Test failures on s390x

[ellert]

• Checked for duplicates

Describe the bug

When running the unit tests on s390x there are several failures:

97% tests passed, 34 tests failed out of 1232
Label Time Summary:
longtest               = 572.89 sec*proc (27 tests)
multithreaded          = 889.21 sec*proc (2 tests)
python_runtime_deps    =  58.00 sec*proc (11 tests)
tutorial               = 4391.45 sec*proc (786 tests)
Total Test time (real) = 2976.41 sec
The following tests FAILED:
	 16 - pyunittests-pyroot-pyz-stl-vector (Failed)
	 57 - pyunittests-pyroot-pyz-rtensor (Failed)
	213 - gtest-roofit-histfactory-test-testHistFactory (Failed)
	219 - gtest-roofit-roofit-test-testRooCrystalBall (Failed)
	220 - gtest-roofit-roofit-test-testRooJohnson (Failed)
	223 - gtest-roofit-roofit-test-testSumW2Error (Failed)
	228 - gtest-roofit-roofitcore-test-testRooBinSamplingPdf (Failed)
	229 - gtest-roofit-roofitcore-test-testRooSimPdfBuilder (Failed)
	230 - gtest-roofit-roofitcore-test-testRooWrapperPdf (Failed)
	231 - gtest-roofit-roofitcore-test-testRooFitDriver (Failed)
	233 - gtest-roofit-roofitcore-test-testRooAbsPdf (Failed)
	237 - gtest-roofit-roofitcore-test-testRooProdPdf (Failed)
	241 - gtest-roofit-roofitcore-test-testTestStatistics (Failed)
	242 - gtest-roofit-roofitcore-test-testRooProductPdf (Failed)
	243 - gtest-roofit-roofitcore-test-testNaNPacker (Failed)
	244 - gtest-roofit-roofitcore-test-testRooSimultaneous (Failed)
	245 - gtest-roofit-roofitcore-test-testRooGradMinimizerFcn (Failed)
	247 - gtest-roofit-roofitcore-test-testLikelihoodSerial (Failed)
	248 - gtest-roofit-roofitcore-test-testRooRealL (Failed)
	249 - gtest-roofit-roofitcore-test-testGlobalObservables (Failed)
	252 - gtest-roofit-roostats-test-testSPlot (Failed)
	274 - test-stresshistogram (Failed)
	275 - test-stresshistogram-interpreted (Failed)
	296 - test-stresshistofit (Failed)
	297 - test-stresshistofit-interpreted (Failed)
	396 - gtest-tree-dataframe-test-datasource-ntuple (Failed)
	402 - gtest-tree-ntuple-v7-test-ntuple-basics (Failed)
	406 - gtest-tree-ntuple-v7-test-ntuple-merger (Failed)
	412 - gtest-tree-ntuple-v7-test-ntuple-serialize (Failed)
	420 - gtest-tree-ntuple-v7-test-ntuple-minifile (Failed)
	423 - gtest-tree-ntuple-v7-test-ntuple-extended (Failed)
	870 - tutorial-roofit-rf612_recoverFromInvalidParameters (Failed)
	1077 - tutorial-dataframe-df006_ranges-py (Failed)
	1106 - tutorial-math-exampleFunction-py (Failed)
Errors while running CTest

With the proposed change in #10303 to not fail on the warning about RooNaNPacker not being implemented for big endian, the list of failures is shorter:

99% tests passed, 17 tests failed out of 1232
Label Time Summary:
longtest               = 540.69 sec*proc (27 tests)
multithreaded          = 678.83 sec*proc (2 tests)
python_runtime_deps    =  60.39 sec*proc (11 tests)
tutorial               = 3802.05 sec*proc (786 tests)
Total Test time (real) = 2732.59 sec
The following tests FAILED:
	 16 - pyunittests-pyroot-pyz-stl-vector (Failed)
	 57 - pyunittests-pyroot-pyz-rtensor (Failed)
	237 - gtest-roofit-roofitcore-test-testRooProdPdf (Failed)
	243 - gtest-roofit-roofitcore-test-testNaNPacker (Failed)
	274 - test-stresshistogram (Failed)
	275 - test-stresshistogram-interpreted (Failed)
	296 - test-stresshistofit (Failed)
	297 - test-stresshistofit-interpreted (Failed)
	396 - gtest-tree-dataframe-test-datasource-ntuple (Failed)
	402 - gtest-tree-ntuple-v7-test-ntuple-basics (Failed)
	406 - gtest-tree-ntuple-v7-test-ntuple-merger (Failed)
	412 - gtest-tree-ntuple-v7-test-ntuple-serialize (Failed)
	420 - gtest-tree-ntuple-v7-test-ntuple-minifile (Failed)
	423 - gtest-tree-ntuple-v7-test-ntuple-extended (Failed)
	870 - tutorial-roofit-rf612_recoverFromInvalidParameters (Failed)
	1077 - tutorial-dataframe-df006_ranges-py (Failed)
	1106 - tutorial-math-exampleFunction-py (Failed)
Errors while running CTest

For both lists the proposed change in #10308 was applied.

Expected behavior

Ideally there should be no test failures.

To Reproduce

Steps to reproduce the behaviour:

1. build root for s390x.
2. run the unit tests.

