ARM TCG conversion 12/16.

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@4149 c046a42c-6fe2-441c-8c8c-71466251a162

7 files changed, 277 insertions, 330 deletions

diff --git a/target-arm/exec.h b/target-arm/exec.h
index d953098..af18181 100644
--- a/target-arm/exec.h
+++ b/target-arm/exec.h
@@ -61,13 +61,6 @@ static inline int cpu_halted(CPUState *env) {
 
 /* In op_helper.c */
 
-void helper_set_cp(CPUState *, uint32_t, uint32_t);
-uint32_t helper_get_cp(CPUState *, uint32_t);
-void helper_set_cp15(CPUState *, uint32_t, uint32_t);
-uint32_t helper_get_cp15(CPUState *, uint32_t);
-uint32_t helper_v7m_mrs(CPUState *env, int reg);
-void helper_v7m_msr(CPUState *env, int reg, uint32_t val);
-
 void helper_mark_exclusive(CPUARMState *, uint32_t addr);
 int helper_test_exclusive(CPUARMState *, uint32_t addr);
 void helper_clrex(CPUARMState *env);
diff --git a/target-arm/helper.c b/target-arm/helper.c
index 257960a..6438882 100644
--- a/target-arm/helper.c
+++ b/target-arm/helper.c
@@ -470,38 +470,38 @@ target_phys_addr_t cpu_get_phys_page_debug(CPUState *env, target_ulong addr)
 }
 
 /* These should probably raise undefined insn exceptions.  */
-void helper_set_cp(CPUState *env, uint32_t insn, uint32_t val)
+void HELPER(set_cp)(CPUState *env, uint32_t insn, uint32_t val)
 {
     int op1 = (insn >> 8) & 0xf;
     cpu_abort(env, "cp%i insn %08x\n", op1, insn);
     return;
 }
 
-uint32_t helper_get_cp(CPUState *env, uint32_t insn)
+uint32_t HELPER(get_cp)(CPUState *env, uint32_t insn)
 {
     int op1 = (insn >> 8) & 0xf;
     cpu_abort(env, "cp%i insn %08x\n", op1, insn);
     return 0;
 }
 
-void helper_set_cp15(CPUState *env, uint32_t insn, uint32_t val)
+void HELPER(set_cp15)(CPUState *env, uint32_t insn, uint32_t val)
 {
     cpu_abort(env, "cp15 insn %08x\n", insn);
 }
 
-uint32_t helper_get_cp15(CPUState *env, uint32_t insn)
+uint32_t HELPER(get_cp15)(CPUState *env, uint32_t insn)
 {
     cpu_abort(env, "cp15 insn %08x\n", insn);
     return 0;
 }
 
 /* These should probably raise undefined insn exceptions.  */
-void helper_v7m_msr(CPUState *env, int reg, uint32_t val)
+void HELPER(v7m_msr)(CPUState *env, uint32_t reg, uint32_t val)
 {
     cpu_abort(env, "v7m_mrs %d\n", reg);
 }
 
-uint32_t helper_v7m_mrs(CPUState *env, int reg)
+uint32_t HELPER(v7m_mrs)(CPUState *env, uint32_t reg)
 {
     cpu_abort(env, "v7m_mrs %d\n", reg);
     return 0;
@@ -1191,7 +1191,7 @@ void helper_clrex(CPUState *env)
     env->mmon_addr = -1;
 }
 
-void helper_set_cp(CPUState *env, uint32_t insn, uint32_t val)
+void HELPER(set_cp)(CPUState *env, uint32_t insn, uint32_t val)
 {
     int cp_num = (insn >> 8) & 0xf;
     int cp_info = (insn >> 5) & 7;
@@ -1203,7 +1203,7 @@ void helper_set_cp(CPUState *env, uint32_t insn, uint32_t val)
                                  cp_info, src, operand, val);
 }
 
-uint32_t helper_get_cp(CPUState *env, uint32_t insn)
+uint32_t HELPER(get_cp)(CPUState *env, uint32_t insn)
 {
     int cp_num = (insn >> 8) & 0xf;
     int cp_info = (insn >> 5) & 7;
@@ -1246,7 +1246,7 @@ static uint32_t extended_mpu_ap_bits(uint32_t val)
     return ret;
 }
 
-void helper_set_cp15(CPUState *env, uint32_t insn, uint32_t val)
+void HELPER(set_cp15)(CPUState *env, uint32_t insn, uint32_t val)
 {
     int op1;
     int op2;
@@ -1530,7 +1530,7 @@ bad_reg:
               (insn >> 16) & 0xf, crm, op1, op2);
 }
 
