sys/ztimer: initial import #11874

kaspar030 · 2019-07-19T13:09:22Z

Contribution description

This PR contains a new timer subsystem meant to eventually replace xtimer. It has been in development for quite some time by @gebart and me.

Please refer to this document for a more detailed problem statement and implementation details.

The PR provides some unittests using a mock clock. This has been used successfully to find quite some bugs.

The PR does not enable or use ztimer for anything but tests. The idea is to get feedback on concept and API, fill the holes, see how it works. Eventually, if deemed better than xtimer, we can convert the codebase. There's an initial coccinelle script that can automatically do that (only fails on 64bit xtimer API calls for lack of ztimer alternative).

ztimer tries to re-use the same defaults and configuration as xtimer. With "USEMODULE += xtimer_on_ztimer", a large part of the code can be tested with ztimer as backend and xtimer on top. *update: this is selected automatically if xtimer and ztimer are in USEMODULE.

Update 2020-02-20

if both "xtimer" and "ztimer" are in USEMODULE, xtimer_on_ztimer will be used automatically.

Update 2019-12-09:

integrated ztimer_extend into the core ztimer code
re-integrated ztimer conversion using fractions (frac module)

Update 2019-12-10:

optimizations left for follow-ups:
- convert_frac uses runtime calculation for determining the magic numbers. This could easily made static for the few we actually use (e.g., 32768/1000000, 32768/1000, 32768/1000, 1024/1000, ...)
- some conversions require only simple shifts / multiplications, e.g., 250000/1000000. A specialized conversion module would save convert_frac's (little) overhead alltogether

Testing procedure

review API and concept
try ~~"USEMODULE += xtimer_on_ztimer"~~ "USEMODULE=ztimer make ..." on your board, with some of the xtimer tests

Issues/PRs references

maribu

The code is disabling type checking by force casting at many instances. Most of the casts are not needed and can be replaced by code that does not disable type checking. Converting from a generic to a specific (e.g. ztimer_dev_t * to ztimer_periph_t) could be moved to a static inline function. That will make refactoring easier, e.g. if the structure layout of ztimer_periph_t changed that ztimer_dev_t no longer is the first member.

Is there any case where a ztimer_base_t * does not point to a ztimer_t? If not, what is the reason to have ztimer_base_t?

sys/ztimer/periph.c

sys/ztimer/rtc.c

sys/ztimer/extend.c

sys/ztimer/init.c

bergzand · 2019-09-06T09:27:38Z

Converting from a generic to a specific (e.g. ztimer_dev_t * to ztimer_periph_t) could be moved to a static inline function. That will make refactoring easier, e.g. if the structure layout of ztimer_periph_t changed that ztimer_dev_t no longer is the first member.

Would the already existing container_of macro be the solution here? It keeps working when the structure layout of ztimer_periph_t changes.

bergzand · 2019-09-06T09:32:33Z

sys/include/ztimer.h

+                            uint32_t period);
+
+/**
+ * @brief   Put the calling thread to sleep for the specified number of ticks


Probably also valid for the other functions: This probably doesn't sleep exactly the number of ticks (or seconds) on every platform. Does this sleep at least ticks or at most ticks. Furthermore, is there a way to get the other behaviour, for example, sleep at least $duration to allow for a peripheral to initialize or sleep at most $duration to not miss a deadline.

I bet that this is impossible, as e.g. a thread with higher priority might get the CPU at the very moment the caller wants to wake up. As far as I know, RIOT never had the goal to provide hard real time features, but soft real time features instead.

Does this sleep at least ticks or at most ticks.

Well, the implementation sleeps "at least x ticks". I guess this information has to be added everywhere.

As far as I know, RIOT never had the goal to provide hard real time features, but soft real time features instead.

I think that is what the website states. Actually, it is more complicated. Parts of RIOT are fully hard real-time capable, some parts are not, some parts are only if certain constraints are honored. With care, it is perfectly possible to do hard real time with RIOT. but, you really have to know the internals, as documentation on the matter is severley lacking.
Still, I like that doing hard real-time with RIOT is possible.

maribu · 2019-09-06T09:50:36Z

Converting from a generic to a specific (e.g. ztimer_dev_t * to ztimer_periph_t) could be moved to a static inline function. That will make refactoring easier, e.g. if the structure layout of ztimer_periph_t changed that ztimer_dev_t no longer is the first member.

