michaelh/zephyr
1
0
Fork 0
Code Releases Activity

tests: latency_measure: Restructure configurations

Browse source 
Restructures the latency_measure benchmark configurations. Most of them
do not need to be executed by default by CI. However, we still want to
make it easy to run the benchmark with various configurations so that
they can be easily compared against the default.

The updated documentation shows how this can be done.

Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
This commit is contained in:
Peter Mitsis 2024-03-05 11:16:25 -05:00 • committed by Johan Hedberg
commit 886098a6c8
7 changed files with 44 additions and 86 deletions
Download patch file Download diff file Expand all files Collapse all files

29
tests/benchmarks/latency_measure/README.rst
View file

@ -26,18 +26,38 @@ including:
* Time it takes to push and pop to/from a k_stack
* Measure average time to alloc memory from heap then free that memory
When userspace is enabled using the prj_user.conf configuration file, this benchmark will
where possible, also test the above capabilities using various configurations involving user
threads:
When userspace is enabled, this benchmark will where possible, also test the
above capabilities using various configurations involving user threads:
* Kernel thread to kernel thread
* Kernel thread to user thread
* User thread to kernel thread
* User thread to user thread
The default configuration builds only for the kernel. However, additional
configurations can be enabled via the use of EXTRA_CONF_FILE.
For example, the following will build this project with userspace support:
EXTRA_CONF_FILE="prj.userspace.conf" west build -p -b <board> <path to project>
The following table summarizes the purposes of the different extra
configuration files that are available to be used with this benchmark.
A tester may mix and match them allowing them different scenarios to
be easily compared the default.
+-----------------------------+------------------------------------+
| prj.canaries.conf | Enable stack canaries |
+-----------------------------+------------------------------------+
| prj.objcore.conf | Enable object cores and statistics |
+-----------------------------+------------------------------------+
| prj.timeslicing.conf | Enable timeslicing |
+-----------------------------+------------------------------------+
| prj.userspace.conf | Enable userspace support |
+-----------------------------+------------------------------------+
Sample output of the benchmark (without userspace enabled)::
*** Booting Zephyr OS build zephyr-v3.5.0-4267-g6ccdc31233a3 ***
thread.yield.preemptive.ctx.k_to_k - Context switch via k_yield : 329 cycles , 2741 ns :
thread.yield.cooperative.ctx.k_to_k - Context switch via k_yield : 329 cycles , 2741 ns :
isr.resume.interrupted.thread.kernel - Return from ISR to interrupted thread : 363 cycles , 3033 ns :
@ -91,7 +111,6 @@ Sample output of the benchmark (without userspace enabled)::
Sample output of the benchmark (with userspace enabled)::
*** Booting Zephyr OS build zephyr-v3.5.0-4268-g6af7a1230a08 ***
thread.yield.preemptive.ctx.k_to_k - Context switch via k_yield : 970 cycles , 8083 ns :
thread.yield.preemptive.ctx.u_to_u - Context switch via k_yield : 1260 cycles , 10506 ns :
thread.yield.preemptive.ctx.k_to_u - Context switch via k_yield : 1155 cycles , 9632 ns :

5
tests/benchmarks/latency_measure/prj.canaries.conf Normal file
View file

@ -0,0 +1,5 @@
# Extra configuration file to enable stack canary support
# Use with EXTRA_CONF_FILE
CONFIG_TEST_RANDOM_GENERATOR=y
CONFIG_STACK_CANARIES=y

2
tests/benchmarks/latency_measure/prj.conf
View file

@ -1,3 +1,5 @@
# Default base configuration file
CONFIG_TEST=y
# eliminate timer interrupts during the benchmark

5
tests/benchmarks/latency_measure/prj.objcore.conf Normal file
View file

@ -0,0 +1,5 @@
# Extra configuration file to enable object core and statistics support
# Use with EXTRA_CONF_FILE
CONFIG_OBJ_CORE=y
CONFIG_OBJ_CORE_STATS=y

