[v1,01/24] eal/loongarch: add atomic operations for LoongArch

Message ID 20220528093311.123946-2-zhoumin@loongson.cn (mailing list archive)
State Superseded, archived
Delegated to: Thomas Monjalon
Headers
Series Support LoongArch architecture |

Checks

Context Check Description
ci/checkpatch warning coding style issues

Commit Message

Min Zhou May 28, 2022, 9:32 a.m. UTC
This patch adds architecture specific atomic operations for
LoongArch architecture.

Signed-off-by: Min Zhou <zhoumin@loongson.cn>
---
 lib/eal/loongarch/include/rte_atomic.h | 401 +++++++++++++++++++++++++
 1 file changed, 401 insertions(+)
 create mode 100644 lib/eal/loongarch/include/rte_atomic.h
  

Comments

Stephen Hemminger May 28, 2022, 3:52 p.m. UTC | #1
On Sat, 28 May 2022 17:32:48 +0800
Min Zhou <zhoumin@loongson.cn> wrote:

> This patch adds architecture specific atomic operations for
> LoongArch architecture.
> 
> Signed-off-by: Min Zhou <zhoumin@loongson.cn>

DPDK has adopted std atomics. Could you please fix your toolchain
to support that instead.
  
Min Zhou May 30, 2022, 1 a.m. UTC | #2
Stephen,

Thanks for your comments!  I'm evaluating the atomic operations of
LoongArch toolchain. An updated patch will come later.

On 2022/5/28 下午11:52, Stephen Hemminger wrote:
> On Sat, 28 May 2022 17:32:48 +0800
> Min Zhou<zhoumin@loongson.cn>  wrote:
>
>> This patch adds architecture specific atomic operations for
>> LoongArch architecture.
>>
>> Signed-off-by: Min Zhou<zhoumin@loongson.cn>
> DPDK has adopted std atomics. Could you please fix your toolchain
> to support that instead.
  

