STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker #7352

bcostm · 2018-06-27T13:12:34Z

Description

This is a fix for Issue #7316

As proposed by @c1728p9 the us_ticker_read() must be used while the SDK is not ready.

Tests:

rtc, time, sleep and tick tests OK on NUCLEO_L073RZ, NUCLEO_F070RB + ARM
HAL_Delay function tested OK inside a basic application with NUCLEO_F103RB
Run rtc, time, sleep and tick tests on all boards

Pull request type

[X] Fix
[ ] Refactor
[ ] New target
[ ] Feature
[ ] Breaking change

c1728p9 · 2018-06-27T13:48:24Z

targets/TARGET_STM/hal_tick_common.c

-    return ticker_read_us(get_us_ticker_data()) / 1000; // 1 ms tick is required for ST HAL
+    // 1 ms tick is required for ST HAL driver
+    if (mbed_sdk_inited) {
+        return (ticker_read_us(get_us_ticker_data()) / 1000);


This is going to start counting from 0 so there will likely be a time jump after the SDK is initialized. As a workaround you may want to record the time returned by us_ticker_read just before mbed_sdk_inited is set and use it as an offset.

Done in commit f785c23

c1728p9 · 2018-06-27T14:09:42Z

targets/TARGET_STM/hal_tick_common.c

+        return (ticker_read_us(get_us_ticker_data()) / 1000);
+    }
+    else {
+        return (us_ticker_read() / 1000);


With the 16 bit timer this will overflow in ~66ms. If anything in the SDK init takes longer than this it may get stuck as HAL_GetTick will never count higher than this during init. As a potential solution you could accumulate elapsed time rather than reading it directly. Something like:

uint32_t new_time = us_ticker_read(); elapsed_time += (new_time - prev_time) & 0xFFFF; // Only use the lower 16 bits prev_time = new_time; return elapsed_time / 1000;

Note - You'll need to make sure prev_time is reset in HAL_InitTick since this resets the timer count.

Done in commit f785c23

jeromecoutant · 2018-07-02T08:29:22Z

Hi
Note that this PR should also solve all PWM failed tests with CI shield.

bcostm · 2018-07-02T09:25:28Z

Today's status:

NUCLEO_F103RB: all RTC tests are OK with or without additional patch requested by Russ
NUCLEO_L073RZ, NUCLEO_F070RB, DISCO_L072CZ: tests-mbed_hal-rtc test is FAIL (RTC-Persist test case) with or without additional patch requested by Russ. On these 3 boards, the LPTICKER+LPTIM is used while no LPTICKER on F103RB. The fail appears after the device enters in deepsleep in RTC-sleep test case.

bcostm · 2018-07-04T08:48:38Z

With commit fcdd529 (thanks @jeromecoutant ), the time, rtc, sleep, tick tests are OK on NUCLEO_L073RZ, NUCLEO_F070RB boards.

We are going to perform more tests on more boards but it looks good now.

jeromecoutant · 2018-07-04T11:39:26Z

targets/TARGET_STM/sleep.c

@@ -161,6 +161,7 @@ void hal_deepsleep(void)
    // Disable IRQs
    core_util_critical_section_enter();

+    // Save the timer counter value in order to reprogram it after deepsleep


I would prefer a new function in hal_tick_common.c which saves TIM_MST needed registers before going into deepsleep:

CNT

CCR1

DIER

done in commit a838cb2

jeromecoutant · 2018-07-04T11:40:05Z

targets/TARGET_STM/sleep.c

    TIM_HandleTypeDef TimMasterHandle;
    TimMasterHandle.Instance = TIM_MST;
    __HAL_TIM_SET_COUNTER(&TimMasterHandle, EnterTimeUS);
+    __HAL_TIM_CLEAR_FLAG(&TimMasterHandle, TIM_FLAG_CC1);


Call here a function to restore TIM_MST registers

done in commit a838cb2

jeromecoutant · 2018-07-04T11:41:25Z

targets/TARGET_STM/hal_tick_common.c

+    else {
+        new_time = us_ticker_read();
+        elapsed_time += (new_time - prev_time) & 0xFFFF; // Only use the lower 16 bits
+        prev_time = new_time;


Is this really needed ?
What about 32b targets ?

commit 9523bcf do this only for 16bits timer

bcostm · 2018-07-05T13:27:18Z

Tests are OK now.

cmonr · 2018-07-05T16:06:03Z

/morph build

mbed-ci · 2018-07-05T17:03:37Z

Build : ABORTED

Build number : 2534
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/7352/

cmonr · 2018-07-05T18:38:30Z

Weird. Trying again.

