sched/clock: Add clock_increase_sched_ticks() and clock_get_sched_ticks() by wangchdo · Pull Request #17192 · apache/nuttx

wangchdo · 2025-10-13T06:18:58Z

In the current NuttX kernel implementation, there is no function
to represent pure scheduler ticks.
The existing clock_systime_ticks() cannot serve this purpose,
because it keeps increasing even when the scheduler is not running
(when interrupt is blocked for a long time for example).
This happens since the returned value reflects real hardware time:
as long as the timer or alarm hardware is active,
clock_systime_ticks() will continuously increase.

This patch introduces two new functions that provide access to
scheduler ticks — a counter that only increases when the scheduler
itself is running.

Note: Please adhere to Contributing Guidelines.

Summary

add two new functions to maintain a pure scheduler related ticks in nuttx

Impact

new functions added, now impact to the existing nuttx functions

Testing

ostest passed on board a2g-tc397-5v-tft

ostest passed on board fvp-armv8r-aarch32

xiaoxiang781216 · 2025-10-13T06:29:16Z

@wangchdo should we update CLOCK_MONOTONIC or CLOCK_BOOTTIME too?

wangchdo · 2025-10-13T10:05:23Z

@wangchdo should we update CLOCK_MONOTONIC or CLOCK_BOOTTIME too?

I added update for clock_systime_timespec() and clock_resynchronize() please check.

xiaoxiang781216 · 2025-10-13T11:30:28Z

@wangchdo should we update CLOCK_MONOTONIC or CLOCK_BOOTTIME too?

I added update for clock_systime_timespec() and clock_resynchronize() please check.

from https://man7.org/linux/man-pages/man3/clock_gettime.3.html:

CLOCK_BOOTTIME (since Linux 2.6.39; Linux-specific)
              A nonsettable system-wide clock that is identical to
              CLOCK_MONOTONIC, except that it also includes any time that
              the system is suspended.  This allows applications to get a
              suspend-aware monotonic clock without having to deal with the
              complications of CLOCK_REALTIME, which may have discontinu‐
              ities if the time is changed using [settimeofday(2)](https://man7.org/linux/man-pages/man2/settimeofday.2.html) or similar.

CLOCK_BOOTTIME contain the suspend time, but CLOCK_MONOTONIC not. Since you add the new api to distinguish these two cases, we could implement CLOCK_MONOTONIC better now.

wangchdo · 2025-10-13T12:53:41Z

@wangchdo should we update CLOCK_MONOTONIC or CLOCK_BOOTTIME too?

I added update for clock_systime_timespec() and clock_resynchronize() please check.

from https://man7.org/linux/man-pages/man3/clock_gettime.3.html:
CLOCK_BOOTTIME (since Linux 2.6.39; Linux-specific)
              A nonsettable system-wide clock that is identical to
              CLOCK_MONOTONIC, except that it also includes any time that
              the system is suspended.  This allows applications to get a
              suspend-aware monotonic clock without having to deal with the
              complications of CLOCK_REALTIME, which may have discontinu‐
              ities if the time is changed using [settimeofday(2)](https://man7.org/linux/man-pages/man2/settimeofday.2.html) or similar.
CLOCK_BOOTTIME contain the suspend time, but CLOCK_MONOTONIC not. Since you add the new api to distinguish these two cases, we could implement CLOCK_MONOTONIC better now.

Hi @xiaoxiang781216

I uploaded a new commit to improve CLOCK_MONOTONIC, please help to check.

tinnedkarma · 2025-10-13T13:36:48Z

+ *
+ ****************************************************************************/
+
+clock_t clock_get_sched_ticks(void)


Are we worried only about the 64 and 32 bit architectures?
What about 16 and 8 bit supported architectures? It seems that this changes will affect all architectures if I am not mistaken.

Good point! NuttX still needs to support 8, 16, 32 and 64-bit arch

hi @acassis @hartmannathan

nuttx cloct_t now is uint32_t or uint64_t as defined in sys/types.h

It looks like nuttx clock module does not consider systems below 32-bit, the new function I added clock_get_sched_ticks() is in align with the current clock module implementation, below is the implementation before. @xiaoxiang781216 How do you think about this?

NuttX require the arch support at least 32bit int, and 64bit is optional, but the recent discuss suggesting to enforce 64bit for POSIX compliant. Please reference the discuss here: #14460

hartmannathan

Is the CI test failing because of lines like this:

====================================================================================
Configuration/Tool: sam4l-xplained/nsh,CONFIG_ARM_TOOLCHAIN_GNU_EABI
2025-10-13 16:53:05
------------------------------------------------------------------------------------
  Cleaning...
  Configuring...
  Disabling CONFIG_ARM_TOOLCHAIN_BUILDROOT_OABI
  Enabling CONFIG_ARM_TOOLCHAIN_GNU_EABI
  Building NuttX...
  [1/1] Normalize sam4l-xplained/nsh
/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found
/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found

(Maybe the script should be updated to output an actual Error: line that will be highlighted in the GitHub CI web interface?)

