Skip to contentSkip to conversation

eal: update public API to use stdatomic atomics

PATCH Superseded Tyler Retzlaff 2 participants 22 messages Jul 31, 2023 – Aug 2, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover eal: update public API to use stdatomic atomics

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not exposing or depending on the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler here to be dropped from this series.

Tyler Retzlaff (4):
build: require minimum c11 compiler
devtools: forbid use of GCC atomic builtins
eal: adapt rte pause APIs to use C11 atomics
eal: adapt rte spinlock and rwlock APIs to use C11 atomics

 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 devtools/checkpatches.sh                 |   8 ++
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/failsafe/meson.build         |   1 -
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   4 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_common_errno.c        |   1 +
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  52 ++++++-------
 lib/eal/include/generic/rte_rwlock.h     |  46 ++++++------
 lib/eal/include/generic/rte_spinlock.h   |  21 +++---
 lib/eal/include/rte_mcslock.h            |  12 +--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 -
 lib/eal/ppc/include/rte_pause.h          |   2 -
 lib/eal/riscv/include/rte_pause.h        |   2 -
 lib/eal/x86/include/rte_spinlock.h       |   2 +-
 lib/ring/rte_ring_core.h                 |   5 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 meson.build                              |   1 +
 37 files changed, 311 insertions(+), 262 deletions(-)
Signature · 1 line
1.8.3.1
#1 lore raw

PATCH 1/4 build: require minimum c11 compiler

Update top level meson.build to require c_std=c11 and remove use of -std=gnu99.

---
 drivers/net/failsafe/meson.build  | 1 -
 lib/eal/common/eal_common_errno.c | 1 +
 meson.build                       | 1 +
 3 files changed, 2 insertions(+), 1 deletion(-)
drivers/net/failsafe/meson.build +0 -1
index 6013e13..c1d3610 100644
@@ -7,7 +7,6 @@ if is_windows
     subdir_done()
 endif
 
-cflags += '-std=gnu99'
 cflags += '-D_DEFAULT_SOURCE'
 cflags += '-D_XOPEN_SOURCE=700'
 cflags += '-pedantic'
lib/eal/common/eal_common_errno.c +1 -0
index ef8f782..d4e2c13 100644
@@ -4,6 +4,7 @@
 
 /* Use XSI-compliant portable version of strerror_r() */
 #undef _GNU_SOURCE
+#define _POSIX_C_SOURCE 200112L
 
 #include <stdio.h>
 #include <string.h>
meson.build +1 -0
index 39cb738..70b54f0 100644
@@ -9,6 +9,7 @@ project('DPDK', 'c',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c11',
             'default_library=static',
             'warning_level=2',
         ],
Signature · 1 line
1.8.3.1
#2 lore raw

PATCH 2/4 devtools: forbid use of GCC atomic builtins

In reply to Tyler Retzlaff

Refrain from using compiler __atomic_xxx builtins DPDK now uses standard C11 atomics atomic_xxx.

---
 devtools/checkpatches.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
devtools/checkpatches.sh +8 -0
index 43f5e36..a32f02e 100755
@@ -102,6 +102,14 @@ check_forbidden_additions() { # <patch>
 		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
 		"$1" || res=1
 
+	# refrain from using compiler __atomic_xxx builtins
+	awk -v FOLDERS="lib drivers app examples" \
+		-v EXPRESSIONS="__atomic_.*\\\(" \
+		-v RET_ON_FAIL=1 \
+		-v MESSAGE='Using __atomic_xxx builtins' \
+		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
+		"$1" || res=1
+
 	# refrain from using compiler __atomic_thread_fence()
 	# It should be avoided on x86 for SMP case.
 	awk -v FOLDERS="lib drivers app examples" \
Signature · 1 line
1.8.3.1
#4 lore raw

PATCH 4/4 eal: adapt rte spinlock and rwlock APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_spinlock.h and rte_rwlock.h APIs to use standard C11 atomics. Update consumers of the spinlock and rwlock APIs for the API break.

---
 lib/eal/include/generic/rte_rwlock.h   | 46 ++++++++++++++++++----------------
 lib/eal/include/generic/rte_spinlock.h | 21 ++++++++--------
 lib/eal/x86/include/rte_spinlock.h     |  2 +-
 3 files changed, 36 insertions(+), 33 deletions(-)
lib/eal/include/generic/rte_rwlock.h +24 -22
index 9e083bb..b659c4c 100644
@@ -22,6 +22,8 @@
  *  https://locklessinc.com/articles/locks/
  */
 
+#include <stdatomic.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -57,7 +59,7 @@
 #define RTE_RWLOCK_READ	 0x4	/* Reader increment */
 
 typedef struct __rte_lockable {
-	int32_t cnt;
+	int32_t _Atomic cnt;
 } rte_rwlock_t;
 
 /**
@@ -92,21 +94,21 @@
 
 	while (1) {
 		/* Wait while writer is present or pending */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed)
 		       & RTE_RWLOCK_MASK)
 			rte_pause();
 
 		/* Try to get read lock */
-		x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-				       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+		x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				       memory_order_acquire) + RTE_RWLOCK_READ;
 
 		/* If no writer, then acquire was successful */
 		if (likely(!(x & RTE_RWLOCK_MASK)))
 			return;
 
 		/* Lost race with writer, backout the change. */
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELAXED);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_relaxed);
 	}
 }
 
@@ -127,20 +129,20 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 	/* fail if write lock is held or writer is pending */
 	if (x & RTE_RWLOCK_MASK)
 		return -EBUSY;
 
 	/* Try to get read lock */
-	x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-			       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+	x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+			       memory_order_acquire) + RTE_RWLOCK_READ;
 
 	/* Back out if writer raced in */
 	if (unlikely(x & RTE_RWLOCK_MASK)) {
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELEASE);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_release);
 
 		return -EBUSY;
 	}
@@ -158,7 +160,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ, memory_order_release);
 }
 
 /**
@@ -178,10 +180,10 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 	if (x < RTE_RWLOCK_WRITE &&
-	    __atomic_compare_exchange_n(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
-					1, __ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+	    atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
+					memory_order_acquire, memory_order_relaxed))
 		return 0;
 	else
 		return -EBUSY;
@@ -201,22 +203,22 @@
 	int32_t x;
 
 	while (1) {
-		x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+		x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 		/* No readers or writers? */
 		if (likely(x < RTE_RWLOCK_WRITE)) {
 			/* Turn off RTE_RWLOCK_WAIT, turn on RTE_RWLOCK_WRITE */
-			if (__atomic_compare_exchange_n(&rwl->cnt, &x, RTE_RWLOCK_WRITE, 1,
-							__ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+			if (atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, RTE_RWLOCK_WRITE,
+							memory_order_acquire, memory_order_relaxed))
 				return;
 		}
 
 		/* Turn on writer wait bit */
 		if (!(x & RTE_RWLOCK_WAIT))
-			__atomic_fetch_or(&rwl->cnt, RTE_RWLOCK_WAIT, __ATOMIC_RELAXED);
+			atomic_fetch_or_explicit(&rwl->cnt, RTE_RWLOCK_WAIT, memory_order_relaxed);
 
 		/* Wait until no readers before trying again */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) > RTE_RWLOCK_WAIT)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) > RTE_RWLOCK_WAIT)
 			rte_pause();
 
 	}
@@ -233,7 +235,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_WRITE, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_WRITE, memory_order_release);
 }
 
 /**
@@ -247,7 +249,7 @@
 static inline int
 rte_rwlock_write_is_locked(rte_rwlock_t *rwl)
 {
-	if (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) & RTE_RWLOCK_WRITE)
+	if (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) & RTE_RWLOCK_WRITE)
 		return 1;
 
 	return 0;
lib/eal/include/generic/rte_spinlock.h +11 -10
index c50ebaa..4cbd22f 100644
@@ -17,6 +17,8 @@
  * All locks must be initialised before use, and only initialised once.
  */
 
+#include <stdatomic.h>
+
 #include <rte_lcore.h>
 #ifdef RTE_FORCE_INTRINSICS
 #include <rte_common.h>
@@ -28,7 +30,7 @@
  * The rte_spinlock_t type.
  */
 typedef struct __rte_lockable {
-	volatile int locked; /**< lock status 0 = unlocked, 1 = locked */
+	volatile int _Atomic locked; /**< lock status 0 = unlocked, 1 = locked */
 } rte_spinlock_t;
 
 /**
@@ -65,10 +67,10 @@
 {
 	int exp = 0;
 
-	while (!__atomic_compare_exchange_n(&sl->locked, &exp, 1, 0,
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED)) {
-		rte_wait_until_equal_32((volatile uint32_t *)&sl->locked,
-			       0, __ATOMIC_RELAXED);
+	while (!atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed)) {
+		rte_wait_until_equal_32((uint32_t _Atomic *)&sl->locked,
+			       0, memory_order_relaxed);
 		exp = 0;
 	}
 }
@@ -89,7 +91,7 @@
 rte_spinlock_unlock(rte_spinlock_t *sl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_store_n(&sl->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&sl->locked, 0, memory_order_release);
 }
 #endif
 
@@ -112,9 +114,8 @@
 	__rte_no_thread_safety_analysis
 {
 	int exp = 0;
-	return __atomic_compare_exchange_n(&sl->locked, &exp, 1,
-				0, /* disallow spurious failure */
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED);
+	return atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed);
 }
 #endif
 
@@ -128,7 +129,7 @@
  */
 static inline int rte_spinlock_is_locked (rte_spinlock_t *sl)
 {
-	return __atomic_load_n(&sl->locked, __ATOMIC_ACQUIRE);
+	return atomic_load_explicit(&sl->locked, memory_order_acquire);
 }
 
 /**
lib/eal/x86/include/rte_spinlock.h +1 -1
index 0b20ddf..c71c54d 100644
@@ -78,7 +78,7 @@ static inline int rte_tm_supported(void)
 }
 
 static inline int
-rte_try_tm(volatile int *lock)
+rte_try_tm(volatile int _Atomic *lock)
 {
 	int i, retries;
 
Signature · 1 line
1.8.3.1
#3 lore raw

PATCH 3/4 eal: adapt rte pause APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_pause.h APIs to use standard C11 atomics. Update consumers of the pause APIs for the API break.

---
 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   4 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  52 ++++++-------
 lib/eal/include/rte_mcslock.h            |  12 +--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 -
 lib/eal/ppc/include/rte_pause.h          |   2 -
 lib/eal/riscv/include/rte_pause.h        |   2 -
 lib/ring/rte_ring_core.h                 |   5 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 30 files changed, 265 insertions(+), 228 deletions(-)
app/test-bbdev/test_bbdev_perf.c +78 -45
index 276bbf0..c3a3a28 100644
@@ -143,7 +143,7 @@ struct test_op_params {
 	uint16_t num_to_process;
 	uint16_t num_lcores;
 	int vector_mask;
-	uint16_t sync;
+	uint16_t _Atomic sync;
 	struct test_buffers q_bufs[RTE_MAX_NUMA_NODES][MAX_QUEUES];
 };
 
@@ -158,7 +158,7 @@ struct thread_params {
 	uint8_t iter_count;
 	double iter_average;
 	double bler;
-	uint16_t nb_dequeued;
+	uint16_t _Atomic nb_dequeued;
 	int16_t processing_status;
 	uint16_t burst_sz;
 	struct test_op_params *op_params;
@@ -3021,27 +3021,32 @@ typedef int (test_case_function)(struct active_device *ad,
 	if (test_vector.op_type == RTE_BBDEV_OP_TURBO_DEC)
 		deq = rte_bbdev_dequeue_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_DEC)
 		deq = rte_bbdev_dequeue_ldpc_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_ENC)
 		deq = rte_bbdev_dequeue_ldpc_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_FFT)
 		deq = rte_bbdev_dequeue_fft_ops(dev_id, queue_id,
 				&tp->fft_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else /*RTE_BBDEV_OP_TURBO_ENC*/
 		deq = rte_bbdev_dequeue_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 
 	if (deq < burst_sz) {
@@ -3052,8 +3057,9 @@ typedef int (test_case_function)(struct active_device *ad,
 		return;
 	}
 
-	if (__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) + deq < num_ops) {
-		__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	if (atomic_load_explicit(&tp->nb_dequeued, memory_order_relaxed) + deq < num_ops) {
+		atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+			memory_order_relaxed);
 		return;
 	}
 
@@ -3126,7 +3132,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	tp->mbps += (((double)(num_ops * tb_len_bits)) / 1000000.0) /
 			((double)total_time / (double)rte_get_tsc_hz());
 
