diff mbox series

[v5,06/14] target/hexagon: introduce new helper functions

Message ID 20210619093713.1845446-7-ale.qemu@rev.ng (mailing list archive)
State New, archived
Headers show
Series target/hexagon: introduce idef-parser | expand

Commit Message

Alessandro Di Federico June 19, 2021, 9:37 a.m. UTC
From: Niccolò Izzo <nizzo@rev.ng>

These helpers will be employed by the idef-parser generated code.

Signed-off-by: Alessandro Di Federico <ale@rev.ng>
Signed-off-by: Niccolò Izzo <nizzo@rev.ng>
---
 target/hexagon/genptr.c | 163 ++++++++++++++++++++++++++++++++++++----
 target/hexagon/genptr.h |  23 ++++++
 target/hexagon/macros.h |   9 +++
 3 files changed, 180 insertions(+), 15 deletions(-)

Comments

Taylor Simpson June 23, 2021, 12:05 p.m. UTC | #1
> -----Original Message-----
> From: Alessandro Di Federico <ale.qemu@rev.ng>
> Sent: Saturday, June 19, 2021 3:37 AM
> To: qemu-devel@nongnu.org
> Cc: Taylor Simpson <tsimpson@quicinc.com>; Brian Cain
> <bcain@quicinc.com>; babush@rev.ng; nizzo@rev.ng; philmd@redhat.com;
> richard.henderson@linaro.org; Alessandro Di Federico <ale@rev.ng>
> Subject: [PATCH v5 06/14] target/hexagon: introduce new helper functions
> 
> From: Niccolò Izzo <nizzo@rev.ng>
> 
> These helpers will be employed by the idef-parser generated code.
> 
> Signed-off-by: Alessandro Di Federico <ale@rev.ng>
> Signed-off-by: Niccolò Izzo <nizzo@rev.ng>
> ---
>  target/hexagon/genptr.c | 163
> ++++++++++++++++++++++++++++++++++++----
>  target/hexagon/genptr.h |  23 ++++++
>  target/hexagon/macros.h |   9 +++
>  3 files changed, 180 insertions(+), 15 deletions(-)
> 
> diff --git a/target/hexagon/genptr.c b/target/hexagon/genptr.c index
> 6f2816f6e2..cf45c28f58 100644
> --- a/target/hexagon/genptr.c


> +++ b/target/hexagon/genptr.c
> +void gen_fbrev(TCGv result, TCGv src)
> +{
> +    TCGv lo = tcg_temp_new();
> +    TCGv tmp1 = tcg_temp_new();
> +    TCGv tmp2 = tcg_temp_new();
> +
> +    /* Bit reversal of low 16 bits */
> +    tcg_gen_extract_tl(lo, src, 0, 16);
> +    tcg_gen_andi_tl(tmp1, lo, 0xaaaa);
> +    tcg_gen_shri_tl(tmp1, tmp1, 1);
> +    tcg_gen_andi_tl(tmp2, lo, 0x5555);
> +    tcg_gen_shli_tl(tmp2, tmp2, 1);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_andi_tl(tmp1, lo, 0xcccc);
> +    tcg_gen_shri_tl(tmp1, tmp1, 2);
> +    tcg_gen_andi_tl(tmp2, lo, 0x3333);
> +    tcg_gen_shli_tl(tmp2, tmp2, 2);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_andi_tl(tmp1, lo, 0xf0f0);
> +    tcg_gen_shri_tl(tmp1, tmp1, 4);
> +    tcg_gen_andi_tl(tmp2, lo, 0x0f0f);
> +    tcg_gen_shli_tl(tmp2, tmp2, 4);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_bswap16_tl(lo, lo);
> +
> +    /* Final tweaks */
> +    tcg_gen_deposit_tl(result, src, lo, 0, 16);
> +    tcg_gen_or_tl(result, result, lo);
> +
> +    tcg_temp_free(lo);
> +    tcg_temp_free(tmp1);
> +    tcg_temp_free(tmp2);
> +}

Remove this function and call gen_helper_fbrev instead.  This was feedback from Richard Henderson on one of my previous patch series.

