STM32 RTC : write RTC time while LPTICKER is enabled #8286

jeromecoutant · 2018-10-01T10:03:47Z

Description

During RTC register write, only the date and time counters are set,
the SubSecond part is reset.
In the previous implementation, there was a quite long loop (up to 1 second).

@LMESTM @c1728p9 @mattbrown015

Pull request type

[X] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Breaking change

c1728p9 · 2018-10-02T14:07:41Z

targets/TARGET_STM/rtc_api.c

    struct tm timeinfo;
+    core_util_critical_section_enter();


Was something calling this outside a critical section?

I added protection, as rtc_write and rtc_read_lp functions can update the same variables

LMESTM

Request for few comments to be added

LMESTM · 2018-10-02T08:29:45Z

targets/TARGET_STM/rtc_api.c

-    /* Save current SSR */
-    uint32_t Read_SubSeconds = (uint32_t)(RTC->SSR);
+    /* RTC SubSeconds counter is reset to PREDIV_S_VALUE */
+    LP_last_RTC_time = (timeStruct.Seconds + timeStruct.Minutes * 60 + timeStruct.Hours * 60 * 60) * PREDIV_S_VALUE;


Does this always fit into a uint32_t ? or would a uint64_t be required ?
Also is LP_last_RTC_time in real seconds ? or µs ?

LMESTM · 2018-10-02T08:32:49Z

targets/TARGET_STM/rtc_api.c

-
-    /* Save current SSR */
-    uint32_t Read_SubSeconds = (uint32_t)(RTC->SSR);
+    /* RTC SubSeconds counter is reset to PREDIV_S_VALUE */


Not clear to me what the comment says and how it relates to the code below: where is Subsecond counter reset to PREDIV_S_VALUE ?

LMESTM · 2018-10-02T14:53:53Z

targets/TARGET_STM/rtc_api.c

@@ -341,11 +334,16 @@ static void RTC_IRQHandler(void)

 uint32_t rtc_read_lp(void)
 {
+    /* LPTICKER time is reading only time RTC register (in second) and add the sub-second part


Where is the sub-second part retrieved from and how is it computed ?

LMESTM · 2018-10-02T14:55:46Z

targets/TARGET_STM/rtc_api.c

+    /* LPTICKER time is reading only time RTC register (in second) and add the sub-second part
+    *  In order to get a continuous time information (each 24h = 86400s = 0x15180),
+    *  current time is added into an internal counter : LP_continuous_time
+    *  In order to get spent time since last counter update, current RTC time is saved : LP_last_RTC_time


what is "last counter update" ? do you mean RTC_write (like application set_time()) ?
Does this comment apply here or in RTC_write function ?

LMESTM · 2018-10-02T14:58:45Z

targets/TARGET_STM/rtc_api.c

@@ -358,17 +356,18 @@ uint32_t rtc_read_lp(void)
    timeinfo.tm_min  = RTC_Bcd2ToByte((uint8_t)((Read_time & (RTC_TR_MNT | RTC_TR_MNU)) >> 8));
    timeinfo.tm_sec  = RTC_Bcd2ToByte((uint8_t)((Read_time & (RTC_TR_ST  | RTC_TR_SU))  >> 0));

-    uint32_t RTC_time_s = timeinfo.tm_sec + timeinfo.tm_min * 60 + timeinfo.tm_hour * 60 * 60; // Max 0x0001-517F => * 8191 + 8191 = 0x2A2E-AE80
+    uint32_t RTC_time_s = (timeinfo.tm_sec + timeinfo.tm_min * 60 + timeinfo.tm_hour * 60 * 60) * PREDIV_S_VALUE + PREDIV_S_VALUE - Read_SubSeconds; // Max 0x0001-517F * 8191 + 8191 = 0x2A2E-AE80


is RTC_time_s still in seconds ? the operation seems to mix it with sub_seconds. If so, need to rename the variable.

LMESTM · 2018-10-02T15:11:43Z

targets/TARGET_STM/rtc_api.c

@@ -358,17 +356,18 @@ uint32_t rtc_read_lp(void)
    timeinfo.tm_min  = RTC_Bcd2ToByte((uint8_t)((Read_time & (RTC_TR_MNT | RTC_TR_MNU)) >> 8));
    timeinfo.tm_sec  = RTC_Bcd2ToByte((uint8_t)((Read_time & (RTC_TR_ST  | RTC_TR_SU))  >> 0));