-	__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+		memory_order_relaxed);
 }
 
 static int
@@ -3165,9 +3172,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3222,10 +3230,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3262,9 +3272,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3313,10 +3324,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3352,9 +3365,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3399,10 +3413,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3439,9 +3455,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3488,10 +3505,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3528,9 +3547,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3575,10 +3595,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3613,7 +3635,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3732,7 +3755,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3867,7 +3891,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3990,7 +4015,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4121,7 +4147,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4222,7 +4249,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4323,7 +4351,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4519,7 +4548,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	else
 		return TEST_SKIPPED;
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4542,7 +4572,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = bler_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4641,7 +4672,8 @@ typedef int (test_case_function)(struct active_device *ad,
 			throughput_function = throughput_pmd_lcore_enc;
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4664,7 +4696,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = throughput_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4694,8 +4727,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	 * Wait for main lcore operations.
 	 */
 	tp = &t_params[0];
-	while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-		op_params->num_to_process) &&
+	while ((atomic_load_explicit(&tp->nb_dequeued,
+		memory_order_relaxed) < op_params->num_to_process) &&
 		(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 		TEST_FAILED))
 		rte_pause();
@@ -4708,8 +4741,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	for (used_cores = 1; used_cores < num_lcores; used_cores++) {
 		tp = &t_params[used_cores];
 
-		while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-			op_params->num_to_process) &&
+		while ((atomic_load_explicit(&tp->nb_dequeued,
+			memory_order_relaxed) < op_params->num_to_process) &&
 			(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 			TEST_FAILED))
 			rte_pause();
app/test/test_func_reentrancy.c +4 -4
index ae9de6f..833ceaf 100644
@@ -54,11 +54,11 @@
 #define MAX_LCORES	(rte_memzone_max_get() / (MAX_ITER_MULTI * 4U))
 
 static uint32_t obj_count;
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 #define WAIT_SYNCHRO_FOR_WORKERS()   do { \
 	if (lcore_self != rte_get_main_lcore())                  \
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED); \
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed); \
 } while(0)
 
 /*
@@ -438,7 +438,7 @@ struct test_case test_cases[] = {
 		return -1;
 
 	__atomic_store_n(&obj_count, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	cores = RTE_MIN(rte_lcore_count(), MAX_LCORES);
 	RTE_LCORE_FOREACH_WORKER(lcore_id) {
@@ -448,7 +448,7 @@ struct test_case test_cases[] = {
 		rte_eal_remote_launch(pt_case->func, pt_case->arg, lcore_id);
 	}
 
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	if (pt_case->func(pt_case->arg) < 0)
 		ret = -1;
app/test/test_mcslock.c +6 -6
index 52e45e7..3c26c69 100644
@@ -42,7 +42,7 @@
 
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_mcslock_per_core(__rte_unused void *arg)
@@ -75,7 +75,7 @@
 	rte_mcslock_t ml_perf_me;
 
 	/* wait synchro */
-	rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -100,14 +100,14 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
 	memset(time_count, 0, sizeof(time_count));
 
 	printf("\nTest with lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	lock = 1;
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
@@ -116,11 +116,11 @@
 
 	printf("\nTest with lock on %u cores...\n", (rte_lcore_count()));
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_mempool_perf.c +4 -4
index ce7c624..06f9fc1 100644
@@ -88,7 +88,7 @@
 static int use_external_cache;
 static unsigned external_cache_size = RTE_MEMPOOL_CACHE_MAX_SIZE;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 /* number of objects in one bulk operation (get or put) */
 static unsigned n_get_bulk;
@@ -188,7 +188,7 @@ struct mempool_test_stats {
 
 	/* wait synchro for workers */
 	if (lcore_id != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	start_cycles = rte_get_timer_cycles();
 
@@ -233,7 +233,7 @@ struct mempool_test_stats {
 	int ret;
 	unsigned cores_save = cores;
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	/* reset stats */
 	memset(stats, 0, sizeof(stats));
@@ -258,7 +258,7 @@ struct mempool_test_stats {
 	}
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	ret = per_lcore_mempool_test(mp);
 
app/test/test_pflock.c +6 -6
index 38da6bc..28addf8 100644
@@ -31,7 +31,7 @@
 
 static rte_pflock_t sl;
 static rte_pflock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_pflock_per_core(__rte_unused void *arg)
@@ -69,7 +69,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -99,7 +99,7 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -107,7 +107,7 @@
 
 	printf("\nTest with phase-fair lock on single core...\n");
 	lock = 1;
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -116,12 +116,12 @@
 	printf("\nPhase-fair test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_pmd_perf.c +5 -5
index 3ef590c..1562bbb 100644
@@ -537,7 +537,7 @@ enum {
 	return 0;
 }
 
-static uint64_t start;
+static uint64_t _Atomic start;
 
 static inline int
 poll_burst(void *args)
@@ -575,7 +575,7 @@ enum {
 		num[portid] = pkt_per_port;
 	}
 
-	rte_wait_until_equal_64(&start, 1, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_64(&start, 1, memory_order_acquire);
 
 	cur_tsc = rte_rdtsc();
 	while (total) {
@@ -629,9 +629,9 @@ enum {
 
 	/* only when polling first */
 	if (flags == SC_BURST_POLL_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 1, memory_order_relaxed);
 	else
-		__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 0, memory_order_relaxed);
 
 	/* start polling thread
 	 * if in POLL_FIRST mode, poll once launched;
@@ -655,7 +655,7 @@ enum {
 
 	/* only when polling second  */
 	if (flags == SC_BURST_XMIT_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELEASE);
+		atomic_store_explicit(&start, 1, memory_order_release);
 
 	/* wait for polling finished */
 	diff_tsc = rte_eal_wait_lcore(lcore);
app/test/test_ring_perf.c +4 -4
index 3972fd9..f0f5ef1 100644
@@ -320,7 +320,7 @@ struct thread_params {
 	return 0;
 }
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 static uint64_t queue_count[RTE_MAX_LCORE];
 
 #define TIME_MS 100
@@ -342,7 +342,7 @@ struct thread_params {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (time_diff < hz * TIME_MS / 1000) {
@@ -397,12 +397,12 @@ struct thread_params {
 		param.r = r;
 
 		/* clear synchro and start workers */
-		__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 		if (rte_eal_mp_remote_launch(lcore_f, &param, SKIP_MAIN) < 0)
 			return -1;
 
 		/* start synchro and launch test on main */
-		__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 		lcore_f(&param);
 
 		rte_eal_mp_wait_lcore();
app/test/test_rwlock.c +4 -4
index 4ae0bf8..dfbd0d6 100644
@@ -35,7 +35,7 @@
 
 static rte_rwlock_t sl;
 static rte_rwlock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 enum {
 	LC_TYPE_RDLOCK,
@@ -101,7 +101,7 @@ struct try_rwlock_lcore {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -134,12 +134,12 @@ struct try_rwlock_lcore {
 	printf("\nRwlock Perf Test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, NULL, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(NULL);
 
 	rte_eal_mp_wait_lcore();
app/test/test_spinlock.c +4 -4
index 3f59372..38724a1 100644
@@ -48,7 +48,7 @@
 static rte_spinlock_recursive_t slr;
 static unsigned count = 0;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_spinlock_per_core(__rte_unused void *arg)
@@ -110,7 +110,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -149,11 +149,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_stack_perf.c +6 -6
index 1eae00a..67510d6 100644
@@ -23,7 +23,7 @@
  */
 static volatile unsigned int bulk_sizes[] = {8, MAX_BURST};
 
-static uint32_t lcore_barrier;
+static uint32_t _Atomic lcore_barrier;
 
 struct lcore_pair {
 	unsigned int c1;
@@ -143,8 +143,8 @@ struct thread_args {
 	s = args->s;
 	size = args->sz;
 
-	__atomic_fetch_sub(&lcore_barrier, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_32(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_fetch_sub_explicit(&lcore_barrier, 1, memory_order_relaxed);
+	rte_wait_until_equal_32(&lcore_barrier, 0, memory_order_relaxed);
 
 	uint64_t start = rte_rdtsc();
 
@@ -173,7 +173,7 @@ struct thread_args {
 	unsigned int i;
 
 	for (i = 0; i < RTE_DIM(bulk_sizes); i++) {
-		__atomic_store_n(&lcore_barrier, 2, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, 2, memory_order_relaxed);
 
 		args[0].sz = args[1].sz = bulk_sizes[i];
 		args[0].s = args[1].s = s;
@@ -206,7 +206,7 @@ struct thread_args {
 		int cnt = 0;
 		double avg;
 
-		__atomic_store_n(&lcore_barrier, n, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, n, memory_order_relaxed);
 
 		RTE_LCORE_FOREACH_WORKER(lcore_id) {
 			if (++cnt >= n)
@@ -300,7 +300,7 @@ struct thread_args {
 	struct lcore_pair cores;
 	struct rte_stack *s;
 
-	__atomic_store_n(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&lcore_barrier, 0, memory_order_relaxed);
 
 	s = rte_stack_create(STACK_NAME, STACK_SIZE, rte_socket_id(), flags);
 	if (s == NULL) {
app/test/test_ticketlock.c +4 -4
index 242c136..f12d1e5 100644
@@ -48,7 +48,7 @@
 static rte_ticketlock_recursive_t tlr;
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_ticketlock_per_core(__rte_unused void *arg)
@@ -111,7 +111,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcore_count[lcore] < MAX_LOOP) {
@@ -153,11 +153,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_timer.c +8 -8
index 0c36dc9..cf89a19 100644
@@ -202,7 +202,7 @@ struct mytimerinfo {
 
 /* Need to synchronize worker lcores through multiple steps. */
 enum { WORKER_WAITING = 1, WORKER_RUN_SIGNAL, WORKER_RUNNING, WORKER_FINISHED };
-static uint16_t lcore_state[RTE_MAX_LCORE];
+static uint16_t _Atomic lcore_state[RTE_MAX_LCORE];
 
 static void
 main_init_workers(void)
@@ -210,7 +210,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_WAITING, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_state[i], WORKER_WAITING, memory_order_relaxed);
 	}
 }
 
@@ -220,10 +220,10 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_RUN_SIGNAL, __ATOMIC_RELEASE);
+		atomic_store_explicit(&lcore_state[i], WORKER_RUN_SIGNAL, memory_order_release);
 	}
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, memory_order_acquire);
 	}
 }
 
@@ -233,7 +233,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, memory_order_acquire);
 	}
 }
 
@@ -242,8 +242,8 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, __ATOMIC_ACQUIRE);
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_RUNNING, __ATOMIC_RELEASE);
+	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, memory_order_acquire);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_RUNNING, memory_order_release);
 }
 
 static void
@@ -251,7 +251,7 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_FINISHED, __ATOMIC_RELEASE);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_FINISHED, memory_order_release);
 }
 
 
drivers/event/opdl/opdl_ring.c +26 -21

120 lines not on this page · raw message

drivers/net/thunderx/nicvf_rxtx.c +3 -2
index defa551..0db9505 100644
@@ -385,9 +385,10 @@
 		ltail++;
 	}
 
-	rte_wait_until_equal_32(&rbdr->tail, next_tail, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&rbdr->tail, next_tail,
+		memory_order_relaxed);
 
-	__atomic_store_n(&rbdr->tail, ltail, __ATOMIC_RELEASE);
+	atomic_store_explicit(&rbdr->tail, ltail, memory_order_release);
 	nicvf_addr_write(door, to_fill);
 	return to_fill;
 }
drivers/net/thunderx/nicvf_struct.h +1 -1
index 13cf8fe..38c72b6 100644
@@ -20,7 +20,7 @@ struct nicvf_rbdr {
 	struct rbdr_entry_t *desc;
 	nicvf_iova_addr_t phys;
 	uint32_t buffsz;
-	uint32_t tail;
+	uint32_t _Atomic tail;
 	uint32_t next_tail;
 	uint32_t head;
 	uint32_t qlen_mask;
lib/bpf/bpf_pkt.c +2 -2
index ffd2db7..f5765c2 100644
@@ -25,7 +25,7 @@
 
 struct bpf_eth_cbi {
 	/* used by both data & control path */
-	uint32_t use;    /*usage counter */
+	uint32_t _Atomic use;    /*usage counter */
 	const struct rte_eth_rxtx_callback *cb;  /* callback handle */
 	struct rte_bpf *bpf;
 	struct rte_bpf_jit jit;
@@ -110,7 +110,7 @@ struct bpf_eth_cbh {
 
 	/* in use, busy wait till current RX/TX iteration is finished */
 	if ((puse & BPF_ETH_CBI_INUSE) != 0) {
-		RTE_WAIT_UNTIL_MASKED((uint32_t *)(uintptr_t)&cbi->use,
+		RTE_WAIT_UNTIL_MASKED(&cbi->use,
 			UINT32_MAX, !=, puse, __ATOMIC_RELAXED);
 	}
 }
lib/distributor/distributor_private.h +1 -1
index 7101f63..a5edb34 100644
@@ -52,7 +52,7 @@
  * Only 64-bits of the memory is actually used though.
  */
 union rte_distributor_buffer_single {
-	volatile int64_t bufptr64;
+	volatile int64_t _Atomic bufptr64;
 	char pad[RTE_CACHE_LINE_SIZE*3];
 } __rte_cache_aligned;
 
lib/distributor/rte_distributor_single.c +22 -22

117 lines not on this page · raw message

lib/eal/arm/include/rte_pause_64.h +14 -14
index 5f70e97..96ad050 100644
@@ -148,13 +148,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_16(volatile uint16_t *addr, uint16_t expected,
-		int memorder)
+rte_wait_until_equal_16(volatile uint16_t _Atomic *addr, uint16_t expected,
+		memory_order memorder)
 {
 	uint16_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_16(addr, value, memorder)
 	if (value != expected) {
@@ -167,13 +167,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_32(volatile uint32_t *addr, uint32_t expected,
-		int memorder)
+rte_wait_until_equal_32(volatile uint32_t _Atomic *addr, uint32_t expected,
+		memory_order memorder)
 {
 	uint32_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_32(addr, value, memorder)
 	if (value != expected) {
@@ -186,13 +186,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_64(volatile uint64_t *addr, uint64_t expected,
-		int memorder)
+rte_wait_until_equal_64(volatile uint64_t _Atomic *addr, uint64_t expected,
+		memory_order memorder)
 {
 	uint64_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_64(addr, value, memorder)
 	if (value != expected) {
@@ -206,8 +206,8 @@ static inline void rte_pause(void)
 
