From 33dc778a34d7b93978efe922bb1b4583d8e6c4bb Mon Sep 17 00:00:00 2001
From: Roger Sayle <roger@nextmovesoftware.com>
Date: Mon, 2 Aug 2021 13:27:53 +0100
Subject: [PATCH 17/35] [Backport] Optimize x ? bswap(x) : 0 in tree-ssa-phiopt

Reference: https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=f9fcf754825a1e01033336f84c18690aaa971a6f

Many thanks again to Jakub Jelinek for a speedy fix for PR 101642.
Interestingly, that test case "bswap16(x) ? : x" also reveals a
missed optimization opportunity.  The resulting "x ? bswap(x) : 0"
can be further simplified to just bswap(x).

Conveniently, tree-ssa-phiopt.c already recognizes/optimizes the
related "x ? popcount(x) : 0", so this patch simply makes that
transformation make general, additionally handling bswap, parity,
ffs and clrsb.  All of the required infrastructure is already
present thanks to Jakub previously adding support for clz/ctz.
To reflect this generalization, the name of the function is changed
from cond_removal_in_popcount_clz_ctz_pattern to the hopefully
equally descriptive cond_removal_in_builtin_zero_pattern.

2021-08-02  Roger Sayle  <roger@nextmovesoftware.com>

gcc/ChangeLog
	* tree-ssa-phiopt.c (cond_removal_in_builtin_zero_pattern):
	Renamed from cond_removal_in_popcount_clz_ctz_pattern.
	Add support for BSWAP, FFS, PARITY and CLRSB builtins.
	(tree_ssa_phiop_worker): Update call to function above.

gcc/testsuite/ChangeLog
	* gcc.dg/tree-ssa/phi-opt-25.c: New test case.
---
 gcc/testsuite/gcc.dg/tree-ssa/phi-opt-25.c | 83 ++++++++++++++++++++++
 gcc/tree-ssa-phiopt.c                      | 37 +++++++---
 2 files changed, 109 insertions(+), 11 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/phi-opt-25.c

diff --git a/gcc/testsuite/gcc.dg/tree-ssa/phi-opt-25.c b/gcc/testsuite/gcc.dg/tree-ssa/phi-opt-25.c
new file mode 100644
index 000000000..c52c92e1d
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/phi-opt-25.c
@@ -0,0 +1,83 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-optimized" } */
+
+unsigned short test_bswap16(unsigned short x)
+{
+  return x ? __builtin_bswap16(x) : 0;
+}
+
+unsigned int test_bswap32(unsigned int x)
+{
+  return x ? __builtin_bswap32(x) : 0;
+}
+
+unsigned long long test_bswap64(unsigned long long x)
+{
+  return x ? __builtin_bswap64(x) : 0;
+}
+
+int test_clrsb(int x)
+{
+  return x ? __builtin_clrsb(x) : (__SIZEOF_INT__*8-1);
+}
+
+int test_clrsbl(long x)
+{
+  return x ? __builtin_clrsbl(x) : (__SIZEOF_LONG__*8-1);
+}
+
+int test_clrsbll(long long x)
+{
+  return x ? __builtin_clrsbll(x) : (__SIZEOF_LONG_LONG__*8-1);
+}
+
+#if 0
+/* BUILT_IN_FFS is transformed by match.pd */
+int test_ffs(unsigned int x)
+{
+  return x ? __builtin_ffs(x) : 0;
+}
+
+int test_ffsl(unsigned long x)
+{
+  return x ? __builtin_ffsl(x) : 0;
+}
+
+int test_ffsll(unsigned long long x)
+{
+  return x ? __builtin_ffsll(x) : 0;
+}
+#endif
+
+int test_parity(int x)
+{
+  return x ? __builtin_parity(x) : 0;
+}
+
+int test_parityl(long x)
+{
+  return x ? __builtin_parityl(x) : 0;
+}
+
+int test_parityll(long long x)
+{
+  return x ? __builtin_parityll(x) : 0;
+}
+
+int test_popcount(int x)
+{
+  return x ? __builtin_popcount(x) : 0;
+}
+
+int test_popcountl(long x)
+{
+  return x ? __builtin_popcountl(x) : 0;
+}
+
+int test_popcountll(long long x)
+{
+  return x ? __builtin_popcountll(x) : 0;
+}
+
+/* { dg-final { scan-tree-dump-not "goto" "optimized" } } */
+
diff --git a/gcc/tree-ssa-phiopt.c b/gcc/tree-ssa-phiopt.c
index 045a7b1b8..21ac08145 100644
--- a/gcc/tree-ssa-phiopt.c
+++ b/gcc/tree-ssa-phiopt.c
@@ -66,9 +66,9 @@ static bool abs_replacement (basic_block, basic_block,
 			     edge, edge, gphi *, tree, tree);
 static bool spaceship_replacement (basic_block, basic_block,
 				   edge, edge, gphi *, tree, tree);
-static bool cond_removal_in_popcount_clz_ctz_pattern (basic_block, basic_block,
-						      edge, edge, gphi *,
-						      tree, tree);
+static bool cond_removal_in_builtin_zero_pattern (basic_block, basic_block,
+						  edge, edge, gphi *,
+						  tree, tree);
 static bool cond_store_replacement (basic_block, basic_block, edge, edge,
 				    hash_set<tree> *);
 static bool cond_if_else_store_replacement (basic_block, basic_block, basic_block);
@@ -355,9 +355,8 @@ tree_ssa_phiopt_worker (bool do_store_elim, bool do_hoist_loads, bool early_p)
 	  else if (abs_replacement (bb, bb1, e1, e2, phi, arg0, arg1))
 	    cfgchanged = true;
 	  else if (!early_p
-		   && cond_removal_in_popcount_clz_ctz_pattern (bb, bb1, e1,
-								e2, phi, arg0,
-								arg1))
+		   && cond_removal_in_builtin_zero_pattern (bb, bb1, e1, e2,
+							    phi, arg0, arg1))
 	    cfgchanged = true;
 	  else if (minmax_replacement (bb, bb1, e1, e2, phi, arg0, arg1))
 	    cfgchanged = true;
@@ -2204,7 +2203,8 @@ spaceship_replacement (basic_block cond_bb, basic_block middle_bb,
   return true;
 }
 
