[v2] RISC-V: Support IMM for operand 0 of ussub pattern

  From: Pan Li <pan2.li@intel.com>

This patch would like to allow IMM for the operand 0 of ussub pattern.
Aka .SAT_SUB(1023, y) as the below example.

Form 1:
  #define DEF_SAT_U_SUB_IMM_FMT_1(T, IMM) \
  T __attribute__((noinline))             \
  sat_u_sub_imm##IMM##_##T##_fmt_1 (T y)  \
  {                                       \
    return (T)IMM >= y ? (T)IMM - y : 0;  \
  }

DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 1023)

Before this patch:
  10   │ sat_u_sub_imm82_uint64_t_fmt_1:
  11   │     li  a5,82
  12   │     bgtu    a0,a5,.L3
  13   │     sub a0,a5,a0
  14   │     ret
  15   │ .L3:
  16   │     li  a0,0
  17   │     ret

After this patch:
  10   │ sat_u_sub_imm82_uint64_t_fmt_1:
  11   │     li  a5,82
  12   │     sltu    a4,a5,a0
  13   │     addi    a4,a4,-1
  14   │     sub a0,a5,a0
  15   │     and a0,a4,a0
  16   │     ret

The below test suites are passed for this patch:
1. The rv64gcv fully regression test.

gcc/ChangeLog:

	* config/riscv/riscv.cc (riscv_gen_unsigned_xmode_reg): Add new
	func impl to gen xmode rtx reg from operand rtx.
	(riscv_expand_ussub): Gen xmode reg for operand 1.
	* config/riscv/riscv.md: Allow const_int for operand 1.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/sat_arith.h: Add test helper macro.
	* gcc.target/riscv/sat_u_sub_imm-1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-1_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-1_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-4.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-3.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-4.c: New test.

Signed-off-by: Pan Li <pan2.li@intel.com>
---
 gcc/config/riscv/riscv.cc                     | 51 ++++++++++++++++-
 gcc/config/riscv/riscv.md                     |  2 +-
 gcc/testsuite/gcc.target/riscv/sat_arith.h    | 10 ++++
 .../gcc.target/riscv/sat_u_sub_imm-1.c        | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-1_1.c      | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-1_2.c      | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2.c        | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2_1.c      | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2_2.c      | 22 ++++++++
 .../gcc.target/riscv/sat_u_sub_imm-3.c        | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-3_1.c      | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-3_2.c      | 22 ++++++++
 .../gcc.target/riscv/sat_u_sub_imm-4.c        | 19 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-1.c    | 56 +++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-2.c    | 56 +++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-3.c    | 55 ++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-4.c    | 48 ++++++++++++++++
 17 files changed, 482 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c
  

This Patch may requires rebase, will send v3 for conflict resolving.

Pan

-----Original Message-----
From: Li, Pan2 <pan2.li@intel.com> 
Sent: Sunday, August 4, 2024 7:48 PM
To: gcc-patches@gcc.gnu.org
Cc: juzhe.zhong@rivai.ai; kito.cheng@gmail.com; jeffreyalaw@gmail.com; rdapp.gcc@gmail.com; Li, Pan2 <pan2.li@intel.com>
Subject: [PATCH v2] RISC-V: Support IMM for operand 0 of ussub pattern

From: Pan Li <pan2.li@intel.com>

This patch would like to allow IMM for the operand 0 of ussub pattern.
Aka .SAT_SUB(1023, y) as the below example.

Form 1:
  #define DEF_SAT_U_SUB_IMM_FMT_1(T, IMM) \
  T __attribute__((noinline))             \
  sat_u_sub_imm##IMM##_##T##_fmt_1 (T y)  \
  {                                       \
    return (T)IMM >= y ? (T)IMM - y : 0;  \
  }

DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 1023)

Before this patch:
  10   │ sat_u_sub_imm82_uint64_t_fmt_1:
  11   │     li  a5,82
  12   │     bgtu    a0,a5,.L3
  13   │     sub a0,a5,a0
  14   │     ret
  15   │ .L3:
  16   │     li  a0,0
  17   │     ret

After this patch:
  10   │ sat_u_sub_imm82_uint64_t_fmt_1:
  11   │     li  a5,82
  12   │     sltu    a4,a5,a0
  13   │     addi    a4,a4,-1
  14   │     sub a0,a5,a0
  15   │     and a0,a4,a0
  16   │     ret

The below test suites are passed for this patch:
1. The rv64gcv fully regression test.

gcc/ChangeLog:

	* config/riscv/riscv.cc (riscv_gen_unsigned_xmode_reg): Add new
	func impl to gen xmode rtx reg from operand rtx.
	(riscv_expand_ussub): Gen xmode reg for operand 1.
	* config/riscv/riscv.md: Allow const_int for operand 1.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/sat_arith.h: Add test helper macro.
	* gcc.target/riscv/sat_u_sub_imm-1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-1_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-1_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-2_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3_1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-3_2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-4.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-1.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-2.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-3.c: New test.
	* gcc.target/riscv/sat_u_sub_imm-run-4.c: New test.