Patch

diff --git a/lib/eal/loongarch/include/rte_atomic.h b/lib/eal/loongarch/include/rte_atomic.h
new file mode 100644
index 0000000000..4c7526b85d
--- /dev/null
+++ b/lib/eal/loongarch/include/rte_atomic.h
@@ -0,0 +1,401 @@ 
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2022 Loongson Technology Corporation Limited
+ */
+
+#ifndef _RTE_ATOMIC_LOONGARCH_H_
+#define _RTE_ATOMIC_LOONGARCH_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdint.h>
+#include "generic/rte_atomic.h"
+
+/**
+ * LoongArch Synchronize
+ */
+static inline void synchronize(void)
+{
+	__asm__ __volatile__("dbar 0":::"memory");
+}
+
+/**
+ * General memory barrier.
+ *
+ * Guarantees that the LOAD and STORE operations generated before the
+ * barrier occur before the LOAD and STORE operations generated after.
+ * This function is architecture dependent.
+ */
+#define rte_mb() synchronize()
+
+/**
+ * Write memory barrier.
+ *
+ * Guarantees that the STORE operations generated before the barrier
+ * occur before the STORE operations generated after.
+ * This function is architecture dependent.
+ */
+#define rte_wmb() synchronize()
+
+/**
+ * Read memory barrier.
+ *
+ * Guarantees that the LOAD operations generated before the barrier
+ * occur before the LOAD operations generated after.
+ * This function is architecture dependent.
+ */
+#define rte_rmb() synchronize()
+
+#define rte_smp_mb() rte_mb()
+
+#define rte_smp_wmb() rte_mb()
+
+#define rte_smp_rmb() rte_mb()
+
+#define rte_io_mb() rte_mb()
+
+#define rte_io_wmb() rte_mb()
+
+#define rte_io_rmb() rte_mb()
+
+#define rte_io_mb() rte_mb()
+
+#define rte_cio_wmb() rte_wmb()
+
+#define rte_cio_rmb() rte_rmb()
+
+#define smp_mb__before_llsc()
+
+#define smp_llsc_mb()
+
+static __rte_always_inline void
+rte_atomic_thread_fence(int memorder)
+{
+	__atomic_thread_fence(memorder);
+}
+
+/*
+ * Helper Functions
+ */
+#define ATOMIC32_OP(op, I, asm_op)                                    \
+static inline void la_atomic32_##op(rte_atomic32_t *v, int32_t i) \
+{                                                                     \
+	__asm__ __volatile__(                                         \
+		"am"#asm_op"_db.w $zero, %[i], %[cnt]\n"     \
+		: [cnt] "+ZB" (v->cnt)                                 \
+		: [i] "r" (I)                                          \
+		: "memory"                                            \
+		);                                                    \
+}
+
+#define ATOMIC32_OP_RETURN(op, I, asm_op)                                             \
+static inline int32_t la_atomic32_##op##_and_return(rte_atomic32_t *v, int32_t i) \
+{                                                                                     \
+	int32_t result;                                                               \
+									\
+	__asm__ __volatile__(                                                         \
+		"am"#asm_op"_db.w %1, %2, %0\n"                               \
+		: "+ZB" (v->cnt), "=&r" (result)                                      \
+		: "r" (I)                                                             \
+		: "memory"                                                            \
+		);                                                                    \
+									\
+	return result + I;                                               \
+}
+ATOMIC32_OP(add, i, add)
+ATOMIC32_OP(sub, -i, add)
+ATOMIC32_OP_RETURN(add, i, add)
+ATOMIC32_OP_RETURN(sub, -i, add)
+
+#define ATOMIC64_OP(op, I, asm_op)                                    \
+static inline void la_atomic64_##op(rte_atomic64_t *v, int64_t i) \
+{                                                                     \
+	__asm__ __volatile__(                                         \
+		"am"#asm_op"_db.d $zero, %[i], %[cnt]\n"     \
+		: [cnt] "+ZB" (v->cnt)                                 \
+		: [i] "r" (I)                                          \
+		: "memory"                                            \
+		);                                                    \
+}
+#define ATOMIC64_OP_RETURN(op, I, asm_op)                                             \
+static inline int64_t la_atomic64_##op##_and_return(rte_atomic64_t *v, int64_t i) \
+{                                                                                     \
+	int64_t result;                                                               \
+									\
+	__asm__ __volatile__(                                                         \
+		"am"#asm_op"_db.d %1, %2, %0\n"                               \
+		: "+ZB" (v->cnt), "=&r" (result)                                      \
+		: "r" (I)                                                             \
+		: "memory"	                                                      \
+		);                                                                    \
+									\
+	return result + I;                                      \
+}
+ATOMIC64_OP(add, i, add)
+ATOMIC64_OP(sub, -i, add)
+ATOMIC64_OP_RETURN(add, i, add)
+ATOMIC64_OP_RETURN(sub, -i, add)
+
+static inline int
+loongarch_atomic32_cmpset(volatile uint32_t *dst, uint32_t exp, uint32_t src)
+{
+	uint32_t res = 0, temp = 0;
+	__asm__ __volatile__(
+		"1: ll.w %[tmp], %[dst]\n"
