On 8/11/11 10:41 AM, Emmanuel Fleury wrote: > >> Aha. Actually, this should have worked on your side because the code >> is currently using -Wall first and later -Wno-unused-parameters to >> silence this part of -Wall. Might this be a bug in gcc 4.6? > > Hmmm, I'm doubtful about it... I didn't heard about somebody "cleaning" > the option parser recently... Though I'm not following gcc-dev very > carefully, I admit. Ah, I think I just misread the whole thing. The subject line of your email very clearly stated -Wunused-but-set-variable, which is a new warning in GCC 4.6. It is different than -Wunused-parameter! I have just committed this giant revision: http://mtn-host.prjek.net/viewmtn/atf/revision/info/a2c38309edfc28608bb371eb937de451e01c0dfb which enables unused parameter tracking and suppresses "false positives". So far I have only tested it in OS X with clang, but will do some more testing later (including Debian sid). Thanks!