Would the already existing container_of macro be the solution here? It keeps working when the structure layout of ztimer_periph_t changes.

The good thing with the static inline funcitons is, if e.g. ztimer_dev_t * and ztimer_periph_t * will at some point no longer be convertible, removing the inline function would break (at compile time) all code that still tries to convert - including some out of tree code of RIOT users. That would be much better than hard to trace runtime bugs. (But using container_of() in that static inline function sounds like a very good idea to me.)

maribu

OK. The general architecture and the user facing API are fine. It adds no more complexity than required to allow different timer bases, and the advantages of this are obvious and outweigh the cost of one additional parameter by means.

Having the proposed plan in mind to bring this to master, so that it can be tested and gradually improved to the point it can replace xtimer, I'd say this PR is generally in the shape to get merged. (One exception: The TODO in the user facing API, see comment below. The function should also be documented properly, as it is part of the user facing API.)

sys/include/ztimer.h

kaspar030 · 2019-09-10T09:00:15Z

The code is disabling type checking by force casting at many instances. Most of the casts are not needed and can be replaced by code that does not disable type checking.

Please let's not have this discussion here. As implemented here, this is idiomatic C. I'd rather keep the style as is, and concentrate on more pressing issues. (I will fix all the unnececessary void * -> specific_t * casts.)

bergzand

Minor issues I've found so far

Makefile.dep

sys/include/ztimer.h

maribu · 2019-09-17T11:11:19Z

I think only four things are left so that this can be merged:

The missing include @bergzand pointed out
The unrelated change @bergzand pointed out
The TODO / missing doc for ztimer_diff()
The unnecessary casts from void *

SemjonWilke

I just found some documentation that's not to clear to me. I did not read the corresponding code yet.

sys/include/ztimer.h

kaspar030 · 2019-12-09T13:33:33Z

I've put quite some work into this. Please see the updated PR description.

There's also a more detailed document on why we actually started this and why I think it should replace xtimer: https://github.com/RIOT-OS/RIOT/wiki/ztimer-problem-statement-and-design-document

miri64 · 2019-12-09T13:42:03Z

There's also a more detailed document on why we actually started this and why I think it should replace xtimer: https://github.com/RIOT-OS/RIOT/wiki/ztimer-problem-statement-and-design-document

Maybe also advertise this on the devel mailinglist.

kaspar030 · 2019-12-09T13:46:43Z

Maybe also advertise this on the devel mailinglist.

ok

Co-authored-by: Joakim Nohlgård <joakim.nohlgard@eistec.se>

This commit adds logic to make xtimer use ztimer_usec as backend (instead of periph_timer). This allows ztimer_usec and xtimer to coexist. It also allows xtimer to profit from eventually implemented power mode blocking in ztimer's periph_timer backend.

jue89 · 2020-03-04T12:00:22Z

The SAMR30 looks fine!

./compile_and_test_for_board.py ../../.. samr30-xpro --with-test-only -j0 --applications="tests/xtimer* tests/ztimer*":

