SH4: Convert remaining non-fp ops to TCG

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5120 c046a42c-6fe2-441c-8c8c-71466251a162

5 files changed, 130 insertions, 153 deletions

diff --git a/target-sh4/exec.h b/target-sh4/exec.h
index 3135874..eb80d92 100644
--- a/target-sh4/exec.h
+++ b/target-sh4/exec.h
@@ -71,10 +71,6 @@ int find_itlb_entry(CPUState * env, target_ulong address,
 		    int use_asid, int update);
 int find_utlb_entry(CPUState * env, target_ulong address, int use_asid);
 
-void helper_div1_T0_T1(void);
-void helper_rotcl(uint32_t * addr);
-void helper_rotcr(uint32_t * addr);
-
 void do_interrupt(CPUState * env);
 
 void cpu_loop_exit(void);
diff --git a/target-sh4/helper.h b/target-sh4/helper.h
index 4dca7ce..c9465a7 100644
--- a/target-sh4/helper.h
+++ b/target-sh4/helper.h
@@ -14,6 +14,7 @@ DEF_HELPER(uint32_t, helper_addc, (uint32_t, uint32_t))
 DEF_HELPER(uint32_t, helper_subv, (uint32_t, uint32_t))
 DEF_HELPER(uint32_t, helper_subc, (uint32_t, uint32_t))
 DEF_HELPER(uint32_t, helper_negc, (uint32_t))
+DEF_HELPER(uint32_t, helper_div1, (uint32_t, uint32_t))
 DEF_HELPER(void, helper_macl, (uint32_t, uint32_t))
 DEF_HELPER(void, helper_macw, (uint32_t, uint32_t))
 
diff --git a/target-sh4/op.c b/target-sh4/op.c
index 7bd2e18..3b25b4a 100644
--- a/target-sh4/op.c
+++ b/target-sh4/op.c
@@ -37,84 +37,6 @@ static inline void cond_t(int cond)
 	clr_t();
 }
 
-void OPPROTO op_cmp_str_T0_T1(void)
-{
-    cond_t((T0 & 0x000000ff) == (T1 & 0x000000ff) ||
-	   (T0 & 0x0000ff00) == (T1 & 0x0000ff00) ||
-	   (T0 & 0x00ff0000) == (T1 & 0x00ff0000) ||
-	   (T0 & 0xff000000) == (T1 & 0xff000000));
-    RETURN();
-}
-
-void OPPROTO op_div0s_T0_T1(void)
-{
-    if (T1 & 0x80000000)
-	env->sr |= SR_Q;
-    else
-	env->sr &= ~SR_Q;
-    if (T0 & 0x80000000)
-	env->sr |= SR_M;
-    else
-	env->sr &= ~SR_M;
-    cond_t((T1 ^ T0) & 0x80000000);
-    RETURN();
-}
-
-void OPPROTO op_div1_T0_T1(void)
-{
-    helper_div1_T0_T1();
-    RETURN();
-}
-
-void OPPROTO op_shad_T0_T1(void)
-{
-    if ((T0 & 0x80000000) == 0)
-	T1 <<= (T0 & 0x1f);
-    else if ((T0 & 0x1f) == 0)
-	T1 = (T1 & 0x80000000)? 0xffffffff : 0;
-    else
-	T1 = ((int32_t) T1) >> ((~T0 & 0x1f) + 1);
-    RETURN();
-}
-
-void OPPROTO op_shld_T0_T1(void)
-{
-    if ((T0 & 0x80000000) == 0)
-	T1 <<= (T0 & 0x1f);
-    else if ((T0 & 0x1f) == 0)
-	T1 = 0;
-    else
-	T1 = ((uint32_t) T1) >> ((~T0 & 0x1f) + 1);
-    RETURN();
-}
-
-void OPPROTO op_rotcl_Rn(void)
-{
-    helper_rotcl(&env->gregs[PARAM1]);
-    RETURN();
-}
-
-void OPPROTO op_rotcr_Rn(void)
-{
-    helper_rotcr(&env->gregs[PARAM1]);
-    RETURN();
-}
-
-void OPPROTO op_rotl_Rn(void) 
-{
-    cond_t(env->gregs[PARAM1] & 0x80000000);
-    env->gregs[PARAM1] = (env->gregs[PARAM1] << 1) | (env->sr & SR_T);
-    RETURN();
-}
-
-void OPPROTO op_rotr_Rn(void)
-{
-    cond_t(env->gregs[PARAM1] & 1);
-    env->gregs[PARAM1] = (env->gregs[PARAM1] >> 1) |
-	((env->sr & SR_T) ? 0x80000000 : 0);
-    RETURN();
-}
-
 void OPPROTO op_fmov_frN_FT0(void)
 {
     FT0 = env->fregs[PARAM1];
diff --git a/target-sh4/op_helper.c b/target-sh4/op_helper.c
index a4498ce..d3bde5c 100644
--- a/target-sh4/op_helper.c
+++ b/target-sh4/op_helper.c
@@ -163,27 +163,27 @@ uint32_t helper_addv(uint32_t arg0, uint32_t arg1)
 #define SETM env->sr |= SR_M
 #define CLRM env->sr &= ~SR_M
 
