[PATCH] arm: Add missing vec_cmp and vcond patterns

Kyrylo Tkachov Kyrylo.Tkachov@arm.com
Thu Oct 1 14:47:15 GMT 2020


Hi Richard,

> -----Original Message-----
> From: Richard Sandiford <richard.sandiford@arm.com>
> Sent: 01 October 2020 15:10
> To: gcc-patches@gcc.gnu.org
> Cc: nickc@redhat.com; Richard Earnshaw <Richard.Earnshaw@arm.com>;
> Ramana Radhakrishnan <Ramana.Radhakrishnan@arm.com>; Kyrylo
> Tkachov <Kyrylo.Tkachov@arm.com>
> Subject: [PATCH] arm: Add missing vec_cmp and vcond patterns
> 
> This patch does several things at once:
> 
> (1) Add vector compare patterns (vec_cmp and vec_cmpu).
> 
> (2) Add vector selects between floating-point modes when the
>     values being compared are integers (affects vcond and vcondu).
> 
> (3) Add vector selects between integer modes when the values being
>     compared are floating-point (affects vcond).
> 
> (4) Add standalone vector select patterns (vcond_mask).
> 
> (5) Tweak the handling of compound comparisons with zeros.
> 
> Unfortunately it proved too difficult (for me) to separate this
> out into a series of smaller patches, since everything is so
> inter-related.  Defining only some of the new patterns does
> not leave things in a happy state.
> 
> The handling of comparisons is mostly taken from the vcond patterns.
> This means that it remains non-compliant with IEEE: “quiet” comparisons
> use signalling instructions.  But that shouldn't matter for floats,
> since we require -funsafe-math-optimizations to vectorize for them
> anyway.
> 
> It remains the case that comparisons and selects aren't implemented
> at all for HF vectors.  Implementing those feels like separate work.
> 
> Tested on arm-linux-gnueabihf and arm-eabi (for MVE).  OK to install?
> 

Ok.
Thanks, this looks good to me.
Kyrill

> Richard
> 
> 
> gcc/
> 	PR target/96528
> 	PR target/97288
> 	* config/arm/arm-protos.h (arm_expand_vector_compare): Declare.
> 	(arm_expand_vcond): Likewise.
> 	* config/arm/arm.c (arm_expand_vector_compare): New function.
> 	(arm_expand_vcond): Likewise.
> 	* config/arm/neon.md (vec_cmp<VDQW:mode><v_cmp_result>):
> New pattern.
> 	(vec_cmpu<VDQW:mode><VDQW:mode>): Likewise.
> 	(vcond<VDQW:mode><VDQW:mode>): Require operand 5 to be a
> register
> 	or zero.  Use arm_expand_vcond.
> 	(vcond<V_cvtto><V32:mode>): New pattern.
> 	(vcondu<VDQIW:mode><VDQIW:mode>): Generalize to...
> 	(vcondu<VDQW:mode><v_cmp_result): ...this.  Require operand 5
> 	to be a register or zero.  Use arm_expand_vcond.
> 	(vcond_mask_<VDQW:mode><v_cmp_result>): New pattern.
> 	(neon_vc<cmp_op><mode>, neon_vc<cmp_op><mode>_insn): Add
> "@" marker.
> 	(neon_vbsl<mode>): Likewise.
> 	(neon_vc<cmp_op>u<mode>): Reexpress as...
> 	(@neon_vc<code><mode>): ...this.
> 
> gcc/testsuite/
> 	* lib/target-supports.exp (check_effective_target_vect_cond_mixed):
> Add
> 	arm neon targets.
> 	* gcc.target/arm/neon-compare-1.c: New test.
> 	* gcc.target/arm/neon-compare-2.c: Likewise.
> 	* gcc.target/arm/neon-compare-3.c: Likewise.
> 	* gcc.target/arm/neon-compare-4.c: Likewise.
> 	* gcc.target/arm/neon-compare-5.c: Likewise.
> 	* gcc.target/arm/neon-vcond-gt.c: Expect comparisons with zero.
> 	* gcc.target/arm/neon-vcond-ltgt.c: Likewise.
> 	* gcc.target/arm/neon-vcond-unordered.c: Likewise.
