riscv: Synthesize all 11-bit-rotate constants with rori

Some constants can be built up using LI+RORI instructions.
The current implementation requires one of the upper 32-bits
to be a zero bit, which is not neccesary.
Let's drop this requirement in order to be able to synthesize
a constant like 0xffffffff00ffffffL.

The tests for LI+RORI are made more strict to detect regression
in the calculation of the LI constant and the rotation amount.

Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>

gcc/ChangeLog:

	* config/riscv/riscv.cc (riscv_build_integer_1): Don't
	require one zero bit in the upper 32 bits for LI+RORI synthesis.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/xtheadbb-li-rotr.c: New tests.
	* gcc.target/riscv/zbb-li-rotr.c: Likewise.

3 files changed, 118 insertions, 12 deletions

diff --git a/gcc/config/riscv/riscv.cc b/gcc/config/riscv/riscv.cc
index ef63079..d8917d7 100644
--- a/gcc/config/riscv/riscv.cc
+++ b/gcc/config/riscv/riscv.cc
@@ -572,10 +572,9 @@ riscv_build_integer_1 (struct riscv_integer_op codes[RISCV_MAX_INTEGER_OPS],
       int trailing_ones = ctz_hwi (~value);
 
       /* If all bits are one except a few that are zero, and the zero bits
-	 are within a range of 11 bits, and at least one of the upper 32-bits
-	 is a zero, then we can generate a constant by loading a small
-	 negative constant and rotating.  */
-      if (leading_ones < 32
+	 are within a range of 11 bits, then we can synthesize a constant
+	 by loading a small negative constant and rotating.  */
+      if (leading_ones < 64
 	  && ((64 - leading_ones - trailing_ones) < 12))
 	{
 	  codes[0].code = UNKNOWN;
diff --git a/gcc/testsuite/gcc.target/riscv/xtheadbb-li-rotr.c b/gcc/testsuite/gcc.target/riscv/xtheadbb-li-rotr.c
index ecd5044..136dcb0 100644
--- a/gcc/testsuite/gcc.target/riscv/xtheadbb-li-rotr.c
+++ b/gcc/testsuite/gcc.target/riscv/xtheadbb-li-rotr.c
@@ -1,34 +1,88 @@
 /* { dg-do compile } */
 /* { dg-options "-march=rv64gc_xtheadbb" } */
+/* { dg-skip-if "" { *-*-* } {"-O0" "-Os" "-Og" "-Oz" "-flto" } } */
+/* { dg-final { check-function-bodies "**" "" } } */
 
+/*
+**li_th_srri_1:
+**     li	a[0-9]+,-18
+**     th.srri	a[0-9]+,a[0-9]+,21
+**     ret
+*/
 long
-li_rori (void)
+li_th_srri_1 (void)
 {
   return 0xffff77ffffffffffL;
 }
 
+/*
+**li_th_srri_2:
+**     li	a[0-9]+,-18
+**     th.srri	a[0-9]+,a[0-9]+,5
+**     ret
+*/
 long
-li_rori_2 (void)
+li_th_srri_2 (void)
 {
   return 0x77ffffffffffffffL;
 }
 
+/*
+**li_th_srri_3:
+**     li	a[0-9]+,-18
+**     th.srri	a[0-9]+,a[0-9]+,36
+**     ret
+*/
 long
-li_rori_3 (void)
+li_th_srri_3 (void)
 {
   return 0xfffffffeefffffffL;
 }
 
+/*
+**li_th_srri_4:
+**     li	a[0-9]+,-86
+**     th.srri	a[0-9]+,a[0-9]+,3
+**     ret
+*/
 long
-li_rori_4 (void)
+li_th_srri_4 (void)
 {
   return 0x5ffffffffffffff5L;
 }
 
+/*
+**li_th_srri_5:
+**     li	a[0-9]+,-86
+**     th.srri	a[0-9]+,a[0-9]+,4
+**     ret
+*/
 long
-li_rori_5 (void)
+li_th_srri_5 (void)
 {
   return 0xaffffffffffffffaL;
 }
 
-/* { dg-final { scan-assembler-times "th.srri\t" 5 } } */
+/*
+**li_th_srri_6:
+**     li	a[0-9]+,-256
+**     th.srri	a[0-9]+,a[0-9]+,40
+**     ret
+*/
+long
+li_th_srri_6 (void)
+{
+  return 0xffffffff00ffffffL;
+}
+
+/*
+**li_th_srri_7:
+**     li	a[0-9]+,-2048
+**     th.srri	a[0-9]+,a[0-9]+,16
+**     ret
+*/
+long
+li_th_srri_7 (void)
+{
+  return 0xf800ffffffffffffL;
+}
diff --git a/gcc/testsuite/gcc.target/riscv/zbb-li-rotr.c b/gcc/testsuite/gcc.target/riscv/zbb-li-rotr.c
index 500264a..594c3dc 100644
--- a/gcc/testsuite/gcc.target/riscv/zbb-li-rotr.c
+++ b/gcc/testsuite/gcc.target/riscv/zbb-li-rotr.c
@@ -1,35 +1,88 @@
 /* { dg-do compile } */
 /* { dg-options "-march=rv64gc_zbb -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } {"-O0" "-Os" "-Og" "-Oz" "-flto" } } */
+/* { dg-final { check-function-bodies "**" "" } } */
 
+/*
+**li_rori_1:
+**     li	a[0-9]+,-18
+**     rori	a[0-9]+,a[0-9]+,21
+**     ret
+*/
 long
-li_rori (void)
+li_rori_1 (void)
 {
   return 0xffff77ffffffffffL;
 }
 
+/*
+**li_rori_2:
+**     li	a[0-9]+,-18
+**     rori	a[0-9]+,a[0-9]+,5
+**     ret
+*/
 long
 li_rori_2 (void)
 {
   return 0x77ffffffffffffffL;
 }
 
+/*
+**li_rori_3:
+**     li	a[0-9]+,-18
+**     rori	a[0-9]+,a[0-9]+,36
+**     ret
+*/
 long
 li_rori_3 (void)
 {
   return 0xfffffffeefffffffL;
 }
 
+/*
+**li_rori_4:
+**     li	a[0-9]+,-86
+**     rori	a[0-9]+,a[0-9]+,3
+**     ret
+*/
 long
 li_rori_4 (void)
 {
   return 0x5ffffffffffffff5L;
 }
 
+/*
+**li_rori_5:
+**     li	a[0-9]+,-86
+**     rori	a[0-9]+,a[0-9]+,4
+**     ret
+*/
 long
 li_rori_5 (void)
 {
   return 0xaffffffffffffffaL;
 }
 
+/*
+**li_rori_6:
+**     li	a[0-9]+,-256
+**     rori	a[0-9]+,a[0-9]+,40
+**     ret
+*/
+long
+li_rori_6 (void)
+{
+  return 0xffffffff00ffffffL;
+}
 
-/* { dg-final { scan-assembler-times "rori\t" 5 } } */
+/*
+**li_rori_7:
+**     li	a[0-9]+,-2048
+**     rori	a[0-9]+,a[0-9]+,16
+**     ret
+*/
+long
+li_rori_7 (void)
+{
+  return 0xf800ffffffffffffL;
+}