diff mbox

[7/7] target-mips: implement DSP (d)append sub-class with TCG

Message ID 1353063863-11446-8-git-send-email-aurelien@aurel32.net
State New
Headers show

Commit Message

Aurelien Jarno Nov. 16, 2012, 11:04 a.m. UTC
DSP instruction from the (d)append sub-class can be implemented with
TCG. Use a different function for these instructions are they are quite
different from compare-pick sub-class.

Fix BALIGN instruction for negative value, where the value should be
zero-extended before being shift to the right.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
---
 target-mips/dsp_helper.c |   67 -----------------------
 target-mips/helper.h     |   13 -----
 target-mips/translate.c  |  133 ++++++++++++++++++++++++++++++----------------
 3 files changed, 87 insertions(+), 126 deletions(-)

Comments

Eric Johnson Dec. 5, 2012, 4:58 a.m. UTC | #1
> -----Original Message-----

> From: qemu-devel-bounces+ericj=mips.com@nongnu.org [mailto:qemu-devel-

> bounces+ericj=mips.com@nongnu.org] On Behalf Of Aurelien Jarno

> Sent: Friday, November 16, 2012 3:04 AM

> To: qemu-devel@nongnu.org

> Cc: Aurelien Jarno

> Subject: [Qemu-devel] [PATCH 7/7] target-mips: implement DSP (d)append

> sub-class with TCG

> 

> DSP instruction from the (d)append sub-class can be implemented with

> TCG. Use a different function for these instructions are they are quite

> different from compare-pick sub-class.

> 

> Fix BALIGN instruction for negative value, where the value should be

> zero-extended before being shift to the right.

> 

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

> ---

>  target-mips/dsp_helper.c |   67 -----------------------

>  target-mips/helper.h     |   13 -----

>  target-mips/translate.c  |  133 ++++++++++++++++++++++++++++++-----------

> -----

>  3 files changed, 87 insertions(+), 126 deletions(-)

> 

> diff --git a/target-mips/dsp_helper.c b/target-mips/dsp_helper.c

> index 474c249..22bbfa1 100644

> --- a/target-mips/dsp_helper.c

> +++ b/target-mips/dsp_helper.c

> @@ -3140,73 +3140,6 @@ PICK_INSN(pick_pw, 2, MIPSDSP_LLO, 32, 0);

>  #endif

>  #undef PICK_INSN

> 

> -#define APPEND_INSN(name, ret_32) \

> -target_ulong helper_##name(target_ulong rt, target_ulong rs, uint32_t sa)

> \

> -{

> \

> -    target_ulong temp;

> \

> -

> \

> -    if (ret_32) {

> \

> -        temp = ((rt & MIPSDSP_LLO) << sa) |

> \

> -               ((rs & MIPSDSP_LLO) & ((0x01 << sa) - 1));

> \

> -        temp = (target_long)(int32_t)(temp & MIPSDSP_LLO);

> \

> -    } else {

> \

> -        temp = (rt << sa) | (rs & ((0x01 << sa) - 1));

> \

> -    }

> \

> -

> \

> -    return temp;

> \

> -}

> -

> -APPEND_INSN(append, 1);

> -#ifdef TARGET_MIPS64

> -APPEND_INSN(dappend, 0);

> -#endif

> -#undef APPEND_INSN

> -

> -#define PREPEND_INSN(name, or_val, ret_32)                    \

> -target_ulong helper_##name(target_ulong rs, target_ulong rt,  \

> -                           uint32_t sa)                       \

> -{                                                             \

> -    sa |= or_val;                                             \

> -                                                              \

> -    if (1) {                                                  \

> -        return (target_long)(int32_t)(uint32_t)               \

> -            (((rs & MIPSDSP_LLO) << (32 - sa)) |              \

> -             ((rt & MIPSDSP_LLO) >> sa));                     \

> -    } else {                                                  \

> -        return (rs << (64 - sa)) | (rt >> sa);                \

> -    }                                                         \

> -}

> -

> -PREPEND_INSN(prepend, 0, 1);

> -#ifdef TARGET_MIPS64

> -PREPEND_INSN(prependw, 0, 0);

> -PREPEND_INSN(prependd, 0x20, 0);

