[PATCH v1] RISC-V: Refactor RVV frm_mode attr for rounding mode intrinsic

[Li, Pan2]

I am not quite sure if I understand it correctly, but I bet below enums are required by RISC-V mode switching, like FRM_MODE_DYN in entry, or FRM_MODE_CALL/EXIT in emit.

> ;; Defines rounding mode of an floating-point operation.
> -(define_attr "frm_mode" "rne,rtz,rdn,rup,rmm,dyn,dyn_exit,dyn_call,none"
> +(define_attr "frm_mode" ""
>  (cond [(eq_attr "type" "vfalu,vfwalu,vfmul,vfdiv,vfwmul,vfdiv,vfwmul")
> -        (const_string "dyn")]
> +        (const_string "FRM_DYN")]
>        (const_string "none")))

Pan

-----Original Message-----
From: Kito Cheng <kito.cheng@gmail.com> 
Sent: Monday, August 7, 2023 11:27 AM
To: Li, Pan2 <pan2.li@intel.com>
Cc: juzhe.zhong@rivai.ai; gcc-patches <gcc-patches@gcc.gnu.org>; Wang, Yanzhang <yanzhang.wang@intel.com>
Subject: Re: RE: [PATCH v1] RISC-V: Refactor RVV frm_mode attr for rounding mode intrinsic

What about using similar way as vlmul?


# NOTE: diff is based on your patch.
[kitoc@hsinchu02 riscv]$ git diff
diff --git a/gcc/config/riscv/riscv-protos.h b/gcc/config/riscv/riscv-protos.h
index 33f7cb1d670..3cb5c23cb09 100644
--- a/gcc/config/riscv/riscv-protos.h
+++ b/gcc/config/riscv/riscv-protos.h
@@ -345,6 +345,7 @@ enum floating_point_rounding_mode
  FRM_DYN = 7, /* Aka 0b111.  */
  FRM_STATIC_MIN = FRM_RNE,
  FRM_STATIC_MAX = FRM_RMM,
+  FRM_NONE = 8,
};

opt_machine_mode vectorize_related_mode (machine_mode, scalar_mode,
diff --git a/gcc/config/riscv/riscv-v.cc b/gcc/config/riscv/riscv-v.cc
index d5fb8611d6e..3d5dc0c11be 100644
--- a/gcc/config/riscv/riscv-v.cc
+++ b/gcc/config/riscv/riscv-v.cc
@@ -112,6 +112,7 @@ public:
  {
    m_has_fp_rounding_mode_p = true;
    m_fp_rounding_mode = mode;
+    gcc_assert (mode != FRM_NONE);
  }

  void add_output_operand (rtx x, machine_mode mode)
diff --git a/gcc/config/riscv/vector.md b/gcc/config/riscv/vector.md
index f966f1ba769..c1a7650fe85 100644
--- a/gcc/config/riscv/vector.md
+++ b/gcc/config/riscv/vector.md
@@ -865,9 +865,9 @@ (define_attr "vxrm_mode" "rnu,rne,rdn,rod,none"
        (const_string "none")))

;; Defines rounding mode of an floating-point operation.
-(define_attr "frm_mode" "rne,rtz,rdn,rup,rmm,dyn,dyn_exit,dyn_call,none"
+(define_attr "frm_mode" ""
  (cond [(eq_attr "type" "vfalu,vfwalu,vfmul,vfdiv,vfwmul,vfdiv,vfwmul")
-        (const_string "dyn")]
+        (const_string "FRM_DYN")]
       (const_string "none")))

;; -----------------------------------------------------------------