-uint32_t helper_get_cp15(CPUState *env, uint32_t insn)
+uint32_t HELPER(get_cp15)(CPUState *env, uint32_t insn)
 {
     int op1;
     int op2;
@@ -1803,7 +1803,7 @@ uint32_t HELPER(get_r13_banked)(CPUState *env, uint32_t mode)
     return env->banked_r13[bank_number(mode)];
 }
 
-uint32_t helper_v7m_mrs(CPUState *env, int reg)
+uint32_t HELPER(v7m_mrs)(CPUState *env, uint32_t reg)
 {
     switch (reg) {
     case 0: /* APSR */
@@ -1840,7 +1840,7 @@ uint32_t helper_v7m_mrs(CPUState *env, int reg)
     }
 }
 
-void helper_v7m_msr(CPUState *env, int reg, uint32_t val)
+void HELPER(v7m_msr)(CPUState *env, uint32_t reg, uint32_t val)
 {
     switch (reg) {
     case 0: /* APSR */
diff --git a/target-arm/helpers.h b/target-arm/helpers.h
index 4a22ef3..08eb590 100644
--- a/target-arm/helpers.h
+++ b/target-arm/helpers.h
@@ -118,6 +118,15 @@ DEF_HELPER_0_0(wfi, void, (void))
 DEF_HELPER_0_2(cpsr_write, void, (uint32_t, uint32_t))
 DEF_HELPER_1_0(cpsr_read, uint32_t, (void))
 
+DEF_HELPER_0_3(v7m_msr, void, (CPUState *, uint32_t, uint32_t))
+DEF_HELPER_1_2(v7m_mrs, uint32_t, (CPUState *, uint32_t))
+
+DEF_HELPER_0_3(set_cp15, void, (CPUState *, uint32_t, uint32_t))
+DEF_HELPER_1_2(get_cp15, uint32_t, (CPUState *, uint32_t))
+
+DEF_HELPER_0_3(set_cp, void, (CPUState *, uint32_t, uint32_t))
+DEF_HELPER_1_2(get_cp, uint32_t, (CPUState *, uint32_t))
+
 DEF_HELPER_1_2(get_r13_banked, uint32_t, (CPUState *, uint32_t))
 DEF_HELPER_0_3(set_r13_banked, void, (CPUState *, uint32_t, uint32_t))
 
@@ -187,6 +196,20 @@ DEF_HELPER_1_2(rsqrte_f32, float32, (float32, CPUState *))
 DEF_HELPER_1_2(recpe_u32, uint32_t, (uint32_t, CPUState *))
 DEF_HELPER_1_2(rsqrte_u32, uint32_t, (uint32_t, CPUState *))
 
+DEF_HELPER_1_2(add_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(adc_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(sub_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(sbc_cc, uint32_t, (uint32_t, uint32_t))
+
+DEF_HELPER_1_2(shl, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(shr, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(sar, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(ror, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(shl_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(shr_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(sar_cc, uint32_t, (uint32_t, uint32_t))
+DEF_HELPER_1_2(ror_cc, uint32_t, (uint32_t, uint32_t))
+
 #undef DEF_HELPER
 #undef DEF_HELPER_0_0
 #undef DEF_HELPER_0_1
diff --git a/target-arm/op.c b/target-arm/op.c
index 7c6d6a1..57086e3 100644
--- a/target-arm/op.c
+++ b/target-arm/op.c
@@ -20,66 +20,6 @@
  */
 #include "exec.h"
 