Signed-off-by: Pan Li <pan2.li@intel.com>
---
 gcc/config/riscv/riscv.cc                     | 51 ++++++++++++++++-
 gcc/config/riscv/riscv.md                     |  2 +-
 gcc/testsuite/gcc.target/riscv/sat_arith.h    | 10 ++++
 .../gcc.target/riscv/sat_u_sub_imm-1.c        | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-1_1.c      | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-1_2.c      | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2.c        | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2_1.c      | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-2_2.c      | 22 ++++++++
 .../gcc.target/riscv/sat_u_sub_imm-3.c        | 20 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-3_1.c      | 21 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-3_2.c      | 22 ++++++++
 .../gcc.target/riscv/sat_u_sub_imm-4.c        | 19 +++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-1.c    | 56 +++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-2.c    | 56 +++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-3.c    | 55 ++++++++++++++++++
 .../gcc.target/riscv/sat_u_sub_imm-run-4.c    | 48 ++++++++++++++++
 17 files changed, 482 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
 create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c

diff --git a/gcc/config/riscv/riscv.cc b/gcc/config/riscv/riscv.cc
index b19d56149e7..5e4e9722729 100644
--- a/gcc/config/riscv/riscv.cc
+++ b/gcc/config/riscv/riscv.cc
@@ -11612,6 +11612,55 @@ riscv_expand_usadd (rtx dest, rtx x, rtx y)
   emit_move_insn (dest, gen_lowpart (mode, xmode_dest));
 }
 
+/* Generate a REG rtx of Xmode from the given rtx and mode.
+   The rtx x can be REG (QI/HI/SI/DI) or const_int.
+   The machine_mode mode is the original mode from define pattern.
+
+   If rtx is REG,  the gen_lowpart of Xmode will be returned.
+
+   If rtx is const_int,  a new REG rtx will be created to hold the value of
+   const_int and then returned.
+
+   According to the gccint doc, the constants generated for modes with fewer
+   bits than in HOST_WIDE_INT must be sign extended to full width.  Thus there
+   will be two cases here,  take QImode as example.
+
+   For .SAT_SUB (127, y) in QImode, we have (const_int 127) and one simple
+   mov from const_int to the new REG rtx is good enough here.
+
+   For .SAT_SUB (254, y) in QImode, we have (const_int -2) after define_expand.
+   Aka 0xfffffffffffffffe in Xmode of RV64 but we actually need 0xfe in Xmode
+   of RV64.  So we need to cleanup the highest 56 bits of the new REG rtx moved
+   from the (const_int -2).
+
+   Then the underlying expanding can perform the code generation based on
+   the REG rtx of Xmode,  instead of taking care of these in expand func.  */
+
+static rtx
+riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
+{
+  if (!CONST_INT_P (x))
+    return gen_lowpart (Xmode, x);
+
+  rtx xmode_x = gen_reg_rtx (Xmode);
+  HOST_WIDE_INT cst = INTVAL (x);
+
+  emit_move_insn (xmode_x, x);
+
+  int xmode_bits = GET_MODE_BITSIZE (Xmode);
+  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
+
+  if (cst < 0 && mode_bits < xmode_bits)
+    {
+      int shift_bits = xmode_bits - mode_bits;
+
+       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
+       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
+    }
+
+  return xmode_x;
+}
+
 /* Implements the unsigned saturation sub standard name usadd for int mode.
 
    z = SAT_SUB(x, y).
@@ -11625,7 +11674,7 @@ void
 riscv_expand_ussub (rtx dest, rtx x, rtx y)
 {
   machine_mode mode = GET_MODE (dest);
-  rtx xmode_x = gen_lowpart (Xmode, x);
+  rtx xmode_x = riscv_gen_unsigned_xmode_reg (x, mode);
   rtx xmode_y = gen_lowpart (Xmode, y);
   rtx xmode_lt = gen_reg_rtx (Xmode);
   rtx xmode_minus = gen_reg_rtx (Xmode);
diff --git a/gcc/config/riscv/riscv.md b/gcc/config/riscv/riscv.md
index d9f6c1765d0..7f57c0c3c89 100644
--- a/gcc/config/riscv/riscv.md
+++ b/gcc/config/riscv/riscv.md
@@ -4306,7 +4306,7 @@ (define_expand "usadd<mode>3"
 
 (define_expand "ussub<mode>3"
   [(match_operand:ANYI 0 "register_operand")
-   (match_operand:ANYI 1 "register_operand")
+   (match_operand:ANYI 1 "reg_or_int_operand")
    (match_operand:ANYI 2 "register_operand")]
   ""
   {
diff --git a/gcc/testsuite/gcc.target/riscv/sat_arith.h b/gcc/testsuite/gcc.target/riscv/sat_arith.h
index 37e0a60f21b..047dd40197c 100644
--- a/gcc/testsuite/gcc.target/riscv/sat_arith.h
+++ b/gcc/testsuite/gcc.target/riscv/sat_arith.h
@@ -201,6 +201,13 @@ sat_u_sub_##T##_fmt_12 (T x, T y)                      \
   return !overflow ? ret : 0;                          \
 }
 