> ---
>  gcc/config/arm/arm-protos.h                   |   2 +
>  gcc/config/arm/arm.c                          | 121 ++++++++
>  gcc/config/arm/neon.md                        | 281 ++++--------------
>  gcc/testsuite/gcc.target/arm/neon-compare-1.c |  84 ++++++
>  gcc/testsuite/gcc.target/arm/neon-compare-2.c |  45 +++
>  gcc/testsuite/gcc.target/arm/neon-compare-3.c |  44 +++
>  gcc/testsuite/gcc.target/arm/neon-compare-4.c |  38 +++
>  gcc/testsuite/gcc.target/arm/neon-compare-5.c |  37 +++
>  gcc/testsuite/gcc.target/arm/neon-vcond-gt.c  |   2 +-
>  .../gcc.target/arm/neon-vcond-ltgt.c          |   3 +-
>  .../gcc.target/arm/neon-vcond-unordered.c     |   4 +-
>  gcc/testsuite/lib/target-supports.exp         |   2 +
>  12 files changed, 442 insertions(+), 221 deletions(-)
>  create mode 100644 gcc/testsuite/gcc.target/arm/neon-compare-1.c
>  create mode 100644 gcc/testsuite/gcc.target/arm/neon-compare-2.c
>  create mode 100644 gcc/testsuite/gcc.target/arm/neon-compare-3.c
>  create mode 100644 gcc/testsuite/gcc.target/arm/neon-compare-4.c
>  create mode 100644 gcc/testsuite/gcc.target/arm/neon-compare-5.c
> 
> diff --git a/gcc/config/arm/arm-protos.h b/gcc/config/arm/arm-protos.h
> index 9bb9c61967b..703d6160c24 100644
> --- a/gcc/config/arm/arm-protos.h
> +++ b/gcc/config/arm/arm-protos.h
> @@ -372,9 +372,11 @@ extern void arm_emit_coreregs_64bit_shift (enum
> rtx_code, rtx, rtx, rtx, rtx,
>  extern bool arm_fusion_enabled_p (tune_params::fuse_ops);
>  extern bool arm_valid_symbolic_address_p (rtx);
>  extern bool arm_validize_comparison (rtx *, rtx *, rtx *);
> +extern bool arm_expand_vector_compare (rtx, rtx_code, rtx, rtx, bool);
>  #endif /* RTX_CODE */
> 
>  extern bool arm_gen_setmem (rtx *);
> +extern void arm_expand_vcond (rtx *, machine_mode);
>  extern void arm_expand_vec_perm (rtx target, rtx op0, rtx op1, rtx sel);
> 
>  extern bool arm_autoinc_modes_ok_p (machine_mode, enum
> arm_auto_incmodes);
> diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c
> index 8105b39e7a4..0e23246c27b 100644
> --- a/gcc/config/arm/arm.c
> +++ b/gcc/config/arm/arm.c
> @@ -30634,6 +30634,127 @@ arm_split_atomic_op (enum rtx_code code,
> rtx old_out, rtx new_out, rtx mem,
>      arm_post_atomic_barrier (model);
>  }
> 
> 
> 
> +/* Expand code to compare vectors OP0 and OP1 using condition CODE.
> +   If CAN_INVERT, store either the result or its inverse in TARGET
> +   and return true if TARGET contains the inverse.  If !CAN_INVERT,
> +   always store the result in TARGET, never its inverse.
> +
> +   Note that the handling of floating-point comparisons is not
> +   IEEE compliant.  */
> +
> +bool
> +arm_expand_vector_compare (rtx target, rtx_code code, rtx op0, rtx op1,
> +			   bool can_invert)
> +{
> +  machine_mode cmp_result_mode = GET_MODE (target);
> +  machine_mode cmp_mode = GET_MODE (op0);
> +
> +  bool inverted;
> +  switch (code)
> +    {
> +    /* For these we need to compute the inverse of the requested
> +       comparison.  */
> +    case UNORDERED:
> +    case UNLT:
> +    case UNLE:
> +    case UNGT:
> +    case UNGE:
> +    case UNEQ:
> +    case NE:
> +      code = reverse_condition_maybe_unordered (code);
> +      if (!can_invert)
> +	{
> +	  /* Recursively emit the inverted comparison into a temporary
> +	     and then store its inverse in TARGET.  This avoids reusing
> +	     TARGET (which for integer NE could be one of the inputs).  */
> +	  rtx tmp = gen_reg_rtx (cmp_result_mode);
> +	  if (arm_expand_vector_compare (tmp, code, op0, op1, true))
> +	    gcc_unreachable ();
> +	  emit_insn (gen_rtx_SET (target, gen_rtx_NOT (cmp_result_mode,
> tmp)));
> +	  return false;
> +	}
> +      inverted = true;
> +      break;
> +
> +    default:
> +      inverted = false;
> +      break;
> +    }
> +
> +  switch (code)
> +    {
> +    /* These are natively supported for zero comparisons, but otherwise
> +       require the operands to be swapped.  */
> +    case LE:
> +    case LT:
> +      if (op1 != CONST0_RTX (cmp_mode))
> +	{
> +	  code = swap_condition (code);
> +	  std::swap (op0, op1);
> +	}
> +      /* Fall through.  */
> +
> +    /* These are natively supported for both register and zero operands.  */
> +    case EQ:
> +    case GE:
> +    case GT:
> +      emit_insn (gen_neon_vc (code, cmp_mode, target, op0, op1));
> +      return inverted;
> +
> +    /* These are natively supported for register operands only.
> +       Comparisons with zero aren't useful and should be folded
> +       or canonicalized by target-independent code.  */
> +    case GEU:
> +    case GTU:
> +      emit_insn (gen_neon_vc (code, cmp_mode, target,
> +			      op0, force_reg (cmp_mode, op1)));
> +      return inverted;
> +
> +    /* These require the operands to be swapped and likewise do not
> +       support comparisons with zero.  */
> +    case LEU:
> +    case LTU:
> +      emit_insn (gen_neon_vc (swap_condition (code), cmp_mode,
> +			      target, force_reg (cmp_mode, op1), op0));
> +      return inverted;
> +
> +    /* These need a combination of two comparisons.  */
> +    case LTGT:
> +    case ORDERED:
> +      {
> +	/* Operands are LTGT iff (a > b || a > b).
