARGET_WIZNET/TARGET_W7500x/gpio_irq_api.c

@@ -141,10 +145,14 @@ void gpio_irq_set(gpio_irq_t *obj, gpio_irq_event event, uint32_t enable)
            obj->rise_null = 0;
        }
        else if (event == IRQ_FALL) {
-            gpio->INTPOLSET &= ~obj->pin_index;
+            //gpio->INTPOLSET &= ~obj->pin_index;


still dead code in here?

Modifications completed

0xc0170 · 2017-04-27T08:26:12Z

targets/TARGET_WIZNET/TARGET_W7500x/pinmap.c

        case PortA:
+            if(pupd != 0)


should be

if () { }

Modifications completed

0xc0170 · 2017-04-27T08:31:24Z

Thanks for sharing the test results, all fine!

PE team and WIZnet have agreed on waiver of RTC and PWM test.

What is with RTC and PWM? Out of curiosity.

I left 2 minor comments, once reviewed, I'll run CI.

andrewc-arm · 2017-04-27T08:36:09Z

RTC: Real Time Clock. For accurate time application like digital clock or constant blip.
PWM: Pulse Width Modulation

modified coding style. -> pinmap.c

khj098765 · 2017-04-27T08:56:00Z

All edited.
Please merge pull request.

0xc0170 · 2017-04-27T09:42:44Z

RTC: Real Time Clock. For accurate time application like digital clock or constant blip.
PWM: Pulse Width Modulation

My question above was why RTC and PWM are waived - what will be added to the board page? The reason

0xc0170 · 2017-04-27T09:44:08Z

/morph test

mbed-bot · 2017-04-27T13:51:50Z

Result: FAILURE

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

/morph test

bridadan · 2017-04-27T13:54:12Z

targets/targets.json

@@ -2460,7 +2460,7 @@
        "supported_toolchains": ["uARM", "ARM"],
        "inherits": ["Target"],
        "device_has": ["ANALOGIN", "I2C", "INTERRUPTIN", "PORTIN", "PORTINOUT", "PORTOUT", "PWMOUT", "RTC", "SERIAL", "SPI", "SPISLAVE", "STDIO_MESSAGES"],
-        "release_versions": ["2"]
+        "release_versions": ["2", "5"]


You must support ARM, GCC_ARM, and the IAR compilers to enable the release version "5" (same goes for the below changes). That's why the CI failed.

No... We don't have time to support GCC_ARM and IAR.
It supports ARM compiler only for now.
We have May 1st contest coming.
Let me call you and others internally.

I asked WIZnet to change the supported toolchains. Let's see how this goes.

Added support for GCC and IAR.

andrewc-arm · 2017-04-28T02:37:33Z

My question above was why RTC and PWM are waived - what will be added to the board page? The reason

RTC:

Currently WIZnet board's capacitance of crystal oscillator need change. Since RTC can be outsourced, PE team decided to put this waiver temporarily.

PWM:

WIZnet's PWM works. However, WIZnet's GPIO does not support Dual Edge Triggering. Theirs only support Single Edge Triggering. Our current PWM test is based on InterruptInt with rise() and fall() handling using Dual Edge Triggering. For this urgent time being, PE team decided to skip the PWM test and support later.

About the Feature Documentation:
Yes. There should be feature documentation warning this limitation in their board page. This is the candidate message.

Due to low need, Dual Edge Triggered GPIO will not be supported.
- InterruptIn class’ rise() or fall() member function will override the previous callback setting. (ex. rise(A) -> fall(B): only fall(B) is active)
RTC will be temporarily unsupported and will be enabled later. (minor HW adjustment may be required)
PWM’s stability could not be tested for internal reasons.

andrewc-arm · 2017-04-28T02:46:18Z

Please ignore my comment about "urgency". I think we can relax and do this PR properly with time. :)

khj098765 · 2017-04-28T05:29:01Z

Added support for GCC and IAR.
completed?

bridadan · 2017-04-28T14:38:12Z

Awesome, thanks for adding that! I'll run the CI again.

