summary refs log tree commit diff stats
path: root/target/arm/tcg/translate-a64.c
diff options
context:
space:
mode:
Diffstat (limited to 'target/arm/tcg/translate-a64.c')
-rw-r--r--target/arm/tcg/translate-a64.c191
1 files changed, 171 insertions, 20 deletions
diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c
index dbf47595db..37bedc3780 100644
--- a/target/arm/tcg/translate-a64.c
+++ b/target/arm/tcg/translate-a64.c
@@ -3237,7 +3237,7 @@ static bool trans_LDXP(DisasContext *s, arg_stxr *a)
 
 static bool trans_CASP(DisasContext *s, arg_CASP *a)
 {
-    if (!dc_isar_feature(aa64_atomics, s)) {
+    if (!dc_isar_feature(aa64_lse, s)) {
         return false;
     }
     if (((a->rt | a->rs) & 1) != 0) {
@@ -3250,7 +3250,7 @@ static bool trans_CASP(DisasContext *s, arg_CASP *a)
 
 static bool trans_CAS(DisasContext *s, arg_CAS *a)
 {
-    if (!dc_isar_feature(aa64_atomics, s)) {
+    if (!dc_isar_feature(aa64_lse, s)) {
         return false;
     }
     gen_compare_and_swap(s, a->rs, a->rt, a->rn, a->sz);
@@ -3743,15 +3743,64 @@ static bool do_atomic_ld(DisasContext *s, arg_atomic *a, AtomicThreeOpFn *fn,
     return true;
 }
 
-TRANS_FEAT(LDADD, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_add_i64, 0, false)
-TRANS_FEAT(LDCLR, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_and_i64, 0, true)
-TRANS_FEAT(LDEOR, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_xor_i64, 0, false)
-TRANS_FEAT(LDSET, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_or_i64, 0, false)
-TRANS_FEAT(LDSMAX, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_smax_i64, MO_SIGN, false)
-TRANS_FEAT(LDSMIN, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_smin_i64, MO_SIGN, false)
-TRANS_FEAT(LDUMAX, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_umax_i64, 0, false)
-TRANS_FEAT(LDUMIN, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_fetch_umin_i64, 0, false)
-TRANS_FEAT(SWP, aa64_atomics, do_atomic_ld, a, tcg_gen_atomic_xchg_i64, 0, false)
+TRANS_FEAT(LDADD, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_add_i64, 0, false)
+TRANS_FEAT(LDCLR, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_and_i64, 0, true)
+TRANS_FEAT(LDEOR, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_xor_i64, 0, false)
+TRANS_FEAT(LDSET, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_or_i64, 0, false)
+TRANS_FEAT(LDSMAX, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_smax_i64, MO_SIGN, false)
+TRANS_FEAT(LDSMIN, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_smin_i64, MO_SIGN, false)
+TRANS_FEAT(LDUMAX, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_umax_i64, 0, false)
+TRANS_FEAT(LDUMIN, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_fetch_umin_i64, 0, false)
+TRANS_FEAT(SWP, aa64_lse, do_atomic_ld, a, tcg_gen_atomic_xchg_i64, 0, false)
+
+typedef void Atomic128ThreeOpFn(TCGv_i128, TCGv_i64, TCGv_i128, TCGArg, MemOp);
+
+static bool do_atomic128_ld(DisasContext *s, arg_atomic128 *a,
+                            Atomic128ThreeOpFn *fn, bool invert)
+{
+    MemOp mop;
+    int rlo, rhi;
+    TCGv_i64 clean_addr, tlo, thi;
+    TCGv_i128 t16;
+
+    if (a->rt == 31 || a->rt2 == 31 || a->rt == a->rt2) {
+        return false;
+    }
+    if (a->rn == 31) {
+        gen_check_sp_alignment(s);
+    }
+    mop = check_atomic_align(s, a->rn, MO_128);
+    clean_addr = gen_mte_check1(s, cpu_reg_sp(s, a->rn), false,
+                                a->rn != 31, mop);
+
+    rlo = (s->be_data == MO_LE ? a->rt : a->rt2);
+    rhi = (s->be_data == MO_LE ? a->rt2 : a->rt);
+
+    tlo = read_cpu_reg(s, rlo, true);
+    thi = read_cpu_reg(s, rhi, true);
+    if (invert) {
+        tcg_gen_not_i64(tlo, tlo);
+        tcg_gen_not_i64(thi, thi);
+    }
+    /*
+     * The tcg atomic primitives are all full barriers.  Therefore we
+     * can ignore the Acquire and Release bits of this instruction.
+     */
+    t16 = tcg_temp_new_i128();
+    tcg_gen_concat_i64_i128(t16, tlo, thi);
+
+    fn(t16, clean_addr, t16, get_mem_index(s), mop);
+
+    tcg_gen_extr_i128_i64(cpu_reg(s, rlo), cpu_reg(s, rhi), t16);
+    return true;
+}
+
+TRANS_FEAT(LDCLRP, aa64_lse128, do_atomic128_ld,
+           a, tcg_gen_atomic_fetch_and_i128, true)
+TRANS_FEAT(LDSETP, aa64_lse128, do_atomic128_ld,
+           a, tcg_gen_atomic_fetch_or_i128, false)
+TRANS_FEAT(SWPP, aa64_lse128, do_atomic128_ld,
+           a, tcg_gen_atomic_xchg_i128, false)
 