> +	   Operands are ORDERED iff (a > b || a <= b).  */
> +	rtx gt_res = gen_reg_rtx (cmp_result_mode);
> +	rtx alt_res = gen_reg_rtx (cmp_result_mode);
> +	rtx_code alt_code = (code == LTGT ? LT : LE);
> +	if (arm_expand_vector_compare (gt_res, GT, op0, op1, true)
> +	    || arm_expand_vector_compare (alt_res, alt_code, op0, op1,
> true))
> +	  gcc_unreachable ();
> +	emit_insn (gen_rtx_SET (target, gen_rtx_IOR (cmp_result_mode,
> +						     gt_res, alt_res)));
> +	return inverted;
> +      }
> +
> +    default:
> +      gcc_unreachable ();
> +    }
> +}
> +
> +/* Expand a vcond or vcondu pattern with operands OPERANDS.
> +   CMP_RESULT_MODE is the mode of the comparison result.  */
> +
> +void
> +arm_expand_vcond (rtx *operands, machine_mode cmp_result_mode)
> +{
> +  rtx mask = gen_reg_rtx (cmp_result_mode);
> +  bool inverted = arm_expand_vector_compare (mask, GET_CODE
> (operands[3]),
> +					     operands[4], operands[5], true);
> +  if (inverted)
> +    std::swap (operands[1], operands[2]);
> +  emit_insn (gen_neon_vbsl (GET_MODE (operands[0]), operands[0],
> +			    mask, operands[1], operands[2]));
> +}
> +
> 
> 
>  #define MAX_VECT_LEN 16
> 
>  struct expand_vec_perm_d
> diff --git a/gcc/config/arm/neon.md b/gcc/config/arm/neon.md
> index 96bf277f501..58832cbf484 100644
> --- a/gcc/config/arm/neon.md
> +++ b/gcc/config/arm/neon.md
> @@ -1530,6 +1530,30 @@ (define_insn "*us_sub<mode>_neon"
>    [(set_attr "type" "neon_qsub<q>")]
>  )
> 
> +(define_expand "vec_cmp<mode><v_cmp_result>"
> +  [(set (match_operand:<V_cmp_result> 0 "s_register_operand")
> +	(match_operator:<V_cmp_result> 1 "comparison_operator"
> +	  [(match_operand:VDQW 2 "s_register_operand")
> +	   (match_operand:VDQW 3 "reg_or_zero_operand")]))]
> +  "TARGET_NEON && (!<Is_float_mode> ||
> flag_unsafe_math_optimizations)"
> +{
> +  arm_expand_vector_compare (operands[0], GET_CODE (operands[1]),
> +			     operands[2], operands[3], false);
> +  DONE;
> +})
> +
> +(define_expand "vec_cmpu<mode><mode>"
> +  [(set (match_operand:VDQIW 0 "s_register_operand")
> +	(match_operator:VDQIW 1 "comparison_operator"
> +	  [(match_operand:VDQIW 2 "s_register_operand")
> +	   (match_operand:VDQIW 3 "reg_or_zero_operand")]))]
> +  "TARGET_NEON"
> +{
> +  arm_expand_vector_compare (operands[0], GET_CODE (operands[1]),
> +			     operands[2], operands[3], false);
> +  DONE;
> +})
> +
>  ;; Conditional instructions.  These are comparisons with conditional moves
> for
>  ;; vectors.  They perform the assignment:
>  ;;
> @@ -1543,230 +1567,53 @@ (define_expand "vcond<mode><mode>"
>  	(if_then_else:VDQW
>  	  (match_operator 3 "comparison_operator"
>  	    [(match_operand:VDQW 4 "s_register_operand")
> -	     (match_operand:VDQW 5 "nonmemory_operand")])
> +	     (match_operand:VDQW 5 "reg_or_zero_operand")])
>  	  (match_operand:VDQW 1 "s_register_operand")
>  	  (match_operand:VDQW 2 "s_register_operand")))]
>    "TARGET_NEON && (!<Is_float_mode> ||
> flag_unsafe_math_optimizations)"
>  {
> -  int inverse = 0;
> -  int use_zero_form = 0;
> -  int swap_bsl_operands = 0;
> -  rtx mask = gen_reg_rtx (<V_cmp_result>mode);
> -  rtx tmp = gen_reg_rtx (<V_cmp_result>mode);
> -
> -  rtx (*base_comparison) (rtx, rtx, rtx);
> -  rtx (*complimentary_comparison) (rtx, rtx, rtx);
> -
> -  switch (GET_CODE (operands[3]))
> -    {
> -    case GE:
> -    case GT:
> -    case LE:
> -    case LT:
> -    case EQ:
> -      if (operands[5] == CONST0_RTX (<MODE>mode))
> -	{
> -	  use_zero_form = 1;
> -	  break;
> -	}
> -      /* Fall through.  */
> -    default:
> -      if (!REG_P (operands[5]))
> -	operands[5] = force_reg (<MODE>mode, operands[5]);
> -    }
> -
> -  switch (GET_CODE (operands[3]))
> -    {
> -    case LT:
> -    case UNLT:
> -      inverse = 1;
> -      /* Fall through.  */
> -    case GE:
> -    case UNGE:
> -    case ORDERED:
> -    case UNORDERED:
> -      base_comparison = gen_neon_vcge<mode>;
> -      complimentary_comparison = gen_neon_vcgt<mode>;
> -      break;
> -    case LE:
> -    case UNLE:
> -      inverse = 1;
> -      /* Fall through.  */
> -    case GT:
> -    case UNGT:
> -      base_comparison = gen_neon_vcgt<mode>;
> -      complimentary_comparison = gen_neon_vcge<mode>;
> -      break;
> -    case EQ:
> -    case NE:
> -    case UNEQ:
> -      base_comparison = gen_neon_vceq<mode>;
> -      complimentary_comparison = gen_neon_vceq<mode>;
> -      break;
> -    default:
> -      gcc_unreachable ();
> -    }
> -
> -  switch (GET_CODE (operands[3]))
> -    {
> -    case LT:
> -    case LE:
> -    case GT:
> -    case GE:
> -    case EQ:
> -      /* The easy case.  Here we emit one of vcge, vcgt or vceq.