Results

INFO:samr30-xpro:Saving toolchain
INFO:samr30-xpro.tests/xtimer_drift:Board supported: True
INFO:samr30-xpro.tests/xtimer_drift:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_drift:Application has test: False
INFO:samr30-xpro.tests/xtimer_hang:Board supported: True
INFO:samr30-xpro.tests/xtimer_hang:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_hang:Application has test: True
INFO:samr30-xpro.tests/xtimer_hang:Run compilation
INFO:samr30-xpro.tests/xtimer_hang:Run test
INFO:samr30-xpro.tests/xtimer_hang:Run test.flash
INFO:samr30-xpro.tests/xtimer_hang:Success
INFO:samr30-xpro.tests/xtimer_longterm:Board supported: True
INFO:samr30-xpro.tests/xtimer_longterm:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_longterm:Application has test: False
INFO:samr30-xpro.tests/xtimer_msg:Board supported: True
INFO:samr30-xpro.tests/xtimer_msg:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_msg:Application has test: True
INFO:samr30-xpro.tests/xtimer_msg:Run compilation
INFO:samr30-xpro.tests/xtimer_msg:Run test
INFO:samr30-xpro.tests/xtimer_msg:Run test.flash
INFO:samr30-xpro.tests/xtimer_msg:Success
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Board supported: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Application has test: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run compilation
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run test
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run test.flash
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Success
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Board supported: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Application has test: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Run compilation
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Run test
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Run test.flash
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Success
INFO:samr30-xpro.tests/xtimer_now64_continuity:Board supported: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Application has test: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Run compilation
INFO:samr30-xpro.tests/xtimer_now64_continuity:Run test
INFO:samr30-xpro.tests/xtimer_now64_continuity:Run test.flash
INFO:samr30-xpro.tests/xtimer_now64_continuity:Success
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Board supported: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Application has test: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run compilation
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run test
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run test.flash
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Success
INFO:samr30-xpro.tests/xtimer_remove:Board supported: True
INFO:samr30-xpro.tests/xtimer_remove:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_remove:Application has test: True
INFO:samr30-xpro.tests/xtimer_remove:Run compilation
INFO:samr30-xpro.tests/xtimer_remove:Run test
INFO:samr30-xpro.tests/xtimer_remove:Run test.flash
INFO:samr30-xpro.tests/xtimer_remove:Success
INFO:samr30-xpro.tests/xtimer_reset:Board supported: True
INFO:samr30-xpro.tests/xtimer_reset:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_reset:Application has test: True
INFO:samr30-xpro.tests/xtimer_reset:Run compilation
INFO:samr30-xpro.tests/xtimer_reset:Run test
INFO:samr30-xpro.tests/xtimer_reset:Run test.flash
INFO:samr30-xpro.tests/xtimer_reset:Success
INFO:samr30-xpro.tests/xtimer_usleep:Board supported: True
INFO:samr30-xpro.tests/xtimer_usleep:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_usleep:Application has test: True
INFO:samr30-xpro.tests/xtimer_usleep:Run compilation
INFO:samr30-xpro.tests/xtimer_usleep:Run test
INFO:samr30-xpro.tests/xtimer_usleep:Run test.flash
INFO:samr30-xpro.tests/xtimer_usleep:Success
INFO:samr30-xpro.tests/xtimer_usleep_short:Board supported: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Application has test: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Run compilation
INFO:samr30-xpro.tests/xtimer_usleep_short:Run test
INFO:samr30-xpro.tests/xtimer_usleep_short:Run test.flash
INFO:samr30-xpro.tests/xtimer_usleep_short:Success
INFO:samr30-xpro.tests/ztimer_msg:Board supported: True
INFO:samr30-xpro.tests/ztimer_msg:Board has enough memory: True
INFO:samr30-xpro.tests/ztimer_msg:Application has test: True
INFO:samr30-xpro.tests/ztimer_msg:Run compilation
INFO:samr30-xpro.tests/ztimer_msg:Run test
INFO:samr30-xpro.tests/ztimer_msg:Run test.flash
INFO:samr30-xpro.tests/ztimer_msg:Success
INFO:samr30-xpro.tests/ztimer_overhead:Board supported: True
INFO:samr30-xpro.tests/ztimer_overhead:Board has enough memory: True
INFO:samr30-xpro.tests/ztimer_overhead:Application has test: True
INFO:samr30-xpro.tests/ztimer_overhead:Run compilation
INFO:samr30-xpro.tests/ztimer_overhead:Run test
INFO:samr30-xpro.tests/ztimer_overhead:Run test.flash
INFO:samr30-xpro.tests/ztimer_overhead:Success
INFO:samr30-xpro:Tests successful

kaspar030 · 2020-03-04T12:54:41Z

The SAMR30 looks fine!

Thanks for running these! Did you somehow export USEMODULE="ztimer ztimer_xtimer_compat"? If not, the xtimer tests just test xtimer. Using "ztimer_xtimer_compat" will make use of the API wrapper. (some tests might not compile due to the incomplete wrapper).

jue89 · 2020-03-04T13:03:37Z

Thanks for running these! Did you somehow export USEMODULE="ztimer ztimer_xtimer_compat"? If not, the xtimer tests just test xtimer. Using "ztimer_xtimer_compat" will make use of the API wrapper. (some tests might not compile due to the incomplete wrapper).