> -#endif

> -#undef PREPEND_INSN

> -

> -#define BALIGN_INSN(name, filter, ret32) \

> -target_ulong helper_##name(target_ulong rs, target_ulong rt, uint32_t bp)

> \

> -{

> \

> -    bp = bp & 0x03;

> \

> -

> \

> -    if ((bp & 1) == 0) {

> \

> -        return rt;

> \

> -    } else {

> \

> -        if (ret32) {

> \

> -            return (target_long)(int32_t)((rt << (8 * bp)) |

> \

> -                                          (rs >> (8 * (4 - bp))));

> \

> -        } else {

> \

> -            return (rt << (8 * bp)) | (rs >> (8 * (8 - bp)));

> \

> -        }

> \

> -    }

> \

> -}

> -

> -BALIGN_INSN(balign, 0x03, 1);

> -#if defined(TARGET_MIPS64)

> -BALIGN_INSN(dbalign, 0x07, 0);

> -#endif

> -#undef BALIGN_INSN

> -

>  target_ulong helper_packrl_ph(target_ulong rs, target_ulong rt)

>  {

>      uint32_t rsl, rth;

> diff --git a/target-mips/helper.h b/target-mips/helper.h

> index acf9ebd..4373ac5 100644

> --- a/target-mips/helper.h

> +++ b/target-mips/helper.h

> @@ -654,19 +654,6 @@ DEF_HELPER_FLAGS_3(pick_ob, 0, tl, tl, tl, env)

>  DEF_HELPER_FLAGS_3(pick_qh, 0, tl, tl, tl, env)

>  DEF_HELPER_FLAGS_3(pick_pw, 0, tl, tl, tl, env)

>  #endif

> -DEF_HELPER_FLAGS_3(append, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#if defined(TARGET_MIPS64)

> -DEF_HELPER_FLAGS_3(dappend, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#endif

> -DEF_HELPER_FLAGS_3(prepend, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#if defined(TARGET_MIPS64)

> -DEF_HELPER_FLAGS_3(prependd, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -DEF_HELPER_FLAGS_3(prependw, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#endif

> -DEF_HELPER_FLAGS_3(balign, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#if defined(TARGET_MIPS64)

> -DEF_HELPER_FLAGS_3(dbalign, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)

> -#endif

>  DEF_HELPER_FLAGS_2(packrl_ph, TCG_CALL_NO_RWG_SE, tl, tl, tl)

>  #if defined(TARGET_MIPS64)

>  DEF_HELPER_FLAGS_2(packrl_pw, TCG_CALL_NO_RWG_SE, tl, tl, tl)

> diff --git a/target-mips/translate.c b/target-mips/translate.c

> index 910dd16..624d5a5 100644

> --- a/target-mips/translate.c

> +++ b/target-mips/translate.c

> @@ -336,7 +336,7 @@ enum {

>      /* DSP Bit/Manipulation Sub-class */

>      OPC_INSV_DSP       = 0x0C | OPC_SPECIAL3,

>      OPC_DINSV_DSP      = 0x0D | OPC_SPECIAL3,

> -    /* MIPS DSP Compare-Pick Sub-class */

> +    /* MIPS DSP Append Sub-class */

>      OPC_APPEND_DSP     = 0x31 | OPC_SPECIAL3,

>      OPC_DAPPEND_DSP    = 0x35 | OPC_SPECIAL3,

>      /* MIPS DSP Accumulator and DSPControl Access Sub-class */

> @@ -543,7 +543,7 @@ enum {

> 

>  #define MASK_APPEND(op) (MASK_SPECIAL3(op) | (op & (0x1F << 6)))

>  enum {

> -    /* MIPS DSP Compare-Pick Sub-class */

> +    /* MIPS DSP Append Sub-class */

>      OPC_APPEND  = (0x00 << 6) | OPC_APPEND_DSP,

>      OPC_PREPEND = (0x01 << 6) | OPC_APPEND_DSP,

>      OPC_BALIGN  = (0x10 << 6) | OPC_APPEND_DSP,

> @@ -667,7 +667,7 @@ enum {

> 

>  #define MASK_DAPPEND(op) (MASK_SPECIAL3(op) | (op & (0x1F << 6)))

