[03/12] target/arm: Stop using cpu_F0s for NEON_2RM_VABS_F

Message ID 20190613163917.28589-4-peter.maydell@linaro.org
State Superseded
Headers show
Series
  • target/arm: VFP decodetree conversion followups
Related show

Commit Message

Peter Maydell June 13, 2019, 4:39 p.m.
Where Neon instructions are floating point operations, we
mostly use the old VFP utility functions like gen_vfp_abs()
which work on the TCG globals cpu_F0s and cpu_F1s. The
Neon for-each-element loop conditionally loads the inputs
into either a plain old TCG temporary for most operations
or into cpu_F0s for float operations, and similarly stores
back either cpu_F0s or the temporary.

Switch NEON_2RM_VABS_F away from using cpu_F0s, and
update neon_2rm_is_float_op() accordingly.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

---
 target/arm/translate.c | 19 ++++++++-----------
 1 file changed, 8 insertions(+), 11 deletions(-)

-- 
2.20.1

Comments

Richard Henderson June 13, 2019, 9:25 p.m. | #1
On 6/13/19 9:39 AM, Peter Maydell wrote:
> Where Neon instructions are floating point operations, we

> mostly use the old VFP utility functions like gen_vfp_abs()

> which work on the TCG globals cpu_F0s and cpu_F1s. The

> Neon for-each-element loop conditionally loads the inputs

> into either a plain old TCG temporary for most operations

> or into cpu_F0s for float operations, and similarly stores

> back either cpu_F0s or the temporary.

> 

> Switch NEON_2RM_VABS_F away from using cpu_F0s, and

> update neon_2rm_is_float_op() accordingly.

> 

> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

> ---

>  target/arm/translate.c | 19 ++++++++-----------

>  1 file changed, 8 insertions(+), 11 deletions(-)


Reviewed-by: Richard Henderson <richard.henderson@linaro.org>



r~

Patch

diff --git a/target/arm/translate.c b/target/arm/translate.c
index c274c8b4602..a10fded1f39 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -1374,14 +1374,6 @@  static TCGv_ptr get_fpstatus_ptr(int neon)
     return statusptr;
 }
 
-static inline void gen_vfp_abs(int dp)
-{
-    if (dp)
-        gen_helper_vfp_absd(cpu_F0d, cpu_F0d);
-    else
-        gen_helper_vfp_abss(cpu_F0s, cpu_F0s);
-}
-
 static inline void gen_vfp_neg(int dp)
 {
     if (dp)
@@ -4191,8 +4183,13 @@  static const uint8_t neon_3r_sizes[] = {
 
 static int neon_2rm_is_float_op(int op)
 {
-    /* Return true if this neon 2reg-misc op is float-to-float */
-    return (op == NEON_2RM_VABS_F || op == NEON_2RM_VNEG_F ||
+    /*
+     * Return true if this neon 2reg-misc op is float-to-float.
+     * This is not a property of the operation but of our code --
+     * what we are asking here is "does the code for this case in
+     * the Neon for-each-pass loop use cpu_F0s?".
+     */
+    return (op == NEON_2RM_VNEG_F ||
             (op >= NEON_2RM_VRINTN && op <= NEON_2RM_VRINTZ) ||
             op == NEON_2RM_VRINTM ||
             (op >= NEON_2RM_VRINTP && op <= NEON_2RM_VCVTMS) ||
@@ -6761,7 +6758,7 @@  static int disas_neon_data_insn(DisasContext *s, uint32_t insn)
                             break;
                         }
                         case NEON_2RM_VABS_F:
-                            gen_vfp_abs(0);
+                            gen_helper_vfp_abss(tmp, tmp);
                             break;
                         case NEON_2RM_VNEG_F:
                             gen_vfp_neg(0);