Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Sender: Vincent Mailhol <vincent.mailhol@gmail.com>
From:   Vincent Mailhol <mailhol.vincent@wanadoo.fr>
To:     x86@kernel.org, Ingo Molnar <mingo@redhat.com>,
        Borislav Petkov <bp@alien8.de>
Cc:     Nick Desaulniers <ndesaulniers@google.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        linux-kernel@vger.kernel.org, Yury Norov <yury.norov@gmail.com>,
        llvm@lists.linux.dev, Vincent Mailhol <mailhol.vincent@wanadoo.fr>,
        Borislav Petkov <bp@suse.de>
Subject: [PATCH v1 2/2] x86/asm/bitops: Use __builtin_clz*() to evaluate constant expressions
Date:   Sun,  6 Nov 2022 18:51:06 +0900
Message-Id: <20221106095106.849154-3-mailhol.vincent@wanadoo.fr>
In-Reply-To: <20221106095106.849154-1-mailhol.vincent@wanadoo.fr>
References: <20221106095106.849154-1-mailhol.vincent@wanadoo.fr>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

GCC and clang offers the __builtin_clz(x) and __builtin_clzll(x)
functions which return the number of leading zero bits in
x. c.f. [1]. By a simple subtraction, we can derive below equivalences:

* For fls:
  Aside of the x = 0 special case, fls(x) is equivalent to
  BITS_PER_TYPE(x) - __builtin_clz(x).

* For fls64:
  Aside of the x = 0 special case, fls64(x) is equivalent to
  BITS_PER_TYPE(x) - __builtin_clzll(x). __builtin_clzll() takes an
  unsigned long long as argument. We choose this version because
  BITS_PER_LONG_LONG is defined as 64 bits for all architecture making
  this flavor the most portable one. A BUILD_BUG_ON() safety net is
  added.

When used on constant expressions, the compiler is only able to fold
the builtin version (c.f. [2]). However, for non constant expressions,
the kernel inline assembly results in better code for both GCC and
clang.

Use __builtin_constant_p() to select between the kernel's
fls()/fls64() __builtin_clz()/__builtin_clzll() depending on whether
the argument is constant or not.

While renaming fls64() to variable_fls64(), change the argument type
from __64 to u64 because we are not in an uapi header.

[1] Built-in Functions Provided by GCC:
https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html

[2] commit 146034fed6ee ("x86/asm/bitops: Use __builtin_ffs() to evaluate constant expressions")

CC: Borislav Petkov <bp@suse.de>
CC: Nick Desaulniers <ndesaulniers@google.com>
CC: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
---
 arch/x86/include/asm/bitops.h | 57 ++++++++++++++++++++++++-----------
 1 file changed, 40 insertions(+), 17 deletions(-)

diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h
index a31453d7686d..58fb2fc49760 100644
--- a/arch/x86/include/asm/bitops.h
+++ b/arch/x86/include/asm/bitops.h
@@ -333,18 +333,15 @@ static __always_inline int variable_ffs(int x)
  */
 #define ffs(x) (__builtin_constant_p(x) ? __builtin_ffs(x) : variable_ffs(x))
 
-/**
- * fls - find last set bit in word
- * @x: the word to search
- *
- * This is defined in a similar way as the libc and compiler builtin
- * ffs, but returns the position of the most significant set bit.
- *
- * fls(value) returns 0 if value is 0 or the position of the last
- * set bit if value is nonzero. The last (most significant) bit is
- * at position 32.
- */
-static __always_inline int fls(unsigned int x)
+static __always_inline int constant_fls(unsigned int x)
+{
+	if (!x)
+		return 0;
+
+	return BITS_PER_TYPE(x) - __builtin_clz(x);
+}
+
+static __always_inline int variable_fls(unsigned int x)
 {
 	int r;
 
@@ -375,18 +372,30 @@ static __always_inline int fls(unsigned int x)
 }
 
 /**
- * fls64 - find last set bit in a 64-bit word
+ * fls - find last set bit in word
  * @x: the word to search
  *
  * This is defined in a similar way as the libc and compiler builtin
- * ffsll, but returns the position of the most significant set bit.
+ * ffs, but returns the position of the most significant set bit.
  *
- * fls64(value) returns 0 if value is 0 or the position of the last
+ * fls(value) returns 0 if value is 0 or the position of the last
  * set bit if value is nonzero. The last (most significant) bit is
- * at position 64.
+ * at position 32.
  */
+#define fls(x) (__builtin_constant_p(x) ? constant_fls(x) : variable_fls(x))
+
 #ifdef CONFIG_X86_64
-static __always_inline int fls64(__u64 x)
+static __always_inline int constant_fls64(u64 x)
+{
+	BUILD_BUG_ON(sizeof(unsigned long long) != sizeof(x));
+
+	if (!x)
+		return 0;
+
+	return BITS_PER_TYPE(x) - __builtin_clzll(x);
+}
+
+static __always_inline int variable_fls64(u64 x)
 {
 	int bitpos = -1;
 	/*
@@ -399,6 +408,20 @@ static __always_inline int fls64(__u64 x)
 	    : "rm" (x));
 	return bitpos + 1;
 }
+
+/**
+ * fls64 - find last set bit in a 64-bit word
+ * @x: the word to search
+ *
+ * This is defined in a similar way as the libc and compiler builtin
+ * ffsll, but returns the position of the most significant set bit.
+ *
+ * fls64(value) returns 0 if value is 0 or the position of the last
+ * set bit if value is nonzero. The last (most significant) bit is
+ * at position 64.
+ */
+#define fls64(x) \
+	(__builtin_constant_p(x) ? constant_fls64(x) : variable_fls64(x))
 #else
 #include <asm-generic/bitops/fls64.h>
 #endif
-- 
2.37.4