As summary says, it's at keywordreq state in bug 272046. Depending on sys-devel/gcc[libffi] isn't a solution, because that gets broken when user does a gcc switch using gcc-config.
+ 04 Jun 2009; Samuli Suominen <ssuominen@gentoo.org> package.mask: + Put jamvm-1.5.0-r3 to libffi mask waiting PowerPC keywording. *jamvm-1.5.0-r3 (04 Jun 2009) 04 Jun 2009; Samuli Suominen <ssuominen@gentoo.org> +jamvm-1.5.0-r3.ebuild: Convert to virtual/libffi wrt #272051 Java Team, please check. I needed to p.mask it for while, because PowerPC ain't playing ball at the moment.. :-)
Tested on amd64 where both dev-libs/libffi and gcc:4.3[libffi] is installed, doesn't look so good I guess: ldd /opt/jamvm/lib64/ldd libjvm.so libffi.so.4 => /usr/lib/gcc/x86_64-pc-linux-gnu/4.3.3/libffi.so.4 (0x00007f8ec67e8000)
(In reply to comment #2) > Tested on amd64 where both dev-libs/libffi and gcc:4.3[libffi] is installed, > doesn't look so good I guess: > > ldd /opt/jamvm/lib64/ldd libjvm.so > libffi.so.4 => /usr/lib/gcc/x86_64-pc-linux-gnu/4.3.3/libffi.so.4 > (0x00007f8ec67e8000) > libffi's ebuild warns you to unset the USE libffi or it might link wrongly. this is expected. that's why we need to package.use.mask the USE="libffi" in gcc soon as possible, and get rid of all the ebuilds using it (and stable what's necessary in 30 days timeframe) the warning in libffi is supposed to take care of this 30 days. please re-emerge gcc without libffi USE
Closing as it's fixed in last revision. Opening a stable bug.