4
tests/benchmarks/latency_measure/prj.timeslicing.conf Normal file
View file

@ -0,0 +1,4 @@
# Extra configuration file to enable timeslicing support
# Use with EXTRA_CONF_FILE
CONFIG_TIMESLICING=y

4
tests/benchmarks/latency_measure/prj.userspace.conf Normal file
View file

@ -0,0 +1,4 @@
# Extra configuration file to enable userspace support
# Use with EXTRA_CONF_FILE
CONFIG_USERSPACE=y

81
tests/benchmarks/latency_measure/testcase.yaml
View file

@ -20,28 +20,6 @@ tests:
regex:
- "PROJECT EXECUTION SUCCESSFUL"
benchmark.kernel.latency.objcore.stats:
# FIXME: no DWT and no RTC_TIMER for qemu_cortex_m0
platform_exclude:
- qemu_cortex_m0
- m2gl025_miv
filter: CONFIG_PRINTK and not CONFIG_SOC_FAMILY_STM32
harness: console
integration_platforms:
- qemu_x86
- qemu_arc/qemu_arc_em
extra_configs:
- CONFIG_OBJ_CORE=y
- CONFIG_OBJ_CORE_STATS=y
harness_config:
type: one_line
record:
regex: "(?P<metric>.*) - (?P<description>.*):(?P<cycles>.*) cycles ,(?P<nanoseconds>.*) ns"
regex:
- "PROJECT EXECUTION SUCCESSFUL"
# Cortex-M has 24bit systick, so default 1 TICK per seconds
# is achievable only if frequency is below 0x00FFFFFF (around 16MHz)
# 20 Ticks per secondes allows a frequency up to 335544300Hz (335MHz)
@ -75,62 +53,3 @@ tests:
regex: "(?P<metric>.*) - (?P<description>.*):(?P<cycles>.*) cycles ,(?P<nanoseconds>.*) ns"
regex:
- "PROJECT EXECUTION SUCCESSFUL"
# Obtain the benchmark results with object core statistics enabled for
# various user thread / kernel thread configurations on platforms that
# support user space
benchmark.kernel.latency.userspace.objcore.stats:
filter: CONFIG_ARCH_HAS_USERSPACE
timeout: 300
harness: console
integration_platforms:
- qemu_x86
- qemu_cortex_a53
extra_configs:
- CONFIG_USERSPACE=y
- CONFIG_OBJ_CORE=y
- CONFIG_OBJ_CORE_STATS=y
harness_config:
type: one_line
record:
regex: "(?P<metric>.*) - (?P<description>.*):(?P<cycles>.*) cycles ,(?P<nanoseconds>.*) ns"
regex:
- "PROJECT EXECUTION SUCCESSFUL"
# Obtain the various kernel benchmark results with time slicing enabled
benchmark.kernel.latency.timeslicing:
# FIXME: no DWT and no RTC_TIMER for qemu_cortex_m0
platform_exclude:
- qemu_cortex_m0
- m2gl025_miv
filter: CONFIG_PRINTK and not CONFIG_SOC_FAMILY_STM32
harness: console
integration_platforms:
- qemu_x86
- qemu_arc/qemu_arc_em
extra_configs:
- CONFIG_TIMESLICING=y
harness_config:
type: one_line
record:
regex: "(?P<metric>.*) - (?P<description>.*):(?P<cycles>.*) cycles ,(?P<nanoseconds>.*) ns"
regex:
- "PROJECT EXECUTION SUCCESSFUL"
# Obtain the various userspace benchmark results with timeslicing enabled
benchmark.kernel.latency.timeslicing.userspace:
filter: CONFIG_ARCH_HAS_USERSPACE
timeout: 300
harness: console
integration_platforms:
- qemu_x86
- qemu_cortex_a53
extra_configs:
- CONFIG_USERSPACE=y
- CONFIG_TIMESLICING=y
harness_config:
type: one_line
record:
regex: "(?P<metric>.*) - (?P<description>.*):(?P<cycles>.*) cycles ,(?P<nanoseconds>.*) ns"
regex:
- "PROJECT EXECUTION SUCCESSFUL"