Thanks,
Taylor
Taylor Simpson June 23, 2021, 6:49 p.m. UTC | #2
> -----Original Message-----
> From: Alessandro Di Federico <ale.qemu@rev.ng>
> Sent: Saturday, June 19, 2021 3:37 AM
> To: qemu-devel@nongnu.org
> Cc: Taylor Simpson <tsimpson@quicinc.com>; Brian Cain
> <bcain@quicinc.com>; babush@rev.ng; nizzo@rev.ng; philmd@redhat.com;
> richard.henderson@linaro.org; Alessandro Di Federico <ale@rev.ng>
> Subject: [PATCH v5 06/14] target/hexagon: introduce new helper functions
> 
> From: Niccolò Izzo <nizzo@rev.ng>
> 
> These helpers will be employed by the idef-parser generated code.
> 
> Signed-off-by: Alessandro Di Federico <ale@rev.ng>
> Signed-off-by: Niccolò Izzo <nizzo@rev.ng>
> ---
>  target/hexagon/genptr.c | 163
> ++++++++++++++++++++++++++++++++++++----
>  target/hexagon/genptr.h |  23 ++++++
>  target/hexagon/macros.h |   9 +++
>  3 files changed, 180 insertions(+), 15 deletions(-)
> 
> diff --git a/target/hexagon/genptr.c b/target/hexagon/genptr.c index
> 6f2816f6e2..cf45c28f58 100644
> --- a/target/hexagon/genptr.c
> +++ b/target/hexagon/genptr.c
> @@ -28,6 +28,12 @@
>  #include "gen_tcg.h"
>  #include "genptr.h"
> 
> +TCGv gen_read_reg(TCGv result, int num) {
> +    tcg_gen_mov_tl(result, hex_gpr[num]);
> +    return result;
> +}
> +
>  TCGv gen_read_preg(TCGv pred, uint8_t num)  {
>      tcg_gen_mov_tl(pred, hex_pred[num]); @@ -396,18 +402,19 @@ static
> inline void gen_store_conditional8(CPUHexagonState *env,
>      tcg_gen_movi_tl(hex_llsc_addr, ~0);  }
> 
> -static inline void gen_store32(TCGv vaddr, TCGv src, int width, int slot)
> +void gen_store32(DisasContext *ctx, TCGv vaddr, TCGv src, tcg_target_long
> width,
> +                 unsigned slot)

uint32_t

>  {
>      tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
>      tcg_gen_movi_tl(hex_store_width[slot], width);
>      tcg_gen_mov_tl(hex_store_val32[slot], src);
> +    ctx->store_width[slot] = width;
>  }
> 
> -static inline void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src,
> -                              DisasContext *ctx, int slot)
> +void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot)
>  {
> -    gen_store32(vaddr, src, 1, slot);
> -    ctx->store_width[slot] = 1;
> +    gen_store32(ctx, vaddr, src, 1, slot);
>  }
> 
>  static inline void gen_store1i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
> @@ -418,11 +425,10 @@ static inline void gen_store1i(TCGv_env cpu_env,
> TCGv vaddr, int32_t src,
>      tcg_temp_free(tmp);
>  }
> 
> -static inline void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src,
> -                              DisasContext *ctx, int slot)
> +void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot)
>  {
> -    gen_store32(vaddr, src, 2, slot);
> -    ctx->store_width[slot] = 2;
> +    gen_store32(ctx, vaddr, src, 2, slot);
>  }
> 
>  static inline void gen_store2i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
> @@ -433,11 +439,10 @@ static inline void gen_store2i(TCGv_env cpu_env,
> TCGv vaddr, int32_t src,
>      tcg_temp_free(tmp);
>  }
> 
> -static inline void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src,
> -                              DisasContext *ctx, int slot)
> +void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot)
>  {
> -    gen_store32(vaddr, src, 4, slot);
> -    ctx->store_width[slot] = 4;
> +    gen_store32(ctx, vaddr, src, 4, slot);
>  }
> 
>  static inline void gen_store4i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
> @@ -448,8 +453,8 @@ static inline void gen_store4i(TCGv_env cpu_env,
> TCGv vaddr, int32_t src,
>      tcg_temp_free(tmp);
>  }
> 
> -static inline void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src,
> -                              DisasContext *ctx, int slot)
> +void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src,
> DisasContext *ctx,
> +                unsigned slot)
>  {
>      tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
>      tcg_gen_movi_tl(hex_store_width[slot], 8); @@ -476,5 +481,133 @@
> static TCGv gen_8bitsof(TCGv result, TCGv value)
>      return result;
>  }
> 
> +void gen_set_usr_field(int field, TCGv val) {
> +    tcg_gen_deposit_tl(hex_gpr[HEX_REG_USR], hex_gpr[HEX_REG_USR],
> val,
> +                       reg_field_info[field].offset,
> +                       reg_field_info[field].width); }
> +
> +void gen_set_usr_fieldi(int field, int x) {
> +    TCGv val = tcg_const_tl(x);
> +    gen_set_usr_field(field, val);
> +    tcg_temp_free(val);
> +}
> +
> +void gen_write_new_pc(TCGv addr)
> +{
> +    /* If there are multiple branches in a packet, ignore the second one */
> +    TCGv zero = tcg_const_tl(0);
> +    tcg_gen_movcond_tl(TCG_COND_NE, hex_next_PC, hex_branch_taken,
> zero,
> +                       hex_next_PC, addr);
> +    tcg_gen_movi_tl(hex_branch_taken, 1);
> +    tcg_temp_free(zero);
> +}
> +
> +void gen_sat_i32(TCGv dest, TCGv source, int width) {
> +    TCGv max_val = tcg_const_i32((1 << (width - 1)) - 1);
> +    TCGv min_val = tcg_const_i32(-(1 << (width - 1)));