 static bool trans_LDAPR(DisasContext *s, arg_LDAPR *a)
 {
@@ -3759,7 +3808,7 @@ static bool trans_LDAPR(DisasContext *s, arg_LDAPR *a)
     TCGv_i64 clean_addr;
     MemOp mop;
 
-    if (!dc_isar_feature(aa64_atomics, s) ||
+    if (!dc_isar_feature(aa64_lse, s) ||
         !dc_isar_feature(aa64_rcpc_8_3, s)) {
         return false;
     }
@@ -4553,6 +4602,50 @@ TRANS(ADDS_i, gen_rri, a, 0, 1, a->sf ? gen_add64_CC : gen_add32_CC)
 TRANS(SUBS_i, gen_rri, a, 0, 1, a->sf ? gen_sub64_CC : gen_sub32_CC)
 
 /*
+ * Min/Max (immediate)
+ */
+
+static void gen_wrap3_i32(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, NeonGenTwoOpFn fn)
+{
+    TCGv_i32 t1 = tcg_temp_new_i32();
+    TCGv_i32 t2 = tcg_temp_new_i32();
+
+    tcg_gen_extrl_i64_i32(t1, n);
+    tcg_gen_extrl_i64_i32(t2, m);
+    fn(t1, t1, t2);
+    tcg_gen_extu_i32_i64(d, t1);
+}
+
+static void gen_smax32_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m)
+{
+    gen_wrap3_i32(d, n, m, tcg_gen_smax_i32);
+}
+
+static void gen_smin32_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m)
+{
+    gen_wrap3_i32(d, n, m, tcg_gen_smin_i32);
+}
+
+static void gen_umax32_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m)
+{
+    gen_wrap3_i32(d, n, m, tcg_gen_umax_i32);
+}
+
+static void gen_umin32_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m)
+{
+    gen_wrap3_i32(d, n, m, tcg_gen_umin_i32);
+}
+
+TRANS_FEAT(SMAX_i, aa64_cssc, gen_rri, a, 0, 0,
+           a->sf ? tcg_gen_smax_i64 : gen_smax32_i64)
+TRANS_FEAT(SMIN_i, aa64_cssc, gen_rri, a, 0, 0,
+           a->sf ? tcg_gen_smin_i64 : gen_smin32_i64)
+TRANS_FEAT(UMAX_i, aa64_cssc, gen_rri, a, 0, 0,
+           a->sf ? tcg_gen_umax_i64 : gen_umax32_i64)
+TRANS_FEAT(UMIN_i, aa64_cssc, gen_rri, a, 0, 0,
+           a->sf ? tcg_gen_umin_i64 : gen_umin32_i64)
+
+/*
  * Add/subtract (immediate, with tags)
  */
 
@@ -8157,6 +8250,28 @@ static bool trans_PACGA(DisasContext *s, arg_rrr *a)
     return false;
 }
 
