target/arm: Convert sha512 and sm3 to gvec helpers

Do not yet convert the helpers to loop over opr_sz, but the
descriptor allows the vector tail to be cleared. Which fixes
an existing bug vs SVE.

Backports commit aaffebd6d3135b8aed7e61932af53b004d261579 from qemu
This commit is contained in:
Richard Henderson 2020-06-14 23:00:24 -04:00 committed by Lioncash
parent 894f2168da
commit 2c6c4da80c
3 changed files with 55 additions and 49 deletions

View file

@ -32,6 +32,19 @@ union CRYPTO_STATE {
#define CR_ST_WORD(state, i) (state.words[i])
#endif
/*
* The caller has not been converted to full gvec, and so only
* modifies the low 16 bytes of the vector register.
*/
static void clear_tail_16(void *vd, uint32_t desc)
{
int opr_sz = simd_oprsz(desc);
int max_sz = simd_maxsz(desc);
assert(opr_sz == 16);
clear_tail(vd, opr_sz, max_sz);
}
static void do_crypto_aese(uint64_t *rd, uint64_t *rn,
uint64_t *rm, bool decrypt)
{
@ -506,7 +519,7 @@ static uint64_t s1_512(uint64_t x)
return ror64(x, 19) ^ ror64(x, 61) ^ (x >> 6);
}
void HELPER(crypto_sha512h)(void *vd, void *vn, void *vm)
void HELPER(crypto_sha512h)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -519,9 +532,11 @@ void HELPER(crypto_sha512h)(void *vd, void *vn, void *vm)
rd[0] = d0;
rd[1] = d1;
clear_tail_16(vd, desc);
}
void HELPER(crypto_sha512h2)(void *vd, void *vn, void *vm)
void HELPER(crypto_sha512h2)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -534,9 +549,11 @@ void HELPER(crypto_sha512h2)(void *vd, void *vn, void *vm)
rd[0] = d0;
rd[1] = d1;
clear_tail_16(vd, desc);
}
void HELPER(crypto_sha512su0)(void *vd, void *vn)
void HELPER(crypto_sha512su0)(void *vd, void *vn, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -548,9 +565,11 @@ void HELPER(crypto_sha512su0)(void *vd, void *vn)
rd[0] = d0;
rd[1] = d1;
clear_tail_16(vd, desc);
}
void HELPER(crypto_sha512su1)(void *vd, void *vn, void *vm)
void HELPER(crypto_sha512su1)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -558,9 +577,11 @@ void HELPER(crypto_sha512su1)(void *vd, void *vn, void *vm)
rd[0] += s1_512(rn[0]) + rm[0];
rd[1] += s1_512(rn[1]) + rm[1];
clear_tail_16(vd, desc);
}
void HELPER(crypto_sm3partw1)(void *vd, void *vn, void *vm)
void HELPER(crypto_sm3partw1)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -593,9 +614,11 @@ void HELPER(crypto_sm3partw1)(void *vd, void *vn, void *vm)
rd[0] = d.l[0];
rd[1] = d.l[1];
clear_tail_16(vd, desc);
}
void HELPER(crypto_sm3partw2)(void *vd, void *vn, void *vm)
void HELPER(crypto_sm3partw2)(void *vd, void *vn, void *vm, uint32_t desc)
{
uint64_t *rd = vd;
uint64_t *rn = vn;
@ -624,6 +647,8 @@ void HELPER(crypto_sm3partw2)(void *vd, void *vn, void *vm)
rd[0] = d.l[0];
rd[1] = d.l[1];
clear_tail_16(vd, desc);
}
void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,

View file

@ -518,14 +518,17 @@ DEF_HELPER_FLAGS_3(crypto_sha256h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_2(crypto_sha256su0, TCG_CALL_NO_RWG, void, ptr, ptr)
DEF_HELPER_FLAGS_3(crypto_sha256su1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_3(crypto_sha512h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_3(crypto_sha512h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_2(crypto_sha512su0, TCG_CALL_NO_RWG, void, ptr, ptr)
DEF_HELPER_FLAGS_3(crypto_sha512su1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_4(crypto_sha512h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha512h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(crypto_sha512su0, TCG_CALL_NO_RWG, void, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sha512su1, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(crypto_sm3tt, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32, i32)
DEF_HELPER_FLAGS_3(crypto_sm3partw1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_3(crypto_sm3partw2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
DEF_HELPER_FLAGS_4(crypto_sm3partw1, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sm3partw2, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sm4e, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_4(crypto_sm4ekey, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)