Use _tl suffix, not _i32.

> +    tcg_gen_smin_tl(dest, source, max_val);
> +    tcg_gen_smax_tl(dest, dest, min_val);
> +    tcg_temp_free_i32(max_val);
> +    tcg_temp_free_i32(min_val);

Ditto

> +}
> +
> +void gen_sat_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width) {

gen_set_i32_ovfl would be a better name - ditto for the other _ext flavors later in this file

> +    gen_sat_i32(dest, source, width);
> +    tcg_gen_setcond_i32(TCG_COND_NE, ovfl, source, dest); }
> +
> +void gen_satu_i32(TCGv dest, TCGv source, int width) {
> +    TCGv max_val = tcg_const_i32((1 << width) - 1);
> +    tcg_gen_movcond_i32(TCG_COND_GTU, dest, source, max_val,
> max_val, source);
> +    TCGv_i32 zero = tcg_const_i32(0);
> +    tcg_gen_movcond_i32(TCG_COND_LT, dest, source, zero, zero, dest);
> +    tcg_temp_free_i32(max_val);
> +    tcg_temp_free_i32(zero);

_tl instead of _i32 for all of the above

For Hexagon, use _tl for all 32 bit operations.

> +}
> +
> +void gen_satu_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width) {
> +    gen_satu_i32(dest, source, width);
> +    tcg_gen_setcond_i32(TCG_COND_NE, ovfl, source, dest); }
> +
> +void gen_sat_i64(TCGv_i64 dest, TCGv_i64 source, int width) {
> +    TCGv_i64 max_val = tcg_const_i64((1 << (width - 1)) - 1);
> +    TCGv_i64 min_val = tcg_const_i64(-(1 << (width - 1)));
> +    tcg_gen_smin_i64(dest, source, max_val);
> +    tcg_gen_smax_i64(dest, dest, min_val);
> +    tcg_temp_free_i64(max_val);
> +    tcg_temp_free_i64(min_val);
> +}
> +
> +void gen_sat_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int
> +width) {
> +    gen_sat_i64(dest, source, width);
> +    TCGv_i64 ovfl_64 = tcg_temp_new_i64();
> +    tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, dest, source);
> +    tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
> +    tcg_temp_free_i64(ovfl_64);
> +}
> +
> +void gen_satu_i64(TCGv_i64 dest, TCGv_i64 source, int width) {
> +    TCGv_i64 max_val = tcg_const_i64((1 << width) - 1);
> +    tcg_gen_movcond_i64(TCG_COND_GTU, dest, source, max_val,
> max_val, source);
> +    TCGv_i64 zero = tcg_const_i64(0);
> +    tcg_gen_movcond_i64(TCG_COND_LT, dest, source, zero, zero, dest);
> +    tcg_temp_free_i64(max_val);
> +    tcg_temp_free_i64(zero);
> +}
> +
> +void gen_satu_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int
> +width) {
> +    gen_sat_i64(dest, source, width);
> +    TCGv_i64 ovfl_64 = tcg_temp_new_i64();
> +    tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, dest, source);
> +    tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
> +    tcg_temp_free_i64(ovfl_64);
> +}
> +
> +void gen_fbrev(TCGv result, TCGv src)
> +{
> +    TCGv lo = tcg_temp_new();
> +    TCGv tmp1 = tcg_temp_new();
> +    TCGv tmp2 = tcg_temp_new();
> +
> +    /* Bit reversal of low 16 bits */
> +    tcg_gen_extract_tl(lo, src, 0, 16);
> +    tcg_gen_andi_tl(tmp1, lo, 0xaaaa);
> +    tcg_gen_shri_tl(tmp1, tmp1, 1);
