ci: failure in win64 unit tests `err:virtual:virtual_alloc_first_teb wine: failed to map the shared user data: c0000018\n0444:err:winediag:nodrv_CreateWindow Application tried to create a window, but no driver could be loaded.\n0444:err:winediag:nodrv_CreateWindow L"Make sure that your X server is running and that $DISPLAY is set correctly."`) #30792

Seen in https://cirrus-ci.com/task/6387064294342656?logs=ci#L2492:

127/135 Test [#129](/bitcoin-bitcoin/129/): spend_tests ..........................   Passed    9.85 sec
128/135 Test [#135](/bitcoin-bitcoin/135/): db_tests .............................   Passed    3.32 sec
129/135 Test   [#1](/bitcoin-bitcoin/1/): util_test_runner .....................***Failed  154.46 sec
2024-09-02 15:49:05,155 - ERROR - Error parsing command output as hex: non-hexadecimal number found in fromhex() arg at position 1
2024-09-02 15:49:30,298 - ERROR - FAILED_TESTCASES:
['Creates a new transaction with one input, one address output and one data (zero value) output']

130/135 Test [#134](/bitcoin-bitcoin/134/): walletload_tests .....................   Passed    6.06 sec
131/135 Test [#131](/bitcoin-bitcoin/131/): wallet_tests .........................   Passed   16.91 sec
132/135 Test [#121](/bitcoin-bitcoin/121/): coinselector_tests ...................   Passed   35.08 sec
133/135 Test   [#5](/bitcoin-bitcoin/5/): noverify_tests .......................   Passed  186.81 sec
134/135 Test   [#6](/bitcoin-bitcoin/6/): tests ................................   Passed  264.45 sec
135/135 Test  [#10](/bitcoin-bitcoin/10/): allocator_tests ......................   Passed  266.16 sec

99% tests passed, 1 tests failed out of 135

Total Test time (real) = 266.20 sec

The following tests FAILED:
	  1 - util_test_runner (Failed)
Errors while running CTest

Interesting. I re-ran it 12 times on the same config/commit, but it passed every time: https://cirrus-ci.com/task/6025886804213760

It would be nice if the test failure was a bit more verbose to show the full traceback, as well as all values involved.

I wonder if it was an intermittent wine related issue, but the output was missing. Similar to something like: https://cirrus-ci.com/task/6309656165875712.

Yeah, possibly. I wonder why there is a presumed increase in those errors recently.

I was able to reproduce locally (in a CI container), but given that there is no useful output, it doesn't help:

# while ( LC_ALL=C.UTF-8 DIR_UNIT_TEST_DATA=/ci_container_base/ci/scratch/qa-assets/unit_test_data/ ctest --test-dir /ci_container_base/ci/scratch/build-x86_64-w64-mingw32 -j $( nproc ) ) ; do ( echo 1 >> /tmp/runs ) ; done

...

# cat /tmp/runs |wc -l
48

The log:

...
1/136 Testing: util_test_runner
1/136 Test: util_test_runner
Command: "/usr/bin/cmake" "-E" "env" "BITCOINUTIL=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-util.exe" "BITCOINTX=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-tx.exe" "/usr/bin/python3" "/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/test/util/test_runner.py"
Directory: /ci_container_base/ci/scratch/build-x86_64-w64-mingw32
"util_test_runner" start time: Oct 02 03:45 UTC
Output:
----------------------------------------------------------
2024-10-02 03:46:17,710 - ERROR - Error parsing command output as hex: non-hexadecimal number found in fromhex() arg at position 1
2024-10-02 03:47:38,991 - ERROR - FAILED_TESTCASES:
['Creates a new transaction with a single empty output script']
<end of output>
Test time = 144.83 sec
----------------------------------------------------------
Test Failed.
"util_test_runner" end time: Oct 02 03:47 UTC
"util_test_runner" time elapsed: 00:02:24
----------------------------------------------------------

134/136 Testing: walletdb_tests
...

Probably the fastest way to reproduce for now inside the CI container:

# while ( LC_ALL=C.UTF-8 BITCOINUTIL=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-util.exe BITCOINTX=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-tx.exe /usr/bin/python3 /ci_container_base/ci/scratch/build-x86_64-w64-mingw32/test/util/test_runner.py ) ; do ( echo '...' && echo 1 >> /tmp/runs ) ; done 
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
2024-10-02 08:34:47,513 - ERROR - Error parsing command output as json: Expecting value: line 1 column 1 (char 0)
2024-10-02 08:35:34,704 - ERROR - FAILED_TESTCASES:
['Parses a transaction with no inputs and a single output script (output in json)']

Couldn't reproduce so far after #31067, but I'll keep trying.

Ok, with the current 31067 (fa43c4f93ca5b40734ec9b3ff91b74acf3ed7cf2):

# while ( LC_ALL=C.UTF-8 BITCOINUTIL=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-util.exe BITCOINTX=/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-tx.exe /usr/bin/python3 /ci_container_base/ci/scratch/build-x86_64-w64-mingw32/test/util/test_runner.py ) ; do ( echo '...' && echo 1 >> /tmp/runs ) ; done 

...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
2024-10-10 20:06:36,340 - ERROR - Error parsing command output as json: 'Expecting value: line 1 column 1 (char 0)'; res: CompletedProcess(args=['/ci_container_base/ci/scratch/build-x86_64-w64-mingw32/src/bitcoin-tx.exe', '-json', '-create', 'outpubkey=0:02a5613bd857b7048924264d1e70e08fb2a7e6527d32b7ab1bb993ac59964ff397:WS', 'nversion=1'], returncode=0, stdout='', stderr='err:virtual:virtual_alloc_first_teb wine: failed to map the shared user data: c0000018\n0444:err:winediag:nodrv_CreateWindow Application tried to create a window, but no driver could be loaded.\n0444:err:winediag:nodrv_CreateWindow L"Make sure that your X server is running and that $DISPLAY is set correctly."\n')
2024-10-10 20:06:52,758 - ERROR - FAILED_TESTCASES:
['Creates a new transaction with a single pay-to-pub-key output, wrapped in P2SH (output as json)']

No idea how to fix this. I'd say to just replace the Wine tests finally: https://github.com/bitcoin/bitcoin/issues/31071

Let's move discussion to #31071