+		"   li.w %[res], 0\n"
+		"   bne %[tmp], %[exp], 2f\n"
+		"   move %[tmp], %[src]\n"
+		"   sc.w %[tmp], %[dst]\n"
+		"   li.w %[res], 1\n"
+		"   beqz %[tmp], 1b\n"
+		"2:\n"
+		: [res] "=&r"(res),
+		  [tmp] "+&r"(temp),
+		  [dst] "+m"(*dst)
+		: [src] "r"(src),
+		  [exp] "r"(exp),
+		  "m"(*dst)
+		: "memory"
+		);
+
+	return res;
+}
+
+static inline int
+loongarch_atomic64_cmpset(volatile uint64_t *dst, uint64_t exp, uint64_t src)
+{
+	uint32_t res = 0;
+	uint64_t temp = 0;
+	__asm__ __volatile__(
+		"1: ll.d %[tmp], %[dst]\n"
+		"   li.d %[res], 0\n"
+		"   bne %[tmp], %[exp], 2f\n"
+		"   move %[tmp], %[src]\n"
+		"   sc.d %[tmp], %[dst]\n"
+		"   li.d %[res], 1\n"
+		"   beqz %[tmp], 1b\n"
+		"2:\n"
+		: [res] "=&r"(res),
+		  [tmp] "+&r"(temp),
+		  [dst] "+m"(*dst)
+		: [src] "r"(src),
+		  [exp] "r"(exp),
+		  "m"(*dst)
+		: "memory"
+		);
+
+	return res;
+}
+
+#ifndef RTE_FORCE_INTRINSICS
+static inline int
+rte_atomic16_cmpset(volatile uint16_t *dst, uint16_t exp, uint16_t src)
+{
+	return __sync_bool_compare_and_swap(dst, exp, src);
+}
+
+static inline int rte_atomic16_test_and_set(rte_atomic16_t *v)
+{
+	return rte_atomic16_cmpset((volatile uint16_t *)&v->cnt, 0, 1);
+}
+
+static inline void
+rte_atomic16_inc(rte_atomic16_t *v)
+{
+	rte_atomic16_add(v, 1);
+}
+
+static inline void
+rte_atomic16_dec(rte_atomic16_t *v)
+{
+	rte_atomic16_sub(v, 1);
+}
+
+static inline int rte_atomic16_inc_and_test(rte_atomic16_t *v)
+{
+	return __sync_add_and_fetch(&v->cnt, 1) == 0;
+}
+
+static inline int rte_atomic16_dec_and_test(rte_atomic16_t *v)
+{
+	return __sync_sub_and_fetch(&v->cnt, 1) == 0;
+}
+
+static inline uint16_t
+rte_atomic16_exchange(volatile uint16_t *dst, uint16_t val)
+{
+	return __atomic_exchange_2(dst, val, __ATOMIC_SEQ_CST);
+}
+
+/*------------------------- 32 bit atomic operations -------------------------*/
+
+static inline int
+rte_atomic32_cmpset(volatile uint32_t *dst, uint32_t exp, uint32_t src)
+{
+	return loongarch_atomic32_cmpset(dst, exp, src);
+}
+
+static inline int rte_atomic32_test_and_set(rte_atomic32_t *v)
+{
+	return rte_atomic32_cmpset((volatile uint32_t *)&v->cnt, 0, 1);
+}
+
+static inline void
+rte_atomic32_inc(rte_atomic32_t *v)
+{
+	la_atomic32_add(v, 1);
+}
+
+static inline void
+rte_atomic32_dec(rte_atomic32_t *v)
+{
+	la_atomic32_sub(v, 1);
+}
+
+static inline int rte_atomic32_inc_and_test(rte_atomic32_t *v)
+{
+	return la_atomic32_add_and_return(v, 1) == 0;
+}
+
+static inline int rte_atomic32_dec_and_test(rte_atomic32_t *v)
+{
+	return la_atomic32_sub_and_return(v, 1) == 0;
+}
+
+static inline uint32_t
+rte_atomic32_exchange(volatile uint32_t *dst, uint32_t val)
+{
+	uint32_t retval;
+
+	__asm__ __volatile__ (
+		"amswap_db.w %1, %z2, %0\n"
+		: "+ZB" (*dst), "=&r" (retval)
+		: "Jr" (val)
+		: "memory"
+		);
+
+	return retval;
+
+}
+/*------------------------- 64 bit atomic operations -------------------------*/
+static inline int
+rte_atomic64_cmpset(volatile uint64_t *dst, uint64_t exp, uint64_t src)
+{
+	return loongarch_atomic64_cmpset(dst, exp, src);
+}
+
+static inline void
+rte_atomic64_init(rte_atomic64_t *v)
+{
+	__asm__ __volatile__(
+		"amand_db.d $zero, $zero, %[rj]\n"
+		: [rj]"+ZB"(v->cnt)
+		);
+}
+
+static inline int64_t
+rte_atomic64_read(rte_atomic64_t *v)
+{
+	int64_t res = 0;
+
+	__asm__ __volatile__(
+		"amadd_db.d %[res], $zero, %[cnt]\n"
+		: [res]"=&r"(res), [cnt]"+ZB"(v->cnt)
+		:
+		);
+	return res;
+
+}
+
+static inline void
+rte_atomic64_set(rte_atomic64_t *v, int64_t new_value)
+{
+	__asm__ __volatile__(
+		"amswap_db.d $zero, %[nv], %[cnt]\n"
+		: [cnt]"+ZB"(v->cnt)
+		: [nv]"r"(new_value)
+		);
+}
+
+static inline void
+rte_atomic64_add(rte_atomic64_t *v, int64_t inc)
+{
+	la_atomic64_add(v, inc);
+}
+
+static inline void
+rte_atomic64_sub(rte_atomic64_t *v, int64_t dec)
+{
+	la_atomic64_sub(v, dec);
+}
+
+static inline void
+rte_atomic64_inc(rte_atomic64_t *v)
+{
+	la_atomic64_add(v, 1);
+}
+
+static inline void
+rte_atomic64_dec(rte_atomic64_t *v)
+{
+	la_atomic64_sub(v, 1);
+}
+
+static inline int64_t
+rte_atomic64_add_return(rte_atomic64_t *v, int64_t inc)
+{
+	return la_atomic64_add_and_return(v, inc);
+}
+
+static inline int64_t
+rte_atomic64_sub_return(rte_atomic64_t *v, int64_t dec)
+{
+	return la_atomic64_sub_and_return(v, dec);
+}
+
+static inline int rte_atomic64_inc_and_test(rte_atomic64_t *v)
+{
+	return la_atomic64_add_and_return(v, 1) == 0;
+}
+
+static inline int rte_atomic64_dec_and_test(rte_atomic64_t *v)
+{
+	return la_atomic64_sub_and_return(v, 1) == 0;
+}
+
+static inline int rte_atomic64_test_and_set(rte_atomic64_t *v)
+{
+	return rte_atomic64_cmpset((volatile uint64_t *)&v->cnt, 0, 1);
+}
+
+static inline void rte_atomic64_clear(rte_atomic64_t *v)
+{
+	rte_atomic64_set(v, 0);
+}
+
+static inline uint64_t __xchg_u64(volatile uint64_t *m, uint64_t val)
+{
+	uint64_t retval;
+	__asm__ __volatile__ (
+		"amswap_db.d %1, %z2, %0\n"
+		: "+ZB" (*m), "=&r" (retval)
+		: "Jr" (val)
+		: "memory"
+		);
+
+	return retval;
+}
+
+static inline uint64_t
+rte_atomic64_exchange(volatile uint64_t *dst, uint64_t val)
+{
+	return __xchg_u64(dst, val);
+}
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _RTE_ATOMIC_LOONGARCH_H_ */