diff mbox series

[v5,6/9] target/arm: Change gen_jmp* to work on displacements

Message ID 20220930220312.135327-7-richard.henderson@linaro.org
State Superseded
Headers show
Series target/arm: pc-relative translation blocks | expand

Commit Message

Richard Henderson Sept. 30, 2022, 10:03 p.m. UTC 
In preparation for TARGET_TB_PCREL, reduce reliance on absolute values.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate.c | 37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

Comments

Peter Maydell Oct. 4, 2022, 3:58 p.m. UTC | #1 
On Fri, 30 Sept 2022 at 23:10, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> In preparation for TARGET_TB_PCREL, reduce reliance on absolute values.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  target/arm/translate.c | 37 +++++++++++++++++++++----------------
>  1 file changed, 21 insertions(+), 16 deletions(-)

> @@ -8368,7 +8372,8 @@ static bool trans_BLX_i(DisasContext *s, arg_BLX_i *a)
>      }
>      tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | s->thumb);
>      store_cpu_field_constant(!s->thumb, thumb);
> -    gen_jmp(s, (read_pc(s) & ~3) + a->imm);
> +    /* This difference computes a page offset so ok for TARGET_TB_PCREL. */
> +    gen_jmp(s, (read_pc(s) & ~3) - s->pc_curr + a->imm);

Could we just calculate the offset of the jump target instead?
read_pc() returns s->pc_curr + a constant, so the s->pc_curr cancels
out anyway:

  (read_pc(s) & ~3) - s->pc_curr + a->imm
==
    (pc_curr + (s->thumb ? 4 : 8) & ~3) - pc_curr + imm
==  pc_curr - pc_curr_low_bits - pc_curr + 4-or-8 + imm
==  imm + 4-or-8 - low_bits_of_pc

That's then more obviously not dependent on the absolute value
of the PC.

-- PMM
Richard Henderson Oct. 4, 2022, 8:57 p.m. UTC | #2 
On 10/4/22 08:58, Peter Maydell wrote:
> On Fri, 30 Sept 2022 at 23:10, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> In preparation for TARGET_TB_PCREL, reduce reliance on absolute values.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>   target/arm/translate.c | 37 +++++++++++++++++++++----------------
>>   1 file changed, 21 insertions(+), 16 deletions(-)
> 
>> @@ -8368,7 +8372,8 @@ static bool trans_BLX_i(DisasContext *s, arg_BLX_i *a)
>>       }
>>       tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | s->thumb);
>>       store_cpu_field_constant(!s->thumb, thumb);
>> -    gen_jmp(s, (read_pc(s) & ~3) + a->imm);
>> +    /* This difference computes a page offset so ok for TARGET_TB_PCREL. */
>> +    gen_jmp(s, (read_pc(s) & ~3) - s->pc_curr + a->imm);
> 
> Could we just calculate the offset of the jump target instead?
> read_pc() returns s->pc_curr + a constant, so the s->pc_curr cancels
> out anyway:
> 
>    (read_pc(s) & ~3) - s->pc_curr + a->imm
> ==
>      (pc_curr + (s->thumb ? 4 : 8) & ~3) - pc_curr + imm
> ==  pc_curr - pc_curr_low_bits - pc_curr + 4-or-8 + imm
> ==  imm + 4-or-8 - low_bits_of_pc
> 
> That's then more obviously not dependent on the absolute value
> of the PC.

Yes, this works:

-    gen_jmp(s, (read_pc(s) & ~3) + a->imm);

+    /* This jump is computed from an aligned PC: subtract off the low bits. */

+    gen_jmp(s, jmp_diff(s, a->imm - (s->pc_curr & 3)));



r~
Peter Maydell Oct. 5, 2022, 2:15 p.m. UTC | #3 
On Tue, 4 Oct 2022 at 21:57, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 10/4/22 08:58, Peter Maydell wrote:
> > On Fri, 30 Sept 2022 at 23:10, Richard Henderson
> > <richard.henderson@linaro.org> wrote:
> >>
> >> In preparation for TARGET_TB_PCREL, reduce reliance on absolute values.
> >>
> >> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> >> ---
> >>   target/arm/translate.c | 37 +++++++++++++++++++++----------------
> >>   1 file changed, 21 insertions(+), 16 deletions(-)
> >
> >> @@ -8368,7 +8372,8 @@ static bool trans_BLX_i(DisasContext *s, arg_BLX_i *a)
> >>       }
> >>       tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | s->thumb);
> >>       store_cpu_field_constant(!s->thumb, thumb);
> >> -    gen_jmp(s, (read_pc(s) & ~3) + a->imm);
> >> +    /* This difference computes a page offset so ok for TARGET_TB_PCREL. */
> >> +    gen_jmp(s, (read_pc(s) & ~3) - s->pc_curr + a->imm);
> >
> > Could we just calculate the offset of the jump target instead?
> > read_pc() returns s->pc_curr + a constant, so the s->pc_curr cancels
> > out anyway:
> >
> >    (read_pc(s) & ~3) - s->pc_curr + a->imm
> > ==
> >      (pc_curr + (s->thumb ? 4 : 8) & ~3) - pc_curr + imm
> > ==  pc_curr - pc_curr_low_bits - pc_curr + 4-or-8 + imm
> > ==  imm + 4-or-8 - low_bits_of_pc
> >
> > That's then more obviously not dependent on the absolute value
> > of the PC.
>
> Yes, this works:
>
> -    gen_jmp(s, (read_pc(s) & ~3) + a->imm);
>
> +    /* This jump is computed from an aligned PC: subtract off the low bits. */
>
> +    gen_jmp(s, jmp_diff(s, a->imm - (s->pc_curr & 3)));

Cool, that looks a lot clearer. With that change,
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM
diff mbox series