> -	 As a LT b <=> b GE a && a LE b <=> b GT a.  Our transformations are:
> -	 a GE b -> a GE b
> -	 a GT b -> a GT b
> -	 a LE b -> b GE a
> -	 a LT b -> b GT a
> -	 a EQ b -> a EQ b
> -	 Note that there also exist direct comparison against 0 forms,
> -	 so catch those as a special case.  */
> -      if (use_zero_form)
> -	{
> -	  inverse = 0;
> -	  switch (GET_CODE (operands[3]))
> -	    {
> -	    case LT:
> -	      base_comparison = gen_neon_vclt<mode>;
> -	      break;
> -	    case LE:
> -	      base_comparison = gen_neon_vcle<mode>;
> -	      break;
> -	    default:
> -	      /* Do nothing, other zero form cases already have the correct
> -		 base_comparison.  */
> -	      break;
> -	    }
> -	}
> -
> -      if (!inverse)
> -	emit_insn (base_comparison (mask, operands[4], operands[5]));
> -      else
> -	emit_insn (complimentary_comparison (mask, operands[5],
> operands[4]));
> -      break;
> -    case UNLT:
> -    case UNLE:
> -    case UNGT:
> -    case UNGE:
> -    case NE:
> -      /* Vector compare returns false for lanes which are unordered, so if we
> use
> -	 the inverse of the comparison we actually want to emit, then
> -	 swap the operands to BSL, we will end up with the correct result.
> -	 Note that a NE NaN and NaN NE b are true for all a, b.
> -
> -	 Our transformations are:
> -	 a GE b -> !(b GT a)
> -	 a GT b -> !(b GE a)
> -	 a LE b -> !(a GT b)
> -	 a LT b -> !(a GE b)
> -	 a NE b -> !(a EQ b)  */
> -
> -      if (inverse)
> -	emit_insn (base_comparison (mask, operands[4], operands[5]));
> -      else
> -	emit_insn (complimentary_comparison (mask, operands[5],
> operands[4]));
> -
> -      swap_bsl_operands = 1;
> -      break;
> -    case UNEQ:
> -      /* We check (a > b ||  b > a).  combining these comparisons give us
> -	 true iff !(a != b && a ORDERED b), swapping the operands to BSL
> -	 will then give us (a == b ||  a UNORDERED b) as intended.  */
> -
> -      emit_insn (gen_neon_vcgt<mode> (mask, operands[4], operands[5]));
> -      emit_insn (gen_neon_vcgt<mode> (tmp, operands[5], operands[4]));
> -      emit_insn (gen_ior<v_cmp_result>3 (mask, mask, tmp));
> -      swap_bsl_operands = 1;
> -      break;
> -    case UNORDERED:
> -       /* Operands are ORDERED iff (a > b || b >= a).