+#define DEF_SAT_U_SUB_IMM_FMT_1(T, IMM) \
+T __attribute__((noinline))             \
+sat_u_sub_imm##IMM##_##T##_fmt_1 (T y)  \
+{                                       \
+  return (T)IMM >= y ? (T)IMM - y : 0;  \
+}
+
 #define RUN_SAT_U_SUB_FMT_1(T, x, y) sat_u_sub_##T##_fmt_1(x, y)
 #define RUN_SAT_U_SUB_FMT_2(T, x, y) sat_u_sub_##T##_fmt_2(x, y)
 #define RUN_SAT_U_SUB_FMT_3(T, x, y) sat_u_sub_##T##_fmt_3(x, y)
@@ -214,6 +221,9 @@ sat_u_sub_##T##_fmt_12 (T x, T y)                      \
 #define RUN_SAT_U_SUB_FMT_11(T, x, y) sat_u_sub_##T##_fmt_11(x, y)
 #define RUN_SAT_U_SUB_FMT_12(T, x, y) sat_u_sub_##T##_fmt_12(x, y)
 
+#define RUN_SAT_U_SUB_IMM_FMT_1(T, IMM, y, expect) \
+  if (sat_u_sub_imm##IMM##_##T##_fmt_1(y) != expect) __builtin_abort ()
+
 /******************************************************************************/
 /* Saturation Truncate (unsigned and signed)                                  */
 /******************************************************************************/
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
new file mode 100644
index 00000000000..3e6fbfe6aca
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm11_uint8_t_fmt_1:
+** li\s+[atx][0-9]+,\s*11
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** andi\s+a0,\s*a0,\s*0xff
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 11)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
new file mode 100644
index 00000000000..586f58a93fc
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm128_uint8_t_fmt_1:
+** li\s+[atx][0-9]+,\s*128
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** andi\s+a0,\s*a0,\s*0xff
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 128)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
new file mode 100644
index 00000000000..463a3ad9578
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm253_uint8_t_fmt_1:
+** li\s+[atx][0-9]+,\s*253
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** andi\s+a0,\s*a0,\s*0xff
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 253)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
new file mode 100644
index 00000000000..df38e126277
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm6_uint16_t_fmt_1:
+** li\s+[atx][0-9]+,\s*6
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** slli\s+a0,\s*a0,\s*48
+** srli\s+a0,\s*a0,\s*48
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 6)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
new file mode 100644
index 00000000000..b714e46300b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm32768_uint16_t_fmt_1:
+** li\s+[atx][0-9]+,\s*32768
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** slli\s+a0,\s*a0,\s*48
+** srli\s+a0,\s*a0,\s*48
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 32768)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
new file mode 100644
index 00000000000..55fe313868e
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
@@ -0,0 +1,22 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm65533_uint16_t_fmt_1:
+** li\s+[atx][0-9]+,\s*65536
+** addi\s+[atx][0-9]+,\s*[atx][0-9]+,\s*-3
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** slli\s+a0,\s*a0,\s*48
+** srli\s+a0,\s*a0,\s*48
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 65533)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
new file mode 100644
index 00000000000..b5c31625868
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm255_uint32_t_fmt_1:
+** li\s+[atx][0-9]+,\s*255
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** sext\.w\s+a0,\s*a0
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 255)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
new file mode 100644
index 00000000000..9762a1fad6d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm2147483648_uint32_t_fmt_1:
+** li\s+[atx][0-9]+,\s*1
+** slli\s+[atx][0-9]+,\s*[atx][0-9]+,\s*31
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** sext\.w\s+a0,\s*a0
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 2147483648)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
new file mode 100644
index 00000000000..bd140d22ec0
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
@@ -0,0 +1,22 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm68719476732_uint32_t_fmt_1:
