Merge patch series "tools: Add barrier implementations for riscv"

Charlie Jenkins <charlie@rivosinc.com> says:

Add support for riscv specific barrier implementations to the tools
tree, so that fence instructions can be emitted for synchronization.

* b4-shazam-merge:
  tools: Optimize ring buffer for riscv
  tools: Add riscv barrier implementation

Link: https://lore.kernel.org/r/20240806-optimize_ring_buffer_read_riscv-v2-0-ca7e193ae198@rivosinc.comSigned-off-by: Palmer Dabbelt <palmer@rivosinc.com>

tools/arch/riscv/include/asm/barrier.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copied from the kernel sources to tools/arch/riscv:
+ *
+ * Copyright (C) 2012 ARM Ltd.
+ * Copyright (C) 2013 Regents of the University of California
+ * Copyright (C) 2017 SiFive
+ */
+
+#ifndef _TOOLS_LINUX_ASM_RISCV_BARRIER_H
+#define _TOOLS_LINUX_ASM_RISCV_BARRIER_H
+
+#include <asm/fence.h>
+#include <linux/compiler.h>
+
+/* These barriers need to enforce ordering on both devices and memory. */
+#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_store_release(p, v)						\
+do {									\
+	RISCV_FENCE(rw, w);						\
+	WRITE_ONCE(*p, v);						\
+} while (0)
+
+#define smp_load_acquire(p)						\
+({									\
+	typeof(*p) ___p1 = READ_ONCE(*p);				\
+	RISCV_FENCE(r, rw);						\
+	___p1;								\
+})
+
+#endif /* _TOOLS_LINUX_ASM_RISCV_BARRIER_H */

tools/arch/riscv/include/asm/fence.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copied from the kernel sources to tools/arch/riscv:
+ */
+
+#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"); })
+
+#endif	/* _ASM_RISCV_FENCE_H */

tools/include/asm/barrier.h

@@ -8,6 +8,8 @@
 #include "../../arch/arm64/include/asm/barrier.h"
 #elif defined(__powerpc__)
 #include "../../arch/powerpc/include/asm/barrier.h"
+#elif defined(__riscv)
+#include "../../arch/riscv/include/asm/barrier.h"
 #elif defined(__s390__)
 #include "../../arch/s390/include/asm/barrier.h"
 #elif defined(__sh__)

tools/include/linux/ring_buffer.h

@@ -55,7 +55,7 @@ static inline u64 ring_buffer_read_head(struct perf_event_mmap_page *base)
  * READ_ONCE() + smp_mb() pair.
  */
 #if defined(__x86_64__) || defined(__aarch64__) || defined(__powerpc64__) || \
-    defined(__ia64__) || defined(__sparc__) && defined(__arch64__)
+    defined(__ia64__) || defined(__sparc__) && defined(__arch64__) || defined(__riscv)
 	return smp_load_acquire(&base->data_head);
 #else
 	u64 head = READ_ONCE(base->data_head);