[PATCH] rs6000: Fix some issues in rs6000_can_inline_p [PR102059]

["Kewen.Lin" <linkw@linux.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 1289 bytes --]

Hi!

This patch is to fix the inconsistent behaviors for non-LTO mode
and LTO mode.  As Martin pointed out, currently the function
rs6000_can_inline_p simply makes it inlinable if callee_tree is
NULL, but it's wrong, we should use the command line options
from target_option_default_node as default.  It also replaces
rs6000_isa_flags with the one from target_option_default_node
when caller_tree is NULL as rs6000_isa_flags could probably
change since initialization.

It also extends the scope of the check for the case that callee
has explicit set options, for test case pr102059-2.c inlining can
happen unexpectedly before, it's fixed accordingly.

As Richi/Mike pointed out, some tuning flags like MASK_P8_FUSION
can be neglected for inlining, this patch also exludes them when
the callee is attributed by always_inline.

Bootstrapped and regtested on powerpc64le-linux-gnu Power9.

BR,
Kewen
-----
gcc/ChangeLog:

	PR ipa/102059
	* config/rs6000/rs6000.c (rs6000_can_inline_p): Adjust with
	target_option_default_node and consider always_inline_safe flags.

gcc/testsuite/ChangeLog:

	PR ipa/102059
	* gcc.target/powerpc/pr102059-1.c: New test.
	* gcc.target/powerpc/pr102059-2.c: New test.
	* gcc.target/powerpc/pr102059-3.c: New test.
	* gcc.target/powerpc/pr102059-4.c: New test.

[-- Attachment #2: 0001-rs6000-Fix-some-issues-in-rs6000_can_inline_p-PR1020.patch --]
[-- Type: text/plain, Size: 9742 bytes --]

---
 gcc/config/rs6000/rs6000.c                    | 87 +++++++++++------
 gcc/testsuite/gcc.target/powerpc/pr102059-1.c | 24 +++++
 gcc/testsuite/gcc.target/powerpc/pr102059-2.c | 20 ++++
 gcc/testsuite/gcc.target/powerpc/pr102059-3.c | 95 +++++++++++++++++++
 gcc/testsuite/gcc.target/powerpc/pr102059-4.c | 22 +++++
 5 files changed, 221 insertions(+), 27 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr102059-1.c
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr102059-2.c
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr102059-3.c
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr102059-4.c

diff --git a/gcc/config/rs6000/rs6000.c b/gcc/config/rs6000/rs6000.c
index 46b8909104e..c2582a3efab 100644
--- a/gcc/config/rs6000/rs6000.c
+++ b/gcc/config/rs6000/rs6000.c
@@ -25058,45 +25058,78 @@ rs6000_generate_version_dispatcher_body (void *node_p)
 static bool
 rs6000_can_inline_p (tree caller, tree callee)
 {
-  bool ret = false;
   tree caller_tree = DECL_FUNCTION_SPECIFIC_TARGET (caller);
   tree callee_tree = DECL_FUNCTION_SPECIFIC_TARGET (callee);
 
-  /* If the callee has no option attributes, then it is ok to inline.  */
+  /* If the caller/callee has option attributes, then use them.
+     Otherwise, use the command line options.  */
   if (!callee_tree)
-    ret = true;
+    callee_tree = target_option_default_node;
+  if (!caller_tree)
+    caller_tree = target_option_default_node;
+
+  struct cl_target_option *caller_opts = TREE_TARGET_OPTION (caller_tree);
+  struct cl_target_option *callee_opts = TREE_TARGET_OPTION (callee_tree);
+  HOST_WIDE_INT caller_isa = caller_opts->x_rs6000_isa_flags;
+  HOST_WIDE_INT callee_isa = callee_opts->x_rs6000_isa_flags;
+
+  bool always_inline =
+    (DECL_DISREGARD_INLINE_LIMITS (callee)
+     && lookup_attribute ("always_inline", DECL_ATTRIBUTES (callee)));
+
+  /* Some flags such as fusion can be tolerated for always inlines.  */
+  unsigned HOST_WIDE_INT always_inline_safe_mask =
+    (MASK_P8_FUSION | MASK_P10_FUSION | OPTION_MASK_SAVE_TOC_INDIRECT
+     | OPTION_MASK_P8_FUSION_SIGN | OPTION_MASK_P10_FUSION_LD_CMPI
+     | OPTION_MASK_P10_FUSION_2LOGICAL | OPTION_MASK_P10_FUSION_LOGADD
+     | OPTION_MASK_P10_FUSION_ADDLOG | OPTION_MASK_P10_FUSION_2ADD
+     | OPTION_MASK_PCREL_OPT);
+
+  if (always_inline) {
+    caller_isa &= ~always_inline_safe_mask;
+    callee_isa &= ~always_inline_safe_mask;
+  }
 