+** li\s+[atx][0-9]+,\s*1
+** slli\s+[atx][0-9]+,\s*[atx][0-9]+,\s*32
+** addi\s+[atx][0-9]+,\s*[atx][0-9]+,\s*-4
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** sext\.w\s+a0,\s*a0
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 68719476732)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
new file mode 100644
index 00000000000..b9f796bb9aa
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
@@ -0,0 +1,19 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc -mabi=lp64d -O3 -fdump-rtl-expand-details -fno-schedule-insns -fno-schedule-insns2" } */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "sat_arith.h"
+
+/*
+** sat_u_sub_imm82_uint64_t_fmt_1:
+** li\s+[atx][0-9]+,\s*82
+** sub\s+[atx][0-9]+,\s*[atx][0-9]+,\s*a0
+** sltu\s+[atx][0-9]+,\s*[atx][0-9]+,\s*[atx][0-9]+
+** addi\s+a0,\s*a0,\s*-1
+** and\s+a0,\s*a0,\s*[atx][0-9]+
+** ret
+*/
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 82)
+
+/* { dg-final { scan-rtl-dump-times ".SAT_SUB " 2 "expand" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
new file mode 100644
index 00000000000..5091872d208
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
@@ -0,0 +1,56 @@
+/* { dg-do run { target { riscv_v } } } */
+/* { dg-additional-options "-std=c99" } */
+
+#include "sat_arith.h"
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 0)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 1)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 5)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 127)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 128)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 254)
+DEF_SAT_U_SUB_IMM_FMT_1(uint8_t, 255)
+
+#define T                       uint8_t
+#define RUN(T, imm, op, expect) RUN_SAT_U_SUB_IMM_FMT_1(T, imm, op, expect)
+
+T d[][3] = {
+  /* arg_0, arg_1, expect */
+  {      0,     0,      0, },
+  {      1,     0,      1, },
+  {      1,   255,      0, },
+  {    254,   254,      0, },
+  {    254,   255,      0, },
+  {    254,     2,    252, },
+  {    255,   254,      1, },
+  {    255,     0,    255, },
+  {      5,     2,      3, },
+  {      5,     6,      0, },
+  {    127,     0,    127, },
+  {    128,   129,      0, },
+};
+
+int
+main ()
+{
+  RUN (T,   0, d[0][1], d[0][2]);
+
+  RUN (T,   1, d[1][1], d[1][2]);
+  RUN (T,   1, d[2][1], d[2][2]);
+
+  RUN (T, 254, d[3][1], d[3][2]);
+  RUN (T, 254, d[4][1], d[4][2]);
+  RUN (T, 254, d[5][1], d[5][2]);
+
+  RUN (T, 255, d[6][1], d[6][2]);
+  RUN (T, 255, d[7][1], d[7][2]);
+
+  RUN (T,   5, d[8][1], d[8][2]);
+  RUN (T,   5, d[9][1], d[9][2]);
+
+  RUN (T, 127, d[10][1], d[10][2]);
+
+  RUN (T, 128, d[11][1], d[11][2]);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
new file mode 100644
index 00000000000..2bc3be3efc5
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
@@ -0,0 +1,56 @@
+/* { dg-do run { target { riscv_v } } } */
+/* { dg-additional-options "-std=c99" } */
+
+#include "sat_arith.h"
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 0)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 1)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 5)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 32767)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 32768)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 65534)
+DEF_SAT_U_SUB_IMM_FMT_1(uint16_t, 65535)
+
+#define T                       uint16_t
+#define RUN(T, imm, op, expect) RUN_SAT_U_SUB_IMM_FMT_1(T, imm, op, expect)
+
+T d[][3] = {
+  /* arg_0, arg_1, expect */
+  {        0,     0,      0, },
+  {        1,     0,      1, },
+  {        1, 65535,      0, },
+  {    65534, 65534,      0, },
+  {    65534, 65535,      0, },
+  {    65534,     2,  65532, },
+  {    65535, 65534,      1, },
+  {    65535,     0,  65535, },
+  {        5,     2,      3, },
+  {        5,     6,      0, },
+  {    32767,     0,  32767, },
+  {    32768, 32767,      1, },
+};
+
+int
+main ()
+{
+  RUN (T,     0, d[0][1], d[0][2]);
+
+  RUN (T,     1, d[1][1], d[1][2]);
+  RUN (T,     1, d[2][1], d[2][2]);
+
+  RUN (T, 65534, d[3][1], d[3][2]);
+  RUN (T, 65534, d[4][1], d[4][2]);
+  RUN (T, 65534, d[5][1], d[5][2]);
+
+  RUN (T, 65535, d[6][1], d[6][2]);
+  RUN (T, 65535, d[7][1], d[7][2]);