-void OPPROTO op_addl_T0_T1_cc(void)
-{
-    unsigned int src1;
-    src1 = T0;
-    T0 += T1;
-    env->NZF = T0;
-    env->CF = T0 < src1;
-    env->VF = (src1 ^ T1 ^ -1) & (src1 ^ T0);
-}
-
-void OPPROTO op_adcl_T0_T1_cc(void)
-{
-    unsigned int src1;
-    src1 = T0;
-    if (!env->CF) {
-        T0 += T1;
-        env->CF = T0 < src1;
-    } else {
-        T0 += T1 + 1;
-        env->CF = T0 <= src1;
-    }
-    env->VF = (src1 ^ T1 ^ -1) & (src1 ^ T0);
-    env->NZF = T0;
-    FORCE_RET();
-}
-
-#define OPSUB(sub, sbc, res, T0, T1)            \
-                                                \
-void OPPROTO op_ ## sub ## l_T0_T1_cc(void)     \
-{                                               \
-    unsigned int src1;                          \
-    src1 = T0;                                  \
-    T0 -= T1;                                   \
-    env->NZF = T0;                              \
-    env->CF = src1 >= T1;                       \
-    env->VF = (src1 ^ T1) & (src1 ^ T0);        \
-    res = T0;                                   \
-}                                               \
-                                                \
-void OPPROTO op_ ## sbc ## l_T0_T1_cc(void)     \
-{                                               \
-    unsigned int src1;                          \
-    src1 = T0;                                  \
-    if (!env->CF) {                             \
-        T0 = T0 - T1 - 1;                       \
-        env->CF = src1 > T1;                    \
-    } else {                                    \
-        T0 = T0 - T1;                           \
-        env->CF = src1 >= T1;                   \
-    }                                           \
-    env->VF = (src1 ^ T1) & (src1 ^ T0);        \
-    env->NZF = T0;                              \
-    res = T0;                                   \
-    FORCE_RET();                                \
-}
-
-OPSUB(sub, sbc, T0, T0, T1)
-
-OPSUB(rsb, rsc, T0, T1, T0)
-
 /* memory access */
 
 #define MEMSUFFIX _raw
@@ -92,164 +32,6 @@ OPSUB(rsb, rsc, T0, T1, T0)
 #include "op_mem.h"
 #endif
 
-void OPPROTO op_clrex(void)
-{
-    cpu_lock();
-    helper_clrex(env);
-    cpu_unlock();
-}
-
-/* T1 based, use T0 as shift count */
-
-void OPPROTO op_shll_T1_T0(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32)
-        T1 = 0;
-    else
-        T1 = T1 << shift;
-    FORCE_RET();
-}
-
-void OPPROTO op_shrl_T1_T0(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32)
-        T1 = 0;
-    else
-        T1 = (uint32_t)T1 >> shift;
-    FORCE_RET();
-}
-
-void OPPROTO op_sarl_T1_T0(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32)
-        shift = 31;
-    T1 = (int32_t)T1 >> shift;
-}
-
-void OPPROTO op_rorl_T1_T0(void)
-{
-    int shift;
-    shift = T0 & 0x1f;
-    if (shift) {
-        T1 = ((uint32_t)T1 >> shift) | (T1 << (32 - shift));
-    }
-    FORCE_RET();
-}
-
-/* T1 based, use T0 as shift count and compute CF */
-
-void OPPROTO op_shll_T1_T0_cc(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32) {
-        if (shift == 32)
-            env->CF = T1 & 1;
-        else
-            env->CF = 0;
-        T1 = 0;
-    } else if (shift != 0) {
-        env->CF = (T1 >> (32 - shift)) & 1;
-        T1 = T1 << shift;
-    }
-    FORCE_RET();
-}
-
-void OPPROTO op_shrl_T1_T0_cc(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32) {
-        if (shift == 32)
-            env->CF = (T1 >> 31) & 1;
-        else
-            env->CF = 0;
-        T1 = 0;
-    } else if (shift != 0) {
-        env->CF = (T1 >> (shift - 1)) & 1;
-        T1 = (uint32_t)T1 >> shift;
-    }
-    FORCE_RET();
-}
-
-void OPPROTO op_sarl_T1_T0_cc(void)
-{
-    int shift;
-    shift = T0 & 0xff;
-    if (shift >= 32) {
-        env->CF = (T1 >> 31) & 1;
-        T1 = (int32_t)T1 >> 31;
-    } else if (shift != 0) {
-        env->CF = (T1 >> (shift - 1)) & 1;
-        T1 = (int32_t)T1 >> shift;
-    }
-    FORCE_RET();
-}
-
-void OPPROTO op_rorl_T1_T0_cc(void)
-{
-    int shift1, shift;
-    shift1 = T0 & 0xff;
-    shift = shift1 & 0x1f;
-    if (shift == 0) {
-        if (shift1 != 0)
-            env->CF = (T1 >> 31) & 1;
-    } else {
-        env->CF = (T1 >> (shift - 1)) & 1;
-        T1 = ((uint32_t)T1 >> shift) | (T1 << (32 - shift));
-    }
-    FORCE_RET();
-}
-
-void OPPROTO op_movl_cp_T0(void)
-{
-    helper_set_cp(env, PARAM1, T0);
-    FORCE_RET();
-}
-
-void OPPROTO op_movl_T0_cp(void)
-{
-    T0 = helper_get_cp(env, PARAM1);
-    FORCE_RET();
-}
-
-void OPPROTO op_movl_cp15_T0(void)
-{
-    helper_set_cp15(env, PARAM1, T0);
-    FORCE_RET();
-}
-
-void OPPROTO op_movl_T0_cp15(void)
-{
-    T0 = helper_get_cp15(env, PARAM1);
-    FORCE_RET();
-}
-
-void OPPROTO op_v7m_mrs_T0(void)
-{
-    T0 = helper_v7m_mrs(env, PARAM1);
-}
-
-void OPPROTO op_v7m_msr_T0(void)
-{
-    helper_v7m_msr(env, PARAM1, T0);
-}
-
-void OPPROTO op_movl_T0_sp(void)
-{
-    if (PARAM1 == env->v7m.current_sp)
-        T0 = env->regs[13];
-    else
-        T0 = env->v7m.other_sp;
-    FORCE_RET();
-}
-
 #include "op_neon.h"
 
