[Bug tree-optimization/119253] RISC-V GCC auto-vectorizes unaligned memory access even if mvector-strict-align is enabled

[Dusan.Stojkovic@rt-rk.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119253

Dusan Stojkovic <Dusan.Stojkovic@rt-rk.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|INVALID                     |FIXED

--- Comment #2 from Dusan Stojkovic <Dusan.Stojkovic@rt-rk.com> ---
(In reply to Andrew Pinski from comment #1)
> >replicates the unaligned 32-bit 
> >(uint32_t *)(flag_buf + (bit_pos >> 3))
> 
> 
> Yes because this is undefined code. You need to use either a type which has
> an alignment of 1 or use memcpy (without a cast to the pointer of an aligned
> type).

Looking at the compiler warnings, this could be something to consider:
* -Wall doesn't warn that this is undefined code;
* -Wextra doesn't warn that this is undefined code;
* -Wcast-align is the only case when this warning is written:
     warning: cast increases required alignment of target type [-Wcast-align]
  122 |             (*(uint32_t *)(flag_buf + (bit_pos >> 3))) >> (bit_pos &
7);

But it's not enabled by default:
# gcc -Q --help=warnings | grep cast-align
  -Wcast-align                          [disabled]
  -Wcast-align=strict                   [disabled]

Maybe something to consider when compiling with -Wall -mstrict-align or
-mvector-strict-align is to have these warnings enabled.