Hi Richard, On 9/22/21 12:35, Richard Henderson wrote:
On 9/21/21 1:18 PM, WANG Xuerui wrote:+ case INDEX_op_eqv_i32: + case INDEX_op_eqv_i64: + if (c2) { + /* guaranteed to fit due to constraint */ + tcg_out_opc_xori(s, a0, a1, ~a2); + } else { + tcg_out_opc_nor(s, a0, a2, TCG_REG_ZERO); + tcg_out_opc_xor(s, a0, a1, a0); + } + break;You don't actually have eqv (xnor), so don't pretend that you do. The middle-end will expand this as xor + not on its own.
Sure; I'll remove support for eqv in v3.
Otherwise, Reviewed-by: Richard Henderson <richard.hender...@linaro.org> r~
