[Bug target/124448] RISC-V: Extract from tuple vector with lmul<1 may cause bad code generation

[rdapp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124448

Robin Dapp <rdapp at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rdapp at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2026-03-11
             Target|                            |riscv

--- Comment #1 from Robin Dapp <rdapp at gcc dot gnu.org> ---
zvl/VLS code should never be worse than VLA but unfortunately it still is. 
That's on my todo pile for next stage 1.  The VLS code here (that's from
sub4x4dct) looks awful and your analysis is correct that it's the subregs that
are causing problems.

Without having had a deeper look, it's possible that we're just lacking some
handling in expand_vector_subreg_extract but it's also possible that it needs
to be tackled at a different level.