Good to know. I set an environment variable accordingly. (BTW: Good test if the compat layer is actually used :D)

env PORT=/dev/ttyACM1 USEMODULE="ztimer_xtimer_compat" ./compile_and_test_for_board.py ../../.. samr30-xpro --with-test-only -j0 --applications="tests/xtimer*"

Results

INFO:samr30-xpro:Saving toolchain
INFO:samr30-xpro.tests/xtimer_drift:Board supported: True
INFO:samr30-xpro.tests/xtimer_drift:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_drift:Application has test: False
INFO:samr30-xpro.tests/xtimer_hang:Board supported: True
INFO:samr30-xpro.tests/xtimer_hang:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_hang:Application has test: True
INFO:samr30-xpro.tests/xtimer_hang:Run compilation
INFO:samr30-xpro.tests/xtimer_hang:Run test
INFO:samr30-xpro.tests/xtimer_hang:Run test.flash
INFO:samr30-xpro.tests/xtimer_hang:Success
INFO:samr30-xpro.tests/xtimer_longterm:Board supported: True
INFO:samr30-xpro.tests/xtimer_longterm:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_longterm:Application has test: False
INFO:samr30-xpro.tests/xtimer_msg:Board supported: True
INFO:samr30-xpro.tests/xtimer_msg:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_msg:Application has test: True
INFO:samr30-xpro.tests/xtimer_msg:Run compilation
INFO:samr30-xpro.tests/xtimer_msg:Run test
INFO:samr30-xpro.tests/xtimer_msg:Run test.flash
INFO:samr30-xpro.tests/xtimer_msg:Success
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Board supported: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Application has test: True
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run compilation
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run test
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Run test.flash
INFO:samr30-xpro.tests/xtimer_msg_receive_timeout:Success
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Board supported: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Application has test: True
INFO:samr30-xpro.tests/xtimer_mutex_lock_timeout:Run compilation
WARNING:samr30-xpro.tests/xtimer_mutex_lock_timeout:make RIOT_CI_BUILD=1 CC_NOCOLOR=1 --no-print-directory -C ../../../tests/xtimer_mutex_lock_timeout clean all --jobs
Building application "tests_xtimer_mutex_lock_timeout" for "samr30-xpro" with MCU "saml21".

/home/jfi/Projects/RIOT/tests/xtimer_mutex_lock_timeout/main.c: In function 'cmd_test_xtimer_mutex_lock_timeout_long_unlocked':
/home/jfi/Projects/RIOT/tests/xtimer_mutex_lock_timeout/main.c:125:9: error: implicit declaration of function 'xtimer_mutex_lock_timeout' [-Werror=implicit-function-declaration]
  125 |     if (xtimer_mutex_lock_timeout(&test_mutex, LONG_MUTEX_TIMEOUT) == 0) {
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
make[1]: *** [/home/jfi/Projects/RIOT/Makefile.base:110: /home/jfi/Projects/RIOT/tests/xtimer_mutex_lock_timeout/bin/samr30-xpro/application_tests_xtimer_mutex_lock_timeout/main.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [/home/jfi/Projects/RIOT/tests/xtimer_mutex_lock_timeout/../../Makefile.include:541: /home/jfi/Projects/RIOT/tests/xtimer_mutex_lock_timeout/bin/samr30-xpro/application_tests_xtimer_mutex_lock_timeout.a] Error 2

Return value: 2

ERROR:samr30-xpro.tests/xtimer_mutex_lock_timeout:Error during compilation, writing to results/samr30-xpro/tests/xtimer_mutex_lock_timeout/compilation.failed
ERROR:samr30-xpro.tests/xtimer_mutex_lock_timeout:Failed during: compilation
INFO:samr30-xpro.tests/xtimer_now64_continuity:Board supported: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Application has test: True
INFO:samr30-xpro.tests/xtimer_now64_continuity:Run compilation
WARNING:samr30-xpro.tests/xtimer_now64_continuity:make RIOT_CI_BUILD=1 CC_NOCOLOR=1 --no-print-directory -C ../../../tests/xtimer_now64_continuity clean all --jobs
Building application "tests_xtimer_now64_continuity" for "samr30-xpro" with MCU "saml21".

