[Bug tree-optimization/123656] [16 Regression] wrong code with unsigned char [2] vector multiplication by 210 since r16-5095

[avinashd at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123656

--- Comment #6 from Avinash Jayakar <avinashd at gcc dot gnu.org> ---
Created attachment 63386
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=63386&action=edit
Reversed the order of subtract as per vector mult synthesis algorithm for
alg_sub_factor case

Tested this patch, it produces the correct results

  _4 = v_1(D) << 3;
  _5 = _4 - v_1(D);
  _6 = _5 << 4;
  _7 = v_1(D) - _4;
  _8 = _6 + _7;
  _9 = _8 << 1;

Will bootstrap and regtest the patch before sending for review.