+
+  RUN (T,     5, d[8][1], d[8][2]);
+  RUN (T,     5, d[9][1], d[9][2]);
+
+  RUN (T, 32767, d[10][1], d[10][2]);
+
+  RUN (T, 32768, d[11][1], d[11][2]);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
new file mode 100644
index 00000000000..b1d1ee33135
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
@@ -0,0 +1,55 @@
+/* { dg-do run { target { riscv_v } } } */
+/* { dg-additional-options "-std=c99" } */
+
+#include "sat_arith.h"
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 0)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 1)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 5)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 2147483647)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 2147483648)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 4294967294)
+DEF_SAT_U_SUB_IMM_FMT_1(uint32_t, 4294967295)
+
+#define T                       uint32_t
+#define RUN(T, imm, op, expect) RUN_SAT_U_SUB_IMM_FMT_1(T, imm, op, expect)
+
+T d[][3] = {
+  /* arg_0, arg_1, expect */
+  {          0,          0,          0, },
+  {          1,          0,          1, },
+  {          1, 4294967295,          0, },
+  { 4294967294, 4294967294,          0, },
+  { 4294967294, 4294967295,          0, },
+  { 4294967294,          2, 4294967292, },
+  { 4294967295, 4294967294,          1, },
+  { 4294967295,          0, 4294967295, },
+  {          5,          2,          3, },
+  {          5,          6,          0, },
+  { 2147483647,          0, 2147483647, },
+  { 2147483648, 2147483647,          1, },
+};
+
+int
+main ()
+{
+  RUN (T,          0, d[0][1], d[0][2]);
+
+  RUN (T,          1, d[1][1], d[1][2]);
+  RUN (T,          1, d[2][1], d[2][2]);
+
+  RUN (T, 4294967294, d[3][1], d[3][2]);
+  RUN (T, 4294967294, d[4][1], d[4][2]);
+  RUN (T, 4294967294, d[5][1], d[5][2]);
+
+  RUN (T, 4294967295, d[6][1], d[6][2]);
+  RUN (T, 4294967295, d[7][1], d[7][2]);
+
+  RUN (T,          5, d[8][1], d[8][2]);
+  RUN (T,          5, d[9][1], d[9][2]);
+
+  RUN (T, 2147483647, d[10][1], d[10][2]);
+  RUN (T, 2147483648, d[11][1], d[11][2]);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c
new file mode 100644
index 00000000000..2539d753c11
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c
@@ -0,0 +1,48 @@
+/* { dg-do run { target { riscv_v } } } */
+/* { dg-additional-options "-std=c99" } */
+
+#include "sat_arith.h"
+
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 0)
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 1)
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 5)
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 18446744073709551614u)
+DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 18446744073709551615u)
+
+#define T                       uint64_t
+#define RUN(T, imm, op, expect) RUN_SAT_U_SUB_IMM_FMT_1(T, imm, op, expect)
+
+T d[][3] = {
+  /* arg_0, arg_1, expect */
+  {                     0,                     0,                     0, },
+  {                     1,                     0,                     1, },
+  {                     1, 18446744073709551615u,                     0, },
+  { 18446744073709551614u, 18446744073709551614u,                     0, },
+  { 18446744073709551614u, 18446744073709551615u,                     0, },
+  { 18446744073709551614u,                     2, 18446744073709551612u, },
+  { 18446744073709551615u, 18446744073709551614u,                     1, },
+  { 18446744073709551615u,                     0, 18446744073709551615u, },
+  {                     5,                     2,                     3, },
+  {                     5,                     6,                     0, },
+};
+
+int
+main ()
+{
+  RUN (T,                     0, d[0][1], d[0][2]);
+
+  RUN (T,                     1, d[1][1], d[1][2]);
+  RUN (T,                     1, d[2][1], d[2][2]);
+
+  RUN (T, 18446744073709551614u, d[3][1], d[3][2]);
+  RUN (T, 18446744073709551614u, d[4][1], d[4][2]);
+  RUN (T, 18446744073709551614u, d[5][1], d[5][2]);
+
+  RUN (T, 18446744073709551615u, d[6][1], d[6][2]);
+  RUN (T, 18446744073709551615u, d[7][1], d[7][2]);
+
+  RUN (T,                     5, d[8][1], d[8][2]);
+  RUN (T,                     5, d[9][1], d[9][2]);
+
+  return 0;
+}
-- 
2.43.0
  