+static bool gen_rrr(DisasContext *s, arg_rrr_sf *a, ArithTwoOp fn)
+{
+    TCGv_i64 tcg_rm = cpu_reg(s, a->rm);
+    TCGv_i64 tcg_rn = cpu_reg(s, a->rn);
+    TCGv_i64 tcg_rd = cpu_reg(s, a->rd);
+
+    fn(tcg_rd, tcg_rn, tcg_rm);
+    if (!a->sf) {
+        tcg_gen_ext32u_i64(tcg_rd, tcg_rd);
+    }
+    return true;
+}
+
+TRANS_FEAT(SMAX, aa64_cssc, gen_rrr, a,
+           a->sf ? tcg_gen_smax_i64 : gen_smax32_i64)
+TRANS_FEAT(SMIN, aa64_cssc, gen_rrr, a,
+           a->sf ? tcg_gen_smin_i64 : gen_smin32_i64)
+TRANS_FEAT(UMAX, aa64_cssc, gen_rrr, a,
+           a->sf ? tcg_gen_umax_i64 : gen_umax32_i64)
+TRANS_FEAT(UMIN, aa64_cssc, gen_rrr, a,
+           a->sf ? tcg_gen_umin_i64 : gen_umin32_i64)
+
 typedef void ArithOneOp(TCGv_i64, TCGv_i64);
 
 static bool gen_rr(DisasContext *s, int rd, int rn, ArithOneOp fn)
@@ -8165,13 +8280,22 @@ static bool gen_rr(DisasContext *s, int rd, int rn, ArithOneOp fn)
     return true;
 }
 
-static void gen_rbit32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+/*
+ * Perform 32-bit operation fn on the low half of n;
+ * the high half of the output is zeroed.
+ */
+static void gen_wrap2_i32(TCGv_i64 d, TCGv_i64 n, NeonGenOneOpFn fn)
 {
-    TCGv_i32 t32 = tcg_temp_new_i32();
+    TCGv_i32 t = tcg_temp_new_i32();
 
-    tcg_gen_extrl_i64_i32(t32, tcg_rn);
-    gen_helper_rbit(t32, t32);
-    tcg_gen_extu_i32_i64(tcg_rd, t32);
+    tcg_gen_extrl_i64_i32(t, n);
+    fn(t, t);
+    tcg_gen_extu_i32_i64(d, t);
+}
+
+static void gen_rbit32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+{
+    gen_wrap2_i32(tcg_rd, tcg_rn, gen_helper_rbit);
 }
 
 static void gen_rev16_xx(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn, TCGv_i64 mask)
@@ -8227,15 +8351,42 @@ static void gen_clz64(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
 
 static void gen_cls32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
 {
+    gen_wrap2_i32(tcg_rd, tcg_rn, tcg_gen_clrsb_i32);
+}
+
+TRANS(CLZ, gen_rr, a->rd, a->rn, a->sf ? gen_clz64 : gen_clz32)
+TRANS(CLS, gen_rr, a->rd, a->rn, a->sf ? tcg_gen_clrsb_i64 : gen_cls32)
+
+static void gen_ctz32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+{
     TCGv_i32 t32 = tcg_temp_new_i32();
 
     tcg_gen_extrl_i64_i32(t32, tcg_rn);
-    tcg_gen_clrsb_i32(t32, t32);
+    tcg_gen_ctzi_i32(t32, t32, 32);
     tcg_gen_extu_i32_i64(tcg_rd, t32);
 }
 
-TRANS(CLZ, gen_rr, a->rd, a->rn, a->sf ? gen_clz64 : gen_clz32)
-TRANS(CLS, gen_rr, a->rd, a->rn, a->sf ? tcg_gen_clrsb_i64 : gen_cls32)
+static void gen_ctz64(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+{
+    tcg_gen_ctzi_i64(tcg_rd, tcg_rn, 64);
+}
+
+static void gen_cnt32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+{
+    gen_wrap2_i32(tcg_rd, tcg_rn, tcg_gen_ctpop_i32);
+}
+
+static void gen_abs32(TCGv_i64 tcg_rd, TCGv_i64 tcg_rn)
+{
+    gen_wrap2_i32(tcg_rd, tcg_rn, tcg_gen_abs_i32);
+}
+
+TRANS_FEAT(CTZ, aa64_cssc, gen_rr, a->rd, a->rn,
+           a->sf ? gen_ctz64 : gen_ctz32)
+TRANS_FEAT(CNT, aa64_cssc, gen_rr, a->rd, a->rn,
+           a->sf ? tcg_gen_ctpop_i64 : gen_cnt32)
+TRANS_FEAT(ABS, aa64_cssc, gen_rr, a->rd, a->rn,
+           a->sf ? tcg_gen_abs_i64 : gen_abs32)
 
 static bool gen_pacaut(DisasContext *s, arg_pacaut *a, NeonGenTwo64OpEnvFn fn)
 {