tests: Bluetooth: tester: Workaround Bus Fault in nRF53x using full RAM #80990
Conversation
zephyrbot
added
area: Bluetooth Qualification
Why is this not causing an out-of-memory error when building? isn't it a problem that we have a silent out-of-memory issue during build? |
There was a problem hiding this comment.
Did you consider applying this to the hci_ipc sample instead?
There are no defined use cases for ISO/Audio that uses multiple BIGs, so this may be a better default.
There was a problem hiding this comment.
This can be done post v4.0 (we we decide to tag this as hotfix) to avoid too much changes/uncertainties to existing test executions.
There was a problem hiding this comment.
Not sure if we can consider this a hotfix as it is not causing an CI issues, or otherwise block anyone from working with Zephyr. This is unlikely to get merged before 4.1
There was a problem hiding this comment.
Dont we want a PTS run this weekend? or we wait for a run after the v4.0 release 2 week later?
There was a problem hiding this comment.
Please give a thumbs up if you want me to revert the overrides here and instead save RAM by reducing heap size?
Need to create a generic bug for not able to use the full RAM. Need to check on other targets and with hello world sample, if have a dummy array to utilize the full RAM causes bus faults due to kernel SYSINITs. |
|
#AutoPTS run zephyr AICS/SR/CP/BV-01-C |
|
Scheduled PR #80990 (comment), board: nrf53, estimated start time: 14:06:13, test case count: 1, estimated duration: 0:01:02 Test cases to be runAICS/SR/CP/BV-01-C |
|
AutoPTS Bot results: Successful tests (1)AICS AICS/SR/CP/BV-01-C PASS |
FYI, #81044 |
|
looking at #81044, maybe we should simply disable heap for hci samples? |
|
Do we need this PR. See this comment on Discord. |
|
maybe we just reduce CONFIG_HEAP_MEM_POOL_SIZE=8192 ? it seems to be 8k for some reason.. I just smoke tested this with 7k heap and it seems to work OK |
Workaround failing autopts weekly run due to tester
application having Bus Fault in the nRF53 NET core hci_ipc
firmware at startup.
Having atleast 32 bytes free in nRF53x hci_ipc sample to
avoid Bus Fault when Zephyr Kernel does SYSINIT calls.
Reduce supported ISO Broadcaster instance and ISO TX buffer
counts in the Controller hci_ipc firmware.
Add sysbuild support for tester application so that the
build is CI verified.
sysbuild commandline for LE audio PTS testing:
# west build --sysbuild -b nrf5340_audio_dk/nrf5340/cpuapp
-d build/tester tests/bluetooth/tester
-DEXTRA_CONF_FILE="overlay-le-audio.conf"
Signed-off-by: Vinayak Kariappa Chettimada <[email protected]>
cvinayak
force-pushed
the
github_tests_bluetooth_tester_fix
branch
from
154c89f to
234255b
Compare
cvinayak
added
the
Hotfix
|
@mmahadevan108 This hotfix PR is needed to be able to run weekly Bluetooth LE Audio Profile Tuning Suite tests hosted by CodeCoup |
7 participants
Workaround failing autopts weekly run due to tester application having Bus Fault in the nRF53 NET core hci_ipc firmware at startup.
Having atleast 32 bytes free in nRF53x hci_ipc sample to avoid Bus Fault when Zephyr Kernel does SYSINIT calls.
Reduce supported ISO Broadcaster instance and ISO TX buffer counts in the Controller hci_ipc firmware.
Add sysbuild support for tester application so that the build is CI verified.
sysbuild commandline for LE audio PTS testing:
Fixes #80852.