>  enum {

> -    /* DSP Compare-Pick Sub-class */

> +    /* DSP Append Sub-class */

>      OPC_DAPPEND  = (0x00 << 6) | OPC_DAPPEND_DSP,

>      OPC_PREPENDD = (0x03 << 6) | OPC_DAPPEND_DSP,

>      OPC_PREPENDW = (0x01 << 6) | OPC_DAPPEND_DSP,

> @@ -13844,7 +13844,6 @@ static void gen_mipsdsp_add_cmp_pick(CPUMIPSState

> *env, DisasContext *ctx,

>                                       int ret, int v1, int v2, int

> check_ret)

>  {

>      const char *opn = "mipsdsp add compare pick";

> -    TCGv_i32 t0;

>      TCGv t1;

>      TCGv v1_t;

>      TCGv v2_t;

> @@ -13855,7 +13854,6 @@ static void gen_mipsdsp_add_cmp_pick(CPUMIPSState

> *env, DisasContext *ctx,

>          return;

>      }

> 

> -    t0 = tcg_temp_new_i32();

>      t1 = tcg_temp_new();

>      v1_t = tcg_temp_new();

>      v2_t = tcg_temp_new();

> @@ -13864,26 +13862,6 @@ static void gen_mipsdsp_add_cmp_pick(CPUMIPSState

> *env, DisasContext *ctx,

>      gen_load_gpr(v2_t, v2);

> 

>      switch (op1) {

> -    case OPC_APPEND_DSP:

> -        switch (op2) {

> -        case OPC_APPEND:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_append(cpu_gpr[ret], cpu_gpr[ret], v1_t, t0);

> -            break;

> -        case OPC_PREPEND:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_prepend(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> -            break;

> -        case OPC_BALIGN:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_balign(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> -            break;

> -        default:            /* Invid */

> -            MIPS_INVAL("MASK APPEND");

> -            generate_exception(ctx, EXCP_RI);

> -            break;

> -        }

> -        break;

>      case OPC_CMPU_EQ_QB_DSP:

>          switch (op2) {

>          case OPC_CMPU_EQ_QB:

> @@ -14041,23 +14019,95 @@ static void

> gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,

>              break;

>          }

>          break;

> +#endif

> +    }

> +

> +    tcg_temp_free(t1);

> +    tcg_temp_free(v1_t);

> +    tcg_temp_free(v2_t);

> +

> +    (void)opn; /* avoid a compiler warning */

> +    MIPS_DEBUG("%s", opn);

> +}

> +

> +static void gen_mipsdsp_append(CPUMIPSState *env, DisasContext *ctx,

> +                               uint32_t op1, int rt, int rs, int sa)

> +{

> +    const char *opn = "mipsdsp append/dappend";

> +    TCGv t0;

> +

> +    check_dspr2(env, ctx);

> +

> +    if (rt == 0) {

> +        /* Treat as NOP. */

> +        MIPS_DEBUG("NOP");

> +        return;

> +    }

> +

> +    t0 = tcg_temp_new();

> +    gen_load_gpr(t0, rs);

> +

> +    switch (op1) {

> +    case OPC_APPEND_DSP:

> +        switch (MASK_APPEND(ctx->opcode)) {

> +        case OPC_APPEND:

> +            if (sa != 0) {

> +                tcg_gen_deposit_tl(cpu_gpr[rt], t0, cpu_gpr[rt], sa, 32 -

> sa);

> +            }

> +            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);

> +            break;

> +        case OPC_PREPEND:

> +            if (sa != 0) {

> +                tcg_gen_ext32u_tl(cpu_gpr[rt], cpu_gpr[rt]);

> +                tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], sa);

> +                tcg_gen_shli_tl(t0, t0, 32 - sa);

> +                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);

> +            }

> +            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);

> +            break;

> +        case OPC_BALIGN:

> +            sa &= 3;

> +            if (sa != 0 && sa != 2) {

> +                tcg_gen_shli_tl(cpu_gpr[rt], cpu_gpr[rt], 8 * sa);

> +                tcg_gen_ext32u_tl(t0, t0);

> +                tcg_gen_shri_tl(t0, t0, 8 * (4 - sa));

> +                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);

> +            }

> +            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);

> +            break;

> +        default:            /* Invalid */

> +            MIPS_INVAL("MASK APPEND");

> +            generate_exception(ctx, EXCP_RI);

> +            break;

> +        }

> +        break;

