ShuriZma
/
xemu
mirror of https://github.com/xemu-project/xemu.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

target/arm: Split helper_crypto_sha1_3reg 

Rather than passing an opcode to a helper, fully decode the
operation at translate time.  Use clear_tail_16 to zap the
balance of the SVE register with the AdvSIMD write.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20200514212831.31248-6-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This commit is contained in:
Richard Henderson 2020-05-14 14:28:30 -07:00 committed by Peter Maydell
parent effa992f15
commit afc8b7d326
5 changed files with 93 additions and 92 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

99
target/arm/crypto_helper.c
View File

@ -24,11 +24,11 @@ union CRYPTO_STATE {
};
#ifdef HOST_WORDS_BIGENDIAN
#define CR_ST_BYTE(state, i) (state.bytes[(15 - (i)) ^ 8])
#define CR_ST_WORD(state, i) (state.words[(3 - (i)) ^ 2])
#define CR_ST_BYTE(state, i) ((state).bytes[(15 - (i)) ^ 8])
#define CR_ST_WORD(state, i) ((state).words[(3 - (i)) ^ 2])
#else
#define CR_ST_BYTE(state, i) (state.bytes[i])
#define CR_ST_WORD(state, i) (state.words[i])
#define CR_ST_BYTE(state, i) ((state).bytes[i])
#define CR_ST_WORD(state, i) ((state).words[i])
#endif
/*
@ -258,49 +258,74 @@ static uint32_t maj(uint32_t x, uint32_t y, uint32_t z)
return (x & y) | ((x | y) & z);
}
void HELPER(crypto_sha1_3reg)(void *vd, void *vn, void *vm, uint32_t op)
void HELPER(crypto_sha1su0)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *d = vd, *n = vn, *m = vm;
uint64_t d0, d1;
d0 = d[1] ^ d[0] ^ m[0];
d1 = n[0] ^ d[1] ^ m[1];
d[0] = d0;
d[1] = d1;
clear_tail_16(vd, desc);
}
static inline void crypto_sha1_3reg(uint64_t *rd, uint64_t *rn,
uint64_t *rm, uint32_t desc,
uint32_t (*fn)(union CRYPTO_STATE *d))
{
uint64_t *rd = vd;
uint64_t *rn = vn;
uint64_t *rm = vm;
union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
int i;
if (op == 3) { /* sha1su0 */
d.l[0] ^= d.l[1] ^ m.l[0];
d.l[1] ^= n.l[0] ^ m.l[1];
} else {
int i;
for (i = 0; i < 4; i++) {
uint32_t t = fn(&d);
for (i = 0; i < 4; i++) {
uint32_t t;
t += rol32(CR_ST_WORD(d, 0), 5) + CR_ST_WORD(n, 0)
+ CR_ST_WORD(m, i);
switch (op) {
case 0: /* sha1c */
t = cho(CR_ST_WORD(d, 1), CR_ST_WORD(d, 2), CR_ST_WORD(d, 3));
break;
case 1: /* sha1p */
t = par(CR_ST_WORD(d, 1), CR_ST_WORD(d, 2), CR_ST_WORD(d, 3));
break;
case 2: /* sha1m */
t = maj(CR_ST_WORD(d, 1), CR_ST_WORD(d, 2), CR_ST_WORD(d, 3));
break;
default:
g_assert_not_reached();
}
t += rol32(CR_ST_WORD(d, 0), 5) + CR_ST_WORD(n, 0)
+ CR_ST_WORD(m, i);
CR_ST_WORD(n, 0) = CR_ST_WORD(d, 3);
CR_ST_WORD(d, 3) = CR_ST_WORD(d, 2);
CR_ST_WORD(d, 2) = ror32(CR_ST_WORD(d, 1), 2);
CR_ST_WORD(d, 1) = CR_ST_WORD(d, 0);
CR_ST_WORD(d, 0) = t;
}
CR_ST_WORD(n, 0) = CR_ST_WORD(d, 3);
CR_ST_WORD(d, 3) = CR_ST_WORD(d, 2);
CR_ST_WORD(d, 2) = ror32(CR_ST_WORD(d, 1), 2);
CR_ST_WORD(d, 1) = CR_ST_WORD(d, 0);
CR_ST_WORD(d, 0) = t;
}
rd[0] = d.l[0];
rd[1] = d.l[1];
clear_tail_16(rd, desc);
}
static uint32_t do_sha1c(union CRYPTO_STATE *d)
{
return cho(CR_ST_WORD(*d, 1), CR_ST_WORD(*d, 2), CR_ST_WORD(*d, 3));
}
void HELPER(crypto_sha1c)(void *vd, void *vn, void *vm, uint32_t desc)
{
crypto_sha1_3reg(vd, vn, vm, desc, do_sha1c);
}
static uint32_t do_sha1p(union CRYPTO_STATE *d)
{
return par(CR_ST_WORD(*d, 1), CR_ST_WORD(*d, 2), CR_ST_WORD(*d, 3));
}
void HELPER(crypto_sha1p)(void *vd, void *vn, void *vm, uint32_t desc)
{
crypto_sha1_3reg(vd, vn, vm, desc, do_sha1p);
}
static uint32_t do_sha1m(union CRYPTO_STATE *d)
{
return maj(CR_ST_WORD(*d, 1), CR_ST_WORD(*d, 2), CR_ST_WORD(*d, 3));
}
void HELPER(crypto_sha1m)(void *vd, void *vn, void *vm, uint32_t desc)
{
crypto_sha1_3reg(vd, vn, vm, desc, do_sha1m);
}
void HELPER(crypto_sha1h)(void *vd, void *vm, uint32_t desc)

