Since 5d66f9fd8e97c05a5dba317d3ad2566e61ead1ff this test has been upassing on Linaro's Windows on Arm lldb bot: https://lab.llvm.org/buildbot/#/builders/219/builds/4320 I can't explain exactly how that happened, but I do see a bunch of QEnvironment packets going by in that test. It is very likely that the order would have been different on Windows. Indeed, when it was xfailed back in df9051e7cfda5519f4584cda22e9ef2006517e94 the reason was not known either.