> +#ifdef TARGET_MIPS64

>      case OPC_DAPPEND_DSP:

> -        switch (op2) {

> +        switch (MASK_DAPPEND(ctx->opcode)) {

>          case OPC_DAPPEND:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_dappend(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> +            if (sa != 0) {

> +                tcg_gen_deposit_tl(cpu_gpr[rt], t0, cpu_gpr[rt], sa, 64 -

> sa);

> +            }

>              break;

>          case OPC_PREPENDD:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_prependd(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> +            tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], 0x20 | sa);

> +            tcg_gen_shli_tl(t0, t0, 64 - (0x20 | sa));

> +            tcg_gen_or_tl(cpu_gpr[rt], t0, t0);

>              break;

>          case OPC_PREPENDW:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_prependw(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> +            if (sa != 0) {

> +                tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], sa);

> +                tcg_gen_shli_tl(t0, t0, 64 - sa);

> +                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);

> +            }

>              break;

>          case OPC_DBALIGN:

> -            tcg_gen_movi_i32(t0, v2);

> -            gen_helper_dbalign(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);

> +            sa &= 7;

> +            if (sa != 0 && sa != 2 && sa != 4) {

> +                tcg_gen_shli_tl(cpu_gpr[rt], cpu_gpr[rt], 8 * sa);

> +                tcg_gen_shri_tl(t0, t0, 8 * (8 - sa));

> +                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);

> +            }

>              break;

>          default:            /* Invalid */

>              MIPS_INVAL("MASK DAPPEND");

> @@ -14067,12 +14117,7 @@ static void gen_mipsdsp_add_cmp_pick(CPUMIPSState

> *env, DisasContext *ctx,

>          break;

>  #endif

>      }

> -

> -    tcg_temp_free_i32(t0);

> -    tcg_temp_free(t1);

> -    tcg_temp_free(v1_t);

> -    tcg_temp_free(v2_t);

> -

> +    tcg_temp_free(t0);

>      (void)opn; /* avoid a compiler warning */

>      MIPS_DEBUG("%s", opn);

>  }

> @@ -14892,9 +14937,7 @@ static void decode_opc (CPUMIPSState *env,

> DisasContext *ctx, int *is_branch)

>              }

>              break;

>          case OPC_APPEND_DSP:

> -            check_dspr2(env, ctx);

> -            op2 = MASK_APPEND(ctx->opcode);

> -            gen_mipsdsp_add_cmp_pick(env, ctx, op1, op2, rt, rs, rd, 1);

> +            gen_mipsdsp_append(env, ctx, op1, rt, rs, rd);

>              break;

>          case OPC_EXTR_W_DSP:

>              op2 = MASK_EXTR_W(ctx->opcode);

> @@ -15068,9 +15111,7 @@ static void decode_opc (CPUMIPSState *env,

> DisasContext *ctx, int *is_branch)

>              }

>              break;

>          case OPC_DAPPEND_DSP:

> -            check_dspr2(ctx);

> -            op2 = MASK_DAPPEND(ctx->opcode);

> -            gen_mipsdsp_add_cmp_pick(ctx, op1, op2, rt, rs, rd, 1);

> +            gen_mipsdsp_append(env, ctx, op1, rt, rs, rd);

>              break;

>          case OPC_DEXTR_W_DSP:

>              op2 = MASK_DEXTR_W(ctx->opcode);

> --

> 1.7.10.4

> 


Reviewed-by: Eric Johnson <ericj@mips.com>
diff mbox

Patch

diff --git a/target-mips/dsp_helper.c b/target-mips/dsp_helper.c
index 474c249..22bbfa1 100644
--- a/target-mips/dsp_helper.c
+++ b/target-mips/dsp_helper.c
@@ -3140,73 +3140,6 @@  PICK_INSN(pick_pw, 2, MIPSDSP_LLO, 32, 0);
 #endif
 #undef PICK_INSN
 