 #define RTE_WAIT_UNTIL_MASKED(addr, mask, cond, expected, memorder) do {  \
 	RTE_BUILD_BUG_ON(!__builtin_constant_p(memorder));                \
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&                  \
-		memorder != __ATOMIC_RELAXED);                            \
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&              \
+		memorder != memory_order_relaxed);                        \
 	const uint32_t size = sizeof(*(addr)) << 3;                       \
 	typeof(*(addr)) expected_value = (expected);                      \
 	typeof(*(addr)) value;                                            \
lib/eal/common/eal_memcfg.h +1 -1
index 8889ba0..29e19b3 100644
@@ -18,7 +18,7 @@
  * Memory configuration shared across multiple processes.
  */
 struct rte_mem_config {
-	volatile uint32_t magic;   /**< Magic number - sanity check. */
+	volatile uint32_t _Atomic magic;   /**< Magic number - sanity check. */
 	uint32_t version;
 	/**< Prevent secondary processes using different DPDK versions. */
 
lib/eal/include/generic/rte_pause.h +26 -26

120 lines not on this page · raw message

lib/eal/include/rte_mcslock.h +6 -6
index a805cb2..7c227ec 100644
@@ -33,7 +33,7 @@
  */
 typedef struct rte_mcslock {
 	struct rte_mcslock *next;
-	int locked; /* 1 if the queue locked, 0 otherwise */
+	uint32_t _Atomic locked; /* 1 if the queue locked, 0 otherwise */
 } rte_mcslock_t;
 
 /**
@@ -53,7 +53,7 @@
 	rte_mcslock_t *prev;
 
 	/* Init me node */
-	__atomic_store_n(&me->locked, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&me->locked, 1, memory_order_relaxed);
 	__atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
 
 	/* If the queue is empty, the exchange operation is enough to acquire
@@ -88,7 +88,7 @@
 	 * to spin on me->locked until the previous lock holder resets
 	 * the me->locked using mcslock_unlock().
 	 */
-	rte_wait_until_equal_32((uint32_t *)&me->locked, 0, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_32(&me->locked, 0, memory_order_acquire);
 }
 
 /**
@@ -120,14 +120,14 @@
 		/* More nodes added to the queue by other CPUs.
 		 * Wait until the next pointer is set.
 		 */
-		uintptr_t *next;
-		next = (uintptr_t *)&me->next;
+		uintptr_t _Atomic *next;
+		next = (uintptr_t _Atomic *)&me->next;
 		RTE_WAIT_UNTIL_MASKED(next, UINTPTR_MAX, !=, 0,
 			__ATOMIC_RELAXED);
 	}
 
 	/* Pass lock to next waiter. */
-	__atomic_store_n(&me->next->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&me->next->locked, 0, memory_order_release);
 }
 
 /**
lib/eal/include/rte_pflock.h +11 -11
index a3f7291..1c32512 100644
@@ -40,8 +40,8 @@
  */
 struct rte_pflock {
 	struct {
-		uint16_t in;
-		uint16_t out;
+		uint16_t _Atomic in;
+		uint16_t _Atomic out;
 	} rd, wr;
 };
 typedef struct rte_pflock rte_pflock_t;
@@ -116,14 +116,14 @@ struct rte_pflock {
 	 * If no writer is present, then the operation has completed
 	 * successfully.
 	 */
-	w = __atomic_fetch_add(&pf->rd.in, RTE_PFLOCK_RINC, __ATOMIC_ACQUIRE)
+	w = atomic_fetch_add_explicit(&pf->rd.in, RTE_PFLOCK_RINC, memory_order_acquire)
 		& RTE_PFLOCK_WBITS;
 	if (w == 0)
 		return;
 
 	/* Wait for current write phase to complete. */
 	RTE_WAIT_UNTIL_MASKED(&pf->rd.in, RTE_PFLOCK_WBITS, !=, w,
-		__ATOMIC_ACQUIRE);
+		memory_order_acquire);
 }
 
 /**
@@ -139,7 +139,7 @@ struct rte_pflock {
 static inline void
 rte_pflock_read_unlock(rte_pflock_t *pf)
 {
-	__atomic_fetch_add(&pf->rd.out, RTE_PFLOCK_RINC, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->rd.out, RTE_PFLOCK_RINC, memory_order_release);
 }
 
 /**
@@ -160,8 +160,8 @@ struct rte_pflock {
 	/* Acquire ownership of write-phase.
 	 * This is same as rte_ticketlock_lock().
 	 */
-	ticket = __atomic_fetch_add(&pf->wr.in, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&pf->wr.out, ticket, __ATOMIC_ACQUIRE);
+	ticket = atomic_fetch_add_explicit(&pf->wr.in, 1, memory_order_relaxed);
+	rte_wait_until_equal_16(&pf->wr.out, ticket, memory_order_acquire);
 
 	/*
 	 * Acquire ticket on read-side in order to allow them
@@ -172,10 +172,10 @@ struct rte_pflock {
 	 * speculatively.
 	 */
 	w = RTE_PFLOCK_PRES | (ticket & RTE_PFLOCK_PHID);
-	ticket = __atomic_fetch_add(&pf->rd.in, w, __ATOMIC_RELAXED);
+	ticket = atomic_fetch_add_explicit(&pf->rd.in, w, memory_order_relaxed);
 
 	/* Wait for any pending readers to flush. */
-	rte_wait_until_equal_16(&pf->rd.out, ticket, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&pf->rd.out, ticket, memory_order_acquire);
 }
 
 /**
@@ -192,10 +192,10 @@ struct rte_pflock {
 rte_pflock_write_unlock(rte_pflock_t *pf)
 {
 	/* Migrate from write phase to read phase. */
-	__atomic_fetch_and(&pf->rd.in, RTE_PFLOCK_LSB, __ATOMIC_RELEASE);
+	atomic_fetch_and_explicit(&pf->rd.in, RTE_PFLOCK_LSB, memory_order_release);
 
 	/* Allow other writers to continue. */
-	__atomic_fetch_add(&pf->wr.out, 1, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->wr.out, 1, memory_order_release);
 }
 
 #ifdef __cplusplus
lib/eal/include/rte_ticketlock.h +4 -4
index 5db0d8a..5206b62 100644
@@ -31,7 +31,7 @@
 typedef union {
 	uint32_t tickets;
 	struct {
-		uint16_t current;
+		uint16_t _Atomic current;
 		uint16_t next;
 	} s;
 } rte_ticketlock_t;
@@ -63,7 +63,7 @@
 rte_ticketlock_lock(rte_ticketlock_t *tl)
 {
 	uint16_t me = __atomic_fetch_add(&tl->s.next, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&tl->s.current, me, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&tl->s.current, me, memory_order_acquire);
 }
 
 /**
@@ -75,8 +75,8 @@
 static inline void
 rte_ticketlock_unlock(rte_ticketlock_t *tl)
 {
-	uint16_t i = __atomic_load_n(&tl->s.current, __ATOMIC_RELAXED);
-	__atomic_store_n(&tl->s.current, i + 1, __ATOMIC_RELEASE);
+	uint16_t i = atomic_load_explicit(&tl->s.current, memory_order_relaxed);
+	atomic_store_explicit(&tl->s.current, i + 1, memory_order_release);
 }
 
 /**
lib/eal/loongarch/include/rte_pause.h +0 -2
index 4302e1b..2987a1b 100644
@@ -9,8 +9,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/eal/ppc/include/rte_pause.h +0 -2
index 16e47ce..54bbbc7 100644
@@ -9,8 +9,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/eal/riscv/include/rte_pause.h +0 -2
index cb8e9ca..ffa7158 100644
@@ -11,8 +11,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/ring/rte_ring_core.h +2 -3
index 82b2370..8356037 100644
@@ -31,7 +31,6 @@
 #include <rte_config.h>
 #include <rte_memory.h>
 #include <rte_lcore.h>
-#include <rte_atomic.h>
 #include <rte_branch_prediction.h>
 #include <rte_memzone.h>
 #include <rte_pause.h>
@@ -66,8 +65,8 @@ enum rte_ring_sync_type {
  * but offset for *sync_type* and *tail* values should remain the same.
  */
 struct rte_ring_headtail {
-	volatile uint32_t head;      /**< prod/consumer head. */
-	volatile uint32_t tail;      /**< prod/consumer tail. */
+	volatile uint32_t _Atomic head;      /**< prod/consumer head. */
+	volatile uint32_t _Atomic tail;      /**< prod/consumer tail. */
 	RTE_STD_C11
 	union {
 		/** sync type of prod/cons */
lib/ring/rte_ring_generic_pvt.h +12 -7
index 5acb6e5..12a3ca8 100644
@@ -89,11 +89,14 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sp)
-			r->prod.head = *new_head, success = 1;
-		else
-			success = rte_atomic32_cmpset(&r->prod.head,
-					*old_head, *new_head);
+		if (is_sp) {
+			r->prod.head = *new_head;
+			success = 1;
+		} else
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->prod.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 	} while (unlikely(success == 0));
 	return n;
 }
@@ -162,8 +165,10 @@
 			rte_smp_rmb();
 			success = 1;
 		} else {
-			success = rte_atomic32_cmpset(&r->cons.head, *old_head,
-					*new_head);
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->cons.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 		}
 	} while (unlikely(success == 0));
 	return n;
lib/ring/rte_ring_peek_elem_pvt.h +1 -1
index bb0a7d5..6707e38 100644
@@ -59,7 +59,7 @@
 
 	pos = tail + num;
 	ht->head = pos;
-	__atomic_store_n(&ht->tail, pos, __ATOMIC_RELEASE);
+	atomic_store_explicit(&ht->tail, pos, memory_order_release);
 }
 
 /**
Signature · 1 line
1.8.3.1
#6 lore raw

PATCH v2 1/4 build: require minimum c11 compiler

In reply to Tyler Retzlaff

Update top level meson.build to require c_std=c11 and remove use of -std=gnu99.

---
 drivers/net/failsafe/meson.build  | 1 -
 lib/eal/common/eal_common_errno.c | 1 +
 meson.build                       | 1 +
 3 files changed, 2 insertions(+), 1 deletion(-)
drivers/net/failsafe/meson.build +0 -1
index 6013e13..c1d3610 100644
@@ -7,7 +7,6 @@ if is_windows
     subdir_done()
 endif
 
-cflags += '-std=gnu99'
 cflags += '-D_DEFAULT_SOURCE'
 cflags += '-D_XOPEN_SOURCE=700'
 cflags += '-pedantic'
lib/eal/common/eal_common_errno.c +1 -0
index ef8f782..d4e2c13 100644
@@ -4,6 +4,7 @@
 
 /* Use XSI-compliant portable version of strerror_r() */
 #undef _GNU_SOURCE
+#define _POSIX_C_SOURCE 200112L
 
 #include <stdio.h>
 #include <string.h>
meson.build +1 -0
index 39cb738..70b54f0 100644
@@ -9,6 +9,7 @@ project('DPDK', 'c',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c11',
             'default_library=static',
             'warning_level=2',
         ],
Signature · 1 line
1.8.3.1
#7 lore raw

PATCH v2 2/4 devtools: forbid use of GCC atomic builtins

In reply to Tyler Retzlaff

Refrain from using compiler __atomic_xxx builtins DPDK now uses standard C11 atomics atomic_xxx.