/morph test

mbed-bot · 2017-04-28T15:09:48Z

Result: FAILURE

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

/morph test

Output

mbed Build Number: 107

Build failed!

0xc0170 · 2017-04-28T15:21:18Z

@khj098765 From the Ci above, you can review the logs with errors. One of them:

```[Fatal Error] RTX_Conf_CM.c@54,0: [Pe035]: #error directive: "no target defined"```` . Can you please test ? I would run mbed test for RTOS tests, seems like the target is not getting in for RTX (some definitions missing?)

bridadan · 2017-04-28T15:24:16Z

targets/TARGET_WIZNET/mbed_rtx.h

+#define OS_CLOCK                20000000
+#endif
+
+#endif


Build is failing because of this extra #endif I believe!

Yes. This should be moved to line 66.

@khj098765 Can you update?

I've fixed and updated this issue.

andrewc-arm · 2017-05-02T07:31:40Z

WIZnet is on vacation for this whole week. So they will be able to respond on next Monday.
Justin is the WIZnet contact person.

Justin, @khj098765, our team did the review of GCC_ARM and I got this comment. Please review.

Their startup code of GCC_ARM needs to call _start function in GCC runtime (then, the _start function calls main function).

https://github.com/ARMmbed/mbed-os/blob/master/targets/TARGET_WIZNET/TARGET_W7500x/TARGET_WIZwiki_W7500/device/TOOLCHAIN_GCC_ARM/startup_W7500.S#L191

https://github.com/ARMmbed/mbed-os/blob/master/targets/TARGET_WIZNET/TARGET_W7500x/TARGET_WIZwiki_W7500ECO/device/TOOLCHAIN_GCC_ARM/startup_W7500.S#L191

https://github.com/ARMmbed/mbed-os/blob/master/targets/TARGET_WIZNET/TARGET_W7500x/TARGET_WIZwiki_W7500P/device/TOOLCHAIN_GCC_ARM/startup_W7500.S

/*bl    _start*/ This code code should comment in (enabled)
bl main             This should be removed

This is explained here:
http://stackoverflow.com/questions/29694564/what-is-the-use-of-start-in-c

And similar issue raised here:
#3622
#889 (comment)

Also, I recommend to remove *.o file to avoid multiple link errors.

https://github.com/ARMmbed/mbed-os/tree/master/targets/TARGET_WIZNET/TARGET_W7500x/TARGET_WIZwiki_W7500ECO/device/TOOLCHAIN_GCC_ARM

theotherjimmy · 2017-05-02T15:19:14Z

Could we get a better name for this? "modify the code to support" is always happening.

khj098765 · 2017-05-08T00:08:31Z

I'm sorry I keep repeating.
There will be no more errors...
thank you

andrewc-arm · 2017-05-08T06:53:19Z

@khj098765, Justin, could you please change the title to "[WIZnet][W7500*] Support for mbed OS 5"? Or you can improvise as you wish. :)

andrewc-arm · 2017-05-08T06:54:40Z

/morph test

khj098765 · 2017-05-08T07:13:38Z

Yes I can change title to "[WIZnet][W7500*] Support for mbed OS 5"

0xc0170 · 2017-05-08T09:37:24Z

/morph test

0xc0170 · 2017-05-08T09:58:45Z

targets/TARGET_WIZNET/TARGET_W7500x/analogin_api.c

-        ADC_Init();
-    }
+    // The ADC initialization
+    ADC_Init();


I believe I asked earlier but cant find my comment. Why are you removing this static init variable? If I have 2 pins, this will reinit ADC 2x (power down and up as I can see in this changeset)? Is this desirable behavior? what was wrong with code as it was?

I talked with WIZnet Justin and he said this

Previously it worked fine. They thought ADC_Init() can be done only once and that was how it was implemented.

During the CI ADC test, they found that previous ADC run had some 'lingering effect'.

Now they realized that ADC should be hardware reset, which is done by calling ADC_Init() all the time when actual init is happening.

This code is now better and more intuitive and correctly resets the ADC hardware.