-#define APPEND_INSN(name, ret_32) \
-target_ulong helper_##name(target_ulong rt, target_ulong rs, uint32_t sa) \
-{                                                                         \
-    target_ulong temp;                                                    \
-                                                                          \
-    if (ret_32) {                                                         \
-        temp = ((rt & MIPSDSP_LLO) << sa) |                               \
-               ((rs & MIPSDSP_LLO) & ((0x01 << sa) - 1));                 \
-        temp = (target_long)(int32_t)(temp & MIPSDSP_LLO);                \
-    } else {                                                              \
-        temp = (rt << sa) | (rs & ((0x01 << sa) - 1));                    \
-    }                                                                     \
-                                                                          \
-    return temp;                                                          \
-}
-
-APPEND_INSN(append, 1);
-#ifdef TARGET_MIPS64
-APPEND_INSN(dappend, 0);
-#endif
-#undef APPEND_INSN
-
-#define PREPEND_INSN(name, or_val, ret_32)                    \
-target_ulong helper_##name(target_ulong rs, target_ulong rt,  \
-                           uint32_t sa)                       \
-{                                                             \
-    sa |= or_val;                                             \
-                                                              \
-    if (1) {                                                  \
-        return (target_long)(int32_t)(uint32_t)               \
-            (((rs & MIPSDSP_LLO) << (32 - sa)) |              \
-             ((rt & MIPSDSP_LLO) >> sa));                     \
-    } else {                                                  \
-        return (rs << (64 - sa)) | (rt >> sa);                \
-    }                                                         \
-}
-
-PREPEND_INSN(prepend, 0, 1);
-#ifdef TARGET_MIPS64
-PREPEND_INSN(prependw, 0, 0);
-PREPEND_INSN(prependd, 0x20, 0);
-#endif
-#undef PREPEND_INSN
-
-#define BALIGN_INSN(name, filter, ret32) \
-target_ulong helper_##name(target_ulong rs, target_ulong rt, uint32_t bp) \
-{                                                                         \
-    bp = bp & 0x03;                                                       \
-                                                                          \
-    if ((bp & 1) == 0) {                                                  \
-        return rt;                                                        \
-    } else {                                                              \
-        if (ret32) {                                                      \
-            return (target_long)(int32_t)((rt << (8 * bp)) |              \
-                                          (rs >> (8 * (4 - bp))));        \
-        } else {                                                          \
-            return (rt << (8 * bp)) | (rs >> (8 * (8 - bp)));             \
-        }                                                                 \
-    }                                                                     \
-}
-
-BALIGN_INSN(balign, 0x03, 1);
-#if defined(TARGET_MIPS64)
-BALIGN_INSN(dbalign, 0x07, 0);
-#endif
-#undef BALIGN_INSN
-
 target_ulong helper_packrl_ph(target_ulong rs, target_ulong rt)
 {
     uint32_t rsl, rth;
diff --git a/target-mips/helper.h b/target-mips/helper.h
index acf9ebd..4373ac5 100644
--- a/target-mips/helper.h
+++ b/target-mips/helper.h
@@ -654,19 +654,6 @@  DEF_HELPER_FLAGS_3(pick_ob, 0, tl, tl, tl, env)
 DEF_HELPER_FLAGS_3(pick_qh, 0, tl, tl, tl, env)
 DEF_HELPER_FLAGS_3(pick_pw, 0, tl, tl, tl, env)
 #endif
-DEF_HELPER_FLAGS_3(append, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#if defined(TARGET_MIPS64)
-DEF_HELPER_FLAGS_3(dappend, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#endif
-DEF_HELPER_FLAGS_3(prepend, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#if defined(TARGET_MIPS64)
-DEF_HELPER_FLAGS_3(prependd, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-DEF_HELPER_FLAGS_3(prependw, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#endif
-DEF_HELPER_FLAGS_3(balign, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#if defined(TARGET_MIPS64)
-DEF_HELPER_FLAGS_3(dbalign, TCG_CALL_NO_RWG_SE, tl, tl, tl, i32)
-#endif
 DEF_HELPER_FLAGS_2(packrl_ph, TCG_CALL_NO_RWG_SE, tl, tl, tl)
 #if defined(TARGET_MIPS64)
 DEF_HELPER_FLAGS_2(packrl_pw, TCG_CALL_NO_RWG_SE, tl, tl, tl)
diff --git a/target-mips/translate.c b/target-mips/translate.c
index 910dd16..624d5a5 100644
--- a/target-mips/translate.c
+++ b/target-mips/translate.c
@@ -336,7 +336,7 @@  enum {
     /* DSP Bit/Manipulation Sub-class */
     OPC_INSV_DSP       = 0x0C | OPC_SPECIAL3,
     OPC_DINSV_DSP      = 0x0D | OPC_SPECIAL3,
-    /* MIPS DSP Compare-Pick Sub-class */
+    /* MIPS DSP Append Sub-class */
     OPC_APPEND_DSP     = 0x31 | OPC_SPECIAL3,
     OPC_DAPPEND_DSP    = 0x35 | OPC_SPECIAL3,
     /* MIPS DSP Accumulator and DSPControl Access Sub-class */
@@ -543,7 +543,7 @@  enum {
 