Patch

diff --git a/target/arm/translate.c b/target/arm/translate.c
index e0b1d415a2..fd35db8c8c 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -270,6 +270,12 @@  static uint32_t read_pc(DisasContext *s)
     return s->pc_curr + (s->thumb ? 4 : 8);
 }
 
+/* The pc_curr difference for an architectural jump. */
+static target_long jmp_diff(DisasContext *s, target_long diff)
+{
+    return diff + (s->thumb ? 4 : 8);
+}
+
 /* Set a variable to the value of a CPU register.  */
 void load_reg_var(DisasContext *s, TCGv_i32 var, int reg)
 {
@@ -2596,7 +2602,7 @@  static void gen_goto_ptr(void)
  * cpu_loop_exec. Any live exit_requests will be processed as we
  * enter the next TB.
  */
-static void gen_goto_tb(DisasContext *s, int n, int diff)
+static void gen_goto_tb(DisasContext *s, int n, target_long diff)
 {
     target_ulong dest = s->pc_curr + diff;
 
@@ -2612,10 +2618,8 @@  static void gen_goto_tb(DisasContext *s, int n, int diff)
 }
 
 /* Jump, specifying which TB number to use if we gen_goto_tb() */
-static inline void gen_jmp_tb(DisasContext *s, uint32_t dest, int tbno)
+static void gen_jmp_tb(DisasContext *s, target_long diff, int tbno)
 {
-    int diff = dest - s->pc_curr;
-
     if (unlikely(s->ss_active)) {
         /* An indirect jump so that we still trigger the debug exception.  */
         gen_update_pc(s, diff);
@@ -2657,9 +2661,9 @@  static inline void gen_jmp_tb(DisasContext *s, uint32_t dest, int tbno)
     }
 }
 
-static inline void gen_jmp(DisasContext *s, uint32_t dest)
+static inline void gen_jmp(DisasContext *s, target_long diff)
 {
-    gen_jmp_tb(s, dest, 0);
+    gen_jmp_tb(s, diff, 0);
 }
 
 static inline void gen_mulxy(TCGv_i32 t0, TCGv_i32 t1, int x, int y)
@@ -8326,7 +8330,7 @@  static bool trans_CLRM(DisasContext *s, arg_CLRM *a)
 
 static bool trans_B(DisasContext *s, arg_i *a)
 {
-    gen_jmp(s, read_pc(s) + a->imm);
+    gen_jmp(s, jmp_diff(s, a->imm));
     return true;
 }
 
@@ -8341,14 +8345,14 @@  static bool trans_B_cond_thumb(DisasContext *s, arg_ci *a)
         return true;
     }
     arm_skip_unless(s, a->cond);
-    gen_jmp(s, read_pc(s) + a->imm);
+    gen_jmp(s, jmp_diff(s, a->imm));
     return true;
 }
 
 static bool trans_BL(DisasContext *s, arg_i *a)
 {
     tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | s->thumb);
-    gen_jmp(s, read_pc(s) + a->imm);
+    gen_jmp(s, jmp_diff(s, a->imm));
     return true;
 }
 
@@ -8368,7 +8372,8 @@  static bool trans_BLX_i(DisasContext *s, arg_BLX_i *a)
     }
     tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | s->thumb);
     store_cpu_field_constant(!s->thumb, thumb);
-    gen_jmp(s, (read_pc(s) & ~3) + a->imm);
+    /* This difference computes a page offset so ok for TARGET_TB_PCREL. */
+    gen_jmp(s, (read_pc(s) & ~3) - s->pc_curr + a->imm);
     return true;
 }
 
@@ -8529,10 +8534,10 @@  static bool trans_WLS(DisasContext *s, arg_WLS *a)
          * when we take this upcoming exit from this TB, so gen_jmp_tb() is OK.
          */
     }
-    gen_jmp_tb(s, s->base.pc_next, 1);
+    gen_jmp_tb(s, curr_insn_len(s), 1);
 
     gen_set_label(nextlabel);
-    gen_jmp(s, read_pc(s) + a->imm);
+    gen_jmp(s, jmp_diff(s, a->imm));
     return true;
 }
 
@@ -8612,7 +8617,7 @@  static bool trans_LE(DisasContext *s, arg_LE *a)
 
     if (a->f) {
         /* Loop-forever: just jump back to the loop start */
-        gen_jmp(s, read_pc(s) - a->imm);
+        gen_jmp(s, jmp_diff(s, -a->imm));
         return true;
     }
 
@@ -8643,7 +8648,7 @@  static bool trans_LE(DisasContext *s, arg_LE *a)
         tcg_temp_free_i32(decr);
     }
     /* Jump back to the loop start */
-    gen_jmp(s, read_pc(s) - a->imm);
+    gen_jmp(s, jmp_diff(s, -a->imm));
 
     gen_set_label(loopend);
     if (a->tp) {
@@ -8651,7 +8656,7 @@  static bool trans_LE(DisasContext *s, arg_LE *a)
         store_cpu_field(tcg_constant_i32(4), v7m.ltpsize);
     }
     /* End TB, continuing to following insn */
-    gen_jmp_tb(s, s->base.pc_next, 1);
+    gen_jmp_tb(s, curr_insn_len(s), 1);
     return true;
 }
 
@@ -8750,7 +8755,7 @@  static bool trans_CBZ(DisasContext *s, arg_CBZ *a)
     tcg_gen_brcondi_i32(a->nz ? TCG_COND_EQ : TCG_COND_NE,
                         tmp, 0, s->condlabel);
     tcg_temp_free_i32(tmp);
-    gen_jmp(s, read_pc(s) + a->imm);
+    gen_jmp(s, jmp_diff(s, a->imm));
     return true;
 }