-  else
+  /* The callee's options must be a subset of the caller's options, i.e.
+     a vsx function may inline an altivec function, but a no-vsx function
+     must not inline a vsx function.  */
+  if ((caller_isa & callee_isa) != callee_isa)
     {
-      HOST_WIDE_INT caller_isa;
-      struct cl_target_option *callee_opts = TREE_TARGET_OPTION (callee_tree);
-      HOST_WIDE_INT callee_isa = callee_opts->x_rs6000_isa_flags;
-      HOST_WIDE_INT explicit_isa = callee_opts->x_rs6000_isa_flags_explicit;
-
-      /* If the caller has option attributes, then use them.
-	 Otherwise, use the command line options.  */
-      if (caller_tree)
-	caller_isa = TREE_TARGET_OPTION (caller_tree)->x_rs6000_isa_flags;
-      else
-	caller_isa = rs6000_isa_flags;
+      if (TARGET_DEBUG_TARGET)
+	fprintf (stderr,
+		 "rs6000_can_inline_p:, caller %s, callee %s, cannot "
+		 "inline since callee's options set isn't a subset of "
+		 "caller's options set.\n",
+		 get_decl_name (caller), get_decl_name (callee));
+      return false;
+    }
 
-      /* The callee's options must be a subset of the caller's options, i.e.
-	 a vsx function may inline an altivec function, but a no-vsx function
-	 must not inline a vsx function.  However, for those options that the
-	 callee has explicitly enabled or disabled, then we must enforce that
-	 the callee's and caller's options match exactly; see PR70010.  */
-      if (((caller_isa & callee_isa) == callee_isa)
-	  && (caller_isa & explicit_isa) == (callee_isa & explicit_isa))
-	ret = true;
+  /* For those options that the callee has explicitly enabled or disabled,
+     then we must enforce that the callee's and caller's options match
+     exactly; see PR70010.  */
+  HOST_WIDE_INT explicit_isa = callee_opts->x_rs6000_isa_flags_explicit;
+  if (always_inline)
+    explicit_isa &= ~always_inline_safe_mask;
+  if ((caller_isa & explicit_isa) != (callee_isa & explicit_isa))
+    {
+      if (TARGET_DEBUG_TARGET)
+	fprintf (stderr,
+		 "rs6000_can_inline_p:, caller %s, callee %s, cannot "
+		 "inline since callee's options set isn't a subset of "
+		 "caller's options set by considering callee's "
+		 "explicitly set options.\n",
+		 get_decl_name (caller), get_decl_name (callee));
+      return false;
     }
 
   if (TARGET_DEBUG_TARGET)
-    fprintf (stderr, "rs6000_can_inline_p:, caller %s, callee %s, %s inline\n",
-	     get_decl_name (caller), get_decl_name (callee),
-	     (ret ? "can" : "cannot"));
+    fprintf (stderr,
+	     "rs6000_can_inline_p:, caller %s, callee %s, can inline.\n",
+	     get_decl_name (caller), get_decl_name (callee));
 
