Received-SPF: pass (google.com: domain of linux-kernel+bounces-61385-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Mon, 12 Feb 2024 10:59:46 +0000
In-Reply-To: <20240212-projector-dangle-7815fa2f7415@wendy>
Precedence: bulk
Mime-Version: 1.0
References: <20240212-projector-dangle-7815fa2f7415@wendy>
Message-ID: <20240212105946.1241100-1-ericchancf@google.com>
Subject: [PATCH v2] riscv/fence: Consolidate fence definitions and define __{mb,rmb,wmb}
From: Eric Chan <ericchancf@google.com>
To: conor.dooley@microchip.com
Cc: aou@eecs.berkeley.edu, ericchancf@google.com, linux-kernel@vger.kernel.org, 
	linux-riscv@lists.infradead.org, palmer@dabbelt.com, paul.walmsley@sifive.com
Content-Type: text/plain; charset="UTF-8"

Disparate fence implementations are consolidated into fence.h.

Introduce __{mb,rmb,wmb}, and rely on the generic definitions
for {mb,rmb,wmb}. A first consequence is that __{mb,rmb,wmb}
map to a compiler barrier on !SMP (while their definition remains
unchanged on SMP).

Introduce RISCV_FULL_BARRIER and use in arch_atomic* function.
like RISCV_ACQUIRE_BARRIER and RISCV_RELEASE_BARRIER, The fence
instruction can be eliminated When SMP is not enabled.

Also clean up the warning with scripts/checkpatch.pl.

Signed-off-by: Eric Chan <ericchancf@google.com>

---
v1 -> v2: makes compilation pass with allyesconfig instead of
defconfig only, also satisfy scripts/checkpatch.pl.
- (__asm__ __volatile__ (RISCV_FENCE_ASM(p, s) : : : "memory"))
+ ({ __asm__ __volatile__ (RISCV_FENCE_ASM(p, s) : : : "memory"); })
Thanks Conor help review it.

 arch/riscv/include/asm/atomic.h  | 24 ++++++++++--------------
 arch/riscv/include/asm/barrier.h | 21 ++++++++++-----------
 arch/riscv/include/asm/cmpxchg.h |  5 ++---
 arch/riscv/include/asm/fence.h   | 11 +++++++++--
 arch/riscv/include/asm/io.h      |  8 ++++----
 arch/riscv/include/asm/mmio.h    |  5 +++--
 arch/riscv/include/asm/mmiowb.h  |  2 +-
 7 files changed, 39 insertions(+), 37 deletions(-)

diff --git a/arch/riscv/include/asm/atomic.h b/arch/riscv/include/asm/atomic.h
index f5dfef6c2153..19050d13b6c1 100644
--- a/arch/riscv/include/asm/atomic.h
+++ b/arch/riscv/include/asm/atomic.h
@@ -17,13 +17,9 @@
 #endif

 #include <asm/cmpxchg.h>
-#include <asm/barrier.h>

-#define __atomic_acquire_fence()					\
-	__asm__ __volatile__(RISCV_ACQUIRE_BARRIER "" ::: "memory")
-
-#define __atomic_release_fence()					\
-	__asm__ __volatile__(RISCV_RELEASE_BARRIER "" ::: "memory");
+#define __atomic_acquire_fence()	RISCV_FENCE(r, rw)
+#define __atomic_release_fence()	RISCV_FENCE(rw, r)

