s390: Implement clz and ctz for SI mode

To properly implement __builtin_ffs for SI mode, implement clz and
(for >= z17) ctz for SI mode.  Otherwise, gcc falls back to a libcall
which causes problems for Linux kernel code.

Also adjust the C?Z_DEFINED_VALUE_AT_ZERO macros to return 2.  Since
the optabs now return exactly the value set by these macros, return
value 2 is more appropriate and leads to better code.

gcc/ChangeLog:

	* config/s390/s390.h (CLZ_DEFINED_VALUE_AT_ZERO): Adjust and
	return 2.
	(CTZ_DEFINED_VALUE_AT_ZERO): Return 2.
	* config/s390/s390.md (clzsi2): Implement.
	(ctzsi2): Implement.

gcc/testsuite/ChangeLog:

	* gcc.dg/vect/pr109011-2.c: Fix expected outcome.
	* gcc.dg/vect/pr109011-4.c: Fix expected outcome.
	* gcc.target/s390/ffs-1.c: New test.

Signed-off-by: Juergen Christ <jchrist@linux.ibm.com>

diff --git a/gcc/config/s390/s390.h b/gcc/config/s390/s390.h
index 8b04bc9a75579efb54eb639591c872b83a13ac71..6478be8c7acdd2aa1532b7d72289d9ac268022ca 100644 (file)
--- a/gcc/config/s390/s390.h
+++ b/gcc/config/s390/s390.h
@@ -1001,10 +1001,10 @@ do {                                                                    \
 #define FUNCTION_MODE QImode
 
 /* Specify the value which is used when clz operand is zero.  */
-#define CLZ_DEFINED_VALUE_AT_ZERO(MODE, VALUE) ((VALUE) = 64, 1)
+#define CLZ_DEFINED_VALUE_AT_ZERO(MODE, VALUE) ((VALUE) = GET_MODE_PRECISION (MODE), 2)
 
 /* Specify the value which is used when ctz operand is zero.  */
-#define CTZ_DEFINED_VALUE_AT_ZERO(MODE, VALUE) ((VALUE) = 64, 1)
+#define CTZ_DEFINED_VALUE_AT_ZERO(MODE, VALUE) ((VALUE) = 64, 2)
 
 /* Machine-specific symbol_ref flags.  */
 #define SYMBOL_FLAG_ALIGN_SHIFT          SYMBOL_FLAG_MACH_DEP_SHIFT

diff --git a/gcc/config/s390/s390.md b/gcc/config/s390/s390.md
index 858387cd85c5c7205f201b079dc0b6365895884a..06876a5563a9d7b03481030c42abe8f4acdc9911 100644 (file)
--- a/gcc/config/s390/s390.md
+++ b/gcc/config/s390/s390.md
@@ -9738,6 +9738,19 @@
   "flogr\t%0,%1"
   [(set_attr "op_type"  "RRE")])
 
+(define_expand "clzsi2"
+  [(match_operand:SI 0 "register_operand" "")
+   (match_operand:SI 1 "register_operand" "")]
+  "TARGET_EXTIMM && TARGET_ZARCH"
+{
+  rtx extreg = gen_reg_rtx (DImode);
+  rtx clzreg = gen_reg_rtx (DImode);
+  emit_insn (gen_zero_extendsidi2 (extreg, operands[1]));
+  emit_insn (gen_clzdi2 (clzreg, extreg));
+  rtx truncreg = gen_lowpart (SImode, clzreg);
+  emit_insn (gen_addsi3 (operands[0], truncreg, GEN_INT (-32)));
+  DONE;
+})
 
 ;;
 ;; Count Trailing Zeros.
@@ -9750,6 +9763,16 @@
   "ctzg\t%0,%1"
   [(set_attr "op_type" "RRE")])
 
+(define_expand "ctzsi2"
+  [(set (match_dup 2)
+       (zero_extend:DI (match_operand:SI 1 "register_operand" "")))
+   (set (match_dup 3) (ctz:DI (match_dup 2)))
+   (set (match_operand:SI 0 "register_operand" "") (subreg:SI (match_dup 3) 4))]
+  "TARGET_Z17 && TARGET_64BIT"
+{
+  operands[2] = gen_reg_rtx (DImode);
+  operands[3] = gen_reg_rtx (DImode);
+})
 
 ;;
 ;;- Rotate instructions.

diff --git a/gcc/testsuite/gcc.dg/vect/pr109011-2.c b/gcc/testsuite/gcc.dg/vect/pr109011-2.c
index 4c7e6ad07a46a9e6f2f0cf15d6cdb48d5fafd46b..dc62d01da5d914649ec800fc4683603685aca5c4 100644 (file)
--- a/gcc/testsuite/gcc.dg/vect/pr109011-2.c
+++ b/gcc/testsuite/gcc.dg/vect/pr109011-2.c
@@ -31,5 +31,4 @@ baz (int *p, int *q)
 
 /* { dg-final { scan-tree-dump-times " = \.CLZ \\\(vect" 3 "optimized" { target { { { { i?86-*-* x86_64-*-* } && avx512cd } && lzcnt } && bmi } } } } */
 /* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 3 "optimized" { target powerpc_vsx } } } */
-/* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 2 "optimized" { target s390_vx } } } */
-/* { dg-final { scan-tree-dump-times " = \.POPCOUNT \\\(vect" 1 "optimized" { target s390_vx } } } */
+/* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 3 "optimized" { target s390_vx } } } */

diff --git a/gcc/testsuite/gcc.dg/vect/pr109011-4.c b/gcc/testsuite/gcc.dg/vect/pr109011-4.c
index 38b2ab4d511e2c827f0ff05ba82ecbc093b210c7..8440ec73080a4defabc6b709e683ee7b10029f4a 100644 (file)
--- a/gcc/testsuite/gcc.dg/vect/pr109011-4.c
+++ b/gcc/testsuite/gcc.dg/vect/pr109011-4.c
@@ -31,5 +31,4 @@ baz (long long *p, long long *q)
 
 /* { dg-final { scan-tree-dump-times " = \.CLZ \\\(vect" 3 "optimized" { target { { { { i?86-*-* x86_64-*-* } && avx512cd } && lzcnt } && bmi } } } } */
 /* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 3 "optimized" { target powerpc_vsx } } } */
-/* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 2 "optimized" { target s390_vx } } } */
-/* { dg-final { scan-tree-dump-times " = \.POPCOUNT \\\(vect" 1 "optimized" { target s390_vx } } } */
+/* { dg-final { scan-tree-dump-times " = \.CTZ \\\(vect" 3 "optimized" { target s390_vx } } } */

diff --git a/gcc/testsuite/gcc.target/s390/ffs-1.c b/gcc/testsuite/gcc.target/s390/ffs-1.c
new file mode 100644 (file)
index 0000000..79774d2
--- /dev/null
+++ b/gcc/testsuite/gcc.target/s390/ffs-1.c
@@ -0,0 +1,18 @@
+/* Check that __builtin_ffs does not expand to libcall.  This is required by
+   Linux kernel code since libcalls are not present there.  */
+/* { dg-do compile } */
+/* { dg-options "-march=z10" } */
+
+long
+fool (long x)
+{
+  return __builtin_ffsl (x);
+}
+
+int
+foo (int x)
+{
+  return __builtin_ffs (x);
+}
+
+/* { dg-final { scan-assembler-not "brasl" } } */