On 8/13/24 8:23 PM, Li, Pan2 wrote:
> This Patch may requires rebase, will send v3 for conflict resolving.
> 
> Pan
> 
> -----Original Message-----
> From: Li, Pan2 <pan2.li@intel.com>
> Sent: Sunday, August 4, 2024 7:48 PM
> To: gcc-patches@gcc.gnu.org
> Cc: juzhe.zhong@rivai.ai; kito.cheng@gmail.com; jeffreyalaw@gmail.com; rdapp.gcc@gmail.com; Li, Pan2 <pan2.li@intel.com>
> Subject: [PATCH v2] RISC-V: Support IMM for operand 0 of ussub pattern
> 
> From: Pan Li <pan2.li@intel.com>
> 
> This patch would like to allow IMM for the operand 0 of ussub pattern.
> Aka .SAT_SUB(1023, y) as the below example.
> 
> Form 1:
>    #define DEF_SAT_U_SUB_IMM_FMT_1(T, IMM) \
>    T __attribute__((noinline))             \
>    sat_u_sub_imm##IMM##_##T##_fmt_1 (T y)  \
>    {                                       \
>      return (T)IMM >= y ? (T)IMM - y : 0;  \
>    }
> 
> DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 1023)
> 
> Before this patch:
>    10   │ sat_u_sub_imm82_uint64_t_fmt_1:
>    11   │     li  a5,82
>    12   │     bgtu    a0,a5,.L3
>    13   │     sub a0,a5,a0
>    14   │     ret
>    15   │ .L3:
>    16   │     li  a0,0
>    17   │     ret
> 
> After this patch:
>    10   │ sat_u_sub_imm82_uint64_t_fmt_1:
>    11   │     li  a5,82
>    12   │     sltu    a4,a5,a0
>    13   │     addi    a4,a4,-1
>    14   │     sub a0,a5,a0
>    15   │     and a0,a4,a0
>    16   │     ret
> 
> The below test suites are passed for this patch:
> 1. The rv64gcv fully regression test.
> 
> gcc/ChangeLog:
> 
> 	* config/riscv/riscv.cc (riscv_gen_unsigned_xmode_reg): Add new
> 	func impl to gen xmode rtx reg from operand rtx.
> 	(riscv_expand_ussub): Gen xmode reg for operand 1.
> 	* config/riscv/riscv.md: Allow const_int for operand 1.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.target/riscv/sat_arith.h: Add test helper macro.
> 	* gcc.target/riscv/sat_u_sub_imm-1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-1_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-1_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-4.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-3.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-4.c: New test.
> 
> Signed-off-by: Pan Li <pan2.li@intel.com>
> ---
>   gcc/config/riscv/riscv.cc                     | 51 ++++++++++++++++-
>   gcc/config/riscv/riscv.md                     |  2 +-
>   gcc/testsuite/gcc.target/riscv/sat_arith.h    | 10 ++++
>   .../gcc.target/riscv/sat_u_sub_imm-1.c        | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-1_1.c      | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-1_2.c      | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2.c        | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2_1.c      | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2_2.c      | 22 ++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3.c        | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3_1.c      | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3_2.c      | 22 ++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-4.c        | 19 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-1.c    | 56 +++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-2.c    | 56 +++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-3.c    | 55 ++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-4.c    | 48 ++++++++++++++++
>   17 files changed, 482 insertions(+), 2 deletions(-)
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c
> 
> diff --git a/gcc/config/riscv/riscv.cc b/gcc/config/riscv/riscv.cc
> index b19d56149e7..5e4e9722729 100644
> --- a/gcc/config/riscv/riscv.cc
> +++ b/gcc/config/riscv/riscv.cc
> @@ -11612,6 +11612,55 @@ riscv_expand_usadd (rtx dest, rtx x, rtx y)
>     emit_move_insn (dest, gen_lowpart (mode, xmode_dest));
>   }
>   
> +/* Generate a REG rtx of Xmode from the given rtx and mode.
> +   The rtx x can be REG (QI/HI/SI/DI) or const_int.
> +   The machine_mode mode is the original mode from define pattern.
> +
> +   If rtx is REG,  the gen_lowpart of Xmode will be returned.
> +
> +   If rtx is const_int,  a new REG rtx will be created to hold the value of
> +   const_int and then returned.
> +
> +   According to the gccint doc, the constants generated for modes with fewer
> +   bits than in HOST_WIDE_INT must be sign extended to full width.  Thus there
> +   will be two cases here,  take QImode as example.
> +
> +   For .SAT_SUB (127, y) in QImode, we have (const_int 127) and one simple
> +   mov from const_int to the new REG rtx is good enough here.
> +
> +   For .SAT_SUB (254, y) in QImode, we have (const_int -2) after define_expand.
> +   Aka 0xfffffffffffffffe in Xmode of RV64 but we actually need 0xfe in Xmode
> +   of RV64.  So we need to cleanup the highest 56 bits of the new REG rtx moved
> +   from the (const_int -2).
> +
> +   Then the underlying expanding can perform the code generation based on
> +   the REG rtx of Xmode,  instead of taking care of these in expand func.  */
> +
> +static rtx
> +riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
> +{
> +  if (!CONST_INT_P (x))
> +    return gen_lowpart (Xmode, x);
> +
> +  rtx xmode_x = gen_reg_rtx (Xmode);
> +  HOST_WIDE_INT cst = INTVAL (x);
> +
> +  emit_move_insn (xmode_x, x);
> +
> +  int xmode_bits = GET_MODE_BITSIZE (Xmode);
> +  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
> +
> +  if (cst < 0 && mode_bits < xmode_bits)
> +    {
> +      int shift_bits = xmode_bits - mode_bits;
> +
> +       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
> +       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
> +    }
Isn't this a zero_extension?
  