-void helper_div1_T0_T1(void)
+uint32_t helper_div1(uint32_t arg0, uint32_t arg1)
 {
     uint32_t tmp0, tmp2;
     uint8_t old_q, tmp1 = 0xff;
 
-    //printf("div1 T0=0x%08x T1=0x%08x M=%d Q=%d T=%d\n", T0, T1, M, Q, T);
+    //printf("div1 arg0=0x%08x arg1=0x%08x M=%d Q=%d T=%d\n", arg0, arg1, M, Q, T);
     old_q = Q;
-    if ((0x80000000 & T1) != 0)
+    if ((0x80000000 & arg1) != 0)
 	SETQ;
     else
 	CLRQ;
-    tmp2 = T0;
-    T1 <<= 1;
-    T1 |= T;
+    tmp2 = arg0;
+    arg1 <<= 1;
+    arg1 |= T;
     switch (old_q) {
     case 0:
 	switch (M) {
 	case 0:
-	    tmp0 = T1;
-	    T1 -= tmp2;
-	    tmp1 = T1 > tmp0;
+	    tmp0 = arg1;
+	    arg1 -= tmp2;
+	    tmp1 = arg1 > tmp0;
 	    switch (Q) {
 	    case 0:
 		if (tmp1)
@@ -200,9 +200,9 @@ void helper_div1_T0_T1(void)
 	    }
 	    break;
 	case 1:
-	    tmp0 = T1;
-	    T1 += tmp2;
-	    tmp1 = T1 < tmp0;
+	    tmp0 = arg1;
+	    arg1 += tmp2;
+	    tmp1 = arg1 < tmp0;
 	    switch (Q) {
 	    case 0:
 		if (tmp1 == 0)
@@ -223,9 +223,9 @@ void helper_div1_T0_T1(void)
     case 1:
 	switch (M) {
 	case 0:
-	    tmp0 = T1;
-	    T1 += tmp2;
-	    tmp1 = T1 < tmp0;
+	    tmp0 = arg1;
+	    arg1 += tmp2;
+	    tmp1 = arg1 < tmp0;
 	    switch (Q) {
 	    case 0:
 		if (tmp1)
@@ -242,9 +242,9 @@ void helper_div1_T0_T1(void)
 	    }
 	    break;
 	case 1:
-	    tmp0 = T1;
-	    T1 -= tmp2;
-	    tmp1 = T1 > tmp0;
+	    tmp0 = arg1;
+	    arg1 -= tmp2;
+	    tmp1 = arg1 > tmp0;
 	    switch (Q) {
 	    case 0:
 		if (tmp1 == 0)
@@ -267,7 +267,8 @@ void helper_div1_T0_T1(void)
 	SETT;
     else
 	CLRT;
-    //printf("Output: T1=0x%08x M=%d Q=%d T=%d\n", T1, M, Q, T);
+    //printf("Output: arg1=0x%08x M=%d Q=%d T=%d\n", arg1, M, Q, T);
+    return arg1;
 }
 
 void helper_macl(uint32_t arg0, uint32_t arg1)
@@ -365,30 +366,6 @@ uint32_t helper_subv(uint32_t arg0, uint32_t arg1)
     return arg1;
 }
 
-void helper_rotcl(uint32_t * addr)
-{
-    uint32_t new;
-
-    new = (*addr << 1) | (env->sr & SR_T);
-    if (*addr & 0x80000000)
-	env->sr |= SR_T;
-    else
-	env->sr &= ~SR_T;
-    *addr = new;
-}
-
-void helper_rotcr(uint32_t * addr)
-{
-    uint32_t new;
-
-    new = (*addr >> 1) | ((env->sr & SR_T) ? 0x80000000 : 0);
-    if (*addr & 1)
-	env->sr |= SR_T;
-    else
-	env->sr &= ~SR_T;
-    *addr = new;
-}
-
 void helper_ld_fpscr(uint32_t val)
 {
     env->fpscr = val & 0x003fffff;
diff --git a/target-sh4/translate.c b/target-sh4/translate.c
index 6e29e8f..a456e45 100644
--- a/target-sh4/translate.c
+++ b/target-sh4/translate.c
@@ -337,6 +337,24 @@ static inline void gen_store_flags(uint32_t flags)
     tcg_gen_ori_i32(cpu_flags, cpu_flags, flags);
 }
 