Interesting, hw reset peripheral for each ADC pin. If this is their hw requirement to restart adc like that for each adc channel initialization, ok.

mbed-bot · 2017-05-08T15:57:12Z

Result: SUCCESS

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

/morph test

Output

mbed Build Number: 165

All builds and test passed!

0xc0170 · 2017-05-09T12:12:44Z

@andrewc-arm happy with this patch as it is now?

@khj098765 Can you please answer ADC question above

andrewc-arm · 2017-05-10T01:22:48Z

@andrewc-arm happy with this patch as it is now?

Yes. But let me clear the ADC question you raised, I will talk to Justin why it was that way before and why it is this way now.

I just posted the reason for the ADC code change and I think this can be patched? :)

Resolved

andrewc-arm · 2017-05-12T07:50:26Z

Could you let us know when this will be merged to mbed-os repo? The WIZnet contest makers are waiting. :)

theotherjimmy · 2017-05-12T18:22:51Z

@andrewc-arm We will review this as a candidate for inclusion in mbed OS on Monday, May 15th. Sorry for the wait.

0xc0170 requested changes Apr 26, 2017

View reviewed changes

0xc0170 added the needs: work label Apr 26, 2017

khj098765 added 2 commits April 27, 2017 08:34

Modified coding style of mbed.

fe4edae

- deleted dead code - deleted test code

modified typo

415f975

khj098765 closed this Apr 26, 2017

khj098765 reopened this Apr 26, 2017

0xc0170 reviewed Apr 27, 2017

View reviewed changes

delete dead code. -> gpio_irq_api.c

6f654ea

modified coding style. -> pinmap.c

0xc0170 approved these changes Apr 27, 2017

View reviewed changes

0xc0170 added needs: CI and removed needs: work labels Apr 27, 2017

bridadan reviewed Apr 27, 2017

View reviewed changes

add to support gcc & iar

c71557c

bridadan previously requested changes Apr 28, 2017

View reviewed changes

0xc0170 added needs: work and removed needs: CI labels May 2, 2017

modify GCC_ARM startup code & mbed_rtx.h syntax error correct.

34da12a

khj098765 changed the title ~~Modifying the code to support mbed OS 5 -> add mbed_rtx.h~~ [WIZnet][W7500*] Support for mbed OS 5 May 8, 2017

0xc0170 reviewed May 8, 2017

View reviewed changes

0xc0170 added ready for merge and removed needs: work labels May 10, 2017

0xc0170 added release-version: 5.4.6 and removed ready for merge labels May 15, 2017

0xc0170 merged commit 444ff09 into ARMmbed:master May 15, 2017

[WIZnet][W7500*] Support for mbed OS 5 #4226

[WIZnet][W7500*] Support for mbed OS 5 #4226

Conversation

khj098765 commented Apr 26, 2017

Description

Status

Migrations

Uh oh!

andrewc-arm commented Apr 26, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Apr 26, 2017

Uh oh!

khj098765 commented Apr 26, 2017

Uh oh!

andrewc-arm commented Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Apr 27, 2017

Uh oh!

andrewc-arm commented Apr 27, 2017

Uh oh!

khj098765 commented Apr 27, 2017

Uh oh!

0xc0170 commented Apr 27, 2017

Uh oh!

0xc0170 commented Apr 27, 2017

Uh oh!

mbed-bot commented Apr 27, 2017

Uh oh!

bridadan Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

andrewc-arm commented Apr 28, 2017

Uh oh!

andrewc-arm commented Apr 28, 2017

Uh oh!

khj098765 commented Apr 28, 2017

Uh oh!

bridadan commented Apr 28, 2017

andrewc-arm commented Apr 27, 2017 •

edited

Loading

0xc0170 Apr 27, 2017 •

edited

Loading

bridadan Apr 27, 2017 •

edited

Loading

0xc0170 commented Apr 28, 2017 •

edited by theotherjimmy

Loading

andrewc-arm commented May 2, 2017 •

edited

Loading

khj098765 commented May 8, 2017 •

edited

Loading