-/* Convert
+/* Optimize x ? __builtin_fun (x) : C, where C is __builtin_fun (0).
+   Convert
 
    <bb 2>
    if (b_4(D) != 0)
@@ -2236,10 +2236,10 @@ spaceship_replacement (basic_block cond_bb, basic_block middle_bb,
    instead of 0 above it uses the value from that macro.  */
 
 static bool
-cond_removal_in_popcount_clz_ctz_pattern (basic_block cond_bb,
-					  basic_block middle_bb,
-					  edge e1, edge e2, gphi *phi,
-					  tree arg0, tree arg1)
+cond_removal_in_builtin_zero_pattern (basic_block cond_bb,
+				      basic_block middle_bb,
+				      edge e1, edge e2, gphi *phi,
+				      tree arg0, tree arg1)
 {
   gimple *cond;
   gimple_stmt_iterator gsi, gsi_from;
@@ -2287,6 +2287,12 @@ cond_removal_in_popcount_clz_ctz_pattern (basic_block cond_bb,
   int val = 0;
   switch (cfn)
     {
+    case CFN_BUILT_IN_BSWAP16:
+    case CFN_BUILT_IN_BSWAP32:
+    case CFN_BUILT_IN_BSWAP64:
+    case CFN_BUILT_IN_BSWAP128:
+    CASE_CFN_FFS:
+    CASE_CFN_PARITY:
     CASE_CFN_POPCOUNT:
       break;
     CASE_CFN_CLZ:
@@ -2315,6 +2321,15 @@ cond_removal_in_popcount_clz_ctz_pattern (basic_block cond_bb,
 	    }
 	}
       return false;
+    case BUILT_IN_CLRSB:
+      val = TYPE_PRECISION (integer_type_node) - 1;
+      break;
+    case BUILT_IN_CLRSBL:
+      val = TYPE_PRECISION (long_integer_type_node) - 1;
+      break;
+    case BUILT_IN_CLRSBLL:
+      val = TYPE_PRECISION (long_long_integer_type_node) - 1;
+      break;
     default:
       return false;
     }
-- 
2.27.0.windows.1