[Lldb-commits] Fwd: buildbot failure in LLVM on lldb-x64-windows-ninja

Pavel Labath via lldb-commits lldb-commits at lists.llvm.org
Fri May 10 02:02:10 PDT 2019


On 10/05/2019 00:20, Greg Clayton via lldb-commits wrote:
> Can anyone that builds the windows lldb compile the following file as 
> the test suite would:
> 
> trunk/packages/Python/lldbsuite/test/python_api/thread/main2.cpp
> 
> and send me the binary that is produced by the test suite?
> 
> I see the windows bot is failing and need to see binary that we produce 
> so I can disassemble and see what is going wrong.
> 
> Greg
> 

I am pretty sure your commit actually "fixed" step over in this test on 
windows. The bot was red because this test was now unexpectedly passing.

What I happened was that due to bad unwinding support on windows, the 
lldb would previously step into some function, and then couldn't figure 
out where it is, so it aborted the step over. Now, it no longer attempts 
to step into the function, and so the step over completes successfully.

I've reverted r360397 and removed the windows XFAILs in 360407.

BTW: I think r360397 was a pretty bad way to "revert" a patch. If you 
think your commit has broken something, you should just revert the whole 
thing.


More information about the lldb-commits mailing list