 #define MASK_APPEND(op) (MASK_SPECIAL3(op) | (op & (0x1F << 6)))
 enum {
-    /* MIPS DSP Compare-Pick Sub-class */
+    /* MIPS DSP Append Sub-class */
     OPC_APPEND  = (0x00 << 6) | OPC_APPEND_DSP,
     OPC_PREPEND = (0x01 << 6) | OPC_APPEND_DSP,
     OPC_BALIGN  = (0x10 << 6) | OPC_APPEND_DSP,
@@ -667,7 +667,7 @@  enum {
 
 #define MASK_DAPPEND(op) (MASK_SPECIAL3(op) | (op & (0x1F << 6)))
 enum {
-    /* DSP Compare-Pick Sub-class */
+    /* DSP Append Sub-class */
     OPC_DAPPEND  = (0x00 << 6) | OPC_DAPPEND_DSP,
     OPC_PREPENDD = (0x03 << 6) | OPC_DAPPEND_DSP,
     OPC_PREPENDW = (0x01 << 6) | OPC_DAPPEND_DSP,
@@ -13844,7 +13844,6 @@  static void gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,
                                      int ret, int v1, int v2, int check_ret)
 {
     const char *opn = "mipsdsp add compare pick";
-    TCGv_i32 t0;
     TCGv t1;
     TCGv v1_t;
     TCGv v2_t;
@@ -13855,7 +13854,6 @@  static void gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,
         return;
     }
 
-    t0 = tcg_temp_new_i32();
     t1 = tcg_temp_new();
     v1_t = tcg_temp_new();
     v2_t = tcg_temp_new();
@@ -13864,26 +13862,6 @@  static void gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,
     gen_load_gpr(v2_t, v2);
 