 /* iwMMXt support */
diff --git a/target-arm/op_helper.c b/target-arm/op_helper.c
index 6748b06..79b3f1a 100644
--- a/target-arm/op_helper.c
+++ b/target-arm/op_helper.c
@@ -304,3 +304,151 @@ void HELPER(set_user_reg)(uint32_t regno, uint32_t val)
     }
 }
 
+/* ??? Flag setting arithmetic is awkward because we need to do comparisons.
+   The only way to do that in TCG is a conditional branch, which clobbers
+   all our temporaries.  For now implement these as helper functions.  */
+
+uint32_t HELPER (add_cc)(uint32_t a, uint32_t b)
+{
+    uint32_t result;
+    result = T0 + T1;
+    env->NZF = result;
+    env->CF = result < a;
+    env->VF = (a ^ b ^ -1) & (a ^ result);
+    return result;
+}
+
+uint32_t HELPER(adc_cc)(uint32_t a, uint32_t b)
+{
+    uint32_t result;
+    if (!env->CF) {
+        result = a + b;
+        env->CF = result < a;
+    } else {
+        result = a + b + 1;
+        env->CF = result <= a;
+    }
+    env->VF = (a ^ b ^ -1) & (a ^ result);
+    env->NZF = result;
+    return result;
+}
+
+uint32_t HELPER(sub_cc)(uint32_t a, uint32_t b)
+{
+    uint32_t result;
+    result = a - b;
+    env->NZF = result;
+    env->CF = a >= b;
+    env->VF = (a ^ b) & (a ^ result);
+    return result;
+}
+
+uint32_t HELPER(sbc_cc)(uint32_t a, uint32_t b)
+{
+    uint32_t result;
+    if (!env->CF) {
+        result = a - b - 1;
+        env->CF = a > b;
+    } else {
+        result = a - b;
+        env->CF = a >= b;
+    }
+    env->VF = (a ^ b) & (a ^ result);
+    env->NZF = result;
+    return result;
+}
+
+/* Similarly for variable shift instructions.  */
+
+uint32_t HELPER(shl)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32)
+        return 0;
+    return x << shift;
+}
+
+uint32_t HELPER(shr)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32)
+        return 0;
+    return (uint32_t)x >> shift;
+}
+
+uint32_t HELPER(sar)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32)
+        shift = 31;
+    return (int32_t)x >> shift;
+}
+
+uint32_t HELPER(ror)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift == 0)
+        return x;
+    return (x >> shift) | (x << (32 - shift));
+}
+
+uint32_t HELPER(shl_cc)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32) {
+        if (shift == 32)
+            env->CF = x & 1;
+        else
+            env->CF = 0;
+        return 0;
+    } else if (shift != 0) {
+        env->CF = (x >> (32 - shift)) & 1;
+        return x << shift;
+    }
+    return x;
+}
+
+uint32_t HELPER(shr_cc)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32) {
+        if (shift == 32)
+            env->CF = (x >> 31) & 1;
+        else
+            env->CF = 0;
+        return 0;
+    } else if (shift != 0) {
+        env->CF = (x >> (shift - 1)) & 1;
+        return x >> shift;
+    }
+    return x;
+}
+
+uint32_t HELPER(sar_cc)(uint32_t x, uint32_t i)
+{
+    int shift = i & 0xff;
+    if (shift >= 32) {
+        env->CF = (x >> 31) & 1;
+        return (int32_t)x >> 31;
+    } else if (shift != 0) {
+        env->CF = (x >> (shift - 1)) & 1;
+        return (int32_t)x >> shift;
+    }
+    return x;
+}
+
+uint32_t HELPER(ror_cc)(uint32_t x, uint32_t i)
+{
+    int shift1, shift;
+    shift1 = i & 0xff;
+    shift = shift1 & 0x1f;
+    if (shift == 0) {
+        if (shift1 != 0)
+            env->CF = (x >> 31) & 1;
+        return x;
+    } else {
+        env->CF = (x >> (shift - 1)) & 1;
+        return ((uint32_t)x >> shift) | (x << (32 - shift));
+    }
+}
+
diff --git a/target-arm/op_mem.h b/target-arm/op_mem.h
index 0e2e7b6..519544d 100644
--- a/target-arm/op_mem.h
+++ b/target-arm/op_mem.h
@@ -1,24 +1,5 @@
 /* ARM memory operations.  */
 