Setup

1. ROOT version 6.26.02.
2. Fedora rawhide for s390x. The list of failures is similar for other Fedora and EPEL releases.
3. Build from source (during package build for Fedora/EPEL).

Additional context

The log

[eguiraud]

Assigning to Axel because I am not sure whether ROOT tries to support s390x at all.

[jblomer]

Regarding the RNTuple failures, one issue is the endianness (big-endian), which is addressed in #10402

[ferdymercury]

@ellert What is the status with current master ?
The PR with the ntuple issues has been merged.
Thanks!

[ferdymercury]

Pinging @ellert

[ellert]

Here is a s390x build using master from yesterday (Jun 18):
https://koji.fedoraproject.org/koji/taskinfo?taskID=134115604

Many RNTuple tests still fail:

The following tests FAILED:

  1 - cppinterop-CppInterOpTests (Failed)               DEPENDS
 13 - pyunittests-bindings-distrdf-backend-distrdf-unit-backend-graph-caching (Failed) python
 43 - pyunittests-bindings-pyroot-pythonizations-pyroot-pyz-stl-vector (Failed) python
 83 - pyunittests-bindings-pyroot-pythonizations-pyroot-pyz-rvec-asrvec (Failed) python python_runtime_deps
 84 - pyunittests-bindings-pyroot-pythonizations-pyroot-pyz-rdataframe-makenumpy (Failed) python python_runtime_deps
 89 - pyunittests-bindings-pyroot-pythonizations-pyroot-pyz-rtensor (Failed) python python_runtime_deps
126 - gtest-core-dictgen-dictgen-base (Failed)
144 - gtest-core-metacling-TClingTest (Failed)
424 - test-stresshistogram (Failed)                     longtest
425 - test-stresshistogram-interpreted (Failed)
436 - test-stresshistofit (Failed)                      longtest
437 - test-stresshistofit-interpreted (Failed)
450 - gtest-tmva-sofie-TestCustomModelsFromONNX (Failed)
543 - gtest-tree-dataframe-dataframe-snapshot-ntuple (Failed)
544 - gtest-tree-dataframe-dataframe-unified-constructor (Failed)
552 - gtest-tree-ntuple-ntuple-basics (Failed)
553 - gtest-tree-ntuple-ntuple-bulk (Failed)
554 - gtest-tree-ntuple-ntuple-cast (Failed)
557 - gtest-tree-ntuple-ntuple-compat (Failed)
562 - gtest-tree-ntuple-ntuple-join-table (Failed)
563 - gtest-tree-ntuple-ntuple-merger (Failed)
565 - gtest-tree-ntuple-ntuple-model (Failed)
566 - gtest-tree-ntuple-ntuple-multi-column (Failed)
567 - gtest-tree-ntuple-ntuple-packing (Timeout)
570 - gtest-tree-ntuple-ntuple-processor (Failed)
571 - gtest-tree-ntuple-ntuple-processor-chain (Failed)
572 - gtest-tree-ntuple-ntuple-processor-join (Failed)
573 - gtest-tree-ntuple-ntuple-project (Failed)
574 - gtest-tree-ntuple-ntuple-modelext (Failed)
577 - gtest-tree-ntuple-ntuple-types (Failed)
578 - gtest-tree-ntuple-ntuple-view (Failed)
581 - gtest-tree-ntuple-rfield-class (Failed)
583 - gtest-tree-ntuple-rfield-variant (Failed)
584 - gtest-tree-ntuple-rfield-vector (Failed)
586 - gtest-tree-ntuple-ntuple-show (Failed)
587 - gtest-tree-ntuple-ntuple-storage (Failed)
588 - gtest-tree-ntuple-ntuple-extended (Failed)
590 - gtest-tree-ntuple-ntuple-largefile2 (Failed)
592 - gtest-tree-ntuple-ntuple-parallel-writer (Failed)
594 - gtest-tree-ntuple-rfield-streamer (Failed)
595 - gtest-tree-ntuple-ntuple-storage-daos (Failed)
598 - gtest-tree-ntupleutil-v7-ntuple-importer (Failed)
737 - tutorial-hist-hist102_TH2_contour_list (Timeout)  tutorial
1266 - tutorial-analysis-dataframe-df006_ranges-py (Failed) tutorial
1290 - tutorial-hist-hist007_TH1_liveupdate-py (Failed)  tutorial
1317 - tutorial-math-exampleFunction-py (Failed)         tutorial
1318 - tutorial-math-fit-NumericalMinimization-py (Failed) python_runtime_deps tutorial
1319 - tutorial-math-fit-combinedFit-py (Failed)         python_runtime_deps tutorial
1475 - tutorial-visualisation-rcanvas-rbox-py (Failed)   tutorial

See the logs in the link above for details.

[jblomer]

I'm looking into them with a VM. It's a slow process so it will take a few days before I have results.

[jblomer]

I created a debug build and the only failing tests I see are the ones connected to quantized floats (@silverweed something we need to look into).

I'll create a release with debug info build to see if the situation changes.

[jblomer]

I get the same result with a RelWithDebugInfo build. Trying again with a Release build.

[ferdymercury]

Maybe related? #19411 and #19420

Also: #14512
#12431
#12429