/morph build

mbed-ci · 2018-07-05T20:42:45Z

Build : ABORTED

Build number : 2537
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/7352/

cmonr · 2018-07-05T22:20:37Z

/morph build

mbed-ci · 2018-07-06T02:51:32Z

Build : SUCCESS

Build number : 2547
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/7352/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build
/morph mbed2-build

mbed-ci · 2018-07-06T04:15:44Z

Exporter Build : SUCCESS

Build number : 2191
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/7352/

mbed-ci · 2018-07-06T08:59:26Z

Test : SUCCESS

Build number : 2302
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/7352/2302

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker

0xc0170 requested a review from a team June 27, 2018 13:54

0xc0170 added the needs: review label Jun 27, 2018

0xc0170 previously approved these changes Jun 27, 2018

View reviewed changes

c1728p9 reviewed Jun 27, 2018

View reviewed changes

0xc0170 added needs: work and removed needs: review labels Jun 28, 2018

jeromecoutant mentioned this pull request Jul 2, 2018

Timer.read_us takes 8.5µs, making wait_us unable to wait for less than that on Nucleo F103RB #7397

Closed

cmonr mentioned this pull request Jul 3, 2018

STM32: Refactor us_ticker files #7290

Merged

bcostm added 3 commits July 4, 2018 10:15

Use us_ticker_read while SDK is not ready

26cb388

HAL_GetTick returns elapsed time

f785c23

Re-enable IT CC1 after deepsleep

fcdd529

bcostm dismissed 0xc0170’s stale review via fcdd529 July 4, 2018 08:45

bcostm force-pushed the fix_rtc_ticker branch from 9c406ef to fcdd529 Compare July 4, 2018 08:45

jeromecoutant suggested changes Jul 4, 2018

View reviewed changes

bcostm added 2 commits July 4, 2018 16:43

Use elapsed time only for 16bit timer

9523bcf

save/restore timer registers before/after deepsleep

a838cb2

jeromecoutant approved these changes Jul 5, 2018

View reviewed changes

cmonr added needs: CI and removed needs: work labels Jul 5, 2018

cmonr added the release-version: 5.9.3 label Jul 5, 2018

0xc0170 requested a review from a team July 9, 2018 10:18

cmonr approved these changes Jul 9, 2018

View reviewed changes

cmonr merged commit bcec185 into ARMmbed:master Jul 9, 2018

0xc0170 removed the needs: CI label Jul 9, 2018

bcostm deleted the fix_rtc_ticker branch July 16, 2018 09:08

pan- pushed a commit to pan-/mbed that referenced this pull request Aug 22, 2018

Merge pull request ARMmbed#7352 from bcostm/fix_rtc_ticker

1b9c9be

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker #7352

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker #7352

bcostm commented Jun 27, 2018 •

edited

Loading

c1728p9 Jun 27, 2018

bcostm Jul 6, 2018

c1728p9 Jun 27, 2018

bcostm Jul 6, 2018

jeromecoutant commented Jul 2, 2018

bcostm commented Jul 2, 2018 •

edited

Loading

bcostm commented Jul 4, 2018 •

edited

Loading

jeromecoutant Jul 4, 2018

bcostm Jul 4, 2018

jeromecoutant Jul 4, 2018

bcostm Jul 4, 2018

jeromecoutant Jul 4, 2018

bcostm Jul 4, 2018

bcostm commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 6, 2018

mbed-ci commented Jul 6, 2018

mbed-ci commented Jul 6, 2018

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker #7352

STM32: Fix RTC test issue on targets using a 16-bit timer for us_ticker #7352

Conversation

bcostm commented Jun 27, 2018 • edited Loading

Description

Pull request type

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

jeromecoutant commented Jul 2, 2018

bcostm commented Jul 2, 2018 • edited Loading

bcostm commented Jul 4, 2018 • 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

bcostm commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 5, 2018

cmonr commented Jul 5, 2018

mbed-ci commented Jul 6, 2018

Triggering tests

mbed-ci commented Jul 6, 2018

mbed-ci commented Jul 6, 2018

bcostm commented Jun 27, 2018 •

edited

Loading

bcostm commented Jul 2, 2018 •

edited

Loading

bcostm commented Jul 4, 2018 •

edited

Loading