>> +static rtx
>> +riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
>> +{
>> +  if (!CONST_INT_P (x))
>> +    return gen_lowpart (Xmode, x);
>> +
>> +  rtx xmode_x = gen_reg_rtx (Xmode);
>> +  HOST_WIDE_INT cst = INTVAL (x);
>> +
>> +  emit_move_insn (xmode_x, x);
>> +
>> +  int xmode_bits = GET_MODE_BITSIZE (Xmode);
>> +  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
>> +
>> +  if (cst < 0 && mode_bits < xmode_bits)
>> +    {
>> +      int shift_bits = xmode_bits - mode_bits;
>> +
>> +       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
>> +       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
>> +    }
> Isn't this a zero_extension?

I am not sure it is valid for zero_extend, given the incoming rtx x is const_int which is DImode(integer promoted)
for ussub<qi>. 
I will rebase this patch after PR116278 commit, and give a try for this.

Pan

-----Original Message-----
From: Jeff Law <jeffreyalaw@gmail.com> 
Sent: Wednesday, August 14, 2024 11:33 AM
To: Li, Pan2 <pan2.li@intel.com>; gcc-patches@gcc.gnu.org
Cc: juzhe.zhong@rivai.ai; kito.cheng@gmail.com; rdapp.gcc@gmail.com
Subject: Re: [PATCH v2] RISC-V: Support IMM for operand 0 of ussub pattern



On 8/13/24 8:23 PM, Li, Pan2 wrote:
> This Patch may requires rebase, will send v3 for conflict resolving.
> 
> Pan
> 
> -----Original Message-----
> From: Li, Pan2 <pan2.li@intel.com>
> Sent: Sunday, August 4, 2024 7:48 PM
> To: gcc-patches@gcc.gnu.org
> Cc: juzhe.zhong@rivai.ai; kito.cheng@gmail.com; jeffreyalaw@gmail.com; rdapp.gcc@gmail.com; Li, Pan2 <pan2.li@intel.com>
> Subject: [PATCH v2] RISC-V: Support IMM for operand 0 of ussub pattern
> 
> From: Pan Li <pan2.li@intel.com>
> 
> This patch would like to allow IMM for the operand 0 of ussub pattern.
> Aka .SAT_SUB(1023, y) as the below example.
> 
> Form 1:
>    #define DEF_SAT_U_SUB_IMM_FMT_1(T, IMM) \
>    T __attribute__((noinline))             \
>    sat_u_sub_imm##IMM##_##T##_fmt_1 (T y)  \
>    {                                       \
>      return (T)IMM >= y ? (T)IMM - y : 0;  \
>    }
> 
> DEF_SAT_U_SUB_IMM_FMT_1(uint64_t, 1023)
> 
> Before this patch:
>    10   │ sat_u_sub_imm82_uint64_t_fmt_1:
>    11   │     li  a5,82
>    12   │     bgtu    a0,a5,.L3
>    13   │     sub a0,a5,a0
>    14   │     ret
>    15   │ .L3:
>    16   │     li  a0,0
>    17   │     ret
> 
> After this patch:
>    10   │ sat_u_sub_imm82_uint64_t_fmt_1:
>    11   │     li  a5,82
>    12   │     sltu    a4,a5,a0
>    13   │     addi    a4,a4,-1
>    14   │     sub a0,a5,a0
>    15   │     and a0,a4,a0
>    16   │     ret
> 
> The below test suites are passed for this patch:
> 1. The rv64gcv fully regression test.
> 
> gcc/ChangeLog:
> 
> 	* config/riscv/riscv.cc (riscv_gen_unsigned_xmode_reg): Add new
> 	func impl to gen xmode rtx reg from operand rtx.
> 	(riscv_expand_ussub): Gen xmode reg for operand 1.
> 	* config/riscv/riscv.md: Allow const_int for operand 1.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.target/riscv/sat_arith.h: Add test helper macro.
> 	* gcc.target/riscv/sat_u_sub_imm-1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-1_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-1_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-2_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3_1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-3_2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-4.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-1.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-2.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-3.c: New test.
> 	* gcc.target/riscv/sat_u_sub_imm-run-4.c: New test.
> 
> Signed-off-by: Pan Li <pan2.li@intel.com>
> ---
>   gcc/config/riscv/riscv.cc                     | 51 ++++++++++++++++-
>   gcc/config/riscv/riscv.md                     |  2 +-
>   gcc/testsuite/gcc.target/riscv/sat_arith.h    | 10 ++++
>   .../gcc.target/riscv/sat_u_sub_imm-1.c        | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-1_1.c      | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-1_2.c      | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2.c        | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2_1.c      | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-2_2.c      | 22 ++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3.c        | 20 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3_1.c      | 21 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-3_2.c      | 22 ++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-4.c        | 19 +++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-1.c    | 56 +++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-2.c    | 56 +++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-3.c    | 55 ++++++++++++++++++
>   .../gcc.target/riscv/sat_u_sub_imm-run-4.c    | 48 ++++++++++++++++
>   17 files changed, 482 insertions(+), 2 deletions(-)
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-1_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-2_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-3_2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-4.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-1.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-2.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-3.c
>   create mode 100644 gcc/testsuite/gcc.target/riscv/sat_u_sub_imm-run-4.c
> 
> diff --git a/gcc/config/riscv/riscv.cc b/gcc/config/riscv/riscv.cc
> index b19d56149e7..5e4e9722729 100644
> --- a/gcc/config/riscv/riscv.cc
> +++ b/gcc/config/riscv/riscv.cc
> @@ -11612,6 +11612,55 @@ riscv_expand_usadd (rtx dest, rtx x, rtx y)
>     emit_move_insn (dest, gen_lowpart (mode, xmode_dest));
>   }
>   
> +/* Generate a REG rtx of Xmode from the given rtx and mode.
> +   The rtx x can be REG (QI/HI/SI/DI) or const_int.
> +   The machine_mode mode is the original mode from define pattern.
> +
> +   If rtx is REG,  the gen_lowpart of Xmode will be returned.
> +
> +   If rtx is const_int,  a new REG rtx will be created to hold the value of
> +   const_int and then returned.
> +
> +   According to the gccint doc, the constants generated for modes with fewer
> +   bits than in HOST_WIDE_INT must be sign extended to full width.  Thus there
> +   will be two cases here,  take QImode as example.
> +
> +   For .SAT_SUB (127, y) in QImode, we have (const_int 127) and one simple
> +   mov from const_int to the new REG rtx is good enough here.
> +
> +   For .SAT_SUB (254, y) in QImode, we have (const_int -2) after define_expand.
> +   Aka 0xfffffffffffffffe in Xmode of RV64 but we actually need 0xfe in Xmode
> +   of RV64.  So we need to cleanup the highest 56 bits of the new REG rtx moved
> +   from the (const_int -2).
> +
> +   Then the underlying expanding can perform the code generation based on
> +   the REG rtx of Xmode,  instead of taking care of these in expand func.  */
> +
> +static rtx
> +riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
> +{
> +  if (!CONST_INT_P (x))
> +    return gen_lowpart (Xmode, x);
> +
> +  rtx xmode_x = gen_reg_rtx (Xmode);
> +  HOST_WIDE_INT cst = INTVAL (x);
> +
> +  emit_move_insn (xmode_x, x);
> +
> +  int xmode_bits = GET_MODE_BITSIZE (Xmode);
> +  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
> +
> +  if (cst < 0 && mode_bits < xmode_bits)
> +    {
> +      int shift_bits = xmode_bits - mode_bits;
> +
> +       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
> +       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
> +    }
Isn't this a zero_extension?
  