---
 devtools/checkpatches.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
devtools/checkpatches.sh +8 -0
index 43f5e36..a32f02e 100755
@@ -102,6 +102,14 @@ check_forbidden_additions() { # <patch>
 		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
 		"$1" || res=1
 
+	# refrain from using compiler __atomic_xxx builtins
+	awk -v FOLDERS="lib drivers app examples" \
+		-v EXPRESSIONS="__atomic_.*\\\(" \
+		-v RET_ON_FAIL=1 \
+		-v MESSAGE='Using __atomic_xxx builtins' \
+		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
+		"$1" || res=1
+
 	# refrain from using compiler __atomic_thread_fence()
 	# It should be avoided on x86 for SMP case.
 	awk -v FOLDERS="lib drivers app examples" \
Signature · 1 line
1.8.3.1
#5 lore raw

v2 cover eal: update public API to use stdatomic atomics

In reply to Tyler Retzlaff

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not do not straddle the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler here to be dropped from this series.

v2:
* remove volatile qualification from variables, structure fields and parameters that have been qualified _Atomic
* restore removal of #include <rte_atomic.h> in rte_ring_core.h the ring library still uses other rte_atomic APIs and types that are not replaced by the standard or compiler intrinsics
* add missing adaption to stdatomic in ring/rte_ring_c11_pvt.h

Tyler Retzlaff (4):
build: require minimum c11 compiler
devtools: forbid use of GCC atomic builtins
eal: adapt rte pause APIs to use C11 atomics
eal: adapt rte spinlock and rwlock APIs to use C11 atomics

 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 devtools/checkpatches.sh                 |   8 ++
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/failsafe/meson.build         |   1 -
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   4 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_common_errno.c        |   1 +
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  52 ++++++-------
 lib/eal/include/generic/rte_rwlock.h     |  46 ++++++------
 lib/eal/include/generic/rte_spinlock.h   |  21 +++---
 lib/eal/include/rte_mcslock.h            |  12 +--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 -
 lib/eal/ppc/include/rte_pause.h          |   2 -
 lib/eal/riscv/include/rte_pause.h        |   2 -
 lib/eal/x86/include/rte_spinlock.h       |   2 +-
 lib/ring/rte_ring_c11_pvt.h              |  28 +++----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 meson.build                              |   1 +
 38 files changed, 326 insertions(+), 274 deletions(-)
Signature · 1 line
1.8.3.1
#9 lore raw

PATCH v2 4/4 eal: adapt rte spinlock and rwlock APIs to use C11 atomics

Adapt rte_spinlock.h and rte_rwlock.h APIs to use standard C11 atomics. Update consumers of the spinlock and rwlock APIs for the API break.

---
 lib/eal/include/generic/rte_rwlock.h   | 46 ++++++++++++++++++----------------
 lib/eal/include/generic/rte_spinlock.h | 21 ++++++++--------
 lib/eal/x86/include/rte_spinlock.h     |  2 +-
 3 files changed, 36 insertions(+), 33 deletions(-)
lib/eal/include/generic/rte_rwlock.h +24 -22
index 9e083bb..b659c4c 100644
@@ -22,6 +22,8 @@
  *  https://locklessinc.com/articles/locks/
  */
 
+#include <stdatomic.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -57,7 +59,7 @@
 #define RTE_RWLOCK_READ	 0x4	/* Reader increment */
 
 typedef struct __rte_lockable {
-	int32_t cnt;
+	int32_t _Atomic cnt;
 } rte_rwlock_t;
 
 /**
@@ -92,21 +94,21 @@
 
 	while (1) {
 		/* Wait while writer is present or pending */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed)
 		       & RTE_RWLOCK_MASK)
 			rte_pause();
 
 		/* Try to get read lock */
-		x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-				       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+		x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				       memory_order_acquire) + RTE_RWLOCK_READ;
 
 		/* If no writer, then acquire was successful */
 		if (likely(!(x & RTE_RWLOCK_MASK)))
 			return;
 
 		/* Lost race with writer, backout the change. */
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELAXED);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_relaxed);
 	}
 }
 
@@ -127,20 +129,20 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 	/* fail if write lock is held or writer is pending */
 	if (x & RTE_RWLOCK_MASK)
 		return -EBUSY;
 
 	/* Try to get read lock */
-	x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-			       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+	x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+			       memory_order_acquire) + RTE_RWLOCK_READ;
 
 	/* Back out if writer raced in */
 	if (unlikely(x & RTE_RWLOCK_MASK)) {
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELEASE);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_release);
 
 		return -EBUSY;
 	}
@@ -158,7 +160,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ, memory_order_release);
 }
 
 /**
@@ -178,10 +180,10 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 	if (x < RTE_RWLOCK_WRITE &&
-	    __atomic_compare_exchange_n(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
-					1, __ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+	    atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
+					memory_order_acquire, memory_order_relaxed))
 		return 0;
 	else
 		return -EBUSY;
@@ -201,22 +203,22 @@
 	int32_t x;
 
 	while (1) {
-		x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+		x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 		/* No readers or writers? */
 		if (likely(x < RTE_RWLOCK_WRITE)) {
 			/* Turn off RTE_RWLOCK_WAIT, turn on RTE_RWLOCK_WRITE */
-			if (__atomic_compare_exchange_n(&rwl->cnt, &x, RTE_RWLOCK_WRITE, 1,
-							__ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+			if (atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, RTE_RWLOCK_WRITE,
+							memory_order_acquire, memory_order_relaxed))
 				return;
 		}
 
 		/* Turn on writer wait bit */
 		if (!(x & RTE_RWLOCK_WAIT))
-			__atomic_fetch_or(&rwl->cnt, RTE_RWLOCK_WAIT, __ATOMIC_RELAXED);
+			atomic_fetch_or_explicit(&rwl->cnt, RTE_RWLOCK_WAIT, memory_order_relaxed);
 
 		/* Wait until no readers before trying again */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) > RTE_RWLOCK_WAIT)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) > RTE_RWLOCK_WAIT)
 			rte_pause();
 
 	}
@@ -233,7 +235,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_WRITE, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_WRITE, memory_order_release);
 }
 
 /**
@@ -247,7 +249,7 @@
 static inline int
 rte_rwlock_write_is_locked(rte_rwlock_t *rwl)
 {
-	if (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) & RTE_RWLOCK_WRITE)
+	if (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) & RTE_RWLOCK_WRITE)
 		return 1;
 
 	return 0;
lib/eal/include/generic/rte_spinlock.h +11 -10
index c50ebaa..d92432d 100644
@@ -17,6 +17,8 @@
  * All locks must be initialised before use, and only initialised once.
  */
 
+#include <stdatomic.h>
+
 #include <rte_lcore.h>
 #ifdef RTE_FORCE_INTRINSICS
 #include <rte_common.h>
@@ -28,7 +30,7 @@
  * The rte_spinlock_t type.
  */
 typedef struct __rte_lockable {
-	volatile int locked; /**< lock status 0 = unlocked, 1 = locked */
+	int _Atomic locked; /**< lock status 0 = unlocked, 1 = locked */
 } rte_spinlock_t;
 
 /**
@@ -65,10 +67,10 @@
 {
 	int exp = 0;
 
-	while (!__atomic_compare_exchange_n(&sl->locked, &exp, 1, 0,
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED)) {
-		rte_wait_until_equal_32((volatile uint32_t *)&sl->locked,
-			       0, __ATOMIC_RELAXED);
+	while (!atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed)) {
+		rte_wait_until_equal_32((uint32_t _Atomic *)&sl->locked,
+			       0, memory_order_relaxed);
 		exp = 0;
 	}
 }
@@ -89,7 +91,7 @@
 rte_spinlock_unlock(rte_spinlock_t *sl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_store_n(&sl->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&sl->locked, 0, memory_order_release);
 }
 #endif
 
@@ -112,9 +114,8 @@
 	__rte_no_thread_safety_analysis
 {
 	int exp = 0;
-	return __atomic_compare_exchange_n(&sl->locked, &exp, 1,
-				0, /* disallow spurious failure */
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED);
+	return atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed);
 }
 #endif
 
@@ -128,7 +129,7 @@
  */
 static inline int rte_spinlock_is_locked (rte_spinlock_t *sl)
 {
-	return __atomic_load_n(&sl->locked, __ATOMIC_ACQUIRE);
+	return atomic_load_explicit(&sl->locked, memory_order_acquire);
 }
 
 /**
lib/eal/x86/include/rte_spinlock.h +1 -1
index 0b20ddf..3e7f8ac 100644
@@ -78,7 +78,7 @@ static inline int rte_tm_supported(void)
 }
 
 static inline int
-rte_try_tm(volatile int *lock)
+rte_try_tm(int _Atomic *lock)
 {
 	int i, retries;
 
Signature · 1 line
1.8.3.1
#8 lore raw

PATCH v2 3/4 eal: adapt rte pause APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_pause.h APIs to use standard C11 atomics. Update consumers of the pause APIs for the API break.

---
 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   4 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  52 ++++++-------
 lib/eal/include/rte_mcslock.h            |  12 +--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 -
 lib/eal/ppc/include/rte_pause.h          |   2 -
 lib/eal/riscv/include/rte_pause.h        |   2 -
 lib/ring/rte_ring_c11_pvt.h              |  28 +++----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 31 files changed, 280 insertions(+), 240 deletions(-)
app/test-bbdev/test_bbdev_perf.c +78 -45
index 276bbf0..c3a3a28 100644
@@ -143,7 +143,7 @@ struct test_op_params {
 	uint16_t num_to_process;
 	uint16_t num_lcores;
 	int vector_mask;
-	uint16_t sync;
+	uint16_t _Atomic sync;
 	struct test_buffers q_bufs[RTE_MAX_NUMA_NODES][MAX_QUEUES];
 };
 
@@ -158,7 +158,7 @@ struct thread_params {
 	uint8_t iter_count;
 	double iter_average;
 	double bler;
-	uint16_t nb_dequeued;
+	uint16_t _Atomic nb_dequeued;
 	int16_t processing_status;
 	uint16_t burst_sz;
 	struct test_op_params *op_params;
@@ -3021,27 +3021,32 @@ typedef int (test_case_function)(struct active_device *ad,
 	if (test_vector.op_type == RTE_BBDEV_OP_TURBO_DEC)
 		deq = rte_bbdev_dequeue_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_DEC)
 		deq = rte_bbdev_dequeue_ldpc_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_ENC)
 		deq = rte_bbdev_dequeue_ldpc_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_FFT)
 		deq = rte_bbdev_dequeue_fft_ops(dev_id, queue_id,
 				&tp->fft_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else /*RTE_BBDEV_OP_TURBO_ENC*/
 		deq = rte_bbdev_dequeue_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 
 	if (deq < burst_sz) {
@@ -3052,8 +3057,9 @@ typedef int (test_case_function)(struct active_device *ad,
 		return;
 	}
 
-	if (__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) + deq < num_ops) {
-		__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	if (atomic_load_explicit(&tp->nb_dequeued, memory_order_relaxed) + deq < num_ops) {
+		atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+			memory_order_relaxed);
 		return;
 	}
 
@@ -3126,7 +3132,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	tp->mbps += (((double)(num_ops * tb_len_bits)) / 1000000.0) /
 			((double)total_time / (double)rte_get_tsc_hz());
 
-	__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+		memory_order_relaxed);
 }
 
 static int
@@ -3165,9 +3172,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3222,10 +3230,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3262,9 +3272,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3313,10 +3324,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3352,9 +3365,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3399,10 +3413,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3439,9 +3455,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3488,10 +3505,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3528,9 +3547,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3575,10 +3595,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3613,7 +3635,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3732,7 +3755,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3867,7 +3891,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3990,7 +4015,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4121,7 +4147,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4222,7 +4249,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4323,7 +4351,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4519,7 +4548,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	else
 		return TEST_SKIPPED;
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4542,7 +4572,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = bler_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4641,7 +4672,8 @@ typedef int (test_case_function)(struct active_device *ad,
 			throughput_function = throughput_pmd_lcore_enc;
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4664,7 +4696,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = throughput_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4694,8 +4727,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	 * Wait for main lcore operations.
 	 */
 	tp = &t_params[0];
-	while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-		op_params->num_to_process) &&
+	while ((atomic_load_explicit(&tp->nb_dequeued,
+		memory_order_relaxed) < op_params->num_to_process) &&
 		(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 		TEST_FAILED))
 		rte_pause();
@@ -4708,8 +4741,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	for (used_cores = 1; used_cores < num_lcores; used_cores++) {
 		tp = &t_params[used_cores];
 
-		while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-			op_params->num_to_process) &&
+		while ((atomic_load_explicit(&tp->nb_dequeued,
+			memory_order_relaxed) < op_params->num_to_process) &&
 			(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 			TEST_FAILED))
 			rte_pause();
app/test/test_func_reentrancy.c +4 -4
index ae9de6f..833ceaf 100644
@@ -54,11 +54,11 @@
 #define MAX_LCORES	(rte_memzone_max_get() / (MAX_ITER_MULTI * 4U))
 
 static uint32_t obj_count;
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 #define WAIT_SYNCHRO_FOR_WORKERS()   do { \
 	if (lcore_self != rte_get_main_lcore())                  \
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED); \
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed); \
 } while(0)
 
 /*
@@ -438,7 +438,7 @@ struct test_case test_cases[] = {
 		return -1;
 
 	__atomic_store_n(&obj_count, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	cores = RTE_MIN(rte_lcore_count(), MAX_LCORES);
 	RTE_LCORE_FOREACH_WORKER(lcore_id) {
@@ -448,7 +448,7 @@ struct test_case test_cases[] = {
 		rte_eal_remote_launch(pt_case->func, pt_case->arg, lcore_id);
 	}
 
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	if (pt_case->func(pt_case->arg) < 0)
 		ret = -1;
app/test/test_mcslock.c +6 -6
index 52e45e7..3c26c69 100644
@@ -42,7 +42,7 @@
 
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_mcslock_per_core(__rte_unused void *arg)
@@ -75,7 +75,7 @@
 	rte_mcslock_t ml_perf_me;
 
