arm64: atomics: fix use of acquire + release for full barrier semantics

Linux requires a number of atomic operations to provide full barrier
semantics, that is no memory accesses after the operation can be
observed before any accesses up to and including the operation in
program order.

On arm64, these operations have been incorrectly implemented as follows:

	// A, B, C are independent memory locations

	<Access [A]>

	// atomic_op (B)
1:	ldaxr	x0, [B]		// Exclusive load with acquire
	<op(B)>
	stlxr	w1, x0, [B]	// Exclusive store with release
	cbnz	w1, 1b

	<Access [C]>

The assumption here being that two half barriers are equivalent to a
full barrier, so the only permitted ordering would be A -> B -> C
(where B is the atomic operation involving both a load and a store).

Unfortunately, this is not the case by the letter of the architecture
and, in fact, the accesses to A and C are permitted to pass their
nearest half barrier resulting in orderings such as Bl -> A -> C -> Bs
or Bl -> C -> A -> Bs (where Bl is the load-acquire on B and Bs is the
store-release on B). This is a clear violation of the full barrier
requirement.

The simple way to fix this is to implement the same algorithm as ARMv7
using explicit barriers:

	<Access [A]>

	// atomic_op (B)
	dmb	ish		// Full barrier
1:	ldxr	x0, [B]		// Exclusive load
	<op(B)>
	stxr	w1, x0, [B]	// Exclusive store
	cbnz	w1, 1b
	dmb	ish		// Full barrier

	<Access [C]>

but this has the undesirable effect of introducing *two* full barrier
instructions. A better approach is actually the following, non-intuitive
sequence:

	<Access [A]>

	// atomic_op (B)
1:	ldxr	x0, [B]		// Exclusive load
	<op(B)>
	stlxr	w1, x0, [B]	// Exclusive store with release
	cbnz	w1, 1b
	dmb	ish		// Full barrier

	<Access [C]>

The simple observations here are:

  - The dmb ensures that no subsequent accesses (e.g. the access to C)
    can enter or pass the atomic sequence.

  - The dmb also ensures that no prior accesses (e.g. the access to A)
    can pass the atomic sequence.

  - Therefore, no prior access can pass a subsequent access, or
    vice-versa (i.e. A is strictly ordered before C).

  - The stlxr ensures that no prior access can pass the store component
    of the atomic operation.

The only tricky part remaining is the ordering between the ldxr and the
access to A, since the absence of the first dmb means that we're now
permitting re-ordering between the ldxr and any prior accesses.

From an (arbitrary) observer's point of view, there are two scenarios:

  1. We have observed the ldxr. This means that if we perform a store to
     [B], the ldxr will still return older data. If we can observe the
     ldxr, then we can potentially observe the permitted re-ordering
     with the access to A, which is clearly an issue when compared to
     the dmb variant of the code. Thankfully, the exclusive monitor will
     save us here since it will be cleared as a result of the store and
     the ldxr will retry. Notice that any use of a later memory
     observation to imply observation of the ldxr will also imply
     observation of the access to A, since the stlxr/dmb ensure strict
     ordering.

  2. We have not observed the ldxr. This means we can perform a store
     and influence the later ldxr. However, that doesn't actually tell
     us anything about the access to [A], so we've not lost anything
     here either when compared to the dmb variant.

This patch implements this solution for our barriered atomic operations,
ensuring that we satisfy the full barrier requirements where they are
needed.

Cc: <stable@vger.kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

3 files changed, 29 insertions, 15 deletions

diff --git a/arch/arm64/include/asm/atomic.h b/arch/arm64/include/asm/atomic.h
index 01de5aaa3edc..e32893e005d4 100644
--- a/arch/arm64/include/asm/atomic.h
+++ b/arch/arm64/include/asm/atomic.h
@@ -64,7 +64,7 @@ static inline int atomic_add_return(int i, atomic_t *v)
 	int result;
 
 	asm volatile("// atomic_add_return\n"
-"1:	ldaxr	%w0, %2\n"
+"1:	ldxr	%w0, %2\n"
 "	add	%w0, %w0, %w3\n"
 "	stlxr	%w1, %w0, %2\n"
 "	cbnz	%w1, 1b"
@@ -72,6 +72,7 @@ static inline int atomic_add_return(int i, atomic_t *v)
 	: "Ir" (i)
 	: "cc", "memory");
 