-/* Swap T0 with memory at address T1.  */
-/* ??? Is this exception safe?  */
-#define MEM_SWP_OP(name, lname) \
-void OPPROTO glue(op_swp##name,MEMSUFFIX)(void) \
-{ \
-    uint32_t tmp; \
-    cpu_lock(); \
-    tmp = glue(ld##lname,MEMSUFFIX)(T1); \
-    glue(st##name,MEMSUFFIX)(T1, T0); \
-    T0 = tmp; \
-    cpu_unlock(); \
-    FORCE_RET(); \
-}
-
-MEM_SWP_OP(b, ub)
-MEM_SWP_OP(l, l)
-
-#undef MEM_SWP_OP
-
 /* Load-locked, store exclusive.  */
 #define EXCLUSIVE_OP(suffix, ldsuffix) \
 void OPPROTO glue(op_ld##suffix##ex,MEMSUFFIX)(void) \
diff --git a/target-arm/translate.c b/target-arm/translate.c
index 369dfab..0220d6a 100644
--- a/target-arm/translate.c
+++ b/target-arm/translate.c
@@ -201,6 +201,13 @@ static void store_reg(DisasContext *s, int reg, TCGv var)
 #define gen_op_subl_T0_T1() tcg_gen_sub_i32(cpu_T[0], cpu_T[0], cpu_T[1])
 #define gen_op_rsbl_T0_T1() tcg_gen_sub_i32(cpu_T[0], cpu_T[1], cpu_T[0])
 
+#define gen_op_addl_T0_T1_cc() gen_helper_add_cc(cpu_T[0], cpu_T[0], cpu_T[1])
+#define gen_op_adcl_T0_T1_cc() gen_helper_adc_cc(cpu_T[0], cpu_T[0], cpu_T[1])
+#define gen_op_subl_T0_T1_cc() gen_helper_sub_cc(cpu_T[0], cpu_T[0], cpu_T[1])
+#define gen_op_sbcl_T0_T1_cc() gen_helper_sbc_cc(cpu_T[0], cpu_T[0], cpu_T[1])
+#define gen_op_rsbl_T0_T1_cc() gen_helper_sub_cc(cpu_T[0], cpu_T[1], cpu_T[0])
+#define gen_op_rscl_T0_T1_cc() gen_helper_sbc_cc(cpu_T[0], cpu_T[1], cpu_T[0])
+
 #define gen_op_andl_T0_T1() tcg_gen_and_i32(cpu_T[0], cpu_T[0], cpu_T[1])
 #define gen_op_xorl_T0_T1() tcg_gen_xor_i32(cpu_T[0], cpu_T[0], cpu_T[1])
 #define gen_op_orl_T0_T1() tcg_gen_or_i32(cpu_T[0], cpu_T[0], cpu_T[1])
@@ -538,6 +545,27 @@ static inline void gen_arm_shift_im(TCGv var, int shiftop, int shift, int flags)
     }
 };
 