> +    tcg_gen_andi_tl(tmp2, lo, 0x5555);
> +    tcg_gen_shli_tl(tmp2, tmp2, 1);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_andi_tl(tmp1, lo, 0xcccc);
> +    tcg_gen_shri_tl(tmp1, tmp1, 2);
> +    tcg_gen_andi_tl(tmp2, lo, 0x3333);
> +    tcg_gen_shli_tl(tmp2, tmp2, 2);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_andi_tl(tmp1, lo, 0xf0f0);
> +    tcg_gen_shri_tl(tmp1, tmp1, 4);
> +    tcg_gen_andi_tl(tmp2, lo, 0x0f0f);
> +    tcg_gen_shli_tl(tmp2, tmp2, 4);
> +    tcg_gen_or_tl(lo, tmp1, tmp2);
> +    tcg_gen_bswap16_tl(lo, lo);
> +
> +    /* Final tweaks */
> +    tcg_gen_deposit_tl(result, src, lo, 0, 16);
> +    tcg_gen_or_tl(result, result, lo);
> +
> +    tcg_temp_free(lo);
> +    tcg_temp_free(tmp1);
> +    tcg_temp_free(tmp2);
> +}
> +

Use gen_helper_fbrev instead of this one.

>  #include "tcg_funcs_generated.c.inc"
>  #include "tcg_func_table_generated.c.inc"
> diff --git a/target/hexagon/genptr.h b/target/hexagon/genptr.h index
> 709b8eb099..48d1a32f49 100644
> --- a/target/hexagon/genptr.h
> +++ b/target/hexagon/genptr.h
> @@ -24,8 +24,31 @@
> 
>  extern const SemanticInsn opcode_genptr[];
> 
> +TCGv gen_read_reg(TCGv result, int num);
>  TCGv gen_read_preg(TCGv pred, uint8_t num);  void gen_log_reg_write(int
> rnum, TCGv val);  void gen_log_pred_write(DisasContext *ctx, int pnum,
> TCGv val);
> +void gen_store32(DisasContext *ctx, TCGv vaddr, TCGv src, tcg_target_long
> width,
> +                 unsigned slot);
> +void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot);
> +void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot);
> +void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext
> *ctx,
> +                unsigned slot);
> +void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src,
> DisasContext *ctx,
> +                unsigned slot);
> +void gen_write_new_pc(TCGv addr);
> +void gen_set_usr_field(int field, TCGv val); void
> +gen_set_usr_fieldi(int field, int x); void gen_sat_i32(TCGv dest, TCGv
> +source, int width); void gen_sat_i32_ext(TCGv ovfl, TCGv dest, TCGv
> +source, int width); void gen_satu_i32(TCGv dest, TCGv source, int
> +width); void gen_satu_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int
> +width); void gen_sat_i64(TCGv_i64 dest, TCGv_i64 source, int width);
> +void gen_sat_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int
> +width); void gen_satu_i64(TCGv_i64 dest, TCGv_i64 source, int width);
> +void gen_satu_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int
> +width); void gen_fbrev(TCGv result, TCGv src);
> 
>  #endif
> diff --git a/target/hexagon/macros.h b/target/hexagon/macros.h index
> eadb7e5d05..7d09501844 100644
> --- a/target/hexagon/macros.h
> +++ b/target/hexagon/macros.h
> @@ -181,7 +181,16 @@
>  #define MEM_STORE8(VA, DATA, SLOT) log_store64(env, VA, DATA, 8,
> SLOT)  #endif
> 
> +#ifdef QEMU_GENERATE
> +static inline void gen_cancel(unsigned slot) {