 	/* wait synchro */
-	rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -100,14 +100,14 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
 	memset(time_count, 0, sizeof(time_count));
 
 	printf("\nTest with lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	lock = 1;
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
@@ -116,11 +116,11 @@
 
 	printf("\nTest with lock on %u cores...\n", (rte_lcore_count()));
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_mempool_perf.c +4 -4
index ce7c624..06f9fc1 100644
@@ -88,7 +88,7 @@
 static int use_external_cache;
 static unsigned external_cache_size = RTE_MEMPOOL_CACHE_MAX_SIZE;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 /* number of objects in one bulk operation (get or put) */
 static unsigned n_get_bulk;
@@ -188,7 +188,7 @@ struct mempool_test_stats {
 
 	/* wait synchro for workers */
 	if (lcore_id != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	start_cycles = rte_get_timer_cycles();
 
@@ -233,7 +233,7 @@ struct mempool_test_stats {
 	int ret;
 	unsigned cores_save = cores;
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	/* reset stats */
 	memset(stats, 0, sizeof(stats));
@@ -258,7 +258,7 @@ struct mempool_test_stats {
 	}
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	ret = per_lcore_mempool_test(mp);
 
app/test/test_pflock.c +6 -6
index 38da6bc..28addf8 100644
@@ -31,7 +31,7 @@
 
 static rte_pflock_t sl;
 static rte_pflock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_pflock_per_core(__rte_unused void *arg)
@@ -69,7 +69,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -99,7 +99,7 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -107,7 +107,7 @@
 
 	printf("\nTest with phase-fair lock on single core...\n");
 	lock = 1;
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -116,12 +116,12 @@
 	printf("\nPhase-fair test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_pmd_perf.c +5 -5
index 3ef590c..1562bbb 100644
@@ -537,7 +537,7 @@ enum {
 	return 0;
 }
 
-static uint64_t start;
+static uint64_t _Atomic start;
 
 static inline int
 poll_burst(void *args)
@@ -575,7 +575,7 @@ enum {
 		num[portid] = pkt_per_port;
 	}
 
-	rte_wait_until_equal_64(&start, 1, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_64(&start, 1, memory_order_acquire);
 
 	cur_tsc = rte_rdtsc();
 	while (total) {
@@ -629,9 +629,9 @@ enum {
 
 	/* only when polling first */
 	if (flags == SC_BURST_POLL_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 1, memory_order_relaxed);
 	else
-		__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 0, memory_order_relaxed);
 
 	/* start polling thread
 	 * if in POLL_FIRST mode, poll once launched;
@@ -655,7 +655,7 @@ enum {
 
 	/* only when polling second  */
 	if (flags == SC_BURST_XMIT_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELEASE);
+		atomic_store_explicit(&start, 1, memory_order_release);
 
 	/* wait for polling finished */
 	diff_tsc = rte_eal_wait_lcore(lcore);
app/test/test_ring_perf.c +4 -4
index 3972fd9..f0f5ef1 100644
@@ -320,7 +320,7 @@ struct thread_params {
 	return 0;
 }
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 static uint64_t queue_count[RTE_MAX_LCORE];
 
 #define TIME_MS 100
@@ -342,7 +342,7 @@ struct thread_params {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (time_diff < hz * TIME_MS / 1000) {
@@ -397,12 +397,12 @@ struct thread_params {
 		param.r = r;
 
 		/* clear synchro and start workers */
-		__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 		if (rte_eal_mp_remote_launch(lcore_f, &param, SKIP_MAIN) < 0)
 			return -1;
 
 		/* start synchro and launch test on main */
-		__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 		lcore_f(&param);
 
 		rte_eal_mp_wait_lcore();
app/test/test_rwlock.c +4 -4
index 4ae0bf8..dfbd0d6 100644
@@ -35,7 +35,7 @@
 
 static rte_rwlock_t sl;
 static rte_rwlock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 enum {
 	LC_TYPE_RDLOCK,
@@ -101,7 +101,7 @@ struct try_rwlock_lcore {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -134,12 +134,12 @@ struct try_rwlock_lcore {
 	printf("\nRwlock Perf Test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, NULL, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(NULL);
 
 	rte_eal_mp_wait_lcore();
app/test/test_spinlock.c +4 -4
index 3f59372..38724a1 100644
@@ -48,7 +48,7 @@
 static rte_spinlock_recursive_t slr;
 static unsigned count = 0;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_spinlock_per_core(__rte_unused void *arg)
@@ -110,7 +110,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -149,11 +149,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_stack_perf.c +6 -6
index 1eae00a..67510d6 100644
@@ -23,7 +23,7 @@
  */
 static volatile unsigned int bulk_sizes[] = {8, MAX_BURST};
 
-static uint32_t lcore_barrier;
+static uint32_t _Atomic lcore_barrier;
 
 struct lcore_pair {
 	unsigned int c1;
@@ -143,8 +143,8 @@ struct thread_args {
 	s = args->s;
 	size = args->sz;
 
-	__atomic_fetch_sub(&lcore_barrier, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_32(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_fetch_sub_explicit(&lcore_barrier, 1, memory_order_relaxed);
+	rte_wait_until_equal_32(&lcore_barrier, 0, memory_order_relaxed);
 
 	uint64_t start = rte_rdtsc();
 
@@ -173,7 +173,7 @@ struct thread_args {
 	unsigned int i;
 
 	for (i = 0; i < RTE_DIM(bulk_sizes); i++) {
-		__atomic_store_n(&lcore_barrier, 2, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, 2, memory_order_relaxed);
 
 		args[0].sz = args[1].sz = bulk_sizes[i];
 		args[0].s = args[1].s = s;
@@ -206,7 +206,7 @@ struct thread_args {
 		int cnt = 0;
 		double avg;
 
-		__atomic_store_n(&lcore_barrier, n, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, n, memory_order_relaxed);
 
 		RTE_LCORE_FOREACH_WORKER(lcore_id) {
 			if (++cnt >= n)
@@ -300,7 +300,7 @@ struct thread_args {
 	struct lcore_pair cores;
 	struct rte_stack *s;
 
-	__atomic_store_n(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&lcore_barrier, 0, memory_order_relaxed);
 
 	s = rte_stack_create(STACK_NAME, STACK_SIZE, rte_socket_id(), flags);
 	if (s == NULL) {
app/test/test_ticketlock.c +4 -4
index 242c136..f12d1e5 100644
@@ -48,7 +48,7 @@
 static rte_ticketlock_recursive_t tlr;
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_ticketlock_per_core(__rte_unused void *arg)
@@ -111,7 +111,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcore_count[lcore] < MAX_LOOP) {
@@ -153,11 +153,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_timer.c +8 -8
index 0c36dc9..cf89a19 100644
@@ -202,7 +202,7 @@ struct mytimerinfo {
 
 /* Need to synchronize worker lcores through multiple steps. */
 enum { WORKER_WAITING = 1, WORKER_RUN_SIGNAL, WORKER_RUNNING, WORKER_FINISHED };
-static uint16_t lcore_state[RTE_MAX_LCORE];
+static uint16_t _Atomic lcore_state[RTE_MAX_LCORE];
 
 static void
 main_init_workers(void)
@@ -210,7 +210,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_WAITING, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_state[i], WORKER_WAITING, memory_order_relaxed);
 	}
 }
 
@@ -220,10 +220,10 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_RUN_SIGNAL, __ATOMIC_RELEASE);
+		atomic_store_explicit(&lcore_state[i], WORKER_RUN_SIGNAL, memory_order_release);
 	}
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, memory_order_acquire);
 	}
 }
 
@@ -233,7 +233,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, memory_order_acquire);
 	}
 }
 
@@ -242,8 +242,8 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, __ATOMIC_ACQUIRE);
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_RUNNING, __ATOMIC_RELEASE);
+	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, memory_order_acquire);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_RUNNING, memory_order_release);
 }
 
 static void
@@ -251,7 +251,7 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_FINISHED, __ATOMIC_RELEASE);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_FINISHED, memory_order_release);
 }
 
 
drivers/event/opdl/opdl_ring.c +26 -21

120 lines not on this page · raw message

drivers/net/thunderx/nicvf_rxtx.c +3 -2
index defa551..0db9505 100644
@@ -385,9 +385,10 @@
 		ltail++;
 	}
 
-	rte_wait_until_equal_32(&rbdr->tail, next_tail, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&rbdr->tail, next_tail,
+		memory_order_relaxed);
 
-	__atomic_store_n(&rbdr->tail, ltail, __ATOMIC_RELEASE);
+	atomic_store_explicit(&rbdr->tail, ltail, memory_order_release);
 	nicvf_addr_write(door, to_fill);
 	return to_fill;
 }
drivers/net/thunderx/nicvf_struct.h +1 -1
index 13cf8fe..38c72b6 100644
@@ -20,7 +20,7 @@ struct nicvf_rbdr {
 	struct rbdr_entry_t *desc;
 	nicvf_iova_addr_t phys;
 	uint32_t buffsz;
-	uint32_t tail;
+	uint32_t _Atomic tail;
 	uint32_t next_tail;
 	uint32_t head;
 	uint32_t qlen_mask;
lib/bpf/bpf_pkt.c +2 -2
index ffd2db7..f5765c2 100644
@@ -25,7 +25,7 @@
 
 struct bpf_eth_cbi {
 	/* used by both data & control path */
-	uint32_t use;    /*usage counter */
+	uint32_t _Atomic use;    /*usage counter */
 	const struct rte_eth_rxtx_callback *cb;  /* callback handle */
 	struct rte_bpf *bpf;
 	struct rte_bpf_jit jit;
@@ -110,7 +110,7 @@ struct bpf_eth_cbh {
 
 	/* in use, busy wait till current RX/TX iteration is finished */
 	if ((puse & BPF_ETH_CBI_INUSE) != 0) {
-		RTE_WAIT_UNTIL_MASKED((uint32_t *)(uintptr_t)&cbi->use,
+		RTE_WAIT_UNTIL_MASKED(&cbi->use,
 			UINT32_MAX, !=, puse, __ATOMIC_RELAXED);
 	}
 }
lib/distributor/distributor_private.h +1 -1
index 7101f63..3b43d3d 100644
@@ -52,7 +52,7 @@
  * Only 64-bits of the memory is actually used though.
  */
 union rte_distributor_buffer_single {
-	volatile int64_t bufptr64;
+	int64_t _Atomic bufptr64;
 	char pad[RTE_CACHE_LINE_SIZE*3];
 } __rte_cache_aligned;
 
lib/distributor/rte_distributor_single.c +22 -22

117 lines not on this page · raw message

lib/eal/arm/include/rte_pause_64.h +14 -14
index 5f70e97..96ad050 100644
@@ -148,13 +148,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_16(volatile uint16_t *addr, uint16_t expected,
-		int memorder)
+rte_wait_until_equal_16(volatile uint16_t _Atomic *addr, uint16_t expected,
+		memory_order memorder)
 {
 	uint16_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_16(addr, value, memorder)
 	if (value != expected) {
@@ -167,13 +167,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_32(volatile uint32_t *addr, uint32_t expected,
-		int memorder)
+rte_wait_until_equal_32(volatile uint32_t _Atomic *addr, uint32_t expected,
+		memory_order memorder)
 {
 	uint32_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_32(addr, value, memorder)
 	if (value != expected) {
@@ -186,13 +186,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_64(volatile uint64_t *addr, uint64_t expected,
-		int memorder)
+rte_wait_until_equal_64(volatile uint64_t _Atomic *addr, uint64_t expected,
+		memory_order memorder)
 {
 	uint64_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_64(addr, value, memorder)
 	if (value != expected) {
@@ -206,8 +206,8 @@ static inline void rte_pause(void)
 
 #define RTE_WAIT_UNTIL_MASKED(addr, mask, cond, expected, memorder) do {  \
 	RTE_BUILD_BUG_ON(!__builtin_constant_p(memorder));                \
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&                  \
-		memorder != __ATOMIC_RELAXED);                            \
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&              \
+		memorder != memory_order_relaxed);                        \
 	const uint32_t size = sizeof(*(addr)) << 3;                       \
 	typeof(*(addr)) expected_value = (expected);                      \
 	typeof(*(addr)) value;                                            \
lib/eal/common/eal_memcfg.h +1 -1
index 8889ba0..0decc29 100644
@@ -18,7 +18,7 @@
  * Memory configuration shared across multiple processes.
  */
 struct rte_mem_config {
-	volatile uint32_t magic;   /**< Magic number - sanity check. */
+	uint32_t _Atomic magic;   /**< Magic number - sanity check. */
 	uint32_t version;
 	/**< Prevent secondary processes using different DPDK versions. */
 
lib/eal/include/generic/rte_pause.h +26 -26

120 lines not on this page · raw message

lib/eal/include/rte_mcslock.h +6 -6
index a805cb2..7c227ec 100644
@@ -33,7 +33,7 @@
  */
 typedef struct rte_mcslock {
 	struct rte_mcslock *next;
-	int locked; /* 1 if the queue locked, 0 otherwise */
+	uint32_t _Atomic locked; /* 1 if the queue locked, 0 otherwise */
 } rte_mcslock_t;
 
 /**
@@ -53,7 +53,7 @@
 	rte_mcslock_t *prev;
 
 	/* Init me node */
-	__atomic_store_n(&me->locked, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&me->locked, 1, memory_order_relaxed);
 	__atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
 