+static inline void gen_arm_shift_reg(TCGv var, int shiftop,
+                                     TCGv shift, int flags)
+{
+    if (flags) {
+        switch (shiftop) {
+        case 0: gen_helper_shl_cc(var, var, shift); break;
+        case 1: gen_helper_shr_cc(var, var, shift); break;
+        case 2: gen_helper_sar_cc(var, var, shift); break;
+        case 3: gen_helper_ror_cc(var, var, shift); break;
+        }
+    } else {
+        switch (shiftop) {
+        case 0: gen_helper_shl(var, var, shift); break;
+        case 1: gen_helper_shr(var, var, shift); break;
+        case 2: gen_helper_sar(var, var, shift); break;
+        case 3: gen_helper_ror(var, var, shift); break;
+        }
+    }
+    dead_tmp(shift);
+}
+
 #define PAS_OP(pfx) \
     switch (op2) {  \
     case 0: gen_pas_helper(glue(pfx,add16)); break; \
@@ -746,20 +774,6 @@ const uint8_t table_logic_cc[16] = {
     1, /* mvn */
 };
 
-static GenOpFunc *gen_shift_T1_T0[4] = {
-    gen_op_shll_T1_T0,
-    gen_op_shrl_T1_T0,
-    gen_op_sarl_T1_T0,
-    gen_op_rorl_T1_T0,
-};
-
-static GenOpFunc *gen_shift_T1_T0_cc[4] = {
-    gen_op_shll_T1_T0_cc,
-    gen_op_shrl_T1_T0_cc,
-    gen_op_sarl_T1_T0_cc,
-    gen_op_rorl_T1_T0_cc,
-};
-
 /* Set PC and Thumb state from an immediate address.  */
 static inline void gen_bx_im(DisasContext *s, uint32_t addr)
 {
@@ -2249,6 +2263,7 @@ static int disas_dsp_insn(CPUState *env, DisasContext *s, uint32_t insn)
    instruction is not defined.  */
 static int disas_cp_insn(CPUState *env, DisasContext *s, uint32_t insn)
 {
+    TCGv tmp;
     uint32_t rd = (insn >> 12) & 0xf;
     uint32_t cp = (insn >> 8) & 0xf;
     if (IS_USER(s)) {
@@ -2258,17 +2273,16 @@ static int disas_cp_insn(CPUState *env, DisasContext *s, uint32_t insn)
     if (insn & ARM_CP_RW_BIT) {
         if (!env->cp[cp].cp_read)
             return 1;
-        gen_op_movl_T0_im((uint32_t) s->pc);
-        gen_set_pc_T0();
-        gen_op_movl_T0_cp(insn);
-        gen_movl_reg_T0(s, rd);
+        gen_set_pc_im(s->pc);
+        tmp = new_tmp();
+        gen_helper_get_cp(tmp, cpu_env, tcg_const_i32(insn));
+        store_reg(s, rd, tmp);
     } else {
         if (!env->cp[cp].cp_write)
             return 1;
-        gen_op_movl_T0_im((uint32_t) s->pc);
-        gen_set_pc_T0();
-        gen_movl_T0_reg(s, rd);
-        gen_op_movl_cp_T0(insn);
+        gen_set_pc_im(s->pc);
+        tmp = load_reg(s, rd);
+        gen_helper_set_cp(cpu_env, tcg_const_i32(insn), tmp);
     }
     return 0;
 }
@@ -2298,6 +2312,7 @@ static int cp15_user_ok(uint32_t insn)
 static int disas_cp15_insn(CPUState *env, DisasContext *s, uint32_t insn)
 {
     uint32_t rd;
+    TCGv tmp;
 
