[Qemu-devel] [PATCH v2 17/42] target/arm: Remove VLDR/VSTR/VLDM/VSTM use of cpu_F0s and cpu_F0d

[Peter Maydell <peter.maydell@linaro.org>]

Expand out the sequences in the new decoder VLDR/VSTR/VLDM/VSTM trans
functions which perform the memory accesses by going via the TCG
globals cpu_F0s and cpu_F0d, to use local TCG temps instead.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate-vfp.inc.c | 46 +++++++++++++++++++++-------------
 target/arm/translate.c         | 18 -------------
 2 files changed, 28 insertions(+), 36 deletions(-)

diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c
index 32a1805e582..9729946d734 100644
--- a/target/arm/translate-vfp.inc.c
+++ b/target/arm/translate-vfp.inc.c
@@ -857,7 +857,7 @@ static bool trans_VMOV_64_dp(DisasContext *s, arg_VMOV_64_sp *a)
 static bool trans_VLDR_VSTR_sp(DisasContext *s, arg_VLDR_VSTR_sp *a)
 {
     uint32_t offset;
-    TCGv_i32 addr;
+    TCGv_i32 addr, tmp;
 
     if (!vfp_access_check(s)) {
         return true;
@@ -876,13 +876,15 @@ static bool trans_VLDR_VSTR_sp(DisasContext *s, arg_VLDR_VSTR_sp *a)
         addr = load_reg(s, a->rn);
     }
     tcg_gen_addi_i32(addr, addr, offset);
+    tmp = tcg_temp_new_i32();
     if (a->l) {
-        gen_vfp_ld(s, false, addr);
-        gen_mov_vreg_F0(false, a->vd);
+        gen_aa32_ld32u(s, tmp, addr, get_mem_index(s));
+        neon_store_reg32(tmp, a->vd);
     } else {
-        gen_mov_F0_vreg(false, a->vd);
-        gen_vfp_st(s, false, addr);
+        neon_load_reg32(tmp, a->vd);
+        gen_aa32_st32(s, tmp, addr, get_mem_index(s));
     }
+    tcg_temp_free_i32(tmp);
     tcg_temp_free_i32(addr);
 
     return true;
@@ -892,6 +894,7 @@ static bool trans_VLDR_VSTR_dp(DisasContext *s, arg_VLDR_VSTR_sp *a)
 {
     uint32_t offset;
     TCGv_i32 addr;
+    TCGv_i64 tmp;
 
     /* UNDEF accesses to D16-D31 if they don't exist */
     if (!dc_isar_feature(aa32_fp_d32, s) && (a->vd & 0x10)) {
@@ -915,13 +918,15 @@ static bool trans_VLDR_VSTR_dp(DisasContext *s, arg_VLDR_VSTR_sp *a)
         addr = load_reg(s, a->rn);
     }
     tcg_gen_addi_i32(addr, addr, offset);
+    tmp = tcg_temp_new_i64();
     if (a->l) {
-        gen_vfp_ld(s, true, addr);
-        gen_mov_vreg_F0(true, a->vd);
+        gen_aa32_ld64(s, tmp, addr, get_mem_index(s));
+        neon_store_reg64(tmp, a->vd);
     } else {
-        gen_mov_F0_vreg(true, a->vd);
-        gen_vfp_st(s, true, addr);
+        neon_load_reg64(tmp, a->vd);
+        gen_aa32_st64(s, tmp, addr, get_mem_index(s));
     }
+    tcg_temp_free_i64(tmp);
     tcg_temp_free_i32(addr);
 
     return true;
@@ -930,7 +935,7 @@ static bool trans_VLDR_VSTR_dp(DisasContext *s, arg_VLDR_VSTR_sp *a)
 static bool trans_VLDM_VSTM_sp(DisasContext *s, arg_VLDM_VSTM_sp *a)
 {
     uint32_t offset;
-    TCGv_i32 addr;
+    TCGv_i32 addr, tmp;
     int i, n;
 
     n = a->imm;
@@ -976,18 +981,20 @@ static bool trans_VLDM_VSTM_sp(DisasContext *s, arg_VLDM_VSTM_sp *a)
     }
 
     offset = 4;
+    tmp = tcg_temp_new_i32();
     for (i = 0; i < n; i++) {
         if (a->l) {
             /* load */
-            gen_vfp_ld(s, false, addr);
-            gen_mov_vreg_F0(false, a->vd + i);
+            gen_aa32_ld32u(s, tmp, addr, get_mem_index(s));
+            neon_store_reg32(tmp, a->vd + i);
         } else {
             /* store */
-            gen_mov_F0_vreg(false, a->vd + i);
-            gen_vfp_st(s, false, addr);
+            neon_load_reg32(tmp, a->vd + i);
+            gen_aa32_st32(s, tmp, addr, get_mem_index(s));
         }
         tcg_gen_addi_i32(addr, addr, offset);
     }
+    tcg_temp_free_i32(tmp);
     if (a->w) {
         /* writeback */
         if (a->p) {
@@ -1006,6 +1013,7 @@ static bool trans_VLDM_VSTM_dp(DisasContext *s, arg_VLDM_VSTM_dp *a)
 {
     uint32_t offset;
     TCGv_i32 addr;
+    TCGv_i64 tmp;
     int i, n;
 
     n = a->imm >> 1;
@@ -1056,18 +1064,20 @@ static bool trans_VLDM_VSTM_dp(DisasContext *s, arg_VLDM_VSTM_dp *a)
     }
 
     offset = 8;
+    tmp = tcg_temp_new_i64();
     for (i = 0; i < n; i++) {
         if (a->l) {
             /* load */
-            gen_vfp_ld(s, true, addr);
-            gen_mov_vreg_F0(true, a->vd + i);
+            gen_aa32_ld64(s, tmp, addr, get_mem_index(s));
+            neon_store_reg64(tmp, a->vd + i);
         } else {
             /* store */
-            gen_mov_F0_vreg(true, a->vd + i);
-            gen_vfp_st(s, true, addr);
+            neon_load_reg64(tmp, a->vd + i);
+            gen_aa32_st64(s, tmp, addr, get_mem_index(s));
         }
         tcg_gen_addi_i32(addr, addr, offset);
     }
+    tcg_temp_free_i64(tmp);
     if (a->w) {
         /* writeback */
         if (a->p) {
diff --git a/target/arm/translate.c b/target/arm/translate.c
index 5a9d0c30d3d..c197084e925 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -1522,24 +1522,6 @@ VFP_GEN_FIX(uhto, )
 VFP_GEN_FIX(ulto, )
 #undef VFP_GEN_FIX
 
-static inline void gen_vfp_ld(DisasContext *s, int dp, TCGv_i32 addr)
-{
-    if (dp) {
-        gen_aa32_ld64(s, cpu_F0d, addr, get_mem_index(s));
-    } else {
-        gen_aa32_ld32u(s, cpu_F0s, addr, get_mem_index(s));
-    }
-}
-
-static inline void gen_vfp_st(DisasContext *s, int dp, TCGv_i32 addr)
-{
-    if (dp) {
-        gen_aa32_st64(s, cpu_F0d, addr, get_mem_index(s));
-    } else {
-        gen_aa32_st32(s, cpu_F0s, addr, get_mem_index(s));
-    }
-}
-
 static inline long vfp_reg_offset(bool dp, unsigned reg)
 {
     if (dp) {
-- 
2.20.1