 	/* If the queue is empty, the exchange operation is enough to acquire
@@ -88,7 +88,7 @@
 	 * to spin on me->locked until the previous lock holder resets
 	 * the me->locked using mcslock_unlock().
 	 */
-	rte_wait_until_equal_32((uint32_t *)&me->locked, 0, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_32(&me->locked, 0, memory_order_acquire);
 }
 
 /**
@@ -120,14 +120,14 @@
 		/* More nodes added to the queue by other CPUs.
 		 * Wait until the next pointer is set.
 		 */
-		uintptr_t *next;
-		next = (uintptr_t *)&me->next;
+		uintptr_t _Atomic *next;
+		next = (uintptr_t _Atomic *)&me->next;
 		RTE_WAIT_UNTIL_MASKED(next, UINTPTR_MAX, !=, 0,
 			__ATOMIC_RELAXED);
 	}
 
 	/* Pass lock to next waiter. */
-	__atomic_store_n(&me->next->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&me->next->locked, 0, memory_order_release);
 }
 
 /**
lib/eal/include/rte_pflock.h +11 -11
index a3f7291..1c32512 100644
@@ -40,8 +40,8 @@
  */
 struct rte_pflock {
 	struct {
-		uint16_t in;
-		uint16_t out;
+		uint16_t _Atomic in;
+		uint16_t _Atomic out;
 	} rd, wr;
 };
 typedef struct rte_pflock rte_pflock_t;
@@ -116,14 +116,14 @@ struct rte_pflock {
 	 * If no writer is present, then the operation has completed
 	 * successfully.
 	 */
-	w = __atomic_fetch_add(&pf->rd.in, RTE_PFLOCK_RINC, __ATOMIC_ACQUIRE)
+	w = atomic_fetch_add_explicit(&pf->rd.in, RTE_PFLOCK_RINC, memory_order_acquire)
 		& RTE_PFLOCK_WBITS;
 	if (w == 0)
 		return;
 
 	/* Wait for current write phase to complete. */
 	RTE_WAIT_UNTIL_MASKED(&pf->rd.in, RTE_PFLOCK_WBITS, !=, w,
-		__ATOMIC_ACQUIRE);
+		memory_order_acquire);
 }
 
 /**
@@ -139,7 +139,7 @@ struct rte_pflock {
 static inline void
 rte_pflock_read_unlock(rte_pflock_t *pf)
 {
-	__atomic_fetch_add(&pf->rd.out, RTE_PFLOCK_RINC, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->rd.out, RTE_PFLOCK_RINC, memory_order_release);
 }
 
 /**
@@ -160,8 +160,8 @@ struct rte_pflock {
 	/* Acquire ownership of write-phase.
 	 * This is same as rte_ticketlock_lock().
 	 */
-	ticket = __atomic_fetch_add(&pf->wr.in, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&pf->wr.out, ticket, __ATOMIC_ACQUIRE);
+	ticket = atomic_fetch_add_explicit(&pf->wr.in, 1, memory_order_relaxed);
+	rte_wait_until_equal_16(&pf->wr.out, ticket, memory_order_acquire);
 
 	/*
 	 * Acquire ticket on read-side in order to allow them
@@ -172,10 +172,10 @@ struct rte_pflock {
 	 * speculatively.
 	 */
 	w = RTE_PFLOCK_PRES | (ticket & RTE_PFLOCK_PHID);
-	ticket = __atomic_fetch_add(&pf->rd.in, w, __ATOMIC_RELAXED);
+	ticket = atomic_fetch_add_explicit(&pf->rd.in, w, memory_order_relaxed);
 
 	/* Wait for any pending readers to flush. */
-	rte_wait_until_equal_16(&pf->rd.out, ticket, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&pf->rd.out, ticket, memory_order_acquire);
 }
 
 /**
@@ -192,10 +192,10 @@ struct rte_pflock {
 rte_pflock_write_unlock(rte_pflock_t *pf)
 {
 	/* Migrate from write phase to read phase. */
-	__atomic_fetch_and(&pf->rd.in, RTE_PFLOCK_LSB, __ATOMIC_RELEASE);
+	atomic_fetch_and_explicit(&pf->rd.in, RTE_PFLOCK_LSB, memory_order_release);
 
 	/* Allow other writers to continue. */
-	__atomic_fetch_add(&pf->wr.out, 1, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->wr.out, 1, memory_order_release);
 }
 
 #ifdef __cplusplus
lib/eal/include/rte_ticketlock.h +4 -4
index 5db0d8a..5206b62 100644
@@ -31,7 +31,7 @@
 typedef union {
 	uint32_t tickets;
 	struct {
-		uint16_t current;
+		uint16_t _Atomic current;
 		uint16_t next;
 	} s;
 } rte_ticketlock_t;
@@ -63,7 +63,7 @@
 rte_ticketlock_lock(rte_ticketlock_t *tl)
 {
 	uint16_t me = __atomic_fetch_add(&tl->s.next, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&tl->s.current, me, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&tl->s.current, me, memory_order_acquire);
 }
 
 /**
@@ -75,8 +75,8 @@
 static inline void
 rte_ticketlock_unlock(rte_ticketlock_t *tl)
 {
-	uint16_t i = __atomic_load_n(&tl->s.current, __ATOMIC_RELAXED);
-	__atomic_store_n(&tl->s.current, i + 1, __ATOMIC_RELEASE);
+	uint16_t i = atomic_load_explicit(&tl->s.current, memory_order_relaxed);
+	atomic_store_explicit(&tl->s.current, i + 1, memory_order_release);
 }
 
 /**
lib/eal/loongarch/include/rte_pause.h +0 -2
index 4302e1b..2987a1b 100644
@@ -9,8 +9,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/eal/ppc/include/rte_pause.h +0 -2
index 16e47ce..54bbbc7 100644
@@ -9,8 +9,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/eal/riscv/include/rte_pause.h +0 -2
index cb8e9ca..ffa7158 100644
@@ -11,8 +11,6 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
-
 #include "generic/rte_pause.h"
 
 static inline void rte_pause(void)
lib/ring/rte_ring_c11_pvt.h +15 -13
index f895950..402f819 100644
@@ -91,14 +91,15 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sp)
-			r->prod.head = *new_head, success = 1;
-		else
+		if (is_sp) {
+			r->prod.head = *new_head;
+			success = 1;
+		} else
 			/* on failure, *old_head is updated */
-			success = __atomic_compare_exchange_n(&r->prod.head,
+			success = atomic_compare_exchange_strong_explicit(&r->prod.head,
 					old_head, *new_head,
-					0, __ATOMIC_RELAXED,
-					__ATOMIC_RELAXED);
+					memory_order_relaxed,
+					memory_order_relaxed);
 	} while (unlikely(success == 0));
 	return n;
 }
@@ -137,7 +138,7 @@
 	int success;
 
 	/* move cons.head atomically */
-	*old_head = __atomic_load_n(&r->cons.head, __ATOMIC_RELAXED);
+	*old_head = atomic_load_explicit(&r->cons.head, memory_order_relaxed);
 	do {
 		/* Restore n as it may change every loop */
 		n = max;
@@ -166,14 +167,15 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sc)
-			r->cons.head = *new_head, success = 1;
-		else
+		if (is_sc) {
+			r->cons.head = *new_head;
+			success = 1;
+		} else
 			/* on failure, *old_head will be updated */
-			success = __atomic_compare_exchange_n(&r->cons.head,
+			success = atomic_compare_exchange_strong_explicit(&r->cons.head,
 							old_head, *new_head,
-							0, __ATOMIC_RELAXED,
-							__ATOMIC_RELAXED);
+							memory_order_relaxed,
+							memory_order_relaxed);
 	} while (unlikely(success == 0));
 	return n;
 }
lib/ring/rte_ring_core.h +2 -2
index 82b2370..cf3cb84 100644
@@ -66,8 +66,8 @@ enum rte_ring_sync_type {
  * but offset for *sync_type* and *tail* values should remain the same.
  */
 struct rte_ring_headtail {
-	volatile uint32_t head;      /**< prod/consumer head. */
-	volatile uint32_t tail;      /**< prod/consumer tail. */
+	uint32_t _Atomic head;      /**< prod/consumer head. */
+	uint32_t _Atomic tail;      /**< prod/consumer tail. */
 	RTE_STD_C11
 	union {
 		/** sync type of prod/cons */
lib/ring/rte_ring_generic_pvt.h +12 -7
index 5acb6e5..12a3ca8 100644
@@ -89,11 +89,14 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sp)
-			r->prod.head = *new_head, success = 1;
-		else
-			success = rte_atomic32_cmpset(&r->prod.head,
-					*old_head, *new_head);
+		if (is_sp) {
+			r->prod.head = *new_head;
+			success = 1;
+		} else
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->prod.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 	} while (unlikely(success == 0));
 	return n;
 }
@@ -162,8 +165,10 @@
 			rte_smp_rmb();
 			success = 1;
 		} else {
-			success = rte_atomic32_cmpset(&r->cons.head, *old_head,
-					*new_head);
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->cons.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 		}
 	} while (unlikely(success == 0));
 	return n;
lib/ring/rte_ring_peek_elem_pvt.h +1 -1
index bb0a7d5..6707e38 100644
@@ -59,7 +59,7 @@
 
 	pos = tail + num;
 	ht->head = pos;
-	__atomic_store_n(&ht->tail, pos, __ATOMIC_RELEASE);
+	atomic_store_explicit(&ht->tail, pos, memory_order_release);
 }
 
 /**
Signature · 1 line
1.8.3.1
#10 lore raw
In reply to Tyler Retzlaff
Quoted 29 lines

From: Tyler Retzlaff [mailto:[email protected]]
Sent: Tuesday, 1 August 2023 07.04

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not do not straddle the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler
here to be dropped from this series.

Quoted 7 lines

v2:
* remove volatile qualification from variables, structure fields and parameters that have been qualified _Atomic
* restore removal of #include <rte_atomic.h> in rte_ring_core.h the ring library still uses other rte_atomic APIs and types that are not replaced by the standard or compiler intrinsics
* add missing adaption to stdatomic in ring/rte_ring_c11_pvt.h

#11 lore raw
In reply to Tyler Retzlaff

top post

just a heads up on this series, please defer review until v3 is submitted.

clang is stricter in atomic type enforcement than gcc so a new revision is needed that removes the direct use of intrinsics replacing them with the standard atomics for all _Atomic types.

thanks

Quoted 81 lines · Tyler Retzlaff

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not do not straddle the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler here to be dropped from this series.

v2:
* remove volatile qualification from variables, structure fields and parameters that have been qualified _Atomic
* restore removal of #include <rte_atomic.h> in rte_ring_core.h the ring library still uses other rte_atomic APIs and types that are not replaced by the standard or compiler intrinsics
* add missing adaption to stdatomic in ring/rte_ring_c11_pvt.h

Tyler Retzlaff (4):
build: require minimum c11 compiler
devtools: forbid use of GCC atomic builtins
eal: adapt rte pause APIs to use C11 atomics
eal: adapt rte spinlock and rwlock APIs to use C11 atomics

 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 devtools/checkpatches.sh                 |   8 ++
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/failsafe/meson.build         |   1 -
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   4 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_common_errno.c        |   1 +
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  52 ++++++-------
 lib/eal/include/generic/rte_rwlock.h     |  46 ++++++------
 lib/eal/include/generic/rte_spinlock.h   |  21 +++---
 lib/eal/include/rte_mcslock.h            |  12 +--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 -
 lib/eal/ppc/include/rte_pause.h          |   2 -
 lib/eal/riscv/include/rte_pause.h        |   2 -
 lib/eal/x86/include/rte_spinlock.h       |   2 +-
 lib/ring/rte_ring_c11_pvt.h              |  28 +++----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 meson.build                              |   1 +
 38 files changed, 326 insertions(+), 274 deletions(-)
Signature · 1 line
1.8.3.1
#13 lore raw

PATCH v3 1/4 build: require minimum c11 compiler

In reply to Tyler Retzlaff

Update top level meson.build to require c_std=c11 and remove use of -std=gnu99.