-    uint32_t RTC_time_s = timeinfo.tm_sec + timeinfo.tm_min * 60 + timeinfo.tm_hour * 60 * 60; // Max 0x0001-517F => * 8191 + 8191 = 0x2A2E-AE80
+    uint32_t RTC_time_s = (timeinfo.tm_sec + timeinfo.tm_min * 60 + timeinfo.tm_hour * 60 * 60) * PREDIV_S_VALUE + PREDIV_S_VALUE - Read_SubSeconds; // Max 0x0001-517F * 8191 + 8191 = 0x2A2E-AE80

    if (LP_last_RTC_time <= RTC_time_s) {
        LP_continuous_time += (RTC_time_s - LP_last_RTC_time);


Would be great to explain in commit message how those 3 variables are used:
LP_continuous_time : is it the time in µsec since start-up ?
RTC_time_s: the current time that was just read from RTC register ?
LP_last_RTC_time: the last time read from RTC register, either in this function, or in RTC_write() ?)
I don't undertstand the "add 24h" case ...

This fix avoid a long waiting loop in rtc_write function, which was not acceptable in TICKLESS context. Implementation comments added. Global variable name has been updated for easier maintenance: - LPTICKER_counter is the U32 continuous tick counter - LPTICKER_RTC_time is the RTC time used to get the time difference between rtc_read_lp() calls

jeromecoutant · 2018-10-04T13:32:33Z

@LMESTM please check updates

NirSonnenschein · 2018-10-07T18:54:29Z

/morph build

mbed-ci · 2018-10-07T19:53:09Z

Build : SUCCESS

Build number : 3274
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/8286/

Triggering tests

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

mbed-ci · 2018-10-07T21:39:29Z

Exporter Build : FAILURE

Build number : 2899
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/8286/

mbed-ci · 2018-10-08T00:19:35Z

Test : SUCCESS

Build number : 3082
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/8286/3082

cmonr · 2018-10-09T00:01:20Z

/morph export-build

mbed-ci · 2018-10-09T01:40:47Z

Exporter Build : SUCCESS

Build number : 2909
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/8286/

0xc0170 added the needs: review label Oct 1, 2018

0xc0170 requested review from c1728p9 and a team October 1, 2018 10:08

jeromecoutant mentioned this pull request Oct 2, 2018

All STM should have tickless mode enabled #7153

Closed

c1728p9 reviewed Oct 2, 2018

View reviewed changes

c1728p9 approved these changes Oct 2, 2018

View reviewed changes

LMESTM reviewed Oct 2, 2018

View reviewed changes

jeromecoutant force-pushed the PR_RTCWRITE branch from 0362e53 to 83caa04 Compare October 4, 2018 13:31

0xc0170 approved these changes Oct 4, 2018

View reviewed changes

LMESTM approved these changes Oct 5, 2018

View reviewed changes

0xc0170 added needs: CI and removed needs: review labels Oct 5, 2018

cmonr added ready for merge release-version: 5.10.2 and removed needs: CI labels Oct 9, 2018

0xc0170 merged commit 18d6131 into ARMmbed:master Oct 9, 2018

0xc0170 removed the ready for merge label Oct 9, 2018

jeromecoutant deleted the PR_RTCWRITE branch October 9, 2018 15:34

mattbrown015 mentioned this pull request Oct 18, 2018

STM32: #8286 Causes Tickless RTC Mode App to Hang #8473

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

STM32 RTC : write RTC time while LPTICKER is enabled #8286

STM32 RTC : write RTC time while LPTICKER is enabled #8286

jeromecoutant commented Oct 1, 2018

c1728p9 Oct 2, 2018

jeromecoutant Oct 4, 2018

LMESTM left a comment

LMESTM Oct 2, 2018

LMESTM Oct 2, 2018

LMESTM Oct 2, 2018

LMESTM Oct 2, 2018

LMESTM Oct 2, 2018

LMESTM Oct 2, 2018

jeromecoutant commented Oct 4, 2018

NirSonnenschein commented Oct 7, 2018

mbed-ci commented Oct 7, 2018

mbed-ci commented Oct 7, 2018

mbed-ci commented Oct 8, 2018

cmonr commented Oct 9, 2018

mbed-ci commented Oct 9, 2018

STM32 RTC : write RTC time while LPTICKER is enabled #8286

STM32 RTC : write RTC time while LPTICKER is enabled #8286

Conversation

jeromecoutant commented Oct 1, 2018

Description

Pull request type

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LMESTM 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

Choose a reason for hiding this comment

jeromecoutant commented Oct 4, 2018

NirSonnenschein commented Oct 7, 2018

mbed-ci commented Oct 7, 2018

Triggering tests

mbed-ci commented Oct 7, 2018

mbed-ci commented Oct 8, 2018

cmonr commented Oct 9, 2018

mbed-ci commented Oct 9, 2018