riscv/barrier: Consolidate fence definitions

Disparate fence implementations are consolidated into fence.h.
Also introduce RISCV_FENCE_ASM to make fence macro more reusable.
Signed-off-by: Eric Chan <ericchancf@google.com>
Reviewed-by: Andrea Parri <parri.andrea@gmail.com>
Reviewed-by: Samuel Holland <samuel.holland@sifive.com>
Tested-by: Samuel Holland <samuel.holland@sifive.com>
Link: https://lore.kernel.org/r/20240217131316.3668927-1-ericchancf@google.comSigned-off-by: Palmer Dabbelt <palmer@rivosinc.com>

arch/riscv/include/asm/atomic.h

@@ -17,7 +17,6 @@
 #endif
 
 #include <asm/cmpxchg.h>
-#include <asm/barrier.h>
 
 #define __atomic_acquire_fence()					\
 	__asm__ __volatile__(RISCV_ACQUIRE_BARRIER "" ::: "memory")

arch/riscv/include/asm/barrier.h

@@ -11,13 +11,12 @@
 #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)

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)					\

arch/riscv/include/asm/fence.h

 #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_FULL_BARRIER		"\tfence rw, rw\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

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

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; })

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>