> -	 Swapping the operands to BSL will give the UNORDERED case.  */
> -     swap_bsl_operands = 1;
> -     /* Fall through.  */
> -    case ORDERED:
> -      emit_insn (gen_neon_vcgt<mode> (tmp, operands[4], operands[5]));
> -      emit_insn (gen_neon_vcge<mode> (mask, operands[5], operands[4]));
> -      emit_insn (gen_ior<v_cmp_result>3 (mask, mask, tmp));
> -      break;
> -    default:
> -      gcc_unreachable ();
> -    }
> +  arm_expand_vcond (operands, <V_cmp_result>mode);
> +  DONE;
> +})
> 
> -  if (swap_bsl_operands)
> -    emit_insn (gen_neon_vbsl<mode> (operands[0], mask, operands[2],
> -				    operands[1]));
> -  else
> -    emit_insn (gen_neon_vbsl<mode> (operands[0], mask, operands[1],
> -				    operands[2]));
> +(define_expand "vcond<V_cvtto><mode>"
> +  [(set (match_operand:<V_CVTTO> 0 "s_register_operand")
> +	(if_then_else:<V_CVTTO>
> +	  (match_operator 3 "comparison_operator"
> +	    [(match_operand:V32 4 "s_register_operand")
> +	     (match_operand:V32 5 "reg_or_zero_operand")])
> +	  (match_operand:<V_CVTTO> 1 "s_register_operand")
> +	  (match_operand:<V_CVTTO> 2 "s_register_operand")))]
> +  "TARGET_NEON && (!<Is_float_mode> ||
> flag_unsafe_math_optimizations)"
> +{
> +  arm_expand_vcond (operands, <V_cmp_result>mode);
>    DONE;
>  })
> 
> -(define_expand "vcondu<mode><mode>"
> -  [(set (match_operand:VDQIW 0 "s_register_operand")
> -	(if_then_else:VDQIW
> +(define_expand "vcondu<mode><v_cmp_result>"
> +  [(set (match_operand:VDQW 0 "s_register_operand")
> +	(if_then_else:VDQW
>  	  (match_operator 3 "arm_comparison_operator"
> -	    [(match_operand:VDQIW 4 "s_register_operand")
> -	     (match_operand:VDQIW 5 "s_register_operand")])
> -	  (match_operand:VDQIW 1 "s_register_operand")
> -	  (match_operand:VDQIW 2 "s_register_operand")))]
> +	    [(match_operand:<V_cmp_result> 4 "s_register_operand")
> +	     (match_operand:<V_cmp_result> 5 "reg_or_zero_operand")])
> +	  (match_operand:VDQW 1 "s_register_operand")
> +	  (match_operand:VDQW 2 "s_register_operand")))]
>    "TARGET_NEON"
>  {
> -  rtx mask;
> -  int inverse = 0, immediate_zero = 0;
> -
> -  mask = gen_reg_rtx (<V_cmp_result>mode);
> -
> -  if (operands[5] == CONST0_RTX (<MODE>mode))
> -    immediate_zero = 1;
> -  else if (!REG_P (operands[5]))
> -    operands[5] = force_reg (<MODE>mode, operands[5]);
> -
> -  switch (GET_CODE (operands[3]))
> -    {
> -    case GEU:
> -      emit_insn (gen_neon_vcgeu<mode> (mask, operands[4], operands[5]));
> -      break;
> -
> -    case GTU:
> -      emit_insn (gen_neon_vcgtu<mode> (mask, operands[4], operands[5]));
> -      break;
> -
> -    case EQ:
> -      emit_insn (gen_neon_vceq<mode> (mask, operands[4], operands[5]));
> -      break;
> -
> -    case LEU:
> -      if (immediate_zero)
> -	emit_insn (gen_neon_vcle<mode> (mask, operands[4], operands[5]));
> -      else
> -	emit_insn (gen_neon_vcgeu<mode> (mask, operands[5],
> operands[4]));
> -      break;
> -
> -    case LTU:
> -      if (immediate_zero)
> -        emit_insn (gen_neon_vclt<mode> (mask, operands[4], operands[5]));
> -      else
> -	emit_insn (gen_neon_vcgtu<mode> (mask, operands[5],
> operands[4]));
> -      break;
> -
> -    case NE:
> -      emit_insn (gen_neon_vceq<mode> (mask, operands[4], operands[5]));
> -      inverse = 1;
> -      break;
> -
> -    default:
> -      gcc_unreachable ();
> -    }
> -
> -  if (inverse)
> -    emit_insn (gen_neon_vbsl<mode> (operands[0], mask, operands[2],
> -				    operands[1]));
> -  else
> -    emit_insn (gen_neon_vbsl<mode> (operands[0], mask, operands[1],
> -				    operands[2]));
> +  arm_expand_vcond (operands, <V_cmp_result>mode);
> +  DONE;
> +})
> 
> +(define_expand "vcond_mask_<mode><v_cmp_result>"
> +  [(set (match_operand:VDQW 0 "s_register_operand")
> +	(if_then_else:VDQW
> +	  (match_operand:<V_cmp_result> 3 "s_register_operand")
> +	  (match_operand:VDQW 1 "s_register_operand")
> +	  (match_operand:VDQW 2 "s_register_operand")))]
> +  "TARGET_NEON"
> +{
> +  emit_insn (gen_neon_vbsl<mode> (operands[0], operands[3], operands[1],
> +				  operands[2]));
>    DONE;
>  })
> 
> @@ -2601,7 +2448,7 @@ (define_insn "neon_v<r>subhn<mode>"
> 
>  ;; These may expand to an UNSPEC pattern when a floating point mode is
> used
>  ;; without unsafe math optimizations.
