Am 12.07.2013 um 00:44 schrieb Eli Friedman: > It would be nice if it mentioned it had something to do with forwarding.
Well, that was a train of thought that I had too, but then discarded it for the reason that it per se has nothing to do with forwarding. It is a different feature that happens to make forwarding for stret possible. > The attached test seems to work as expected for me on trunk (the > NOSTRET part passes, and the STRET part fails on the call). That's interesting - for me, it's exactly the other way around. It complains that it can't even find the define void @test0(). Manually compiling the file with -emit-llvm however gives the correct results and contains the lines listed in the test. I don't get why as soon as both tests are in one file, the second test does not match anymore, even though it is there if you manually compile it. Any ideas how to debug this? This seems to be a problem with matching, as if you manually generate the code and look at it, it looks fine. -- Jonathan _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