---
 drivers/net/failsafe/meson.build  | 1 -
 lib/eal/common/eal_common_errno.c | 1 +
 meson.build                       | 1 +
 3 files changed, 2 insertions(+), 1 deletion(-)
drivers/net/failsafe/meson.build +0 -1
index 6013e13..c1d3610 100644
@@ -7,7 +7,6 @@ if is_windows
     subdir_done()
 endif
 
-cflags += '-std=gnu99'
 cflags += '-D_DEFAULT_SOURCE'
 cflags += '-D_XOPEN_SOURCE=700'
 cflags += '-pedantic'
lib/eal/common/eal_common_errno.c +1 -0
index ef8f782..d4e2c13 100644
@@ -4,6 +4,7 @@
 
 /* Use XSI-compliant portable version of strerror_r() */
 #undef _GNU_SOURCE
+#define _POSIX_C_SOURCE 200112L
 
 #include <stdio.h>
 #include <string.h>
meson.build +1 -0
index 39cb738..70b54f0 100644
@@ -9,6 +9,7 @@ project('DPDK', 'c',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c11',
             'default_library=static',
             'warning_level=2',
         ],
Signature · 1 line
1.8.3.1
#12 lore raw

v3 cover eal: update public API to use stdatomic atomics

In reply to Tyler Retzlaff

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not do not straddle the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler here to be dropped from this series.

v3:
* restore incorrectly dropped architecture specific #include "rte_paused.h" for loongarch, ppc, riscv
* convert parameters to stdatomic in RTE_WAIT_UNTIL_MASKED macro usage (missed previously)
* lib/ring missed some conversion to stdatomic on struct rte_ring _Atomic fields (picked up by clang which is more strict than gcc)
* drivers/mlx5 use stdatomic operations on struct rte_ring _Atomic fields (picked up by clang which is more strict than gcc)

v2:
* remove volatile qualification from variables, structure fields and parameters that have been qualified _Atomic
* restore removal of #include <rte_atomic.h> in rte_ring_core.h the ring library still uses other rte_atomic APIs and types that are not replaced by the standard or compiler intrinsics
* add missing adaption to stdatomic in ring/rte_ring_c11_pvt.h

Tyler Retzlaff (4):
build: require minimum c11 compiler
devtools: forbid use of GCC atomic builtins
eal: adapt rte pause APIs to use C11 atomics
eal: adapt rte spinlock and rwlock APIs to use C11 atomics

 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 devtools/checkpatches.sh                 |   8 ++
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/failsafe/meson.build         |   1 -
 drivers/net/mlx5/mlx5_hws_cnt.h          |   4 +-
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   6 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_common_errno.c        |   1 +
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  51 ++++++-------
 lib/eal/include/generic/rte_rwlock.h     |  46 ++++++------
 lib/eal/include/generic/rte_spinlock.h   |  21 +++---
 lib/eal/include/rte_mcslock.h            |  14 ++--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 +-
 lib/eal/ppc/include/rte_pause.h          |   2 +-
 lib/eal/riscv/include/rte_pause.h        |   2 +-
 lib/eal/x86/include/rte_spinlock.h       |   2 +-
 lib/ring/rte_ring_c11_pvt.h              |  42 ++++++-----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 meson.build                              |   1 +
 39 files changed, 340 insertions(+), 281 deletions(-)
Signature · 1 line
1.8.3.1
#14 lore raw

PATCH v3 2/4 devtools: forbid use of GCC atomic builtins

Refrain from using compiler __atomic_xxx builtins DPDK now uses standard C11 atomics atomic_xxx.

---
 devtools/checkpatches.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
devtools/checkpatches.sh +8 -0
index 43f5e36..a32f02e 100755
@@ -102,6 +102,14 @@ check_forbidden_additions() { # <patch>
 		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
 		"$1" || res=1
 
+	# refrain from using compiler __atomic_xxx builtins
+	awk -v FOLDERS="lib drivers app examples" \
+		-v EXPRESSIONS="__atomic_.*\\\(" \
+		-v RET_ON_FAIL=1 \
+		-v MESSAGE='Using __atomic_xxx builtins' \
+		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
+		"$1" || res=1
+
 	# refrain from using compiler __atomic_thread_fence()
 	# It should be avoided on x86 for SMP case.
 	awk -v FOLDERS="lib drivers app examples" \
Signature · 1 line
1.8.3.1
#15 lore raw

PATCH v3 3/4 eal: adapt rte pause APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_pause.h APIs to use standard C11 atomics. Update consumers of the pause APIs for the API break.

---
 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/mlx5/mlx5_hws_cnt.h          |   4 +-
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   6 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  51 ++++++-------
 lib/eal/include/rte_mcslock.h            |  14 ++--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/loongarch/include/rte_pause.h    |   2 +-
 lib/eal/ppc/include/rte_pause.h          |   2 +-
 lib/eal/riscv/include/rte_pause.h        |   2 +-
 lib/ring/rte_ring_c11_pvt.h              |  42 ++++++-----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 32 files changed, 294 insertions(+), 247 deletions(-)
app/test-bbdev/test_bbdev_perf.c +78 -45
index 276bbf0..c3a3a28 100644
@@ -143,7 +143,7 @@ struct test_op_params {
 	uint16_t num_to_process;
 	uint16_t num_lcores;
 	int vector_mask;
-	uint16_t sync;
+	uint16_t _Atomic sync;
 	struct test_buffers q_bufs[RTE_MAX_NUMA_NODES][MAX_QUEUES];
 };
 
@@ -158,7 +158,7 @@ struct thread_params {
 	uint8_t iter_count;
 	double iter_average;
 	double bler;
-	uint16_t nb_dequeued;
+	uint16_t _Atomic nb_dequeued;
 	int16_t processing_status;
 	uint16_t burst_sz;
 	struct test_op_params *op_params;
@@ -3021,27 +3021,32 @@ typedef int (test_case_function)(struct active_device *ad,
 	if (test_vector.op_type == RTE_BBDEV_OP_TURBO_DEC)
 		deq = rte_bbdev_dequeue_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_DEC)
 		deq = rte_bbdev_dequeue_ldpc_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_ENC)
 		deq = rte_bbdev_dequeue_ldpc_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_FFT)
 		deq = rte_bbdev_dequeue_fft_ops(dev_id, queue_id,
 				&tp->fft_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else /*RTE_BBDEV_OP_TURBO_ENC*/
 		deq = rte_bbdev_dequeue_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 
 	if (deq < burst_sz) {
@@ -3052,8 +3057,9 @@ typedef int (test_case_function)(struct active_device *ad,
 		return;
 	}
 
-	if (__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) + deq < num_ops) {
-		__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	if (atomic_load_explicit(&tp->nb_dequeued, memory_order_relaxed) + deq < num_ops) {
+		atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+			memory_order_relaxed);
 		return;
 	}
 
@@ -3126,7 +3132,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	tp->mbps += (((double)(num_ops * tb_len_bits)) / 1000000.0) /
 			((double)total_time / (double)rte_get_tsc_hz());
 
-	__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+		memory_order_relaxed);
 }
 
 static int
@@ -3165,9 +3172,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3222,10 +3230,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3262,9 +3272,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3313,10 +3324,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3352,9 +3365,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3399,10 +3413,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3439,9 +3455,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3488,10 +3505,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3528,9 +3547,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3575,10 +3595,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3613,7 +3635,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3732,7 +3755,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3867,7 +3891,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3990,7 +4015,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4121,7 +4147,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4222,7 +4249,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4323,7 +4351,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4519,7 +4548,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	else
 		return TEST_SKIPPED;
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4542,7 +4572,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = bler_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4641,7 +4672,8 @@ typedef int (test_case_function)(struct active_device *ad,
 			throughput_function = throughput_pmd_lcore_enc;
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4664,7 +4696,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = throughput_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4694,8 +4727,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	 * Wait for main lcore operations.
 	 */
 	tp = &t_params[0];
-	while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-		op_params->num_to_process) &&
+	while ((atomic_load_explicit(&tp->nb_dequeued,
+		memory_order_relaxed) < op_params->num_to_process) &&
 		(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 		TEST_FAILED))
 		rte_pause();
@@ -4708,8 +4741,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	for (used_cores = 1; used_cores < num_lcores; used_cores++) {
 		tp = &t_params[used_cores];
 
-		while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-			op_params->num_to_process) &&
+		while ((atomic_load_explicit(&tp->nb_dequeued,
+			memory_order_relaxed) < op_params->num_to_process) &&
 			(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 			TEST_FAILED))
 			rte_pause();
app/test/test_func_reentrancy.c +4 -4
index ae9de6f..833ceaf 100644
@@ -54,11 +54,11 @@
 #define MAX_LCORES	(rte_memzone_max_get() / (MAX_ITER_MULTI * 4U))
 
 static uint32_t obj_count;
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 #define WAIT_SYNCHRO_FOR_WORKERS()   do { \
 	if (lcore_self != rte_get_main_lcore())                  \
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED); \
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed); \
 } while(0)
 
 /*
@@ -438,7 +438,7 @@ struct test_case test_cases[] = {
 		return -1;
 
 	__atomic_store_n(&obj_count, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	cores = RTE_MIN(rte_lcore_count(), MAX_LCORES);
 	RTE_LCORE_FOREACH_WORKER(lcore_id) {
@@ -448,7 +448,7 @@ struct test_case test_cases[] = {
 		rte_eal_remote_launch(pt_case->func, pt_case->arg, lcore_id);
 	}
 
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	if (pt_case->func(pt_case->arg) < 0)
 		ret = -1;
app/test/test_mcslock.c +6 -6
index 52e45e7..3c26c69 100644
@@ -42,7 +42,7 @@
 
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_mcslock_per_core(__rte_unused void *arg)
@@ -75,7 +75,7 @@
 	rte_mcslock_t ml_perf_me;
 
 	/* wait synchro */
-	rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -100,14 +100,14 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
 	memset(time_count, 0, sizeof(time_count));
 
 	printf("\nTest with lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	lock = 1;
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
@@ -116,11 +116,11 @@
 
 	printf("\nTest with lock on %u cores...\n", (rte_lcore_count()));
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_mempool_perf.c +4 -4
index ce7c624..06f9fc1 100644
@@ -88,7 +88,7 @@
 static int use_external_cache;
 static unsigned external_cache_size = RTE_MEMPOOL_CACHE_MAX_SIZE;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 /* number of objects in one bulk operation (get or put) */
 static unsigned n_get_bulk;
@@ -188,7 +188,7 @@ struct mempool_test_stats {
 
 	/* wait synchro for workers */
 	if (lcore_id != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	start_cycles = rte_get_timer_cycles();
 
@@ -233,7 +233,7 @@ struct mempool_test_stats {
 	int ret;
 	unsigned cores_save = cores;
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	/* reset stats */
 	memset(stats, 0, sizeof(stats));
@@ -258,7 +258,7 @@ struct mempool_test_stats {
 	}
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	ret = per_lcore_mempool_test(mp);
 
app/test/test_pflock.c +6 -6
index 38da6bc..28addf8 100644
@@ -31,7 +31,7 @@
 
 static rte_pflock_t sl;
 static rte_pflock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_pflock_per_core(__rte_unused void *arg)
@@ -69,7 +69,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -99,7 +99,7 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -107,7 +107,7 @@
 
 	printf("\nTest with phase-fair lock on single core...\n");
 	lock = 1;
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -116,12 +116,12 @@
 	printf("\nPhase-fair test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_pmd_perf.c +5 -5
index 3ef590c..1562bbb 100644
@@ -537,7 +537,7 @@ enum {
 	return 0;
 }
 
-static uint64_t start;
+static uint64_t _Atomic start;
 
 static inline int
 poll_burst(void *args)
@@ -575,7 +575,7 @@ enum {
 		num[portid] = pkt_per_port;
 	}
 
-	rte_wait_until_equal_64(&start, 1, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_64(&start, 1, memory_order_acquire);
 
 	cur_tsc = rte_rdtsc();
 	while (total) {
@@ -629,9 +629,9 @@ enum {
 
 	/* only when polling first */
 	if (flags == SC_BURST_POLL_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 1, memory_order_relaxed);
 	else
-		__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 0, memory_order_relaxed);
 
 	/* start polling thread
 	 * if in POLL_FIRST mode, poll once launched;
@@ -655,7 +655,7 @@ enum {
 
 	/* only when polling second  */
 	if (flags == SC_BURST_XMIT_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELEASE);
+		atomic_store_explicit(&start, 1, memory_order_release);
 
 	/* wait for polling finished */
 	diff_tsc = rte_eal_wait_lcore(lcore);
app/test/test_ring_perf.c +4 -4
index 3972fd9..f0f5ef1 100644
@@ -320,7 +320,7 @@ struct thread_params {
 	return 0;
 }
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 static uint64_t queue_count[RTE_MAX_LCORE];
 
 #define TIME_MS 100
@@ -342,7 +342,7 @@ struct thread_params {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (time_diff < hz * TIME_MS / 1000) {
@@ -397,12 +397,12 @@ struct thread_params {
 		param.r = r;
 
 		/* clear synchro and start workers */
-		__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 		if (rte_eal_mp_remote_launch(lcore_f, &param, SKIP_MAIN) < 0)
 			return -1;
 
 		/* start synchro and launch test on main */
-		__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 		lcore_f(&param);
 
 		rte_eal_mp_wait_lcore();
app/test/test_rwlock.c +4 -4
index 4ae0bf8..dfbd0d6 100644
@@ -35,7 +35,7 @@
 
 static rte_rwlock_t sl;
 static rte_rwlock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 enum {
 	LC_TYPE_RDLOCK,
@@ -101,7 +101,7 @@ struct try_rwlock_lcore {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -134,12 +134,12 @@ struct try_rwlock_lcore {
 	printf("\nRwlock Perf Test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, NULL, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(NULL);
 
 	rte_eal_mp_wait_lcore();
app/test/test_spinlock.c +4 -4
index 3f59372..38724a1 100644
@@ -48,7 +48,7 @@
 static rte_spinlock_recursive_t slr;
 static unsigned count = 0;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_spinlock_per_core(__rte_unused void *arg)
@@ -110,7 +110,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -149,11 +149,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_stack_perf.c +6 -6
index 1eae00a..67510d6 100644
@@ -23,7 +23,7 @@
  */
 static volatile unsigned int bulk_sizes[] = {8, MAX_BURST};
 