> -(define_expand "neon_vc<cmp_op><mode>"
> +(define_expand "@neon_vc<cmp_op><mode>"
>    [(match_operand:<V_cmp_result> 0 "s_register_operand")
>       (neg:<V_cmp_result>
>         (COMPARISONS:VDQW (match_operand:VDQW 1 "s_register_operand")
> @@ -2641,7 +2488,7 @@ (define_expand "neon_vc<cmp_op><mode>"
>    }
>  )
> 
> -(define_insn "neon_vc<cmp_op><mode>_insn"
> +(define_insn "@neon_vc<cmp_op><mode>_insn"
>    [(set (match_operand:<V_cmp_result> 0 "s_register_operand" "=w,w")
>          (neg:<V_cmp_result>
>            (COMPARISONS:<V_cmp_result>
> @@ -2685,7 +2532,7 @@ (define_insn
> "neon_vc<cmp_op_unsp><mode>_insn_unspec"
>    [(set_attr "type" "neon_fp_compare_s<q>")]
>  )
> 
> -(define_expand "neon_vc<cmp_op><mode>"
> +(define_expand "@neon_vc<cmp_op><mode>"
>   [(match_operand:<V_cmp_result> 0 "s_register_operand")
>    (neg:<V_cmp_result>
>     (COMPARISONS:VH
> @@ -2751,7 +2598,7 @@ (define_insn
> "neon_vc<cmp_op_unsp><mode>_fp16insn_unspec"
>  }
>   [(set_attr "type" "neon_fp_compare_s<q>")])
> 
> -(define_insn "neon_vc<cmp_op>u<mode>"
> +(define_insn "@neon_vc<code><mode>"
>    [(set (match_operand:<V_cmp_result> 0 "s_register_operand" "=w")
>          (neg:<V_cmp_result>
>            (GTUGEU:<V_cmp_result>
> @@ -4708,7 +4555,7 @@ (define_insn "neon_vbsl<mode>_internal"
>    [(set_attr "type" "neon_bsl<q>")]
>  )
> 
> -(define_expand "neon_vbsl<mode>"
> +(define_expand "@neon_vbsl<mode>"
>    [(set (match_operand:VDQX 0 "s_register_operand")
>          (unspec:VDQX [(match_operand:<V_cmp_result> 1
> "s_register_operand")
>                        (match_operand:VDQX 2 "s_register_operand")
> diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-
> supports.exp
> index 8314e443c43..69a2b8d838d 100644
> --- a/gcc/testsuite/lib/target-supports.exp
> +++ b/gcc/testsuite/lib/target-supports.exp
> @@ -7233,6 +7233,8 @@ proc check_effective_target_vect_cond_mixed { }
> {
>        expr { [istarget i?86-*-*] || [istarget x86_64-*-*]
>  	     || [istarget aarch64*-*-*]
>  	     || [istarget powerpc*-*-*]
> +	     || ([istarget arm*-*-*]
> +		 && [check_effective_target_arm_neon_ok])
>  	     || ([istarget mips*-*-*]
>  		 && [et-is-effective-target mips_msa])
>  	     || ([istarget s390*-*-*]
> diff --git a/gcc/testsuite/gcc.target/arm/neon-compare-1.c
> b/gcc/testsuite/gcc.target/arm/neon-compare-1.c
> new file mode 100644
> index 00000000000..c915eca6dbe
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/neon-compare-1.c
> @@ -0,0 +1,84 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-O1" }  */
> +/* { dg-add-options arm_neon } */
> +
> +#define COMPARE_REG(NAME, OP, TYPE) \
> +  TYPE \
> +  cmp_##NAME##_##TYPE##_reg (TYPE a, TYPE b) \
> +  { \
> +    return a OP b; \
> +  }
> +
> +#define COMPARE_REG_AND_ZERO(NAME, OP, TYPE) \
> +  COMPARE_REG (NAME, OP, TYPE) \
> +  \
> +  TYPE \
> +  cmp_##NAME##_##TYPE##_zero (TYPE a) \
> +  { \
> +    return a OP (TYPE) {}; \
> +  }
> +
> +#define COMPARE_TYPE(TYPE, COMPARE_ORDERED) \
> +  COMPARE_REG_AND_ZERO (eq, ==, TYPE) \
> +  COMPARE_REG_AND_ZERO (ne, !=, TYPE) \
> +  COMPARE_ORDERED (lt, <, TYPE) \
> +  COMPARE_ORDERED (le, <=, TYPE) \
> +  COMPARE_ORDERED (gt, >, TYPE) \
> +  COMPARE_ORDERED (ge, >=, TYPE)
> +
> +#define TEST_TYPE(NAME, ELEM, COMPARE_ORDERED) \
> +  typedef ELEM NAME __attribute__((vector_size(16))); \
> +  COMPARE_TYPE (NAME, COMPARE_ORDERED)
> +
> +TEST_TYPE (vs8, __INT8_TYPE__, COMPARE_REG_AND_ZERO)
> +TEST_TYPE (vu8, __UINT8_TYPE__, COMPARE_REG)
> +TEST_TYPE (vs16, __INT16_TYPE__, COMPARE_REG_AND_ZERO)
> +TEST_TYPE (vu16, __UINT16_TYPE__, COMPARE_REG)
> +TEST_TYPE (vs32, __INT32_TYPE__, COMPARE_REG_AND_ZERO)
> +TEST_TYPE (vu32, __UINT32_TYPE__, COMPARE_REG)
> +
> +/* { s8, u8 } x { eq, ne }.
> +/* { dg-final { scan-assembler-times {\tvceq.i8\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 4 } } */
> +/* { dg-final { scan-assembler-times {\tvceq.i8\tq[0-9]+, q[0-9]+, #0\n} 4 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.s8\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.s8\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.s8\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.s8\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.s8\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvcle.s8\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.u8\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.u8\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +
> +/* { s16, u16 } x { eq, ne }.
> +/* { dg-final { scan-assembler-times {\tvceq.i16\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 4 } } */
> +/* { dg-final { scan-assembler-times {\tvceq.i16\tq[0-9]+, q[0-9]+, #0\n} 4 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.s16\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.s16\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.s16\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.s16\tq[0-9]+, q[0-9]+, q[0-
> 9]+\n} 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.s16\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvcle.s16\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.u16\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.u16\tq[0-9]+, q[0-9]+, q[0-
> 9]+\n} 2 } } */
> +
> +/* { s32, u32 } x { eq, ne }.
> +/* { dg-final { scan-assembler-times {\tvceq.i32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 4 } } */
> +/* { dg-final { scan-assembler-times {\tvceq.i32\tq[0-9]+, q[0-9]+, #0\n} 4 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.s32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.s32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.s32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.s32\tq[0-9]+, q[0-9]+, q[0-
> 9]+\n} 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.s32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvcle.s32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.u32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.u32\tq[0-9]+, q[0-9]+, q[0-
> 9]+\n} 2 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-compare-2.c
> b/gcc/testsuite/gcc.target/arm/neon-compare-2.c
> new file mode 100644
> index 00000000000..559c5e5d8fc
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/neon-compare-2.c
> @@ -0,0 +1,45 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-O1 -funsafe-math-optimizations" }  */
> +/* { dg-add-options arm_neon } */
> +
> +#ifndef ELEM_TYPE
> +#define ELEM_TYPE float
> +#endif
> +#ifndef INT_ELEM_TYPE
> +#define INT_ELEM_TYPE __INT32_TYPE__
> +#endif
> +
> +#define COMPARE(NAME, OP) \
> +  int_vec \
> +  cmp_##NAME##_reg (vec a, vec b) \
> +  { \
> +    return a OP b; \
> +  } \
> +  \
> +  int_vec \
> +  cmp_##NAME##_zero (vec a) \
> +  { \
> +    return a OP (vec) {}; \
> +  }
> +
> +typedef INT_ELEM_TYPE int_vec __attribute__((vector_size(16)));
> +typedef ELEM_TYPE vec __attribute__((vector_size(16)));
> +
> +COMPARE (eq, ==)
> +COMPARE (ne, !=)
> +COMPARE (lt, <)
> +COMPARE (le, <=)
> +COMPARE (gt, >)
> +COMPARE (ge, >=)
> +
> +/* { dg-final { scan-assembler-times {\tvceq.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvceq.f32\tq[0-9]+, q[0-9]+, #0\n} 2 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvcle.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-compare-3.c
> b/gcc/testsuite/gcc.target/arm/neon-compare-3.c
> new file mode 100644
> index 00000000000..efbe79728a9
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/neon-compare-3.c
> @@ -0,0 +1,44 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-O1 -ftree-vectorize -funsafe-math-optimizations" }  */
> +/* { dg-add-options arm_neon } */
> +
> +#define ult(a, b) (!__builtin_isgreaterequal (a, b))
> +#define ule(a, b) (!__builtin_isgreater (a, b))
> +#define ugt(a, b) (!__builtin_islessequal (a, b))
> +#define uge(a, b) (!__builtin_isless (a, b))
> +
> +int x[16];
> +float a[16];
> +float b[16];
> +
> +#define COMPARE(NAME) \
> +  void \
> +  cmp_##NAME##_reg (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], b[i]) ? 2 : 0; \
> +  } \
> +  \
> +  void \
> +  cmp_##NAME##_zero (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], 0) ? 2 : 0; \
> +  }
> +
> +typedef int int_vec __attribute__((vector_size(16)));
> +typedef float vec __attribute__((vector_size(16)));
> +
> +COMPARE (ult)
> +COMPARE (ule)
> +COMPARE (ugt)
> +COMPARE (uge)
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcge.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> +/* { dg-final { scan-assembler-times {\tvcle.f32\tq[0-9]+, q[0-9]+, #0\n} 1 } }
> */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-compare-4.c
> b/gcc/testsuite/gcc.target/arm/neon-compare-4.c
> new file mode 100644
> index 00000000000..3f8cc906c66
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/neon-compare-4.c
> @@ -0,0 +1,38 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-O1 -ftree-vectorize -funsafe-math-optimizations" }  */
> +/* { dg-add-options arm_neon } */
> +
> +#define ordered(a, b) (!__builtin_isunordered (a, b))
> +#define unordered(a, b) (__builtin_isunordered (a, b))
> +
> +int x[16];
> +float a[16];
> +float b[16];
> +
> +#define COMPARE(NAME) \
> +  void \
> +  cmp_##NAME##_reg (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], b[i]) ? 2 : 0; \
> +  } \
> +  \
> +  void \
> +  cmp_##NAME##_zero (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], 0) ? 2 : 0; \
> +  }
> +
> +typedef int int_vec __attribute__((vector_size(16)));
> +typedef float vec __attribute__((vector_size(16)));
> +
> +COMPARE (ordered)
> +COMPARE (unordered)
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, #0\n} 2 } }
> */
> +
> +/* { dg-final { scan-assembler-times {\tvcge.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 2 } } */
> +/* { dg-final { scan-assembler-times {\tvcle.f32\tq[0-9]+, q[0-9]+, #0\n} 2 } }
> */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-compare-5.c
> b/gcc/testsuite/gcc.target/arm/neon-compare-5.c
> new file mode 100644
> index 00000000000..cb6428d3f08
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/neon-compare-5.c
> @@ -0,0 +1,37 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-O1 -ftree-vectorize -funsafe-math-optimizations" }  */
> +/* { dg-add-options arm_neon } */
> +
> +#define uneq(a, b) (!__builtin_islessgreater (a, b))
> +/* RTL's LTGT is a signaling comparison.  */
> +#define ltgt(a, b) (a < b || b < a)
> +
> +int x[16];
> +float a[16];
> +float b[16];
> +
> +#define COMPARE(NAME) \
> +  void \
> +  cmp_##NAME##_reg (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], b[i]) ? 2 : 0; \
> +  } \
> +  \
> +  void \
> +  cmp_##NAME##_zero (void) \
> +  { \
> +    for (int i = 0; i < 16; ++i) \
> +      x[i] = NAME (a[i], 0) ? 2 : 0; \
> +  }
> +
> +typedef int int_vec __attribute__((vector_size(16)));
> +typedef float vec __attribute__((vector_size(16)));
> +
> +COMPARE (uneq)
> +COMPARE (ltgt)
> +
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, q[0-9]+\n}
> 4 } } */
> +/* { dg-final { scan-assembler-times {\tvcgt.f32\tq[0-9]+, q[0-9]+, #0\n} 2 } }
> */
> +/* { dg-final { scan-assembler-times {\tvclt.f32\tq[0-9]+, q[0-9]+, #0\n} 2 } }
> */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-vcond-gt.c
> b/gcc/testsuite/gcc.target/arm/neon-vcond-gt.c
> index 8e9f3785169..9f601a169d1 100644
> --- a/gcc/testsuite/gcc.target/arm/neon-vcond-gt.c
> +++ b/gcc/testsuite/gcc.target/arm/neon-vcond-gt.c
> @@ -13,5 +13,5 @@ void foo (int ilast,float* w, float* w2)
>    }
>  }
> 
> -/* { dg-final { scan-assembler "vcgt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */
> +/* { dg-final { scan-assembler "vclt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*#0" } } */
>  /* { dg-final { scan-assembler "vbsl|vbit|vbif\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-vcond-ltgt.c
> b/gcc/testsuite/gcc.target/arm/neon-vcond-ltgt.c
> index c8306e364a3..74bc22046ad 100644
> --- a/gcc/testsuite/gcc.target/arm/neon-vcond-ltgt.c
> +++ b/gcc/testsuite/gcc.target/arm/neon-vcond-ltgt.c
> @@ -13,6 +13,7 @@ void foo (int ilast,float* w, float* w2)
>    }
>  }
> 
> -/* { dg-final { scan-assembler-times "vcgt\\.f32\[\\t \]*q\[0-9\]+,\[\\t
> \]*q\[0-9\]+,\[\\t \]*q\[0-9\]+" 2 } } */
> +/* { dg-final { scan-assembler "vclt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*#0" } } */
> +/* { dg-final { scan-assembler "vcgt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*#0" } } */
>  /* { dg-final { scan-assembler "vorr\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-9\]+,\[\\t
> \]*q\[0-9\]+" } } */
>  /* { dg-final { scan-assembler "vbsl|vbit|vbif\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */
> diff --git a/gcc/testsuite/gcc.target/arm/neon-vcond-unordered.c
> b/gcc/testsuite/gcc.target/arm/neon-vcond-unordered.c
> index 3bb67d3afe3..8d3187541e2 100644
> --- a/gcc/testsuite/gcc.target/arm/neon-vcond-unordered.c
> +++ b/gcc/testsuite/gcc.target/arm/neon-vcond-unordered.c
> @@ -13,7 +13,7 @@ void foo (int ilast,float* w, float* w2)
>    }
>  }
> 
> -/* { dg-final { scan-assembler "vcgt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */
> -/* { dg-final { scan-assembler "vcge\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */
> +/* { dg-final { scan-assembler "vcgt\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*#0" } } */
> +/* { dg-final { scan-assembler "vcle\\.f32\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*#0" } } */
>  /* { dg-final { scan-assembler "vorr\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-9\]+,\[\\t
> \]*q\[0-9\]+" } } */
>  /* { dg-final { scan-assembler "vbsl|vbit|vbif\[\\t \]*q\[0-9\]+,\[\\t \]*q\[0-
> 9\]+,\[\\t \]*q\[0-9\]+" } } */


More information about the Gcc-patches mailing list