riscv: advance xperm8 output positions by one byte

[carlosqwqqwq]

riscv: advance xperm8 output positions by one byte

Fixes #275

Commit message

riscv: advance xperm8 output positions by one byte

Description

riscv_bit_xperm8() advances its output position by four bits although each xperm8 element is one byte. The helper therefore writes overlapping byte results. Advance by eight bits so the identity-permutation and general byte controls have the architectural result.

Validation

• The RV64 Zbkx witness, QEMU, and the recorded native/reference controls agree after the fix.
• The proposed change is one source-line change in src/cpu/riscv_base.h.
• The full result matrix is attached with the witness outputs.