5
target/arm/helper.h
View File

@ -513,7 +513,10 @@ DEF_HELPER_FLAGS_2(neon_qzip32, TCG_CALL_NO_RWG, void, ptr, ptr)
DEF_HELPER_FLAGS_4(crypto_aese, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(crypto_aesmc, TCG_CALL_NO_RWG, void, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha1_3reg, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha1su0, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha1c, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha1p, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha1m, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(crypto_sha1h, TCG_CALL_NO_RWG, void, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(crypto_sha1su1, TCG_CALL_NO_RWG, void, ptr, ptr, i32)

6
target/arm/neon-dp.decode
View File

@ -168,8 +168,10 @@ VQRDMLAH_3s 1111 001 1 0 . .. .... .... 1011 ... 1 .... @3same
@3same_crypto .... .... .... .... .... .... .... .... \
&3same vm=%vm_dp vn=%vn_dp vd=%vd_dp size=0 q=1
SHA1_3s 1111 001 0 0 . optype:2 .... .... 1100 . 1 . 0 .... \
vm=%vm_dp vn=%vn_dp vd=%vd_dp
SHA1C_3s 1111 001 0 0 . 00 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA1P_3s 1111 001 0 0 . 01 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA1M_3s 1111 001 0 0 . 10 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA1SU0_3s 1111 001 0 0 . 11 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA256H_3s 1111 001 1 0 . 00 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA256H2_3s 1111 001 1 0 . 01 .... .... 1100 . 1 . 0 .... @3same_crypto
SHA256SU1_3s 1111 001 1 0 . 10 .... .... 1100 . 1 . 0 .... @3same_crypto

