https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103771
--- Comment #4 from Hongtao.liu <crazylht at gmail dot com> ---
(In reply to Hongtao.liu from comment #3)
> for patt_42 = (<signed-boolean:1>) patt_40;
>
> vectype_in (QImode:nunits 4)
>
> <vector_type 0x7fffea18de70
> type <boolean_type 0x7fffea18ddc8 public QI
> size <integer_cst 0x7fffea2e2e40 constant 8>
> unit-size <integer_cst 0x7fffea2e2e58 constant 1>
> align:8 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
> 0x7fffea18ddc8 precision:1 min <integer_cst 0x7fffea196ab0 -1> max
> <integer_cst 0x7fffea196b70 0>>
> QI size <integer_cst 0x7fffea2e2e40 8> unit-size <integer_cst
> 0x7fffea2e2e58 1>
> align:8 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
> 0x7fffea18de70 nunits:4>
>
> vectype_out(HImode)
>
> <vector_type 0x7fffea18df18
> type <boolean_type 0x7fffea18ddc8 public QI
> size <integer_cst 0x7fffea2e2e40 constant 8>
> unit-size <integer_cst 0x7fffea2e2e58 constant 1>
> align:8 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
> 0x7fffea18ddc8 precision:1 min <integer_cst 0x7fffea196ab0 -1> max
> <integer_cst 0x7fffea196b70 0>>
> HI
> size <integer_cst 0x7fffea2e2f00 type <integer_type 0x7fffea2fd0a8
> bitsizetype> constant 16>
> unit-size <integer_cst 0x7fffea2e2f18 type <integer_type 0x7fffea2fd000
> sizetype> constant 2>
> align:16 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
> 0x7fffea18df18 nunits:16>
>
> And ‘vec_pack_sbool_trunc_m’ only handle situation when input and output
> have same mode.
And GCC vectorizer only handle 2X elements, but not 4X,8X,...
/* For scalar masks we may have different boolean
vector types having the same QImode. Thus we
add additional check for elements number. */
if (known_eq (TYPE_VECTOR_SUBPARTS (vectype) * 2,
TYPE_VECTOR_SUBPARTS (narrow_vectype)))
