On Thu, Jun 25, 2026 at 11:53 AM Christopher Bazley <[email protected]> wrote: > > Hi, > > This small change was split from the patch series '[PATCH v11 00/12] > Extend BB SLP vectorization to use predicated tails' as requested by > Richard B. in this email: > https://inbox.sourceware.org/gcc-patches/cafiyyc210cha8jaywrtwg__9w4uhaatvaicbuaskox7daq7...@mail.gmail.com/ > > On 16/06/2026 17:12, Christopher Bazley via Sourceware Forge wrote: > > Christopher Bazley (chris.bazley) <[email protected]>) commented on the > > code: > > > > > >> +++ gcc/expr.cc > >> @@ -8078,0 +8082,4 @@ > >> + else > >> + { > >> + /* Handle variable-length vector types. */ > >> + icode = convert_optab_handler (vec_init_optab, mode, eltmode); > > The requested assertion that icode != CODE_FOR_nothing cannot be added. > > See > > https://inbox.sourceware.org/gcc-patches/[email protected]/
But still, we do not want to fall thru to whatever "generic" handling there's in place for VLA vectors because generic code cannot ensure zero-padding for the remaining vector nor did you show that actuall fallback - when vec_init isn't implemented - is correct on all paths taken. "I didn't find any issues" isn't something I'm happy with. Richard. > > > > -- > > https://forge.sourceware.org/gcc/gcc-TEST/pulls/177#issuecomment-6339 > > I have since created PR 125941 for the issue of bad code generation > triggered by RUNTESTFLAGS="aarch64-sve-acle.exp=cops_bool.c": > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125941 > > Since bad code generation is an existing issue, I do not think it need > block this patch. > > Thanks, > -- > Christopher Bazley > Staff Software Engineer, GNU Tools Team. > Arm Ltd, 110 Fulbourn Road, Cambridge, CB1 9NJ, UK. > http://www.arm.com/