29
target/arm/translate-a64.c
View File

@ -13475,10 +13475,19 @@ static void disas_crypto_three_reg_sha(DisasContext *s, uint32_t insn)
switch (opcode) {
case 0: /* SHA1C */
genfn = gen_helper_crypto_sha1c;
feature = dc_isar_feature(aa64_sha1, s);
break;
case 1: /* SHA1P */
genfn = gen_helper_crypto_sha1p;
feature = dc_isar_feature(aa64_sha1, s);
break;
case 2: /* SHA1M */
genfn = gen_helper_crypto_sha1m;
feature = dc_isar_feature(aa64_sha1, s);
break;
case 3: /* SHA1SU0 */
genfn = NULL;
genfn = gen_helper_crypto_sha1su0;
feature = dc_isar_feature(aa64_sha1, s);
break;
case 4: /* SHA256H */
@ -13506,23 +13515,7 @@ static void disas_crypto_three_reg_sha(DisasContext *s, uint32_t insn)
if (!fp_access_check(s)) {
return;
}
if (genfn) {
gen_gvec_op3_ool(s, true, rd, rn, rm, 0, genfn);
} else {
TCGv_i32 tcg_opcode = tcg_const_i32(opcode);
TCGv_ptr tcg_rd_ptr = vec_full_reg_ptr(s, rd);
TCGv_ptr tcg_rn_ptr = vec_full_reg_ptr(s, rn);
TCGv_ptr tcg_rm_ptr = vec_full_reg_ptr(s, rm);
gen_helper_crypto_sha1_3reg(tcg_rd_ptr, tcg_rn_ptr,
tcg_rm_ptr, tcg_opcode);
tcg_temp_free_i32(tcg_opcode);
tcg_temp_free_ptr(tcg_rd_ptr);
tcg_temp_free_ptr(tcg_rn_ptr);
tcg_temp_free_ptr(tcg_rm_ptr);
}
gen_gvec_op3_ool(s, true, rd, rn, rm, 0, genfn);
}
/* Crypto two-reg SHA

46
target/arm/translate-neon.inc.c
View File

@ -693,42 +693,20 @@ static bool trans_VMUL_p_3s(DisasContext *s, arg_3same *a)
DO_VQRDMLAH(VQRDMLAH, gen_gvec_sqrdmlah_qc)
DO_VQRDMLAH(VQRDMLSH, gen_gvec_sqrdmlsh_qc)
static bool trans_SHA1_3s(DisasContext *s, arg_SHA1_3s *a)
{
TCGv_ptr ptr1, ptr2, ptr3;
TCGv_i32 tmp;
if (!arm_dc_feature(s, ARM_FEATURE_NEON) ||
!dc_isar_feature(aa32_sha1, s)) {
return false;
#define DO_SHA1(NAME, FUNC) \
WRAP_OOL_FN(gen_##NAME##_3s, FUNC) \
static bool trans_##NAME##_3s(DisasContext *s, arg_3same *a) \
{ \
if (!dc_isar_feature(aa32_sha1, s)) { \
return false; \
} \
return do_3same(s, a, gen_##NAME##_3s); \
}
/* UNDEF accesses to D16-D31 if they don't exist. */
if (!dc_isar_feature(aa32_simd_r32, s) &&
((a->vd | a->vn | a->vm) & 0x10)) {
return false;
}
if ((a->vn | a->vm | a->vd) & 1) {
return false;
}
if (!vfp_access_check(s)) {
return true;
}
ptr1 = vfp_reg_ptr(true, a->vd);
ptr2 = vfp_reg_ptr(true, a->vn);
ptr3 = vfp_reg_ptr(true, a->vm);
tmp = tcg_const_i32(a->optype);
gen_helper_crypto_sha1_3reg(ptr1, ptr2, ptr3, tmp);
tcg_temp_free_i32(tmp);
tcg_temp_free_ptr(ptr1);
tcg_temp_free_ptr(ptr2);
tcg_temp_free_ptr(ptr3);
return true;
}
DO_SHA1(SHA1C, gen_helper_crypto_sha1c)
DO_SHA1(SHA1P, gen_helper_crypto_sha1p)
DO_SHA1(SHA1M, gen_helper_crypto_sha1m)
DO_SHA1(SHA1SU0, gen_helper_crypto_sha1su0)
#define DO_SHA2(NAME, FUNC) \
WRAP_OOL_FN(gen_##NAME##_3s, FUNC) \