Revert "Fix pal tests build on osx-arm64"

[AaronRobinsonMSFT]

Checking #82052 regression.

Reverts #81604

[AaronRobinsonMSFT]

/azp list

[azure-pipelines]

CI/CD Pipelines for this repository:

• runtime-coreclr outerloop
• runtime-coreclr jitstress
• runtime-coreclr jitstressregs
• runtime-coreclr jitstress2-jitstressregs
• runtime-coreclr gcstress0x3-gcstress0xc
• runtime-coreclr gcstress-extra
• runtime-coreclr r2r-extra
• runtime-coreclr jitstress-isas-x86
• runtime-coreclr jitstress-isas-arm
• runtime-coreclr jitstressregs-x86
• runtime-coreclr libraries-jitstressregs
• runtime-coreclr libraries-jitstress2-jitstressregs
• runtime-coreclr r2r
• runtime-coreclr runincontext
• runtime-coreclr crossgen2
• runtime-libraries-coreclr outerloop
• runtime-libraries-coreclr outerloop-windows
• runtime-libraries-coreclr outerloop-linux
• runtime-libraries-coreclr outerloop-osx
• runtime
• runtime-libraries enterprise-linux
• runtime-libraries stress-http
• runtime-libraries stress-ssl
• runtime-dev-innerloop
• runtime-coreclr crossgen2 outerloop
• coreclr-release-outerloop-nightly
• runtime-coreclr crossgen2-composite
• runtime-jit-experimental
• runtime-coreclr libraries-jitstress
• dotnet-linker-tests
• runtime-coreclr ilasm
• runtime-coreclr crossgen2-composite gcstress
• runtime-staging
• runtime-richnav
• runtime-coreclr pgo
• runtime-coreclr libraries-pgo
• Antigen
• runtime-community
• Fuzzlyn
• runtime-coreclr superpmi-replay
• runtime-wasm
• runtime-coreclr superpmi-diffs
• runtime-coreclr superpmi-asmdiffs-checked-release
• runtime-extra-platforms
• jit-cfg
• runtime-wasm-perf
• runtime-llvm
• runtime-cet
• runtime-coreclr jitstress-random
• runtime-coreclr libraries-jitstress-random
• runtime-android-grpc-client-tests
• runtime-wasm-libtests
• runtime-wasm-non-libtests
• runtime-android
• runtime-androidemulator
• runtime-ioslike
• runtime-ioslikesimulator
• runtime-linuxbionic
• runtime-maccatalyst
• runtime-coreclr pgostress
• runtime-coreclr jitstress-isas-avx512
• runtime-tools-tests
• runtime-libraries-mono outerloop
• runtime-sanitized
• runtime-wasm-dbgtests
• runtime-wasm-optional

[AaronRobinsonMSFT]

/azp run runtime-coreclr outerloop

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[AaronRobinsonMSFT]

/cc @trylek

[AaronRobinsonMSFT]

@am11 All legs appear to be passing here after being reverted. Am I missing something?

[am11]

Am I missing something?

One of two reasons: 1) CI doesn't run PAL tests on osx-arm64 or 2) CI is using Xcode version < 14.2.

[AaronRobinsonMSFT]

Am I missing something?

One of two reasons: 1) CI doesn't run PAL tests on osx-arm64 or 2) CI is using Xcode version < 14.2.

Okay. I run an osx-arm64. I will give it a shot.

[am11]

The error #81604 (comment) which I was getting while building clr.paltests is the same which we saw earlier during src/tests build #77950 (comment).

This nopal test is basically testing system implementation of sprintf which we have avoided in places where PAL is not available. It was strange that switching to snprintf breaks NamedMutex tests on ARM. Maybe it's better to just delete test_sprintf, since we are not using this API from the system any longer? (we are using the one from PAL in a few places)

[janvorli]

It was strange that switching to snprintf breaks NamedMutex tests on ARM

@am11 it seems it actually fails on all architectures, see #82068

[am11]

@janvorli yup, I commented here yesterday: #82052 (comment) it's failing on arm{64} and linux-musl (all arch). So yes, it is not failing "everywhere". e.g. linux-x64 is green, but linux-musl-x64 isn't in this run: https://dev.azure.com/dnceng-public/public/_build/results?buildId=169947&view=results (both pipelines executed PAL tests)

[janvorli]

@am11 I am looking into why it fails.

[janvorli]

I have tried Ubuntu arm64 debug / checked and x64 Alpine debug/checked. I cannot repro the issue (neither the test failure nor the

paltests <PrintPalTests|TestName>
Either print list of all paltests by passing PrintPalTests, or run a single PAL test.

[janvorli]

Neither Release build reproes that for me. For Alpine, I was using the very same container that the CI uses. So I wonder if the fact that I am using the runpaltests.sh while the CI uses the runpaltestshelix.sh can have any effect. Trying to run that script now.

[janvorli]

Ok, so downloading the build artifacts from CI and running the test with them finally reproduced the problem (both the strange message and the error). I keep investigating.

[AaronRobinsonMSFT]

Maybe it's better to just delete test_sprintf, since we are not using this API from the system any longer? (we are using the one from PAL in a few places)

I would agree with this sentiment. The only scenario I think we need to continue to validate for printf APIs are those involving the "secure" (that is, _s suffix) APIs that perform truncation.

[janvorli]

This is wrong:

   196      processCommandLinePathLength += test_snprintf(&g_processCommandLinePath[processCommandLinePathLength], processCommandLinePathLength, "%s ", "threading/NamedMutex/test1/paltest_namedmutex_test1");

The length passed to that is not correct, it should be 4096 - processCommandLinePathLength. I'm not sure though how that would cause the failure.

The funny thing is that the reason why it failed only for the artifacts from the CI and not my local build from the same checkin on the same platform is the path length. I've put the CI artifacts to /home/helixbot/test while my build had them in /home/helixbot/runtime/artifacts/bin/coreclr/linux.x64.Debug/. When I copied my build artifacts to /home/helixbot/test, the issue started to repro too.

[am11]

@janvorli, thank you for looking into it. I was able to repro it with your steps. Could you please take a look at the fix #82122? I have used precise measurements.

[AaronRobinsonMSFT]

Deferring to proposed fix in #82122