     /* M profile cores use memory mapped registers instead of cp15.  */
     if (arm_feature(env, ARM_FEATURE_M))
@@ -2321,20 +2336,23 @@ static int disas_cp15_insn(CPUState *env, DisasContext *s, uint32_t insn)
     if ((insn & 0x0fff0fff) == 0x0e070f90
         || (insn & 0x0fff0fff) == 0x0e070f58) {
         /* Wait for interrupt.  */
-        gen_op_movl_T0_im((long)s->pc);
-        gen_set_pc_T0();
+        gen_set_pc_im(s->pc);
         s->is_jmp = DISAS_WFI;
         return 0;
     }
     rd = (insn >> 12) & 0xf;
     if (insn & ARM_CP_RW_BIT) {
-        gen_op_movl_T0_cp15(insn);
+        tmp = new_tmp();
+        gen_helper_get_cp15(tmp, cpu_env, tcg_const_i32(insn));
         /* If the destination register is r15 then sets condition codes.  */
         if (rd != 15)
-            gen_movl_reg_T0(s, rd);
+            store_reg(s, rd, tmp);
+        else
+            dead_tmp(tmp);
     } else {
-        gen_movl_T0_reg(s, rd);
-        gen_op_movl_cp15_T0(insn);
+        tmp = load_reg(s, rd);
+        gen_helper_set_cp15(cpu_env, tcg_const_i32(insn), tmp);
+        dead_tmp(tmp);
         /* Normally we would always end the TB here, but Linux
          * arch/arm/mach-pxa/sleep.S expects two instructions following
          * an MMU enable to execute from cache.  Imitate this behaviour.  */
@@ -3052,12 +3070,10 @@ static inline void gen_goto_tb(DisasContext *s, int n, uint32_t dest)
     tb = s->tb;
     if ((tb->pc & TARGET_PAGE_MASK) == (dest & TARGET_PAGE_MASK)) {
         tcg_gen_goto_tb(n);
-        gen_op_movl_T0_im(dest);
-        gen_set_pc_T0();
+        gen_set_pc_im(dest);
         tcg_gen_exit_tb((long)tb + n);
     } else {
-        gen_op_movl_T0_im(dest);
-        gen_set_pc_T0();
+        gen_set_pc_im(dest);
         tcg_gen_exit_tb(0);
     }
 }
@@ -3173,8 +3189,7 @@ static void gen_nop_hint(DisasContext *s, int val)
 {
     switch (val) {
     case 3: /* wfi */
-        gen_op_movl_T0_im((long)s->pc);
-        gen_set_pc_T0();
+        gen_set_pc_im(s->pc);
         s->is_jmp = DISAS_WFI;
         break;
     case 2: /* wfe */
@@ -5770,12 +5785,8 @@ static void disas_arm_insn(CPUState * env, DisasContext *s)
                 gen_arm_shift_im(cpu_T[1], shiftop, shift, logic_cc);
             } else {
                 rs = (insn >> 8) & 0xf;
-                gen_movl_T0_reg(s, rs);
-                if (logic_cc) {
-                    gen_shift_T1_T0_cc[shiftop]();
-                } else {
-                    gen_shift_T1_T0[shiftop]();
-                }
+                tmp = load_reg(s, rs);
+                gen_arm_shift_reg(cpu_T[1], shiftop, tmp, logic_cc);
             }
         }
         if (op1 != 0x0f && op1 != 0x0d) {
@@ -5977,14 +5988,20 @@ static void disas_arm_insn(CPUState * env, DisasContext *s)
                         /* SWP instruction */
                         rm = (insn) & 0xf;
 