+	smp_mb();
 	return result;
 }
 
@@ -96,7 +97,7 @@ static inline int atomic_sub_return(int i, atomic_t *v)
 	int result;
 
 	asm volatile("// atomic_sub_return\n"
-"1:	ldaxr	%w0, %2\n"
+"1:	ldxr	%w0, %2\n"
 "	sub	%w0, %w0, %w3\n"
 "	stlxr	%w1, %w0, %2\n"
 "	cbnz	%w1, 1b"
@@ -104,6 +105,7 @@ static inline int atomic_sub_return(int i, atomic_t *v)
 	: "Ir" (i)
 	: "cc", "memory");
 
+	smp_mb();
 	return result;
 }
 
@@ -112,17 +114,20 @@ static inline int atomic_cmpxchg(atomic_t *ptr, int old, int new)
 	unsigned long tmp;
 	int oldval;
 
+	smp_mb();
+
 	asm volatile("// atomic_cmpxchg\n"
-"1:	ldaxr	%w1, %2\n"
+"1:	ldxr	%w1, %2\n"
 "	cmp	%w1, %w3\n"
 "	b.ne	2f\n"
-"	stlxr	%w0, %w4, %2\n"
+"	stxr	%w0, %w4, %2\n"
 "	cbnz	%w0, 1b\n"
 "2:"
 	: "=&r" (tmp), "=&r" (oldval), "+Q" (ptr->counter)
 	: "Ir" (old), "r" (new)
 	: "cc", "memory");
 
+	smp_mb();
 	return oldval;
 }
 
@@ -183,7 +188,7 @@ static inline long atomic64_add_return(long i, atomic64_t *v)
 	unsigned long tmp;
 
 	asm volatile("// atomic64_add_return\n"
-"1:	ldaxr	%0, %2\n"
+"1:	ldxr	%0, %2\n"
 "	add	%0, %0, %3\n"
 "	stlxr	%w1, %0, %2\n"
 "	cbnz	%w1, 1b"
@@ -191,6 +196,7 @@ static inline long atomic64_add_return(long i, atomic64_t *v)
 	: "Ir" (i)
 	: "cc", "memory");
 
+	smp_mb();
 	return result;
 }
 
@@ -215,7 +221,7 @@ static inline long atomic64_sub_return(long i, atomic64_t *v)
 	unsigned long tmp;
 
 	asm volatile("// atomic64_sub_return\n"
-"1:	ldaxr	%0, %2\n"
+"1:	ldxr	%0, %2\n"
 "	sub	%0, %0, %3\n"
 "	stlxr	%w1, %0, %2\n"
 "	cbnz	%w1, 1b"
@@ -223,6 +229,7 @@ static inline long atomic64_sub_return(long i, atomic64_t *v)
 	: "Ir" (i)
 	: "cc", "memory");
 
+	smp_mb();
 	return result;
 }
 
@@ -231,17 +238,20 @@ static inline long atomic64_cmpxchg(atomic64_t *ptr, long old, long new)
 	long oldval;
 	unsigned long res;
 
+	smp_mb();
+
 	asm volatile("// atomic64_cmpxchg\n"
-"1:	ldaxr	%1, %2\n"
+"1:	ldxr	%1, %2\n"
 "	cmp	%1, %3\n"
 "	b.ne	2f\n"
-"	stlxr	%w0, %4, %2\n"
+"	stxr	%w0, %4, %2\n"
 "	cbnz	%w0, 1b\n"
 "2:"
 	: "=&r" (res), "=&r" (oldval), "+Q" (ptr->counter)
 	: "Ir" (old), "r" (new)
 	: "cc", "memory");
 
