<div dir="ltr">Hi all,<div><br></div><div style>There were some discussions about target specific tests breaking on other targets, so I did a simple test: compiled LLVM with a single target and ran all the tests.</div><div style>
<br></div><div style>Here are some results for random targets:</div><div style><br></div><div style>ARM: Unexpected Failures: 334</div><div style>PowerPC: Unexpected Failures: 340</div><div style>Mips: Unexpected Failures: 334</div>
<div style>X86: Unexpected Failures: 0</div><div><br></div><div style>Most of them LLVM-Unit, Execution Engine, Codegen, DebugInfo, Clang. Most of them exactly the same for all targets. Most of them are just bad tests, ie. target specific tests in generic directories, or target-independent tests expecting target specific behaviour.</div>
<div style><br></div><div style>It's safe to assume that each non-x86 target will have similar failure rate when compiled alone.</div><div style><br></div><div style>While it's important to test LLVM on the specific targets (compile and test on them), it's also important (and sometimes a lot easier and faster) to test cross-compilation. This would also guarantee that cross-compilation is not just possible, but thoroughly tested across all targets.</div>
<div style><br></div><div style>Call for action:</div><div style><br></div><div style>Would be good to have some buildbots doing cross-compilation for each of our targets, but also those interested on their tests passing (Clang, Debug info, MCJIT folks, etc) should also probably run them locally on their machines and try to move their tests to the best category they think it makes sense.</div>
<div style><br></div><div style>cheers,</div><div style>--renato</div></div>