-                        gen_movl_T0_reg(s, rm);
-                        gen_movl_T1_reg(s, rn);
+                        /* ??? This is not really atomic.  However we know
+                           we never have multiple CPUs running in parallel,
+                           so it is good enough.  */
+                        addr = load_reg(s, rn);
+                        tmp = load_reg(s, rm);
                         if (insn & (1 << 22)) {
-                            gen_ldst(swpb, s);
+                            tmp2 = gen_ld8u(addr, IS_USER(s));
+                            gen_st8(tmp, addr, IS_USER(s));
                         } else {
-                            gen_ldst(swpl, s);
+                            tmp2 = gen_ld32(addr, IS_USER(s));
+                            gen_st32(tmp, addr, IS_USER(s));
                         }
-                        gen_movl_reg_T0(s, rd);
+                        dead_tmp(addr);
+                        store_reg(s, rd, tmp2);
                     }
                 }
             } else {
@@ -6903,18 +6920,16 @@ static int disas_thumb2_insn(CPUState *env, DisasContext *s, uint16_t insn_hw1)
             goto illegal_op;
         switch (op) {
         case 0: /* Register controlled shift.  */
-            gen_movl_T0_reg(s, rm);
-            gen_movl_T1_reg(s, rn);
+            tmp = load_reg(s, rn);
+            tmp2 = load_reg(s, rm);
             if ((insn & 0x70) != 0)
                 goto illegal_op;
             op = (insn >> 21) & 3;
-            if (insn & (1 << 20)) {
-                gen_shift_T1_T0_cc[op]();
-                gen_op_logic_T1_cc();
-            } else {
-                gen_shift_T1_T0[op]();
-            }
-            gen_movl_reg_T1(s, rd);
+            logic_cc = (insn & (1 << 20)) != 0;
+            gen_arm_shift_reg(tmp, op, tmp2, logic_cc);
+            if (logic_cc)
+                gen_logic_CC(tmp);
+            store_reg(s, rd, tmp);
             break;
         case 1: /* Sign/zero extend.  */
             tmp = load_reg(s, rm);
@@ -7208,8 +7223,9 @@ static int disas_thumb2_insn(CPUState *env, DisasContext *s, uint16_t insn_hw1)
                     switch (op) {
                     case 0: /* msr cpsr.  */
                         if (IS_M(env)) {
-                            gen_op_v7m_msr_T0(insn & 0xff);
-                            gen_movl_reg_T0(s, rn);
+                            tmp = load_reg(s, rn);
+                            addr = tcg_const_i32(insn & 0xff);
+                            gen_helper_v7m_msr(cpu_env, addr, tmp);
                             gen_lookup_tb(s);
                             break;
                         }
@@ -7276,12 +7292,14 @@ static int disas_thumb2_insn(CPUState *env, DisasContext *s, uint16_t insn_hw1)
                         /* Unpredictable in user mode.  */
                         goto illegal_op;
                     case 6: /* mrs cpsr.  */
+                        tmp = new_tmp();
                         if (IS_M(env)) {
-                            gen_op_v7m_mrs_T0(insn & 0xff);
+                            addr = tcg_const_i32(insn & 0xff);
+                            gen_helper_v7m_mrs(tmp, cpu_env, addr);
                         } else {
-                            gen_helper_cpsr_read(cpu_T[0]);
+                            gen_helper_cpsr_read(tmp);
                         }
-                        gen_movl_reg_T0(s, rd);
+                        store_reg(s, rd, tmp);
                         break;
                     case 7: /* mrs spsr.  */
                         /* Not accessible in user mode.  */
@@ -7753,25 +7771,25 @@ static void disas_thumb_insn(CPUState *env, DisasContext *s)
             break;
         case 0x2: /* lsl */
             if (s->condexec_mask) {
-                gen_op_shll_T1_T0();
+                gen_helper_shl(cpu_T[1], cpu_T[1], cpu_T[0]);
             } else {
-                gen_op_shll_T1_T0_cc();
+                gen_helper_shl_cc(cpu_T[1], cpu_T[1], cpu_T[0]);
                 gen_op_logic_T1_cc();
             }
             break;
         case 0x3: /* lsr */
             if (s->condexec_mask) {
-                gen_op_shrl_T1_T0();
+                gen_helper_shr(cpu_T[1], cpu_T[1], cpu_T[0]);
             } else {
-                gen_op_shrl_T1_T0_cc();
+                gen_helper_shr_cc(cpu_T[1], cpu_T[1], cpu_T[0]);
                 gen_op_logic_T1_cc();
             }
             break;
         case 0x4: /* asr */
             if (s->condexec_mask) {
-                gen_op_sarl_T1_T0();
+                gen_helper_sar(cpu_T[1], cpu_T[1], cpu_T[0]);
             } else {
-                gen_op_sarl_T1_T0_cc();
+                gen_helper_sar_cc(cpu_T[1], cpu_T[1], cpu_T[0]);
                 gen_op_logic_T1_cc();
             }
             break;
@@ -7789,9 +7807,9 @@ static void disas_thumb_insn(CPUState *env, DisasContext *s)
             break;
         case 0x7: /* ror */
             if (s->condexec_mask) {
-                gen_op_rorl_T1_T0();
+                gen_helper_ror(cpu_T[1], cpu_T[1], cpu_T[0]);
             } else {
-                gen_op_rorl_T1_T0_cc();
+                gen_helper_ror_cc(cpu_T[1], cpu_T[1], cpu_T[0]);
                 gen_op_logic_T1_cc();
             }
             break;
@@ -8118,15 +8136,17 @@ static void disas_thumb_insn(CPUState *env, DisasContext *s)
             if (IS_USER(s))
                 break;
             if (IS_M(env)) {
-                val = (insn & (1 << 4)) != 0;
-                gen_op_movl_T0_im(val);
+                tmp = tcg_const_i32((insn & (1 << 4)) != 0);
                 /* PRIMASK */
-                if (insn & 1)
-                    gen_op_v7m_msr_T0(16);
+                if (insn & 1) {
+                    addr = tcg_const_i32(16);
+                    gen_helper_v7m_msr(cpu_env, addr, tmp);
+                }
                 /* FAULTMASK */
-                if (insn & 2)
-                    gen_op_v7m_msr_T0(17);
-
+                if (insn & 2) {
+                    addr = tcg_const_i32(17);
+                    gen_helper_v7m_msr(cpu_env, addr, tmp);
+                }
                 gen_lookup_tb(s);
             } else {
                 if (insn & (1 << 4))