miri64
commented
4 years ago Closed miri64 closed 4 years ago
miri64
commented
4 years ago miri64
commented
4 years ago miri64
commented
4 years ago miri64
commented
4 years ago miri64
commented
4 years ago 04. Task 04 - ICMPv6 echo with iotlab-m3/samr21-xpro 15 minutes
FAILED
The problem with running tests as root ^^"
miri64
commented
4 years ago Commenting did not work for some reason for the success, but it marked it above, so :woman_shrugging:
miri64
commented
4 years ago No surprises for 2.1. They PASS except for the sudo tests and they also succeed when run with sudo
for test in tests/{gnrc_dhcpv6_client,gnrc_ipv6_ext,gnrc_ipv6_nib,gnrc_rpl_srh,gnrc_tcp}*; do
RIOT_CI_BUILD=1 BOARD=native make --no-print-directory -C ${test} -j16 flash && \
sudo BOARD=native make --no-print-directory -C ${test} test || break
doneThe usual suspects for 2.2 2.3. I'm a bit confused about tests/shell failing. I thought this was fixed by https://github.com/RIOT-OS/RIOT/pull/14537 :confused:
bergzand
commented
4 years ago The usual suspects for 2.2. I'm a bit confused about
tests/shellfailing. I thought this was fixed by RIOT-OS/RIOT#14537 confused
The samr21-xpro is not affected by the idle thread removal. The idle thread removal only affects the cortex-m3, m4 and m7's (dynamic priority modifications during the execution of the interrupt is required and is not supported on the m0(+)). I'm afraid this is unrelated to that issue/fix
miri64
commented
4 years ago I see. I just noticed also that, while I can reproduce this in 2020.07-RC1 with my local nodes (I used those this time), it does not show up in my test results from then. Those I did in the testbed. Will recheck that.
miri64
commented
4 years ago Ok.. it works in the testbed, so I suspect it might be just the revision of the board? oO
miri64
commented
4 years ago Same host machine, but different samr21-xpro deployments:
bergzand
commented
4 years ago Bisected this to commit https://github.com/RIOT-OS/RIOT/commit/68a35f53de2c881f82edb09ae5cd6f0d5a5f22ee, Not sure if this is an issue with EDBG or with the atsam itself.
fjmolinas
commented
4 years ago Seems like it might be realated to https://github.com/RIOT-OS/RIOT/issues/10639
miri64
commented
4 years ago More observations. It does not fail in the nightlies:
It does not fail when using an FTDI adapter connected to the samr21-xpro's UART0 (after flashing reconnect the USB to the TARGET USB port and reset manually when the test starts, otherwise you might get garbage):
Accessing EDBG via a Raspberry Pi (basically the setup in the nightlies):
So it seems to be an EDBG issue or some buffering issue with EDBG when the host machine is too fast (raspberry pi vs non-raspberry pi) or the latency on the connection between the host machine and the node is high (testbed).
bergzand
commented
4 years ago I can also reproduce the issue on the same54-xpro (also using EDBG).
miri64
commented
4 years ago The usual suspects for 2.2. I'm a bit confused about
tests/shellfailing. I thought this was fixed by RIOT-OS/RIOT#14537 confused
And lastly no surprises in 2.2.
miri64
commented
4 years ago If anyone wants to test the remaining test cases, feel free to do so. However, I consider the testing efforts for RC2 finished. So let's wait for the release notes (and maybe a potential fix for https://github.com/RIOT-OS/RIOT/issues/14548).
fjmolinas
commented
4 years ago Sorry for all the delays with the results on multiple boards. I had parallelized the builds and found some concurrency issues with the new pkg system, where with the -clean-after option called clean-pkg which erased the shared pkg tree. This means that the run had a lot of noise, I'll re-run wihthout --clean-after before uploading to the results directory, nonetheless here are the results. I went through all unexpected failures that didn't pass on individual re-runs. Those that where not documented in previous issues are in the following summary
tests/pkg_libfixmath_unittests seem to not be working on any board that can run it.
NOTE: in the build I excluded all tests requiring sudo. examples/micropython is excluded as well.
I'll open issues for the ones I don't find a fix before friday.
fjmolinas
commented
4 years ago Regarding:
### openmote-b/failuresummary
Failures during test:
- [tests/unittests](archive/openmote-b/tests/unittests/test.failed)Its actually a failure on tests/unittests/tests-rtt_rtc/tests-rtt_rtc.c.
fjmolinas
commented
4 years ago Regarding:
### openmote-b/failuresummary Failures during test: - [tests/unittests](archive/openmote-b/tests/unittests/test.failed)Its actually a failure on
tests/unittests/tests-rtt_rtc/tests-rtt_rtc.c.
Seems to be the common problem among boards failing the unittests.
fjmolinas
commented
4 years ago The two failures to flash on nrf52840dk are related to the usage of jlink, they work with openocd:
fjmolinas
commented
4 years ago All ffs failures on nrf52840dk seem to be related to my local board, all three pass on an iotlab board except for a timeout issue (Timeout doesn't seem very deterministic I had to increase it a lot for every test, took as much as 11min on one of them).
fjmolinas
commented
4 years ago tests/ssp on openmote-b fails just because the boards shut-downs before printing the message.
fjmolinas
commented
4 years ago tests/periph_wdt: succeeds for me locally on arduino-mega2560
PASSED target time: 128[ms], actual_time: 167.4[ms]
PASSED target time: 512[ms], actual_time: 669.716[ms]
PASSED target time: 1024[ms], actual_time: 1339.476[ms]
PASSED target time: 8192[ms], actual_time: 10019.32[ms]
TEST PASSEDClosing in favor of #175
This issue lists the status of all tests for the Release Candidate 2 of the 2020.07 release.
Specs tested: