[PATCH 3/5] xtensa: Add some dedicated patterns that correspond to GIMPLE canonicalizations

[PATCH 3/5] xtensa: Add some dedicated patterns that correspond to GIMPLE canonicalizations

[Takayuki 'January June' Suwa]

This patch offers better RTL representations against straightforward
derivations from some tree optimizers' canonicalized forms.

- rounding up to even, such as '(x + (x & 1))', is canonicalized to
  '((x + 1) & -2)', but the former is one instruction less than the latter
  in Xtensa ISA.
- signed greater or equal to zero as logical value '((signed)x >= 0)',
  is canonicalized to '((unsigned)(x ^ -1) >> 31)', but the equivalent
  '(((signed)x >> 31) + 1)' is one instruction less.

gcc/ChangeLog:

	* config/xtensa/xtensa.md (*round_up_to_even):
	New insn-and-split pattern.
	(*signed_ge_zero): Ditto.
---
 gcc/config/xtensa/xtensa.md | 45 +++++++++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)

diff --git a/gcc/config/xtensa/xtensa.md b/gcc/config/xtensa/xtensa.md
index 181f935e3c3..9588a829136 100644
--- a/gcc/config/xtensa/xtensa.md
+++ b/gcc/config/xtensa/xtensa.md
@@ -2679,3 +2679,48 @@
   xtensa_expand_atomic (<CODE>, operands[0], operands[1], operands[2], true);
   DONE;
 })
+
+(define_insn_and_split "*round_up_to_even"
+  [(set (match_operand:SI 0 "register_operand" "=a")
+	(and:SI (plus:SI (match_operand:SI 1 "register_operand" "r")
+			 (const_int 1))
+		(const_int -2)))]
+  ""
+  "#"
+  "can_create_pseudo_p ()"
+  [(set (match_dup 2)
+	(and:SI (match_dup 1)
+		(const_int 1)))
+   (set (match_dup 0)
+	(plus:SI (match_dup 2)
+		 (match_dup 1)))]
+{
+  operands[2] = gen_reg_rtx (SImode);
+}
+  [(set_attr "type"	"arith")
+   (set_attr "mode"	"SI")
+   (set (attr "length")
+	(if_then_else (match_test "TARGET_DENSITY")
+		      (const_int 5)
+		      (const_int 6)))])
+
+(define_insn_and_split "*signed_ge_zero"
+  [(set (match_operand:SI 0 "register_operand" "=a")
+	(ge:SI (match_operand:SI 1 "register_operand" "r")
+	       (const_int 0)))]
+  ""
+  "#"
+  ""
+  [(set (match_dup 0)
+	(ashiftrt:SI (match_dup 1)
+		     (const_int 31)))
+   (set (match_dup 0)
+	(plus:SI (match_dup 0)
+		 (const_int 1)))]
+  ""
+  [(set_attr "type"	"arith")
+   (set_attr "mode"	"SI")
+   (set (attr "length")
+	(if_then_else (match_test "TARGET_DENSITY")
+		      (const_int 5)
+		      (const_int 6)))])
-- 
2.20.1

Re: [PATCH 3/5] xtensa: Add some dedicated patterns that correspond to GIMPLE canonicalizations

[Max Filippov]

On Mon, Jun 13, 2022 at 8:54 PM Takayuki 'January June' Suwa
<jjsuwa_sys3175@yahoo.co.jp> wrote:
>
> This patch offers better RTL representations against straightforward
> derivations from some tree optimizers' canonicalized forms.
>
> - rounding up to even, such as '(x + (x & 1))', is canonicalized to
>   '((x + 1) & -2)', but the former is one instruction less than the latter
>   in Xtensa ISA.
> - signed greater or equal to zero as logical value '((signed)x >= 0)',
>   is canonicalized to '((unsigned)(x ^ -1) >> 31)', but the equivalent
>   '(((signed)x >> 31) + 1)' is one instruction less.
>
> gcc/ChangeLog:
>
>         * config/xtensa/xtensa.md (*round_up_to_even):
>         New insn-and-split pattern.
>         (*signed_ge_zero): Ditto.
> ---
>  gcc/config/xtensa/xtensa.md | 45 +++++++++++++++++++++++++++++++++++++
>  1 file changed, 45 insertions(+)

Regtested for target=xtensa-linux-uclibc, no new regressions.
Committed to master.

-- 
Thanks.
-- Max