…ks() In the current NuttX kernel implementation, there is no function to represent pure scheduler ticks. The existing clock_systime_ticks() cannot serve this purpose, because it keeps increasing even when the scheduler is not running. This happens since the returned value reflects real hardware time: as long as the timer or alarm hardware is active, clock_systime_ticks() will continuously increase. This patch introduces two new functions that provide access to scheduler ticks — a counter that only increases when the scheduler itself is running. Signed-off-by: Chengdong Wang wangchengdong@lixiang.com

Replace direct usage of g_system_ticks with clock_increase_sched_ticks() and clock_get_sched_ticks() Signed-off-by: Chengdong Wang wangchengdong@lixiang.com

Improve nxclock_gettime(), to get the system time excluding the time that the system is suspended, when the clockid is CLOCK_MONOTONIC Signed-off-by: Chengdong Wang wangchengdong@lixiang.com

wangchdo · 2025-10-14T01:15:44Z

Is the CI test failing because of lines like this:

====================================================================================
Configuration/Tool: sam4l-xplained/nsh,CONFIG_ARM_TOOLCHAIN_GNU_EABI
2025-10-13 16:53:05
------------------------------------------------------------------------------------
  Cleaning...
  Configuring...
  Disabling CONFIG_ARM_TOOLCHAIN_BUILDROOT_OABI
  Enabling CONFIG_ARM_TOOLCHAIN_GNU_EABI
  Building NuttX...
  [1/1] Normalize sam4l-xplained/nsh
/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found
/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found

(Maybe the script should be updated to output an actual Error: line that will be highlighted in the GitHub CI web interface?)

Hi @hartmannathan,

Thank you for catching that.
I'm not very familiar with CI, but I think the build failure is unrelated to my PR.

hartmannathan · 2025-10-14T02:37:47Z

Is the CI test failing because of lines like this:
====================================================================================

Configuration/Tool: sam4l-xplained/nsh,CONFIG_ARM_TOOLCHAIN_GNU_EABI

2025-10-13 16:53:05

Cleaning...

Configuring...

Disabling CONFIG_ARM_TOOLCHAIN_BUILDROOT_OABI

Enabling CONFIG_ARM_TOOLCHAIN_GNU_EABI

Building NuttX...

[1/1] Normalize sam4l-xplained/nsh

/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found

/usr/bin/bash: line 1: arm-nuttx-elf-gcc: command not found
(Maybe the script should be updated to output an actual Error: line that will be highlighted in the GitHub CI web interface?)

Hi @hartmannathan,

Thank you for catching that.

I'm not very familiar with CI, but I think the build failure is unrelated to my PR.

I think you are correct: it seems the CI is missing the compiler for some platforms, but I don't know how this can be. I am out of time for today but hopefully someone who understands the CI better can help.

Enlarge kflash memory size for the sam3u-ek board to resolve the build error. Signed-off-by: Chengdong Wang <wangchengdong@lixiang.com>

jerpelea

according to the datasheet the flash size is 256k
https://ww1.microchip.com/downloads/en/DeviceDoc/doc6478.pdf

wangchdo · 2025-10-14T09:33:26Z

according to the datasheet the flash size is 256k https://ww1.microchip.com/downloads/en/DeviceDoc/doc6478.pdf

Hi @jerpelea,

The flash memory is divided into three regions. In this update, I have enlarged the kflash region and reduced the uflash region accordingly to fix the build issue. The total flash size remains unchanged as 256k .

Before:
  kflash (rx)  : ORIGIN = 0x00080000, LENGTH = 64K
  uflash (rx)  : ORIGIN = 0x00090000, LENGTH = 64K
  xflash (rx)  : ORIGIN = 0x000a0000, LENGTH = 128K

After:
  kflash (rx)  : ORIGIN = 0x00080000, LENGTH = 65K
  uflash (rx)  : ORIGIN = 0x00090400, LENGTH = 63K
  xflash (rx)  : ORIGIN = 0x000a0000, LENGTH = 128K

wangchdo · 2025-10-17T05:50:39Z

according to the datasheet the flash size is 256k https://ww1.microchip.com/downloads/en/DeviceDoc/doc6478.pdf

