use crate::backend::aes128::Aes128;
#[cfg(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
any(feature = "std", all(target_feature = "aes", target_feature = "sse2"))
))]
use crate::backend::tree::finalize_merge4_lk;
#[cfg(any(
feature = "cpu",
all(target_arch = "aarch64", target_endian = "little"),
not(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
target_feature = "aes",
target_feature = "sse2"
))
))]
use crate::backend::tree::finalize_merge4_pre;
pub(crate) trait AesMode<L: Aes128> {
unsafe fn r1_zero(ctx: L, a: L::V, b: L::V) -> L::V;
unsafe fn f_zero(ctx: L, s: L::V) -> L::V;
unsafe fn chain_x01<const N: usize>(ctx: L, x0: L::V, x1: L::V, rk_li: [L::V; N]) -> L::V;
unsafe fn acc_init_1(ctx: L, key: L::V, data: L::V) -> L::V;
unsafe fn acc_absorb_1(ctx: L, acc: L::V, data: L::V) -> L::V;
unsafe fn ctr_prep_1(ctx: L, ctr: L::V, rk0: L::V) -> L::V;
unsafe fn ctr_mid_1(ctx: L, s: L::V, rk: L::V) -> L::V;
unsafe fn ctr_fin_1(ctx: L, s: L::V, rk3: L::V, rk0: L::V) -> L::V;
unsafe fn compress_17_32_core(
ctx: L,
b0: L::V,
k0: L::V,
b1: L::V,
k1: L::V,
lk: L::V,
) -> (L::V, L::V);
unsafe fn compress_33_64_core<const HI: bool>(
ctx: L,
b_li: [L::V; 4],
key_li: [L::V; 4],
lk: L::V,
) -> (L::V, L::V);
unsafe fn merge4_128<const HI: bool>(ctx: L, acc_li: [L::V; 4], lk: L::V) -> (L::V, L::V);
unsafe fn merge8_128<const HI: bool>(ctx: L, acc_li: [L::V; 8], lk: L::V) -> (L::V, L::V);
unsafe fn merge16_128<const HI: bool>(
ctx: L,
lo_li: [L::V; 8],
hi_li: [L::V; 8],
lk: L::V,
) -> (L::V, L::V);
}
#[cfg(any(
feature = "cpu",
all(target_arch = "aarch64", target_endian = "little"),
not(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
target_feature = "aes",
target_feature = "sse2"
))
))]
mod pre_xor {
use super::*;
pub(crate) trait PreXorAes: Aes128 {
unsafe fn pre_r1_zero(self, a: Self::V, b: Self::V) -> Self::V;
unsafe fn pre_fl_zero(self, a: Self::V, b: Self::V) -> Self::V;
#[inline(always)]
unsafe fn pre_enc(self, state: Self::V, rk: Self::V) -> Self::V {
unsafe { self.xor(self.pre_r1_zero(state, self.zero()), rk) }
}
}
pub(crate) struct PreXor;
impl<L: PreXorAes> AesMode<L> for PreXor {
#[inline(always)]
unsafe fn r1_zero(ctx: L, a: L::V, b: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(a, b) }
}
#[inline(always)]
unsafe fn f_zero(ctx: L, s: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(s, ctx.zero()) }
}
#[inline(always)]
unsafe fn chain_x01<const N: usize>(ctx: L, x0: L::V, x1: L::V, rk_li: [L::V; N]) -> L::V {
unsafe {
let mut s = ctx.pre_r1_zero(x0, x1);
for k in rk_li {
s = ctx.pre_r1_zero(s, k);
}
s
}
}
#[inline(always)]
unsafe fn acc_init_1(ctx: L, key: L::V, data: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(key, data) }
}
#[inline(always)]
unsafe fn acc_absorb_1(ctx: L, acc: L::V, data: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(acc, data) }
}
#[inline(always)]
unsafe fn ctr_prep_1(ctx: L, ctr: L::V, rk0: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(ctr, rk0) }
}
#[inline(always)]
unsafe fn ctr_mid_1(ctx: L, s: L::V, rk: L::V) -> L::V {
unsafe { ctx.pre_r1_zero(s, rk) }
}
#[inline(always)]
unsafe fn ctr_fin_1(ctx: L, s: L::V, rk3: L::V, rk0: L::V) -> L::V {
unsafe { ctx.xor(ctx.pre_fl_zero(s, rk3), rk0) }
}
#[inline(always)]
unsafe fn compress_17_32_core(
ctx: L,
b0: L::V,
k0: L::V,
b1: L::V,
k1: L::V,
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let a = ctx.pre_r1_zero(b0, k0);
let c = ctx.pre_r1_zero(b1, k1);
let a2 = ctx.pre_r1_zero(a, lk);
(ctx.chain_x01_ff(a2, c, []), c)
}
}
#[inline(always)]
unsafe fn compress_33_64_core<const HI: bool>(
ctx: L,
b_li: [L::V; 4],
key_li: [L::V; 4],
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let h_li = unroll4!(I => ctx.pre_r1_zero(b_li[I], key_li[I]));
finalize_merge4_pre::<L, HI>(ctx, h_li, lk)
}
}
#[inline(always)]
unsafe fn merge4_128<const HI: bool>(ctx: L, acc_li: [L::V; 4], lk: L::V) -> (L::V, L::V) {
unsafe { finalize_merge4_pre::<L, HI>(ctx, acc_li, lk) }
}
#[inline(always)]
unsafe fn merge8_128<const HI: bool>(ctx: L, acc_li: [L::V; 8], lk: L::V) -> (L::V, L::V) {
unsafe {