uint32_t

> +    tcg_gen_ori_tl(hex_slot_cancelled, hex_slot_cancelled, 1 << slot);
> +}
> +
> +#define CANCEL gen_cancel(slot);
> +#else
>  #define CANCEL cancel_slot(env, slot)
> +#endif
> 
>  #define LOAD_CANCEL(EA) do { CANCEL; } while (0)
> 
> --
> 2.31.1
diff mbox series

Patch

diff --git a/target/hexagon/genptr.c b/target/hexagon/genptr.c
index 6f2816f6e2..cf45c28f58 100644
--- a/target/hexagon/genptr.c
+++ b/target/hexagon/genptr.c
@@ -28,6 +28,12 @@ 
 #include "gen_tcg.h"
 #include "genptr.h"
 
+TCGv gen_read_reg(TCGv result, int num)
+{
+    tcg_gen_mov_tl(result, hex_gpr[num]);
+    return result;
+}
+
 TCGv gen_read_preg(TCGv pred, uint8_t num)
 {
     tcg_gen_mov_tl(pred, hex_pred[num]);
@@ -396,18 +402,19 @@  static inline void gen_store_conditional8(CPUHexagonState *env,
     tcg_gen_movi_tl(hex_llsc_addr, ~0);
 }
 
-static inline void gen_store32(TCGv vaddr, TCGv src, int width, int slot)
+void gen_store32(DisasContext *ctx, TCGv vaddr, TCGv src, tcg_target_long width,
+                 unsigned slot)
 {
     tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
     tcg_gen_movi_tl(hex_store_width[slot], width);
     tcg_gen_mov_tl(hex_store_val32[slot], src);
+    ctx->store_width[slot] = width;
 }
 
-static inline void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src,
-                              DisasContext *ctx, int slot)
+void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot)
 {
-    gen_store32(vaddr, src, 1, slot);
-    ctx->store_width[slot] = 1;
+    gen_store32(ctx, vaddr, src, 1, slot);
 }
 
 static inline void gen_store1i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
@@ -418,11 +425,10 @@  static inline void gen_store1i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
     tcg_temp_free(tmp);
 }
 
-static inline void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src,
-                              DisasContext *ctx, int slot)
+void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot)
 {
-    gen_store32(vaddr, src, 2, slot);
-    ctx->store_width[slot] = 2;
+    gen_store32(ctx, vaddr, src, 2, slot);
 }
 
 static inline void gen_store2i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
@@ -433,11 +439,10 @@  static inline void gen_store2i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
     tcg_temp_free(tmp);
 }
 
-static inline void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src,
-                              DisasContext *ctx, int slot)
+void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot)
 {
-    gen_store32(vaddr, src, 4, slot);
-    ctx->store_width[slot] = 4;
+    gen_store32(ctx, vaddr, src, 4, slot);
 }
 
 static inline void gen_store4i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
@@ -448,8 +453,8 @@  static inline void gen_store4i(TCGv_env cpu_env, TCGv vaddr, int32_t src,
     tcg_temp_free(tmp);
 }
 
-static inline void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src,
-                              DisasContext *ctx, int slot)
+void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src, DisasContext *ctx,
+                unsigned slot)
 {
     tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
     tcg_gen_movi_tl(hex_store_width[slot], 8);
@@ -476,5 +481,133 @@  static TCGv gen_8bitsof(TCGv result, TCGv value)
     return result;
 }
 