 static __always_inline int arch_atomic_read(const atomic_t *v)
 {
@@ -207,7 +203,7 @@ static __always_inline int arch_atomic_fetch_add_unless(atomic_t *v, int a, int
 		"	add      %[rc], %[p], %[a]\n"
 		"	sc.w.rl  %[rc], %[rc], %[c]\n"
 		"	bnez     %[rc], 0b\n"
-		"	fence    rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		: [a]"r" (a), [u]"r" (u)
@@ -228,7 +224,7 @@ static __always_inline s64 arch_atomic64_fetch_add_unless(atomic64_t *v, s64 a,
 		"	add      %[rc], %[p], %[a]\n"
 		"	sc.d.rl  %[rc], %[rc], %[c]\n"
 		"	bnez     %[rc], 0b\n"
-		"	fence    rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		: [a]"r" (a), [u]"r" (u)
@@ -248,7 +244,7 @@ static __always_inline bool arch_atomic_inc_unless_negative(atomic_t *v)
 		"	addi      %[rc], %[p], 1\n"
 		"	sc.w.rl   %[rc], %[rc], %[c]\n"
 		"	bnez      %[rc], 0b\n"
-		"	fence     rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
@@ -268,7 +264,7 @@ static __always_inline bool arch_atomic_dec_unless_positive(atomic_t *v)
 		"	addi      %[rc], %[p], -1\n"
 		"	sc.w.rl   %[rc], %[rc], %[c]\n"
 		"	bnez      %[rc], 0b\n"
-		"	fence     rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
@@ -288,7 +284,7 @@ static __always_inline int arch_atomic_dec_if_positive(atomic_t *v)
 		"	bltz     %[rc], 1f\n"
 		"	sc.w.rl  %[rc], %[rc], %[c]\n"
 		"	bnez     %[rc], 0b\n"
-		"	fence    rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
@@ -310,7 +306,7 @@ static __always_inline bool arch_atomic64_inc_unless_negative(atomic64_t *v)
 		"	addi      %[rc], %[p], 1\n"
 		"	sc.d.rl   %[rc], %[rc], %[c]\n"
 		"	bnez      %[rc], 0b\n"
-		"	fence     rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
@@ -331,7 +327,7 @@ static __always_inline bool arch_atomic64_dec_unless_positive(atomic64_t *v)
 		"	addi      %[rc], %[p], -1\n"
 		"	sc.d.rl   %[rc], %[rc], %[c]\n"
 		"	bnez      %[rc], 0b\n"
-		"	fence     rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
@@ -352,7 +348,7 @@ static __always_inline s64 arch_atomic64_dec_if_positive(atomic64_t *v)
 		"	bltz     %[rc], 1f\n"
 		"	sc.d.rl  %[rc], %[rc], %[c]\n"
 		"	bnez     %[rc], 0b\n"
-		"	fence    rw, rw\n"
+		RISCV_FULL_BARRIER
 		"1:\n"
 		: [p]"=&r" (prev), [rc]"=&r" (rc), [c]"+A" (v->counter)
 		:
diff --git a/arch/riscv/include/asm/barrier.h b/arch/riscv/include/asm/barrier.h
index 110752594228..880b56d8480d 100644
--- a/arch/riscv/include/asm/barrier.h
+++ b/arch/riscv/include/asm/barrier.h
@@ -11,28 +11,27 @@
 #define _ASM_RISCV_BARRIER_H

 #ifndef __ASSEMBLY__
+#include <asm/fence.h>

 #define nop()		__asm__ __volatile__ ("nop")
 #define __nops(n)	".rept	" #n "\nnop\n.endr\n"
 #define nops(n)		__asm__ __volatile__ (__nops(n))

-#define RISCV_FENCE(p, s) \
-	__asm__ __volatile__ ("fence " #p "," #s : : : "memory")

 /* These barriers need to enforce ordering on both devices or memory. */
-#define mb()		RISCV_FENCE(iorw,iorw)
-#define rmb()		RISCV_FENCE(ir,ir)
-#define wmb()		RISCV_FENCE(ow,ow)
+#define __mb()		RISCV_FENCE(iorw, iorw)
+#define __rmb()		RISCV_FENCE(ir, ir)
+#define __wmb()		RISCV_FENCE(ow, ow)

 /* These barriers do not need to enforce ordering on devices, just memory. */
-#define __smp_mb()	RISCV_FENCE(rw,rw)
-#define __smp_rmb()	RISCV_FENCE(r,r)
-#define __smp_wmb()	RISCV_FENCE(w,w)
+#define __smp_mb()	RISCV_FENCE(rw, rw)
+#define __smp_rmb()	RISCV_FENCE(r, r)
+#define __smp_wmb()	RISCV_FENCE(w, w)

 #define __smp_store_release(p, v)					\
 do {									\
 	compiletime_assert_atomic_type(*p);				\
-	RISCV_FENCE(rw,w);						\
+	RISCV_FENCE(rw, w);						\
 	WRITE_ONCE(*p, v);						\
 } while (0)

@@ -40,7 +39,7 @@ do {									\
 ({									\
 	typeof(*p) ___p1 = READ_ONCE(*p);				\
 	compiletime_assert_atomic_type(*p);				\
-	RISCV_FENCE(r,rw);						\
+	RISCV_FENCE(r, rw);						\
 	___p1;								\
 })

@@ -69,7 +68,7 @@ do {									\
  * instances the scheduler pairs this with an mb(), so nothing is necessary on
  * the new hart.
  */
-#define smp_mb__after_spinlock()	RISCV_FENCE(iorw,iorw)
+#define smp_mb__after_spinlock()	RISCV_FENCE(iorw, iorw)

 #include <asm-generic/barrier.h>

diff --git a/arch/riscv/include/asm/cmpxchg.h b/arch/riscv/include/asm/cmpxchg.h
index 2f4726d3cfcc..2fee65cc8443 100644
--- a/arch/riscv/include/asm/cmpxchg.h
+++ b/arch/riscv/include/asm/cmpxchg.h
@@ -8,7 +8,6 @@