On 8/13/24 9:47 PM, Li, Pan2 wrote:
>>> +static rtx
>>> +riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
>>> +{
>>> +  if (!CONST_INT_P (x))
>>> +    return gen_lowpart (Xmode, x);
>>> +
>>> +  rtx xmode_x = gen_reg_rtx (Xmode);
>>> +  HOST_WIDE_INT cst = INTVAL (x);
>>> +
>>> +  emit_move_insn (xmode_x, x);
>>> +
>>> +  int xmode_bits = GET_MODE_BITSIZE (Xmode);
>>> +  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
>>> +
>>> +  if (cst < 0 && mode_bits < xmode_bits)
>>> +    {
>>> +      int shift_bits = xmode_bits - mode_bits;
>>> +
>>> +       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
>>> +       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
>>> +    }
>> Isn't this a zero_extension?
> 
> I am not sure it is valid for zero_extend, given the incoming rtx x is const_int which is DImode(integer promoted)
> for ussub<qi>.
> I will rebase this patch after PR116278 commit, and give a try for this.
But you're shifting a REG, not a CONST_INT.

Jeff
  

> But you're shifting a REG, not a CONST_INT.
I see, we can make a QImode REG to be moved to, and then zero_extend.
Thanks Jeff for enlightening me, and will send v3 for this.

Pan

-----Original Message-----
From: Jeff Law <jeffreyalaw@gmail.com> 
Sent: Wednesday, August 14, 2024 11:52 AM
To: Li, Pan2 <pan2.li@intel.com>; gcc-patches@gcc.gnu.org
Cc: juzhe.zhong@rivai.ai; kito.cheng@gmail.com; rdapp.gcc@gmail.com
Subject: Re: [PATCH v2] RISC-V: Support IMM for operand 0 of ussub pattern



On 8/13/24 9:47 PM, Li, Pan2 wrote:
>>> +static rtx
>>> +riscv_gen_unsigned_xmode_reg (rtx x, machine_mode mode)
>>> +{
>>> +  if (!CONST_INT_P (x))
>>> +    return gen_lowpart (Xmode, x);
>>> +
>>> +  rtx xmode_x = gen_reg_rtx (Xmode);
>>> +  HOST_WIDE_INT cst = INTVAL (x);
>>> +
>>> +  emit_move_insn (xmode_x, x);
>>> +
>>> +  int xmode_bits = GET_MODE_BITSIZE (Xmode);
>>> +  int mode_bits = GET_MODE_BITSIZE (mode).to_constant ();
>>> +
>>> +  if (cst < 0 && mode_bits < xmode_bits)
>>> +    {
>>> +      int shift_bits = xmode_bits - mode_bits;
>>> +
>>> +       riscv_emit_binary (ASHIFT, xmode_x, xmode_x, GEN_INT (shift_bits));
>>> +       riscv_emit_binary (LSHIFTRT, xmode_x, xmode_x, GEN_INT (shift_bits));
>>> +    }
>> Isn't this a zero_extension?
> 
> I am not sure it is valid for zero_extend, given the incoming rtx x is const_int which is DImode(integer promoted)
> for ussub<qi>.
> I will rebase this patch after PR116278 commit, and give a try for this.
But you're shifting a REG, not a CONST_INT.

Jeff