let p_li = unroll4!(I => ctx.pre_enc(acc_li[I], acc_li[I + 4]));
finalize_merge4_pre::<L, HI>(ctx, p_li, lk)
}
}
#[inline(always)]
unsafe fn merge16_128<const HI: bool>(
ctx: L,
lo_li: [L::V; 8],
hi_li: [L::V; 8],
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let m_li = unroll8!(I => ctx.pre_enc(lo_li[I], hi_li[I]));
let p_li = unroll4!(I => ctx.pre_enc(m_li[I], m_li[I + 4]));
finalize_merge4_pre::<L, HI>(ctx, p_li, lk)
}
}
}
}
#[cfg(any(
feature = "cpu",
all(target_arch = "aarch64", target_endian = "little"),
not(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
target_feature = "aes",
target_feature = "sse2"
))
))]
pub(crate) use pre_xor::{PreXor, PreXorAes};
#[cfg(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
any(feature = "std", all(target_feature = "aes", target_feature = "sse2"))
))]
mod post_xor {
use super::*;
pub(crate) trait PostXorAes: Aes128 {
unsafe fn post_enc(self, state: Self::V, rk: Self::V) -> Self::V;
unsafe fn post_enclast(self, state: Self::V, rk: Self::V) -> Self::V;
}
pub(crate) struct PostXor;
impl<L: PostXorAes> AesMode<L> for PostXor {
#[inline(always)]
unsafe fn r1_zero(ctx: L, a: L::V, b: L::V) -> L::V {
unsafe { ctx.post_enc(ctx.xor(a, b), ctx.zero()) }
}
#[inline(always)]
unsafe fn f_zero(ctx: L, s: L::V) -> L::V {
unsafe { ctx.post_enc(s, ctx.zero()) }
}
#[inline(always)]
unsafe fn chain_x01<const N: usize>(ctx: L, x0: L::V, x1: L::V, rk_li: [L::V; N]) -> L::V {
unsafe {
let mut s = ctx.xor(x0, x1);
for k in rk_li {
s = ctx.post_enc(s, k);
}
ctx.post_enc(s, ctx.zero())
}
}
#[inline(always)]
unsafe fn acc_init_1(ctx: L, key: L::V, data: L::V) -> L::V {
unsafe { ctx.xor(key, data) }
}
#[inline(always)]
unsafe fn acc_absorb_1(ctx: L, acc: L::V, data: L::V) -> L::V {
unsafe { ctx.post_enc(acc, data) }
}
#[inline(always)]
unsafe fn ctr_prep_1(ctx: L, ctr: L::V, rk0: L::V) -> L::V {
unsafe { ctx.xor(ctr, rk0) }
}
#[inline(always)]
unsafe fn ctr_mid_1(ctx: L, s: L::V, rk: L::V) -> L::V {
unsafe { ctx.post_enc(s, rk) }
}
#[inline(always)]
unsafe fn ctr_fin_1(ctx: L, s: L::V, rk3: L::V, rk0: L::V) -> L::V {
unsafe { ctx.post_enclast(ctx.post_enc(s, rk3), rk0) }
}
#[inline(always)]
unsafe fn compress_17_32_core(
ctx: L,
b0: L::V,
k0: L::V,
b1: L::V,
k1: L::V,
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let z = ctx.zero();
let a_lk = ctx.post_enc(ctx.xor(b0, k0), lk);
let c = ctx.post_enc(ctx.xor(b1, k1), z);
let s = ctx.post_enc(a_lk, c);
let u = ctx.post_enc(s, z);
let fin = ctx.post_enc(u, z);
(ctx.add_u64(fin, u), c)
}
}
#[inline(always)]
unsafe fn compress_33_64_core<const HI: bool>(
ctx: L,
b_li: [L::V; 4],
key_li: [L::V; 4],
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let z = ctx.zero();
let l_li =
unroll4!(I => ctx.post_enc(ctx.xor(b_li[I], key_li[I]), if I == 0 { lk } else { z }));
finalize_merge4_lk::<L, HI>(ctx, l_li, lk)
}
}
#[inline(always)]
unsafe fn merge4_128<const HI: bool>(ctx: L, acc_li: [L::V; 4], lk: L::V) -> (L::V, L::V) {
unsafe {
let z = ctx.zero();
let l_li = unroll4!(I => ctx.post_enc(acc_li[I], if I == 0 { lk } else { z }));
finalize_merge4_lk::<L, HI>(ctx, l_li, lk)
}
}
#[inline(always)]
unsafe fn merge8_128<const HI: bool>(ctx: L, acc_li: [L::V; 8], lk: L::V) -> (L::V, L::V) {
unsafe {
let z = ctx.zero();
let p_li = unroll4!(I => {
let hi_end = ctx.post_enc(acc_li[I + 4], if I == 0 { lk } else { z });
ctx.post_enc(ctx.post_enc(acc_li[I], z), hi_end)
});
finalize_merge4_lk::<L, HI>(ctx, p_li, lk)
}
}
#[inline(always)]
unsafe fn merge16_128<const HI: bool>(
ctx: L,
lo_li: [L::V; 8],
hi_li: [L::V; 8],
lk: L::V,
) -> (L::V, L::V) {
unsafe {
let z = ctx.zero();
let m_li = unroll8!(I => {
let hi_end = ctx.post_enc(hi_li[I], if I == 4 { lk } else { z });
ctx.post_enc(ctx.post_enc(lo_li[I], z), hi_end)
});
let p_li = unroll4!(I => ctx.post_enc(m_li[I], m_li[I + 4]));
finalize_merge4_lk::<L, HI>(ctx, p_li, lk)
}
}
}
}
#[cfg(all(
any(target_arch = "x86", target_arch = "x86_64"),
target_endian = "little",
any(feature = "std", all(target_feature = "aes", target_feature = "sse2"))
))]
pub(crate) use post_xor::{PostXor, PostXorAes};