[Bug tree-optimization/68955] [6 Regression] wrong code at -O3 on x86-64-linux-gnu in 32-bit mode

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68955

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
It uses the wrong PHI arg value for the exit PHI of the unswitched loop
It's unswitched on two conditions thus gets 4 loop copies but appearantly
the m_lsm.28 exit PHI gets its arg from the PHI of the condition we unswitched
on rather than the correct one (in two cases).