Il 12/11/2013 16:32, Peter Maydell ha scritto: > > Is this FUD or do you have examples of bad debuggability of -O1 code? > > The clang manpage says specifically "Note that Clang debug > information works best at -O0. ", and I see no reason to > disbelieve it. In particular, they don't say "we definitely > will never add an optimization to -O1 that makes the debug > info much worse".
This doesn't quite answer my question. It looks like another bug that should be reported to clang. "-O1 is somewhere between -O0 and -O2" (quoted from the man page) is a joke, it's not documentation. Every time I look at clang, it seems to me that they are still relying on the "buzz" from their "better syntax errors" blog posts (undeserved these days), and from clang-analyzer (deserved). I don't really see a reason why QEMU should give clang more weight than Windows or Mac OS X. Paolo
