makefiles: quiet messages on CI

[MrKevinWeiss]

Contribution description

After introducing #18423 there are occasional messages that still happen.
These messages cause a diff output when testing with TEST_KCONFIG=1.
This then causes a failure when comparing make/kconfig modules and packages.

Testing procedure

rm -f dist/tools/kconfiglib/menuconfig.py && /bin/bash -c "source .murdock; compile tests/shell native:gnu"

Issues/PRs references

Introduced in #18423
Discovered by #18466

[MrKevinWeiss]

ping @maribu @benpicco @leandrolanzieri

[MrKevinWeiss]

Hmmm will that work or will it still show

git-cache: cloning from cache. tag=commit061e71f7d78cb057762d88de088055361863deff-1639

On the output failure we see it:

-- running on worker riotbuild-4 thread 2, build number 30999.
make: Entering directory '/tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/tests/test_tools'
make: Leaving directory '/tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/tests/test_tools'
make: Entering directory '/tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/tests/test_tools'
Building application "tests_test_tools" for "dwm1001" with MCU "nrf52".

sha1sum /tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/tests/test_tools/tests/01-run.py /tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/build/tests_test_tools.bin > /tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/build/test-input-hash.sha1
   text	   data	    bss	    dec	    hex	filename
  10088	    128	   2632	  12848	   3230	/tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/build/tests_test_tools.elf
make: Leaving directory '/tmp/dwq.0.6542732782276176/7c47d59c9617da36c7b13142d81eb830/tests/test_tools'
0a1,3
> [INFO] Kconfiglib not found - getting it
> git-cache: cloning from cache. tag=commit061e71f7d78cb057762d88de088055361863deff-1639
> [INFO] Kconfiglib downloaded
{"build/": 1864}

But locally I don't. Let me keep testing.

[MrKevinWeiss]

Nope still doesn't work... let me keep trying...

[MrKevinWeiss]

OK better test and better results. Now I am happy.

[MrKevinWeiss]

Heh, well that super didn't work... I will have to revisit it when I have time.

[MrKevinWeiss]

Strange error now:

-- running on worker mobi3 thread 17, build number 131395.
make: Entering directory '/tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/examples/suit_update'
# Reset package to checkout state.
rm -rf /tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/build/pkg/c25519
rm -rf /tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/build/pkg-build/libcose
rm -rf /tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/build/pkg-build/littlefs2
rm -rf /tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/build/pkg-build/nanocbor
make: Leaving directory '/tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/examples/suit_update'
make: Entering directory '/tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/examples/suit_update'
make: *** No rule to make target '/tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/examples/suit_update/bin/native/suit_update/fw.1661156187.bin', needed by 'link'.  Stop.
make: *** Waiting for unfinished jobs....
Building application "suit_update" for "native" with MCU "native".

make: Leaving directory '/tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/examples/suit_update'
cat: /tmp/dwq.0.5019332222865925/1890ed54c1c4f32158d9d853786b47a4/build/test-input-hash.sha1: No such file or directory

rerunning

[MrKevinWeiss]

Maybe a mobi3 specific problem? Running locally seems to work.

[MrKevinWeiss]

As my make knowledge is not so great maybe it would be worth looking at the code. I simply want to quiet the output if RIOT_CI_BUILD is 1, but still clone and build the kconfig library if it is missing.

If this passes it is probably good to merge in before starting other builds as this would prevent the false positives. My only concern is that maybe all the conditions are not hit in this run (ie, all the libs are build but if we clean out a worker it will throw different errors).

[benpicco]

What is : supposed to do?

[MrKevinWeiss]

Apparently the make way to optionally quiet the echos...

[MrKevinWeiss]

Very open to suggestions.

[benpicco]

huh TIL, I only knew about @ to silence echoing the command

[MrKevinWeiss]

We could also hit it from diff side. The whole issues is that there is extra output when using info-modules info-packages, I just do a simple diff. Maybe we can either strip extras or have different targets that are a bit more robust or so.

This is getting to one of those issues where in theory it is simple but reality becomes more and more complex with a whole bunch of small corner cases.

[MrKevinWeiss]

Pushed the fixes and squashed

[MrKevinWeiss]

Nice, thanks @benpicco. Let's hope that is the last of it.