/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c: In function 'main':
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:41:33: error: implicit declaration of function 'xtimer_diff64'; did you mean 'xtimer_now64'? [-Werror=implicit-function-declaration]
   41 |         xtimer_ticks64_t diff = xtimer_diff64(now, before);
      |                                 ^~~~~~~~~~~~~
      |                                 xtimer_now64
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:42:17: error: request for member 'ticks64' in something not a structure or union
   42 |         if (diff.ticks64 > diff_max) {
      |                 ^
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:43:28: error: request for member 'ticks64' in something not a structure or union
   43 |             diff_max = diff.ticks64;
      |                            ^
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:45:17: error: request for member 'ticks64' in something not a structure or union
   45 |         if (diff.ticks64 < diff_min) {
      |                 ^
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:46:28: error: request for member 'ticks64' in something not a structure or union
   46 |             diff_min = diff.ticks64;
      |                            ^
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:48:25: error: request for member 'ticks64' in something not a structure or union
   48 |         diff_sum += diff.ticks64;
      |                         ^
/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/main.c:41:26: error: variable 'diff' set but not used [-Werror=unused-but-set-variable]
   41 |         xtimer_ticks64_t diff = xtimer_diff64(now, before);
      |                          ^~~~
cc1: all warnings being treated as errors
make[1]: *** [/home/jfi/Projects/RIOT/Makefile.base:110: /home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/bin/samr30-xpro/application_tests_xtimer_now64_continuity/main.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [/home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/../../Makefile.include:541: /home/jfi/Projects/RIOT/tests/xtimer_now64_continuity/bin/samr30-xpro/application_tests_xtimer_now64_continuity.a] Error 2

Return value: 2

ERROR:samr30-xpro.tests/xtimer_now64_continuity:Error during compilation, writing to results/samr30-xpro/tests/xtimer_now64_continuity/compilation.failed
ERROR:samr30-xpro.tests/xtimer_now64_continuity:Failed during: compilation
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Board supported: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Application has test: True
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run compilation
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run test
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Run test.flash
INFO:samr30-xpro.tests/xtimer_periodic_wakeup:Success
INFO:samr30-xpro.tests/xtimer_remove:Board supported: True
INFO:samr30-xpro.tests/xtimer_remove:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_remove:Application has test: True
INFO:samr30-xpro.tests/xtimer_remove:Run compilation
INFO:samr30-xpro.tests/xtimer_remove:Run test
INFO:samr30-xpro.tests/xtimer_remove:Run test.flash
INFO:samr30-xpro.tests/xtimer_remove:Success
INFO:samr30-xpro.tests/xtimer_reset:Board supported: True
INFO:samr30-xpro.tests/xtimer_reset:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_reset:Application has test: True
INFO:samr30-xpro.tests/xtimer_reset:Run compilation
INFO:samr30-xpro.tests/xtimer_reset:Run test
INFO:samr30-xpro.tests/xtimer_reset:Run test.flash
INFO:samr30-xpro.tests/xtimer_reset:Success
INFO:samr30-xpro.tests/xtimer_usleep:Board supported: True
INFO:samr30-xpro.tests/xtimer_usleep:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_usleep:Application has test: True
INFO:samr30-xpro.tests/xtimer_usleep:Run compilation
INFO:samr30-xpro.tests/xtimer_usleep:Run test
INFO:samr30-xpro.tests/xtimer_usleep:Run test.flash
INFO:samr30-xpro.tests/xtimer_usleep:Success
INFO:samr30-xpro.tests/xtimer_usleep_short:Board supported: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Board has enough memory: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Application has test: True
INFO:samr30-xpro.tests/xtimer_usleep_short:Run compilation
INFO:samr30-xpro.tests/xtimer_usleep_short:Run test
INFO:samr30-xpro.tests/xtimer_usleep_short:Run test.flash
INFO:samr30-xpro.tests/xtimer_usleep_short:Success
ERROR:samr30-xpro:Tests failed: 2
Failures during compilation:
- [tests/xtimer_mutex_lock_timeout](tests/xtimer_mutex_lock_timeout/compilation.failed)
- [tests/xtimer_now64_continuity](tests/xtimer_now64_continuity/compilation.failed)

fjmolinas · 2020-03-04T15:56:37Z

Re ran with USEMODULE+="ztimer ztimer_xtimer_compat"

compile_and_test_for_board.py . BOARD --with-test-only --incremental -j0 --applications="tests/xtimer*

pba-d-01-kw2x/failuresummary

Failures during compilation:

nucleo-f207zg/failuresummary

Failures during compilation:

hifive1b/failuresummary

Failures during compilation:

nrf52dk/failuresummary

Failures during compilation:

nrf52840dk/failuresummary

Failures during compilation:

iotlab-m3/failuresummary

Failures during compilation:

frdm-k64f/failuresummary

Failures during compilation:

nucleo-l152re/failuresummary

Failures during compilation:

b-l475e-iot01a/failuresummary

Failures during compilation:

nucleo-f410rb/failuresummary

Failures during compilation:

arduino-uno/failuresummary

Failures during compilation:

tests/xtimer_now64_continuity

nucleo-f767zi/failuresummary

Failures during compilation:

arduino-mega2560/failuresummary

Failures during compilation:

i-nucleo-lrwan1/failuresummary

Failures during compilation:

openmote-b/failuresummary

Failures during compilation:

z1/failuresummary

Failures during compilation:

nucleo-f091rc/failuresummary

Failures during compilation:

arduino-zero/failuresummary

Failures during compilation:

nucleo-f103rb/failuresummary

Failures during compilation:

b-l072z-lrwan1/failuresummary

Failures during compilation:

nucleo-f303re/failuresummary

Failures during compilation:

slstk3402a/failuresummary

Failures during compilation:

nucleo-l073rz/failuresummary

Failures during compilation:

Only compilation failures for:

    tests/xtimer_mutex_lock_timeout
    tests/xtimer_now64_continuity

kaspar030 · 2020-03-04T16:05:24Z

Only compilation failures for:

    tests/xtimer_mutex_lock_timeout
    tests/xtimer_now64_continuity

So the other xtimer tests pass. They don't really check for quality of the results. This basically means that the difficult corner cases that are checked don't hang or crash with ztimer (which is nice).

bergzand

ztimer is fully optional and not intrusive. Default RIOT behavior is not affected. Tests showed no issues and the code is in good shape. Documentation is in good enough shape for this to be merged.

ACK!

kaspar030 mentioned this pull request Jul 19, 2019

esp32: define PERIPH_TIMER_PROVIDES_SET #11878

Merged

kaspar030 force-pushed the ztimer branch 2 times, most recently from 10af470 to 4b12eb5 Compare July 21, 2019 19:55

kaspar030 mentioned this pull request Aug 19, 2019

sys: make xtimer isr-safe #5428

Closed

maribu previously requested changes Sep 6, 2019

View reviewed changes

bergzand reviewed Sep 6, 2019

View reviewed changes

maribu reviewed Sep 10, 2019

View reviewed changes

sys/include/ztimer.h Outdated Show resolved Hide resolved

bergzand requested changes Sep 10, 2019

View reviewed changes

Makefile.dep Show resolved Hide resolved

sys/include/ztimer.h Show resolved Hide resolved

SemjonWilke reviewed Oct 29, 2019

View reviewed changes

sys/include/ztimer.h Outdated Show resolved Hide resolved

sys/include/ztimer.h Outdated Show resolved Hide resolved

sys/include/ztimer.h Outdated Show resolved Hide resolved

kaspar030 mentioned this pull request Oct 30, 2019

tests/periph_timer_short_relative_set: initial commit #12610

Merged

kaspar030 mentioned this pull request Nov 18, 2019

sys: xtimer concurrency/robustness improvement #9530

Merged

kaspar030 force-pushed the ztimer branch from d08071b to 1eeef8c Compare December 6, 2019 13:40

benpicco mentioned this pull request Dec 7, 2019

cpu/atmega256rfr2: symbol counter based RTT support #12852

Merged

kaspar030 removed Discussion: RFC The issue/PR is used as a discussion starting point about the item of the issue/PR State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet labels Dec 9, 2019

kaspar030 force-pushed the ztimer branch from 257872c to 93569bc Compare December 9, 2019 13:32

kaspar030 force-pushed the ztimer branch from 0863b12 to 0e2b3a7 Compare December 13, 2019 12:57

