------- Additional Comments From rth at gcc dot gnu dot org 2005-09-02 22:08 ------- I'm sure that the patches listed in comments #6 and #7 are bogus.
But we've had problems before with unwinding with -mno-accumulate-outgoing-args (implied by -mtune=i486, implied by i486-linux), plus -fomit-frame-pointer. I assume this problem has returned. But since essentially no one uses this combination regularly, we don't know exactly when it failed. I'm trying to get reduced unwind test cases to fail. If that works, it'll be easier than looking at libjava itself. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23602