     switch (op1) {
-    case OPC_APPEND_DSP:
-        switch (op2) {
-        case OPC_APPEND:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_append(cpu_gpr[ret], cpu_gpr[ret], v1_t, t0);
-            break;
-        case OPC_PREPEND:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_prepend(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
-            break;
-        case OPC_BALIGN:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_balign(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
-            break;
-        default:            /* Invid */
-            MIPS_INVAL("MASK APPEND");
-            generate_exception(ctx, EXCP_RI);
-            break;
-        }
-        break;
     case OPC_CMPU_EQ_QB_DSP:
         switch (op2) {
         case OPC_CMPU_EQ_QB:
@@ -14041,23 +14019,95 @@  static void gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,
             break;
         }
         break;
+#endif
+    }
+
+    tcg_temp_free(t1);
+    tcg_temp_free(v1_t);
+    tcg_temp_free(v2_t);
+
+    (void)opn; /* avoid a compiler warning */
+    MIPS_DEBUG("%s", opn);
+}
+
+static void gen_mipsdsp_append(CPUMIPSState *env, DisasContext *ctx,
+                               uint32_t op1, int rt, int rs, int sa)
+{
+    const char *opn = "mipsdsp append/dappend";
+    TCGv t0;
+
+    check_dspr2(env, ctx);
+
+    if (rt == 0) {
+        /* Treat as NOP. */
+        MIPS_DEBUG("NOP");
+        return;
+    }
+
+    t0 = tcg_temp_new();
+    gen_load_gpr(t0, rs);
+
+    switch (op1) {
+    case OPC_APPEND_DSP:
+        switch (MASK_APPEND(ctx->opcode)) {
+        case OPC_APPEND:
+            if (sa != 0) {
+                tcg_gen_deposit_tl(cpu_gpr[rt], t0, cpu_gpr[rt], sa, 32 - sa);
+            }
+            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);
+            break;
+        case OPC_PREPEND:
+            if (sa != 0) {
+                tcg_gen_ext32u_tl(cpu_gpr[rt], cpu_gpr[rt]);
+                tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], sa);
+                tcg_gen_shli_tl(t0, t0, 32 - sa);
+                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);
+            }
+            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);
+            break;
+        case OPC_BALIGN:
+            sa &= 3;
+            if (sa != 0 && sa != 2) {
+                tcg_gen_shli_tl(cpu_gpr[rt], cpu_gpr[rt], 8 * sa);
+                tcg_gen_ext32u_tl(t0, t0);
+                tcg_gen_shri_tl(t0, t0, 8 * (4 - sa));
+                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);
+            }
+            tcg_gen_ext32s_tl(cpu_gpr[rt], cpu_gpr[rt]);
+            break;
+        default:            /* Invalid */
+            MIPS_INVAL("MASK APPEND");
+            generate_exception(ctx, EXCP_RI);
+            break;
+        }
+        break;
+#ifdef TARGET_MIPS64
     case OPC_DAPPEND_DSP:
-        switch (op2) {
+        switch (MASK_DAPPEND(ctx->opcode)) {
         case OPC_DAPPEND:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_dappend(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
+            if (sa != 0) {
+                tcg_gen_deposit_tl(cpu_gpr[rt], t0, cpu_gpr[rt], sa, 64 - sa);
+            }
             break;
         case OPC_PREPENDD:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_prependd(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
+            tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], 0x20 | sa);
+            tcg_gen_shli_tl(t0, t0, 64 - (0x20 | sa));
+            tcg_gen_or_tl(cpu_gpr[rt], t0, t0);
             break;
         case OPC_PREPENDW:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_prependw(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
+            if (sa != 0) {
+                tcg_gen_shri_tl(cpu_gpr[rt], cpu_gpr[rt], sa);
+                tcg_gen_shli_tl(t0, t0, 64 - sa);
+                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);
+            }
             break;
         case OPC_DBALIGN:
-            tcg_gen_movi_i32(t0, v2);
-            gen_helper_dbalign(cpu_gpr[ret], v1_t, cpu_gpr[ret], t0);
+            sa &= 7;
+            if (sa != 0 && sa != 2 && sa != 4) {
+                tcg_gen_shli_tl(cpu_gpr[rt], cpu_gpr[rt], 8 * sa);
+                tcg_gen_shri_tl(t0, t0, 8 * (8 - sa));
+                tcg_gen_or_tl(cpu_gpr[rt], cpu_gpr[rt], t0);
+            }
             break;
         default:            /* Invalid */
             MIPS_INVAL("MASK DAPPEND");
@@ -14067,12 +14117,7 @@  static void gen_mipsdsp_add_cmp_pick(CPUMIPSState *env, DisasContext *ctx,
         break;
 #endif
     }
-
-    tcg_temp_free_i32(t0);
-    tcg_temp_free(t1);
-    tcg_temp_free(v1_t);
-    tcg_temp_free(v2_t);
-
+    tcg_temp_free(t0);
     (void)opn; /* avoid a compiler warning */
     MIPS_DEBUG("%s", opn);
 }
@@ -14892,9 +14937,7 @@  static void decode_opc (CPUMIPSState *env, DisasContext *ctx, int *is_branch)
             }
             break;
         case OPC_APPEND_DSP:
-            check_dspr2(env, ctx);
-            op2 = MASK_APPEND(ctx->opcode);
-            gen_mipsdsp_add_cmp_pick(env, ctx, op1, op2, rt, rs, rd, 1);
+            gen_mipsdsp_append(env, ctx, op1, rt, rs, rd);
             break;
         case OPC_EXTR_W_DSP:
             op2 = MASK_EXTR_W(ctx->opcode);
@@ -15068,9 +15111,7 @@  static void decode_opc (CPUMIPSState *env, DisasContext *ctx, int *is_branch)
             }
             break;
         case OPC_DAPPEND_DSP:
-            check_dspr2(ctx);
-            op2 = MASK_DAPPEND(ctx->opcode);
-            gen_mipsdsp_add_cmp_pick(ctx, op1, op2, rt, rs, rd, 1);
+            gen_mipsdsp_append(env, ctx, op1, rt, rs, rd);
             break;
         case OPC_DEXTR_W_DSP:
             op2 = MASK_DEXTR_W(ctx->opcode);