kaspar030 and others added 15 commits March 4, 2020 12:44

sys: add ztimer subsystem

6dd7936

Co-authored-by: Joakim Nohlgård <joakim.nohlgard@eistec.se>

unittests: Add ztimer test suite

39980d5

tests: add ztimer test applications

779e479

sys/ztimer: add coccinelle script for xtimer -> ztimer conversion

eccb6e8

sys/ztimer: introduce xtimer wrapper

616daff

tests/xtimer_drift: ztimer_xtimer_compat fix

4d3c103

boards/common/arduino-atmega: add ztimer configuration

eafba65

boards/samr21-xpro: add ztimer config

73adecc

boards/*-kw41z: add ztimer_msec config

84365e3

boards/native: add ztimer configuration

a2c0823

boards/ek-lm4f120xl: add ztimer config, increase ztimer_usec_min

f49e756

sys/ztimer: move dependencies to sys/ztimer/Makefile.dep

8c1e577

sys/ztimer: adapt to DEFAULT_MODULE based auto_init scheme

21613b0

sys/ztimer: add xtimer_on_ztimer

dd21833

This commit adds logic to make xtimer use ztimer_usec as backend (instead of periph_timer). This allows ztimer_usec and xtimer to coexist. It also allows xtimer to profit from eventually implemented power mode blocking in ztimer's periph_timer backend.

CODEOWNERS: add @kaspar030 and @bergzand as owners of ztimer

471f938

kaspar030 force-pushed the ztimer branch from 7b6fd0f to 471f938 Compare March 4, 2020 11:44

bergzand approved these changes Mar 6, 2020

View reviewed changes

bergzand merged commit ed31630 into RIOT-OS:master Mar 6, 2020

leandrolanzieri added this to the Release 2020.04 milestone Mar 6, 2020

jue89 mentioned this pull request Mar 6, 2020

sys/ztimer: fix RTT backend #13574

Merged

jue89 mentioned this pull request Mar 19, 2020

sys/evtimer: introduce ZTIMER_MSEC as timer backend #13661

Merged

kaspar030 deleted the ztimer branch March 30, 2020 19:28

leandrolanzieri added the Area: sys Area: System label Apr 24, 2020

sys/ztimer: initial import #11874

sys/ztimer: initial import #11874

Conversation

kaspar030 commented Jul 19, 2019 • edited Loading

Contribution description

Testing procedure

Issues/PRs references

maribu left a comment

Choose a reason for hiding this comment

bergzand commented Sep 6, 2019 • edited Loading

bergzand Sep 6, 2019

Choose a reason for hiding this comment

maribu Sep 10, 2019

Choose a reason for hiding this comment

kaspar030 Sep 10, 2019

Choose a reason for hiding this comment

maribu commented Sep 6, 2019 • edited Loading

maribu left a comment • edited Loading

Choose a reason for hiding this comment

kaspar030 commented Sep 10, 2019

bergzand left a comment

Choose a reason for hiding this comment

maribu commented Sep 17, 2019

SemjonWilke left a comment

Choose a reason for hiding this comment

kaspar030 commented Dec 9, 2019

miri64 commented Dec 9, 2019

kaspar030 commented Dec 9, 2019

jue89 commented Mar 4, 2020

kaspar030 commented Mar 4, 2020

jue89 commented Mar 4, 2020

fjmolinas commented Mar 4, 2020

pba-d-01-kw2x/failuresummary

nucleo-f207zg/failuresummary

hifive1b/failuresummary

nrf52dk/failuresummary

nrf52840dk/failuresummary

iotlab-m3/failuresummary

frdm-k64f/failuresummary

nucleo-l152re/failuresummary

b-l475e-iot01a/failuresummary

nucleo-f410rb/failuresummary

arduino-uno/failuresummary

nucleo-f767zi/failuresummary

arduino-mega2560/failuresummary

i-nucleo-lrwan1/failuresummary

openmote-b/failuresummary

z1/failuresummary

nucleo-f091rc/failuresummary

arduino-zero/failuresummary

nucleo-f103rb/failuresummary

b-l072z-lrwan1/failuresummary

nucleo-f303re/failuresummary

slstk3402a/failuresummary

nucleo-l073rz/failuresummary

kaspar030 commented Mar 4, 2020

bergzand left a comment

Choose a reason for hiding this comment

kaspar030 commented Jul 19, 2019 •

edited

Loading

bergzand commented Sep 6, 2019 •

edited

Loading

maribu commented Sep 6, 2019 •

edited

Loading

maribu left a comment •

edited

Loading