+static inline void gen_copy_bit_i32(TCGv t0, int p0, TCGv t1, int p1)
+{
+    TCGv tmp = tcg_temp_new(TCG_TYPE_I32);
+
+    p0 &= 0x1f;
+    p1 &= 0x1f;
+
+    tcg_gen_andi_i32(tmp, t1, (1 << p1));
+    tcg_gen_andi_i32(t0, t0, ~(1 << p0));
+    if (p0 < p1)
+        tcg_gen_shri_i32(tmp, tmp, p1 - p0);
+    else if (p0 > p1)
+        tcg_gen_shli_i32(tmp, tmp, p0 - p1);
+    tcg_gen_or_i32(t0, t0, tmp);
+
+    tcg_temp_free(tmp);
+}
+
 #define B3_0 (ctx->opcode & 0xf)
 #define B6_4 ((ctx->opcode >> 4) & 0x7)
 #define B7_4 ((ctx->opcode >> 4) & 0xf)
@@ -643,20 +661,33 @@ void _decode_opc(DisasContext * ctx)
 	gen_cmp(TCG_COND_GEU, cpu_T[0], cpu_T[1]);
 	return;
     case 0x200c:		/* cmp/str Rm,Rn */
-	tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
-	tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]);
-	gen_op_cmp_str_T0_T1();
+	{
+	    int label1 = gen_new_label();
+	    int label2 = gen_new_label();
+	    tcg_gen_xor_i32(cpu_T[0], cpu_gregs[REG(B7_4)], cpu_gregs[REG(B11_8)]);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0xff000000);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label1);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0x00ff0000);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label1);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0x0000ff00);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label1);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0x000000ff);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label1);
+	    tcg_gen_andi_i32(cpu_sr, cpu_sr, ~SR_T);
+	    tcg_gen_br(label2);
+	    gen_set_label(label1);
+	    tcg_gen_ori_i32(cpu_sr, cpu_sr, SR_T);
+	    gen_set_label(label2);
+	}
 	return;
     case 0x2007:		/* div0s Rm,Rn */
-	tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
-	tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]);
-	gen_op_div0s_T0_T1();
+	gen_copy_bit_i32(cpu_sr, 8, cpu_gregs[REG(B11_8)], 31);	/* SR_Q */
+	gen_copy_bit_i32(cpu_sr, 9, cpu_gregs[REG(B7_4)], 31);	/* SR_M */
+	tcg_gen_xor_i32(cpu_T[0], cpu_gregs[REG(B7_4)], cpu_gregs[REG(B11_8)]);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_T[0], 31);		/* SR_T */
 	return;
     case 0x3004:		/* div1 Rm,Rn */
-	tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
-	tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]);
-	gen_op_div1_T0_T1();
-	tcg_gen_mov_i32(cpu_gregs[REG(B11_8)], cpu_T[1]);
+	tcg_gen_helper_1_2(helper_div1, cpu_gregs[REG(B11_8)], cpu_gregs[REG(B7_4)], cpu_gregs[REG(B11_8)]);
 	return;
     case 0x300d:		/* dmuls.l Rm,Rn */
 	{
@@ -758,16 +789,59 @@ void _decode_opc(DisasContext * ctx)
 	tcg_gen_or_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_gregs[REG(B7_4)]);
 	return;
     case 0x400c:		/* shad Rm,Rn */
-	tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
-	tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]);
-	gen_op_shad_T0_T1();
-	tcg_gen_mov_i32(cpu_gregs[REG(B11_8)], cpu_T[1]);
+	{
+	    int label1 = gen_new_label();
+	    int label2 = gen_new_label();
+	    int label3 = gen_new_label();
+	    int label4 = gen_new_label();
+	    tcg_gen_brcondi_i32(TCG_COND_LT, cpu_gregs[REG(B7_4)], 0, label1);
+	    /* Rm positive, shift to the left */
+	    tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B7_4)], 0x1f);
+	    tcg_gen_shl_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_T[0]);
+	    tcg_gen_br(label4);
+	    /* Rm negative, shift to the right */
+	    gen_set_label(label1);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B7_4)], 0x1f);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label2);
+	    tcg_gen_not_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0x1f);
+	    tcg_gen_addi_i32(cpu_T[0], cpu_T[0], 1);
+	    tcg_gen_sar_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_T[0]);
+	    tcg_gen_br(label4);