+	smp_mb();
 	return oldval;
 }
 
@@ -253,11 +263,12 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	unsigned long tmp;
 
 	asm volatile("// atomic64_dec_if_positive\n"
-"1:	ldaxr	%0, %2\n"
+"1:	ldxr	%0, %2\n"
 "	subs	%0, %0, #1\n"
 "	b.mi	2f\n"
 "	stlxr	%w1, %0, %2\n"
 "	cbnz	%w1, 1b\n"
+"	dmb	ish\n"
 "2:"
 	: "=&r" (result), "=&r" (tmp), "+Q" (v->counter)
 	:
diff --git a/arch/arm64/include/asm/cmpxchg.h b/arch/arm64/include/asm/cmpxchg.h
index 56166d7f4a25..189390ce8653 100644
--- a/arch/arm64/include/asm/cmpxchg.h
+++ b/arch/arm64/include/asm/cmpxchg.h
@@ -29,7 +29,7 @@ static inline unsigned long __xchg(unsigned long x, volatile void *ptr, int size
 	switch (size) {
 	case 1:
 		asm volatile("//	__xchg1\n"
-		"1:	ldaxrb	%w0, %2\n"
+		"1:	ldxrb	%w0, %2\n"
 		"	stlxrb	%w1, %w3, %2\n"
 		"	cbnz	%w1, 1b\n"
 			: "=&r" (ret), "=&r" (tmp), "+Q" (*(u8 *)ptr)
@@ -38,7 +38,7 @@ static inline unsigned long __xchg(unsigned long x, volatile void *ptr, int size
 		break;
 	case 2:
 		asm volatile("//	__xchg2\n"
-		"1:	ldaxrh	%w0, %2\n"
+		"1:	ldxrh	%w0, %2\n"
 		"	stlxrh	%w1, %w3, %2\n"
 		"	cbnz	%w1, 1b\n"
 			: "=&r" (ret), "=&r" (tmp), "+Q" (*(u16 *)ptr)
@@ -47,7 +47,7 @@ static inline unsigned long __xchg(unsigned long x, volatile void *ptr, int size
 		break;
 	case 4:
 		asm volatile("//	__xchg4\n"
-		"1:	ldaxr	%w0, %2\n"
+		"1:	ldxr	%w0, %2\n"
 		"	stlxr	%w1, %w3, %2\n"
 		"	cbnz	%w1, 1b\n"
 			: "=&r" (ret), "=&r" (tmp), "+Q" (*(u32 *)ptr)
@@ -56,7 +56,7 @@ static inline unsigned long __xchg(unsigned long x, volatile void *ptr, int size
 		break;
 	case 8:
 		asm volatile("//	__xchg8\n"
-		"1:	ldaxr	%0, %2\n"
+		"1:	ldxr	%0, %2\n"
 		"	stlxr	%w1, %3, %2\n"
 		"	cbnz	%w1, 1b\n"
 			: "=&r" (ret), "=&r" (tmp), "+Q" (*(u64 *)ptr)
@@ -67,6 +67,7 @@ static inline unsigned long __xchg(unsigned long x, volatile void *ptr, int size
 		BUILD_BUG();
 	}
 
+	smp_mb();
 	return ret;
 }
 
diff --git a/arch/arm64/include/asm/futex.h b/arch/arm64/include/asm/futex.h
index 78cc3aba5d69..572193d0005d 100644
--- a/arch/arm64/include/asm/futex.h
+++ b/arch/arm64/include/asm/futex.h
@@ -24,10 +24,11 @@
 
 #define __futex_atomic_op(insn, ret, oldval, uaddr, tmp, oparg)		\
 	asm volatile(							\
-"1:	ldaxr	%w1, %2\n"						\
+"1:	ldxr	%w1, %2\n"						\
 	insn "\n"							\
 "2:	stlxr	%w3, %w0, %2\n"						\
 "	cbnz	%w3, 1b\n"						\
+"	dmb	ish\n"							\
 "3:\n"									\
 "	.pushsection .fixup,\"ax\"\n"					\
 "	.align	2\n"							\
@@ -111,11 +112,12 @@ futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
 		return -EFAULT;
 
 	asm volatile("// futex_atomic_cmpxchg_inatomic\n"
-"1:	ldaxr	%w1, %2\n"
+"1:	ldxr	%w1, %2\n"
 "	sub	%w3, %w1, %w4\n"
 "	cbnz	%w3, 3f\n"
 "2:	stlxr	%w3, %w5, %2\n"
 "	cbnz	%w3, 1b\n"
+"	dmb	ish\n"
 "3:\n"
 "	.pushsection .fixup,\"ax\"\n"
 "4:	mov	%w0, %w6\n"