Cannot call a function in the debugee on Windows.

[llvmbot]

Bugzilla Link	21765
Version	unspecified
OS	Windows NT
Blocks	#22140
Reporter	LLVM Bugzilla Contributor
CC	@amccarth-google,@stella.stamenova

Extended Description

This can be reproduced via the following command line:

d:\src\llvm\build\ninja>D:\python_src\Python-2.7.8\PCbuild\python_d.exe D:\src\llvm\tools\lldb\test/dotest.py -t -q --arch=i686 --executable D:/src/llvm/build/ninja/bin/lldb.exe -s D:/src/llvm/build/ninja/lldb-test-traces -u CXXFLAGS -u CFLAGS -C D:\src\llvm\build\ninja\bin\clang.exe -p TestCallWithTimeout.py D:\src\llvm\tools\lldb\test\expression_command\timeout

At some point during the run of the test, the test tries to call a function that exists in the debugee. Parsing this expression fails and it says it cannot find the symbol in the target.

[llvmbot]

assigned to @amccarth-google

[llvmbot]

Assigning to majnemer as it might be a mangling issue, which he has some expertise in.

You can put the breakpoint in all overloads of SBFrame::EvaluateExpression in SBFrame.cpp to step through the failing method.

[llvmbot]

This is also affecting the following test suites.

TestCPPStaticMethods.py
TestCStrings.py
TestExprHelpExamples.py
TestExprs.py
TestFormatters.py
TestFunctionTypes.py
TestNestedPersistentTypes.py
TestPersistentTypes.py
TestPrintfAfterUp.py
TestSetValues.py

[amccarth-google]

The test that references this bug is passing, and the bug this one supposedly blocks is fixed, so I think this is fixed. I'm about to un-XFAIL the test (TestCallStdSTringFunction.test), at which point I'll close this one.

[amccarth-google]

The test is passing for the wrong reasons, so maybe this is still an issue. Needs more investigation.

[stella.stamenova]

TestCallOverriddenMethod
TestExprsBug35310

are also failing because of this bug

[llvmbot]

Can you try the patch here:

llvm/llvm-bugzilla-archive#37485

And see if it fixes the issue? This bug just came through today, I expect it will fix some kind of issues with the test suite, but it's hard to know what without trying it.

[stella.stamenova]

All of these tests are still failing with the patch

[llvmbot]

Oh well, was worth a try. Thanks for checking

[stella.stamenova]

This is the full list of tests that are marked as failing because of this bug right now:

TestCallWithTimeout.py
TestFormatters.py
TestFunctionTypes.py
TestCallOverriddenMethod.py
TestCallStdStringFunction.py
TestTargetAPI.py
TestNameLookup.py
TestSymbolAPI.py
TestDisasmAPI.py
TestCPPStaticMembers.py
TestXValuePrinting.py
TestIRInterpreter.py
TestExpressionInSyscall.py

Several that had been marked are now passing. It is possible that some of them are false positives, but there's a large number of unexpectedly passing tests on Windows now, so I am bulk un-xfailing them, so that we can make the Buildbot green.

[llvmbot]

mentioned in issue #22140

[dwblaikie]

Looks like maybe one instance of this got fixed by some ABI changes I made recently...
un-xfail: 40501f1
Probably causal change: 4769976