Hi @jerpelea,

The flash memory is divided into three regions. In this update, I have enlarged the kflash region and reduced the uflash region accordingly to fix the build issue. The total flash size remains unchanged as 256k .
Before:
  kflash (rx)  : ORIGIN = 0x00080000, LENGTH = 64K
  uflash (rx)  : ORIGIN = 0x00090000, LENGTH = 64K
  xflash (rx)  : ORIGIN = 0x000a0000, LENGTH = 128K

After:
  kflash (rx)  : ORIGIN = 0x00080000, LENGTH = 65K
  uflash (rx)  : ORIGIN = 0x00090400, LENGTH = 63K
  xflash (rx)  : ORIGIN = 0x000a0000, LENGTH = 128K

Hi @jerpelea

Do you have any further comments?

wangchdo · 2025-10-20T05:29:24Z

This PR duplicates PR #17199
. Closing it now.

wangchdo requested review from GUIDINGLI, jerpelea, raiden00pl, simbit18 and xiaoxiang781216 as code owners October 13, 2025 06:18

github-actions Bot added Area: OS Components OS Components issues Size: M The size of the change in this PR is medium labels Oct 13, 2025

wangchdo force-pushed the add_sched_ticks branch from 263eb50 to 8270190 Compare October 13, 2025 10:01

wangchdo force-pushed the add_sched_ticks branch from 8270190 to 643c871 Compare October 13, 2025 12:52

wangchdo force-pushed the add_sched_ticks branch 3 times, most recently from 34ce476 to 892eeb5 Compare October 13, 2025 12:57

tinnedkarma reviewed Oct 13, 2025

View reviewed changes

hartmannathan reviewed Oct 13, 2025

View reviewed changes

wangchdo added 2 commits October 14, 2025 08:59

sched/clock: Improve clock implementation with sched ticks functions

8cc1be4

Replace direct usage of g_system_ticks with clock_increase_sched_ticks() and clock_get_sched_ticks() Signed-off-by: Chengdong Wang wangchengdong@lixiang.com

wangchdo force-pushed the add_sched_ticks branch from 892eeb5 to eef64e8 Compare October 14, 2025 00:59

wangchdo requested review from acassis, hartmannathan and tinnedkarma October 14, 2025 01:09

sched/clock: Improve CLOCK_MONOTONIC nxclock_gettime()

fa1a28a

Improve nxclock_gettime(), to get the system time excluding the time that the system is suspended, when the clockid is CLOCK_MONOTONIC Signed-off-by: Chengdong Wang wangchengdong@lixiang.com

wangchdo force-pushed the add_sched_ticks branch from eef64e8 to fa1a28a Compare October 14, 2025 01:11

board/sam3u-ek: Adjust memory.ld to fix build error

4738c71

Enlarge kflash memory size for the sam3u-ek board to resolve the build error. Signed-off-by: Chengdong Wang <wangchengdong@lixiang.com>

wangchdo requested a review from anchao as a code owner October 14, 2025 04:48

wangchdo requested a review from gustavonihei as a code owner October 14, 2025 04:48

github-actions Bot added the Board: arm label Oct 14, 2025

jerpelea requested changes Oct 14, 2025

View reviewed changes

wangchdo requested a review from jerpelea October 14, 2025 09:33

xiaoxiang781216 approved these changes Oct 14, 2025

View reviewed changes

wangchdo closed this Oct 20, 2025

wangchdo deleted the add_sched_ticks branch October 20, 2025 05:46

Conversation

wangchdo commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Impact

Testing

Uh oh!

xiaoxiang781216 commented Oct 13, 2025

Uh oh!

wangchdo commented Oct 13, 2025

Uh oh!

xiaoxiang781216 commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

wangchdo commented Oct 13, 2025

Uh oh!

tinnedkarma Oct 13, 2025

Choose a reason for hiding this comment

Uh oh!

acassis Oct 13, 2025

Choose a reason for hiding this comment

Uh oh!

wangchdo Oct 14, 2025 • 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!

wangchdo Oct 14, 2025 • 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!

xiaoxiang781216 Oct 14, 2025

Choose a reason for hiding this comment

Uh oh!

hartmannathan left a comment

Choose a reason for hiding this comment

Uh oh!

wangchdo commented Oct 14, 2025

Uh oh!

hartmannathan commented Oct 14, 2025

Uh oh!

jerpelea left a comment

Choose a reason for hiding this comment

Uh oh!

wangchdo commented Oct 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

wangchdo commented Oct 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

wangchdo commented Oct 20, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

wangchdo commented Oct 13, 2025 •

edited

Loading

xiaoxiang781216 commented Oct 13, 2025 •

edited

Loading

wangchdo Oct 14, 2025 •

edited

Loading

wangchdo Oct 14, 2025 •

edited

Loading

wangchdo commented Oct 14, 2025 •

edited

Loading

wangchdo commented Oct 17, 2025 •

edited

Loading