+void gen_set_usr_field(int field, TCGv val)
+{
+    tcg_gen_deposit_tl(hex_gpr[HEX_REG_USR], hex_gpr[HEX_REG_USR], val,
+                       reg_field_info[field].offset,
+                       reg_field_info[field].width);
+}
+
+void gen_set_usr_fieldi(int field, int x)
+{
+    TCGv val = tcg_const_tl(x);
+    gen_set_usr_field(field, val);
+    tcg_temp_free(val);
+}
+
+void gen_write_new_pc(TCGv addr)
+{
+    /* If there are multiple branches in a packet, ignore the second one */
+    TCGv zero = tcg_const_tl(0);
+    tcg_gen_movcond_tl(TCG_COND_NE, hex_next_PC, hex_branch_taken, zero,
+                       hex_next_PC, addr);
+    tcg_gen_movi_tl(hex_branch_taken, 1);
+    tcg_temp_free(zero);
+}
+
+void gen_sat_i32(TCGv dest, TCGv source, int width)
+{
+    TCGv max_val = tcg_const_i32((1 << (width - 1)) - 1);
+    TCGv min_val = tcg_const_i32(-(1 << (width - 1)));
+    tcg_gen_smin_tl(dest, source, max_val);
+    tcg_gen_smax_tl(dest, dest, min_val);
+    tcg_temp_free_i32(max_val);
+    tcg_temp_free_i32(min_val);
+}
+
+void gen_sat_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width)
+{
+    gen_sat_i32(dest, source, width);
+    tcg_gen_setcond_i32(TCG_COND_NE, ovfl, source, dest);
+}
+
+void gen_satu_i32(TCGv dest, TCGv source, int width)
+{
+    TCGv max_val = tcg_const_i32((1 << width) - 1);
+    tcg_gen_movcond_i32(TCG_COND_GTU, dest, source, max_val, max_val, source);
+    TCGv_i32 zero = tcg_const_i32(0);
+    tcg_gen_movcond_i32(TCG_COND_LT, dest, source, zero, zero, dest);
+    tcg_temp_free_i32(max_val);
+    tcg_temp_free_i32(zero);
+}
+
+void gen_satu_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width)
+{
+    gen_satu_i32(dest, source, width);
+    tcg_gen_setcond_i32(TCG_COND_NE, ovfl, source, dest);
+}
+
+void gen_sat_i64(TCGv_i64 dest, TCGv_i64 source, int width)
+{
+    TCGv_i64 max_val = tcg_const_i64((1 << (width - 1)) - 1);
+    TCGv_i64 min_val = tcg_const_i64(-(1 << (width - 1)));
+    tcg_gen_smin_i64(dest, source, max_val);
+    tcg_gen_smax_i64(dest, dest, min_val);
+    tcg_temp_free_i64(max_val);
+    tcg_temp_free_i64(min_val);
+}
+
+void gen_sat_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width)
+{
+    gen_sat_i64(dest, source, width);
+    TCGv_i64 ovfl_64 = tcg_temp_new_i64();
+    tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, dest, source);
+    tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
+    tcg_temp_free_i64(ovfl_64);
+}
+
+void gen_satu_i64(TCGv_i64 dest, TCGv_i64 source, int width)
+{
+    TCGv_i64 max_val = tcg_const_i64((1 << width) - 1);
+    tcg_gen_movcond_i64(TCG_COND_GTU, dest, source, max_val, max_val, source);
+    TCGv_i64 zero = tcg_const_i64(0);
+    tcg_gen_movcond_i64(TCG_COND_LT, dest, source, zero, zero, dest);
+    tcg_temp_free_i64(max_val);
+    tcg_temp_free_i64(zero);
+}
+
+void gen_satu_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width)
+{
+    gen_sat_i64(dest, source, width);
+    TCGv_i64 ovfl_64 = tcg_temp_new_i64();
+    tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, dest, source);
+    tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
+    tcg_temp_free_i64(ovfl_64);
+}
+
+void gen_fbrev(TCGv result, TCGv src)
+{
+    TCGv lo = tcg_temp_new();
+    TCGv tmp1 = tcg_temp_new();
+    TCGv tmp2 = tcg_temp_new();
+
+    /* Bit reversal of low 16 bits */
+    tcg_gen_extract_tl(lo, src, 0, 16);
+    tcg_gen_andi_tl(tmp1, lo, 0xaaaa);
+    tcg_gen_shri_tl(tmp1, tmp1, 1);
+    tcg_gen_andi_tl(tmp2, lo, 0x5555);
+    tcg_gen_shli_tl(tmp2, tmp2, 1);
+    tcg_gen_or_tl(lo, tmp1, tmp2);
+    tcg_gen_andi_tl(tmp1, lo, 0xcccc);
+    tcg_gen_shri_tl(tmp1, tmp1, 2);
+    tcg_gen_andi_tl(tmp2, lo, 0x3333);
+    tcg_gen_shli_tl(tmp2, tmp2, 2);
+    tcg_gen_or_tl(lo, tmp1, tmp2);
+    tcg_gen_andi_tl(tmp1, lo, 0xf0f0);
+    tcg_gen_shri_tl(tmp1, tmp1, 4);
+    tcg_gen_andi_tl(tmp2, lo, 0x0f0f);
+    tcg_gen_shli_tl(tmp2, tmp2, 4);
+    tcg_gen_or_tl(lo, tmp1, tmp2);
+    tcg_gen_bswap16_tl(lo, lo);
+
+    /* Final tweaks */
+    tcg_gen_deposit_tl(result, src, lo, 0, 16);
+    tcg_gen_or_tl(result, result, lo);
+
+    tcg_temp_free(lo);
+    tcg_temp_free(tmp1);
+    tcg_temp_free(tmp2);
+}
+
 #include "tcg_funcs_generated.c.inc"
 #include "tcg_func_table_generated.c.inc"
diff --git a/target/hexagon/genptr.h b/target/hexagon/genptr.h
index 709b8eb099..48d1a32f49 100644
--- a/target/hexagon/genptr.h
+++ b/target/hexagon/genptr.h
@@ -24,8 +24,31 @@ 
 