+	    /* Rm = -32 */
+	    gen_set_label(label2);
+	    tcg_gen_brcondi_i32(TCG_COND_LT, cpu_gregs[REG(B11_8)], 0, label3);
+	    tcg_gen_movi_i32(cpu_gregs[REG(B11_8)], 0);
+	    tcg_gen_br(label4);
+	    gen_set_label(label3);
+	    tcg_gen_movi_i32(cpu_gregs[REG(B11_8)], 0xffffffff);
+	    gen_set_label(label4);
+	}
 	return;
     case 0x400d:		/* shld Rm,Rn */
-	tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
-	tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]);
-	gen_op_shld_T0_T1();
-	tcg_gen_mov_i32(cpu_gregs[REG(B11_8)], cpu_T[1]);
+	{
+	    int label1 = gen_new_label();
+	    int label2 = gen_new_label();
+	    int label3 = gen_new_label();
+	    tcg_gen_brcondi_i32(TCG_COND_LT, cpu_gregs[REG(B7_4)], 0, label1);
+	    /* Rm positive, shift to the left */
+	    tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B7_4)], 0x1f);
+	    tcg_gen_shl_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_T[0]);
+	    tcg_gen_br(label3);
+	    /* Rm negative, shift to the right */
+	    gen_set_label(label1);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B7_4)], 0x1f);
+	    tcg_gen_brcondi_i32(TCG_COND_EQ, cpu_T[0], 0, label2);
+	    tcg_gen_not_i32(cpu_T[0], cpu_gregs[REG(B7_4)]);
+	    tcg_gen_andi_i32(cpu_T[0], cpu_T[0], 0x1f);
+	    tcg_gen_addi_i32(cpu_T[0], cpu_T[0], 1);
+	    tcg_gen_shr_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_T[0]);
+	    tcg_gen_br(label3);
+	    /* Rm = -32 */
+	    gen_set_label(label2);
+	    tcg_gen_movi_i32(cpu_gregs[REG(B11_8)], 0);
+	    gen_set_label(label3);
+	}
 	return;
     case 0x3008:		/* sub Rm,Rn */
 	tcg_gen_sub_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], cpu_gregs[REG(B7_4)]);
@@ -1213,31 +1287,38 @@ void _decode_opc(DisasContext * ctx)
     case 0x0083:		/* pref @Rn */
 	return;
     case 0x4024:		/* rotcl Rn */
-	gen_op_rotcl_Rn(REG(B11_8));
+	tcg_gen_mov_i32(cpu_T[0], cpu_sr);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 31);
+	tcg_gen_shli_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
+	gen_copy_bit_i32(cpu_gregs[REG(B11_8)], 0, cpu_T[0], 0);
 	return;
     case 0x4025:		/* rotcr Rn */
-	gen_op_rotcr_Rn(REG(B11_8));
+	tcg_gen_mov_i32(cpu_T[0], cpu_sr);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 0);
+	tcg_gen_shri_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
+	gen_copy_bit_i32(cpu_gregs[REG(B11_8)], 31, cpu_T[0], 0);
 	return;
     case 0x4004:		/* rotl Rn */
-	gen_op_rotl_Rn(REG(B11_8));
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 31);
+	tcg_gen_shli_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
+	gen_copy_bit_i32(cpu_gregs[REG(B11_8)], 0, cpu_sr, 0);
 	return;
     case 0x4005:		/* rotr Rn */
-	gen_op_rotr_Rn(REG(B11_8));
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 0);
+	tcg_gen_shri_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
+	gen_copy_bit_i32(cpu_gregs[REG(B11_8)], 31, cpu_sr, 0);
 	return;
     case 0x4000:		/* shll Rn */
     case 0x4020:		/* shal Rn */
-	tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B11_8)], 0x80000000);
-	gen_cmp_imm(TCG_COND_NE, cpu_T[0], 0);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 31);
 	tcg_gen_shli_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
 	return;
     case 0x4021:		/* shar Rn */
-	tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B11_8)], 1);
-	gen_cmp_imm(TCG_COND_NE, cpu_T[0], 0);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 0);
 	tcg_gen_sari_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
 	return;
     case 0x4001:		/* shlr Rn */
-	tcg_gen_andi_i32(cpu_T[0], cpu_gregs[REG(B11_8)], 1);
-	gen_cmp_imm(TCG_COND_NE, cpu_T[0], 0);
+	gen_copy_bit_i32(cpu_sr, 0, cpu_gregs[REG(B11_8)], 0);
 	tcg_gen_shri_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 1);
 	return;
     case 0x4008:		/* shll2 Rn */