------- Comment #3 from hp at gcc dot gnu dot org 2008-09-02 10:34 ------- Yep, it's revision 139762 that exposed this FAIL, which of course (appears to) just change behaviour of some optimization levels. Incidentally, that also exposed some of the FAILs in PR37315 (I think I fumble-fingered the failing revision range in that PR): gcc.c-torture/execute/931018-1.c, gcc.c-torture/execute/ieee/inf-2.c, gcc.c-torture/execute/ieee/mzero6.c. I think it'd be proper to let this PR depend on that one, particularly since it makes sense that "re-using" a cfun->machine could cause this and this PR seeming nasty to track down (PCH and all).
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37307
