[Bug rtl-optimization/102785] [12 Regression] {smul,umul}_highpart changes break bfin-elf

[roger at nextmovesoftware dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102785

Roger Sayle <roger at nextmovesoftware dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-10-15
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1

--- Comment #1 from Roger Sayle <roger at nextmovesoftware dot com> ---
I'm investigating, but the part of the patch that's relevant is the evaluation
of saturating addition/subtraction at compile-time.  The good news is that this
test should now be optimized away, the bad news is that it should reduce to
"return 0".  My initial guess is that the saturating plus/minus isn't correctly
represented in the RTL (perhaps the wrong sign), but I'll know more soon.
Sorry for the inconvenience.