__mmask16 _kshiftli_mask16 (__mmask16 a, unsigned int count)
__mmask32 _kshiftli_mask32 (__mmask32 a, unsigned int count)
__mmask64 _kshiftli_mask64 (__mmask64 a, unsigned int count)
__mmask8 _kshiftli_mask8 (__mmask8 a, unsigned int count)
__mmask16 _kshiftri_mask16 (__mmask16 a, unsigned int count)
__mmask32 _kshiftri_mask32 (__mmask32 a, unsigned int count)
__mmask64 _kshiftri_mask64 (__mmask64 a, unsigned int count)
__mmask8 _kshiftri_mask8 (__mmask8 a, unsigned int count)
These might be able to reuse the same callbacks as used by clang::X86::BI__builtin_ia32_prldi128\BI__builtin_ia32_psrldi128etc - they have similar out of bounds handling.