[Bug rtl-optimization/64164] [4.9/5/6 Regression] one more stack slot used due to one less inlining level

[aoliva at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64164

Alexandre Oliva <aoliva at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |ASSIGNED
         Resolution|FIXED                       |---
            Summary|[4.9/5 Regression] one more |[4.9/5/6 Regression] one
                   |stack slot used due to one  |more stack slot used due to
                   |less inlining level         |one less inlining level

--- Comment #39 from Alexandre Oliva <aoliva at gcc dot gnu.org> ---
At least the sparc regression was caused by the change richi requested to
disregard the underlying decl in promote_ssa_mode.  I didn't realize this could
cause a mismatch between the mode of the partition created for the QImode parm
default def and the promoted mode for the parm decl expected by the
parm-assignment code in function.c.  This will likely take some time to sort
out, so I'm reverting the patch for now.