 extern const SemanticInsn opcode_genptr[];
 
+TCGv gen_read_reg(TCGv result, int num);
 TCGv gen_read_preg(TCGv pred, uint8_t num);
 void gen_log_reg_write(int rnum, TCGv val);
 void gen_log_pred_write(DisasContext *ctx, int pnum, TCGv val);
+void gen_store32(DisasContext *ctx, TCGv vaddr, TCGv src, tcg_target_long width,
+                 unsigned slot);
+void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot);
+void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot);
+void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src, DisasContext *ctx,
+                unsigned slot);
+void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src, DisasContext *ctx,
+                unsigned slot);
+void gen_write_new_pc(TCGv addr);
+void gen_set_usr_field(int field, TCGv val);
+void gen_set_usr_fieldi(int field, int x);
+void gen_sat_i32(TCGv dest, TCGv source, int width);
+void gen_sat_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width);
+void gen_satu_i32(TCGv dest, TCGv source, int width);
+void gen_satu_i32_ext(TCGv ovfl, TCGv dest, TCGv source, int width);
+void gen_sat_i64(TCGv_i64 dest, TCGv_i64 source, int width);
+void gen_sat_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width);
+void gen_satu_i64(TCGv_i64 dest, TCGv_i64 source, int width);
+void gen_satu_i64_ext(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width);
+void gen_fbrev(TCGv result, TCGv src);
 
 #endif
diff --git a/target/hexagon/macros.h b/target/hexagon/macros.h
index eadb7e5d05..7d09501844 100644
--- a/target/hexagon/macros.h
+++ b/target/hexagon/macros.h
@@ -181,7 +181,16 @@ 
 #define MEM_STORE8(VA, DATA, SLOT) log_store64(env, VA, DATA, 8, SLOT)
 #endif
 
+#ifdef QEMU_GENERATE
+static inline void gen_cancel(unsigned slot)
+{
+    tcg_gen_ori_tl(hex_slot_cancelled, hex_slot_cancelled, 1 << slot);
+}
+
+#define CANCEL gen_cancel(slot);
+#else
 #define CANCEL cancel_slot(env, slot)
+#endif
 
 #define LOAD_CANCEL(EA) do { CANCEL; } while (0)