-  return ret;
+  return true;
 }
+
 \f
 /* Allocate a stack temp and fixup the address so it meets the particular
    memory requirements (either offetable or REG+REG addressing).  */
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-1.c b/gcc/testsuite/gcc.target/powerpc/pr102059-1.c
new file mode 100644
index 00000000000..d2a002cf141
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-1.c
@@ -0,0 +1,24 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target powerpc_htm_ok } */
+/* { dg-options "-O2 -mdejagnu-cpu=power8" } */
+
+/* Verify it emits inlining error msg at non-LTO mode.  */
+
+#include <htmintrin.h>
+
+static inline int __attribute__ ((always_inline))
+foo (int *b) /* { dg-error "inlining failed in call to .* target specific option mismatch" } */
+{
+  int res = _HTM_STATE(__builtin_ttest());
+  *b += res;
+  return *b;
+}
+
+#pragma GCC target "cpu=power10"
+int
+bar (int *a)
+{
+  *a = foo (a); /* { dg-message "called from here" } */
+  return 0;
+}
+
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-2.c b/gcc/testsuite/gcc.target/powerpc/pr102059-2.c
new file mode 100644
index 00000000000..1d5d6c38bf3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-2.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -mno-vsx" } */
+
+/* Verify it emits inlining error msg when the callee has explicit
+   disabling option from command line.  */
+
+vector int c, a, b;
+
+static inline void __attribute__ ((__always_inline__))
+foo () /* { dg-error "inlining failed in call to .* target specific option mismatch" } */
+{
+  c = a + b;
+}
+
+__attribute__ ((target ("vsx")))
+int main ()
+{
+  foo (); /* { dg-message "called from here" } */
+  c = a + b;
+}
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-3.c b/gcc/testsuite/gcc.target/powerpc/pr102059-3.c
new file mode 100644
index 00000000000..9684cab986a
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-3.c
@@ -0,0 +1,95 @@
+/* { dg-do compile } */
+/* -Wno-attributes suppresses always_inline warnings.  */
+/* { dg-options "-O2 -mdejagnu-cpu=power8 -Wno-attributes" } */
+
+/* Verify it doesn't emit inlining error msg since some mismatched
+   features are considered as safe for always_inline.  */
+
+/* 1. Callee enables Power8 fusion implicitly, while caller
+      with Power9 doesn't support power8 fusion at all.  */
+
+__attribute__ ((always_inline))
+int callee1 (int *b)
+{
+  *b += 1;
+  return *b;
+}
+
+#pragma GCC target "cpu=power9"
+int caller1 (int *a)
+{
+  *a = callee1 (a);
+  return 0;
+}
+
+/* 2. Caller enables indirect toc save feature while callee
+      disables it explicitly.  */
+
+#pragma GCC target "save-toc-indirect"
+__attribute__ ((always_inline))
+int callee2 (int *b)
+{
+  *b += 2;
+  return *b;
+}
+
+#pragma GCC target "no-save-toc-indirect"
+int caller2 (int *a)
+{
+  *a = callee2 (a);
+  return 0;
+}
+
+/* 3. Caller disables Power10 fusion explicitly, while callee
+      still supports it as Power10 turns it on by default.  */
+
+#pragma GCC target "cpu=power10"
+__attribute__ ((always_inline))
+int callee3 (int *b)
+{
+  *b += 3;
+  return *b;
+}
+
+#pragma GCC target "cpu=power10,no-power10-fusion"
+int caller3 (int *a)
+{
+  *a = callee3 (a);
+  return 0;
+}
+
+/* 4. Caller enables Power10 fusion implicitly, while callee
+      disables it explicitly.  */
+
+#pragma GCC target "no-power10-fusion"
+__attribute__ ((always_inline))
+int callee4 (int *b)
+{
+  *b += 4;
+  return *b;
+}
+
+#pragma GCC target "cpu=power10"
+int caller4 (int *a)
+{
+  *a = callee4 (a);
+  return 0;
+}
+
+/* 5. Caller disables pcrel-opt while callee enables it explicitly.  */
+
+#pragma GCC target "cpu=power10,no-pcrel-opt"
+__attribute__ ((always_inline))
+int callee5 (int *b)
+{
+  *b += 5;
+  return *b;
+}
+
+#pragma GCC target "cpu=power10,pcrel-opt"
+int caller5 (int *a)
+{
+  *a = callee5 (a);
+  return 0;
+}
+
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-4.c b/gcc/testsuite/gcc.target/powerpc/pr102059-4.c
new file mode 100644
index 00000000000..0f27f2ce7d7
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-4.c
@@ -0,0 +1,22 @@
+/* { dg-do compile } */
+/* -Wno-attributes suppresses always_inline warnings.  */
+/* { dg-options "-O2 -mdejagnu-cpu=power8 -mno-power8-fusion -Wno-attributes" } */
+
+/* Verify it doesn't emit inlining error msg since the flag power8
+   fusion is considered as safe for always_inline, it's still safe
+   even the flag is set explicitly.  */
+
+__attribute__ ((always_inline))
+int foo (int *b)
+{
+  *b += 10;
+  return *b;
+}
+
+#pragma GCC target "power8-fusion"
+int bar (int *a)
+{
+  *a = foo (a);
+  return 0;
+}
+
-- 
2.17.1