 #include <linux/bug.h>

-#include <asm/barrier.h>
 #include <asm/fence.h>

 #define __xchg_relaxed(ptr, new, size)					\
@@ -313,7 +312,7 @@
 			"	bne  %0, %z3, 1f\n"			\
 			"	sc.w.rl %1, %z4, %2\n"			\
 			"	bnez %1, 0b\n"				\
-			"	fence rw, rw\n"				\
+			RISCV_FULL_BARRIER				\
 			"1:\n"						\
 			: "=&r" (__ret), "=&r" (__rc), "+A" (*__ptr)	\
 			: "rJ" ((long)__old), "rJ" (__new)		\
@@ -325,7 +324,7 @@
 			"	bne %0, %z3, 1f\n"			\
 			"	sc.d.rl %1, %z4, %2\n"			\
 			"	bnez %1, 0b\n"				\
-			"	fence rw, rw\n"				\
+			RISCV_FULL_BARRIER				\
 			"1:\n"						\
 			: "=&r" (__ret), "=&r" (__rc), "+A" (*__ptr)	\
 			: "rJ" (__old), "rJ" (__new)			\
diff --git a/arch/riscv/include/asm/fence.h b/arch/riscv/include/asm/fence.h
index 2b443a3a487f..c0fd82d72a0e 100644
--- a/arch/riscv/include/asm/fence.h
+++ b/arch/riscv/include/asm/fence.h
@@ -1,12 +1,19 @@
 #ifndef _ASM_RISCV_FENCE_H
 #define _ASM_RISCV_FENCE_H

+#define RISCV_FENCE_ASM(p, s) \
+	"\tfence " #p "," #s "\n"
+#define RISCV_FENCE(p, s) \
+	({ __asm__ __volatile__ (RISCV_FENCE_ASM(p, s) : : : "memory"); })
+
 #ifdef CONFIG_SMP
-#define RISCV_ACQUIRE_BARRIER		"\tfence r , rw\n"
-#define RISCV_RELEASE_BARRIER		"\tfence rw,  w\n"
+#define RISCV_ACQUIRE_BARRIER		RISCV_FENCE_ASM(r, rw)
+#define RISCV_RELEASE_BARRIER		RISCV_FENCE_ASM(rw, w)
+#define RISCV_FULL_BARRIER		RISCV_FENCE_ASM(rw, rw)
 #else
 #define RISCV_ACQUIRE_BARRIER
 #define RISCV_RELEASE_BARRIER
+#define RISCV_FULL_BARRIER
 #endif

 #endif	/* _ASM_RISCV_FENCE_H */
diff --git a/arch/riscv/include/asm/io.h b/arch/riscv/include/asm/io.h
index 42497d487a17..1c5c641075d2 100644
--- a/arch/riscv/include/asm/io.h
+++ b/arch/riscv/include/asm/io.h
@@ -47,10 +47,10 @@
  * sufficient to ensure this works sanely on controllers that support I/O
  * writes.
  */
-#define __io_pbr()	__asm__ __volatile__ ("fence io,i"  : : : "memory");
-#define __io_par(v)	__asm__ __volatile__ ("fence i,ior" : : : "memory");
-#define __io_pbw()	__asm__ __volatile__ ("fence iow,o" : : : "memory");
-#define __io_paw()	__asm__ __volatile__ ("fence o,io"  : : : "memory");
+#define __io_pbr()	RISCV_FENCE(io, i)
+#define __io_par(v)	RISCV_FENCE(i, ior)
+#define __io_pbw()	RISCV_FENCE(iow, o)
+#define __io_paw()	RISCV_FENCE(o, io)

 /*
  * Accesses from a single hart to a single I/O address must be ordered.  This
diff --git a/arch/riscv/include/asm/mmio.h b/arch/riscv/include/asm/mmio.h
index 4c58ee7f95ec..06cadfd7a237 100644
--- a/arch/riscv/include/asm/mmio.h
+++ b/arch/riscv/include/asm/mmio.h
@@ -12,6 +12,7 @@
 #define _ASM_RISCV_MMIO_H

 #include <linux/types.h>
+#include <asm/fence.h>
 #include <asm/mmiowb.h>

 /* Generic IO read/write.  These perform native-endian accesses. */
@@ -131,8 +132,8 @@ static inline u64 __raw_readq(const volatile void __iomem *addr)
  * doesn't define any ordering between the memory space and the I/O space.
  */
 #define __io_br()	do {} while (0)
-#define __io_ar(v)	({ __asm__ __volatile__ ("fence i,ir" : : : "memory"); })
-#define __io_bw()	({ __asm__ __volatile__ ("fence w,o" : : : "memory"); })
+#define __io_ar(v)	RISCV_FENCE(i, ir)
+#define __io_bw()	RISCV_FENCE(w, o)
 #define __io_aw()	mmiowb_set_pending()

 #define readb(c)	({ u8  __v; __io_br(); __v = readb_cpu(c); __io_ar(__v); __v; })
diff --git a/arch/riscv/include/asm/mmiowb.h b/arch/riscv/include/asm/mmiowb.h
index 0b2333e71fdc..52ce4a399d9b 100644
--- a/arch/riscv/include/asm/mmiowb.h
+++ b/arch/riscv/include/asm/mmiowb.h
@@ -7,7 +7,7 @@
  * "o,w" is sufficient to ensure that all writes to the device have completed
  * before the write to the spinlock is allowed to commit.
  */
-#define mmiowb()	__asm__ __volatile__ ("fence o,w" : : : "memory");
+#define mmiowb()	RISCV_FENCE(o, w)

 #include <linux/smp.h>
 #include <asm-generic/mmiowb.h>
--
2.43.0.687.g38aa6559b0-goog