View file

@ -13923,14 +13923,12 @@ void gen_gvec_rax1(TCGContext *s, unsigned vece, uint32_t rd_ofs, uint32_t rn_of
*/
static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
{
TCGContext *tcg_ctx = s->uc->tcg_ctx;
int opcode = extract32(insn, 10, 2);
int o = extract32(insn, 14, 1);
int rm = extract32(insn, 16, 5);
int rn = extract32(insn, 5, 5);
int rd = extract32(insn, 0, 5);
bool feature;
CryptoThreeOpFn *genfn = NULL;
gen_helper_gvec_3 *oolfn = NULL;
GVecGen3Fn *gvecfn = NULL;
@ -13938,15 +13936,15 @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
switch (opcode) {
case 0: /* SHA512H */
feature = dc_isar_feature(aa64_sha512, s);
genfn = gen_helper_crypto_sha512h;
oolfn = gen_helper_crypto_sha512h;
break;
case 1: /* SHA512H2 */
feature = dc_isar_feature(aa64_sha512, s);
genfn = gen_helper_crypto_sha512h2;
oolfn = gen_helper_crypto_sha512h2;
break;
case 2: /* SHA512SU1 */
feature = dc_isar_feature(aa64_sha512, s);
genfn = gen_helper_crypto_sha512su1;
oolfn = gen_helper_crypto_sha512su1;
break;
case 3: /* RAX1 */
feature = dc_isar_feature(aa64_sha3, s);
@ -13959,11 +13957,11 @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
switch (opcode) {
case 0: /* SM3PARTW1 */
feature = dc_isar_feature(aa64_sm3, s);
genfn = gen_helper_crypto_sm3partw1;
oolfn = gen_helper_crypto_sm3partw1;
break;
case 1: /* SM3PARTW2 */
feature = dc_isar_feature(aa64_sm3, s);
genfn = gen_helper_crypto_sm3partw2;
oolfn = gen_helper_crypto_sm3partw2;
break;
case 2: /* SM4EKEY */
feature = dc_isar_feature(aa64_sm4, s);
@ -13986,20 +13984,8 @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
if (oolfn) {
gen_gvec_op3_ool(s, true, rd, rn, rm, 0, oolfn);
} else if (gvecfn) {
gen_gvec_fn3(s, true, rd, rn, rm, gvecfn, MO_64);
} else {
TCGv_ptr tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr;
tcg_rd_ptr = vec_full_reg_ptr(s, rd);
tcg_rn_ptr = vec_full_reg_ptr(s, rn);
tcg_rm_ptr = vec_full_reg_ptr(s, rm);
genfn(tcg_ctx, tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr);
tcg_temp_free_ptr(tcg_ctx, tcg_rd_ptr);
tcg_temp_free_ptr(tcg_ctx, tcg_rn_ptr);
tcg_temp_free_ptr(tcg_ctx, tcg_rm_ptr);
gen_gvec_fn3(s, true, rd, rn, rm, gvecfn, MO_64);
}
}
@ -14011,23 +13997,17 @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
*/
static void disas_crypto_two_reg_sha512(DisasContext *s, uint32_t insn)
{
TCGContext *tcg_ctx = s->uc->tcg_ctx;
int opcode = extract32(insn, 10, 2);
int rn = extract32(insn, 5, 5);
int rd = extract32(insn, 0, 5);
TCGv_ptr tcg_rd_ptr, tcg_rn_ptr;
bool feature;
CryptoTwoOpFn *genfn;
gen_helper_gvec_3 *oolfn = NULL;
switch (opcode) {
case 0: /* SHA512SU0 */
feature = dc_isar_feature(aa64_sha512, s);
genfn = gen_helper_crypto_sha512su0;
break;
case 1: /* SM4E */
feature = dc_isar_feature(aa64_sm4, s);
oolfn = gen_helper_crypto_sm4e;
break;
default:
unallocated_encoding(s);
@ -14043,18 +14023,16 @@ static void disas_crypto_two_reg_sha512(DisasContext *s, uint32_t insn)
return;
}
if (oolfn) {
gen_gvec_op3_ool(s, true, rd, rd, rn, 0, oolfn);
return;
switch (opcode) {
case 0: /* SHA512SU0 */
gen_gvec_op2_ool(s, true, rd, rn, 0, gen_helper_crypto_sha512su0);
break;
case 1: /* SM4E */
gen_gvec_op3_ool(s, true, rd, rd, rn, 0, gen_helper_crypto_sm4e);
break;
default:
g_assert_not_reached();
}
tcg_rd_ptr = vec_full_reg_ptr(s, rd);
tcg_rn_ptr = vec_full_reg_ptr(s, rn);
genfn(tcg_ctx, tcg_rd_ptr, tcg_rn_ptr);
tcg_temp_free_ptr(tcg_ctx, tcg_rd_ptr);
tcg_temp_free_ptr(tcg_ctx, tcg_rn_ptr);
}
/* Crypto four-register