-static uint32_t lcore_barrier;
+static uint32_t _Atomic lcore_barrier;
 
 struct lcore_pair {
 	unsigned int c1;
@@ -143,8 +143,8 @@ struct thread_args {
 	s = args->s;
 	size = args->sz;
 
-	__atomic_fetch_sub(&lcore_barrier, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_32(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_fetch_sub_explicit(&lcore_barrier, 1, memory_order_relaxed);
+	rte_wait_until_equal_32(&lcore_barrier, 0, memory_order_relaxed);
 
 	uint64_t start = rte_rdtsc();
 
@@ -173,7 +173,7 @@ struct thread_args {
 	unsigned int i;
 
 	for (i = 0; i < RTE_DIM(bulk_sizes); i++) {
-		__atomic_store_n(&lcore_barrier, 2, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, 2, memory_order_relaxed);
 
 		args[0].sz = args[1].sz = bulk_sizes[i];
 		args[0].s = args[1].s = s;
@@ -206,7 +206,7 @@ struct thread_args {
 		int cnt = 0;
 		double avg;
 
-		__atomic_store_n(&lcore_barrier, n, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, n, memory_order_relaxed);
 
 		RTE_LCORE_FOREACH_WORKER(lcore_id) {
 			if (++cnt >= n)
@@ -300,7 +300,7 @@ struct thread_args {
 	struct lcore_pair cores;
 	struct rte_stack *s;
 
-	__atomic_store_n(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&lcore_barrier, 0, memory_order_relaxed);
 
 	s = rte_stack_create(STACK_NAME, STACK_SIZE, rte_socket_id(), flags);
 	if (s == NULL) {
app/test/test_ticketlock.c +4 -4
index 242c136..f12d1e5 100644
@@ -48,7 +48,7 @@
 static rte_ticketlock_recursive_t tlr;
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_ticketlock_per_core(__rte_unused void *arg)
@@ -111,7 +111,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcore_count[lcore] < MAX_LOOP) {
@@ -153,11 +153,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_timer.c +8 -8
index 0c36dc9..cf89a19 100644
@@ -202,7 +202,7 @@ struct mytimerinfo {
 
 /* Need to synchronize worker lcores through multiple steps. */
 enum { WORKER_WAITING = 1, WORKER_RUN_SIGNAL, WORKER_RUNNING, WORKER_FINISHED };
-static uint16_t lcore_state[RTE_MAX_LCORE];
+static uint16_t _Atomic lcore_state[RTE_MAX_LCORE];
 
 static void
 main_init_workers(void)
@@ -210,7 +210,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_WAITING, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_state[i], WORKER_WAITING, memory_order_relaxed);
 	}
 }
 
@@ -220,10 +220,10 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_RUN_SIGNAL, __ATOMIC_RELEASE);
+		atomic_store_explicit(&lcore_state[i], WORKER_RUN_SIGNAL, memory_order_release);
 	}
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, memory_order_acquire);
 	}
 }
 
@@ -233,7 +233,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, memory_order_acquire);
 	}
 }
 
@@ -242,8 +242,8 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, __ATOMIC_ACQUIRE);
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_RUNNING, __ATOMIC_RELEASE);
+	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, memory_order_acquire);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_RUNNING, memory_order_release);
 }
 
 static void
@@ -251,7 +251,7 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_FINISHED, __ATOMIC_RELEASE);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_FINISHED, memory_order_release);
 }
 
 
drivers/event/opdl/opdl_ring.c +26 -21

120 lines not on this page · raw message

drivers/net/mlx5/mlx5_hws_cnt.h +2 -2
index f462665..99977a6 100644
@@ -386,7 +386,7 @@ struct mlx5_hws_age_param {
 
 	MLX5_ASSERT(r->prod.sync_type == RTE_RING_SYNC_ST);
 	MLX5_ASSERT(r->cons.sync_type == RTE_RING_SYNC_ST);
-	current_head = __atomic_load_n(&r->prod.head, __ATOMIC_RELAXED);
+	current_head = atomic_load_explicit(&r->prod.head, memory_order_relaxed);
 	MLX5_ASSERT(n <= r->capacity);
 	MLX5_ASSERT(n <= rte_ring_count(r));
 	revert2head = current_head - n;
@@ -394,7 +394,7 @@ struct mlx5_hws_age_param {
 	__rte_ring_get_elem_addr(r, revert2head, sizeof(cnt_id_t), n,
 			&zcd->ptr1, &zcd->n1, &zcd->ptr2);
 	/* Update tail */
-	__atomic_store_n(&r->prod.tail, revert2head, __ATOMIC_RELEASE);
+	atomic_store_explicit(&r->prod.tail, revert2head, memory_order_release);
 	return n;
 }
 
drivers/net/thunderx/nicvf_rxtx.c +3 -2
index defa551..0db9505 100644
@@ -385,9 +385,10 @@
 		ltail++;
 	}
 
-	rte_wait_until_equal_32(&rbdr->tail, next_tail, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&rbdr->tail, next_tail,
+		memory_order_relaxed);
 
-	__atomic_store_n(&rbdr->tail, ltail, __ATOMIC_RELEASE);
+	atomic_store_explicit(&rbdr->tail, ltail, memory_order_release);
 	nicvf_addr_write(door, to_fill);
 	return to_fill;
 }
drivers/net/thunderx/nicvf_struct.h +1 -1
index 13cf8fe..38c72b6 100644
@@ -20,7 +20,7 @@ struct nicvf_rbdr {
 	struct rbdr_entry_t *desc;
 	nicvf_iova_addr_t phys;
 	uint32_t buffsz;
-	uint32_t tail;
+	uint32_t _Atomic tail;
 	uint32_t next_tail;
 	uint32_t head;
 	uint32_t qlen_mask;
lib/bpf/bpf_pkt.c +3 -3
index ffd2db7..d0d50cd 100644
@@ -25,7 +25,7 @@
 
 struct bpf_eth_cbi {
 	/* used by both data & control path */
-	uint32_t use;    /*usage counter */
+	uint32_t _Atomic use;    /*usage counter */
 	const struct rte_eth_rxtx_callback *cb;  /* callback handle */
 	struct rte_bpf *bpf;
 	struct rte_bpf_jit jit;
@@ -110,8 +110,8 @@ struct bpf_eth_cbh {
 
 	/* in use, busy wait till current RX/TX iteration is finished */
 	if ((puse & BPF_ETH_CBI_INUSE) != 0) {
-		RTE_WAIT_UNTIL_MASKED((uint32_t *)(uintptr_t)&cbi->use,
-			UINT32_MAX, !=, puse, __ATOMIC_RELAXED);
+		RTE_WAIT_UNTIL_MASKED(&cbi->use,
+			UINT32_MAX, !=, puse, memory_order_relaxed);
 	}
 }
 
lib/distributor/distributor_private.h +1 -1
index 7101f63..3b43d3d 100644
@@ -52,7 +52,7 @@
  * Only 64-bits of the memory is actually used though.
  */
 union rte_distributor_buffer_single {
-	volatile int64_t bufptr64;
+	int64_t _Atomic bufptr64;
 	char pad[RTE_CACHE_LINE_SIZE*3];
 } __rte_cache_aligned;
 
lib/distributor/rte_distributor_single.c +22 -22

117 lines not on this page · raw message

lib/eal/arm/include/rte_pause_64.h +14 -14
index 5f70e97..96ad050 100644
@@ -148,13 +148,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_16(volatile uint16_t *addr, uint16_t expected,
-		int memorder)
+rte_wait_until_equal_16(volatile uint16_t _Atomic *addr, uint16_t expected,
+		memory_order memorder)
 {
 	uint16_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_16(addr, value, memorder)
 	if (value != expected) {
@@ -167,13 +167,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_32(volatile uint32_t *addr, uint32_t expected,
-		int memorder)
+rte_wait_until_equal_32(volatile uint32_t _Atomic *addr, uint32_t expected,
+		memory_order memorder)
 {
 	uint32_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_32(addr, value, memorder)
 	if (value != expected) {
@@ -186,13 +186,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_64(volatile uint64_t *addr, uint64_t expected,
-		int memorder)
+rte_wait_until_equal_64(volatile uint64_t _Atomic *addr, uint64_t expected,
+		memory_order memorder)
 {
 	uint64_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_64(addr, value, memorder)
 	if (value != expected) {
@@ -206,8 +206,8 @@ static inline void rte_pause(void)
 
 #define RTE_WAIT_UNTIL_MASKED(addr, mask, cond, expected, memorder) do {  \
 	RTE_BUILD_BUG_ON(!__builtin_constant_p(memorder));                \
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&                  \
-		memorder != __ATOMIC_RELAXED);                            \
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&              \
+		memorder != memory_order_relaxed);                        \
 	const uint32_t size = sizeof(*(addr)) << 3;                       \
 	typeof(*(addr)) expected_value = (expected);                      \
 	typeof(*(addr)) value;                                            \
lib/eal/common/eal_memcfg.h +1 -1
index 8889ba0..0decc29 100644
@@ -18,7 +18,7 @@
  * Memory configuration shared across multiple processes.
  */
 struct rte_mem_config {
-	volatile uint32_t magic;   /**< Magic number - sanity check. */
+	uint32_t _Atomic magic;   /**< Magic number - sanity check. */
 	uint32_t version;
 	/**< Prevent secondary processes using different DPDK versions. */
 
lib/eal/include/generic/rte_pause.h +26 -25

116 lines not on this page · raw message

lib/eal/include/rte_mcslock.h +7 -7
index a805cb2..da25be4 100644
@@ -33,7 +33,7 @@
  */
 typedef struct rte_mcslock {
 	struct rte_mcslock *next;
-	int locked; /* 1 if the queue locked, 0 otherwise */
+	uint32_t _Atomic locked; /* 1 if the queue locked, 0 otherwise */
 } rte_mcslock_t;
 
 /**
@@ -53,7 +53,7 @@
 	rte_mcslock_t *prev;
 
 	/* Init me node */
-	__atomic_store_n(&me->locked, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&me->locked, 1, memory_order_relaxed);
 	__atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
 
 	/* If the queue is empty, the exchange operation is enough to acquire
@@ -88,7 +88,7 @@
 	 * to spin on me->locked until the previous lock holder resets
 	 * the me->locked using mcslock_unlock().
 	 */
-	rte_wait_until_equal_32((uint32_t *)&me->locked, 0, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_32(&me->locked, 0, memory_order_acquire);
 }
 
 /**
@@ -120,14 +120,14 @@
 		/* More nodes added to the queue by other CPUs.
 		 * Wait until the next pointer is set.
 		 */
-		uintptr_t *next;
-		next = (uintptr_t *)&me->next;
+		uintptr_t _Atomic *next;
+		next = (uintptr_t _Atomic *)&me->next;
 		RTE_WAIT_UNTIL_MASKED(next, UINTPTR_MAX, !=, 0,
-			__ATOMIC_RELAXED);
+			memory_order_relaxed);
 	}
 
 	/* Pass lock to next waiter. */
-	__atomic_store_n(&me->next->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&me->next->locked, 0, memory_order_release);
 }
 
 /**
lib/eal/include/rte_pflock.h +11 -11
index a3f7291..1c32512 100644
@@ -40,8 +40,8 @@
  */
 struct rte_pflock {
 	struct {
-		uint16_t in;
-		uint16_t out;
+		uint16_t _Atomic in;
+		uint16_t _Atomic out;
 	} rd, wr;
 };
 typedef struct rte_pflock rte_pflock_t;
@@ -116,14 +116,14 @@ struct rte_pflock {
 	 * If no writer is present, then the operation has completed
 	 * successfully.
 	 */
-	w = __atomic_fetch_add(&pf->rd.in, RTE_PFLOCK_RINC, __ATOMIC_ACQUIRE)
+	w = atomic_fetch_add_explicit(&pf->rd.in, RTE_PFLOCK_RINC, memory_order_acquire)
 		& RTE_PFLOCK_WBITS;
 	if (w == 0)
 		return;
 
 	/* Wait for current write phase to complete. */
 	RTE_WAIT_UNTIL_MASKED(&pf->rd.in, RTE_PFLOCK_WBITS, !=, w,
-		__ATOMIC_ACQUIRE);
+		memory_order_acquire);
 }
 
 /**
@@ -139,7 +139,7 @@ struct rte_pflock {
 static inline void
 rte_pflock_read_unlock(rte_pflock_t *pf)
 {
-	__atomic_fetch_add(&pf->rd.out, RTE_PFLOCK_RINC, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->rd.out, RTE_PFLOCK_RINC, memory_order_release);
 }
 
 /**
@@ -160,8 +160,8 @@ struct rte_pflock {
 	/* Acquire ownership of write-phase.
 	 * This is same as rte_ticketlock_lock().
 	 */
-	ticket = __atomic_fetch_add(&pf->wr.in, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&pf->wr.out, ticket, __ATOMIC_ACQUIRE);
+	ticket = atomic_fetch_add_explicit(&pf->wr.in, 1, memory_order_relaxed);
+	rte_wait_until_equal_16(&pf->wr.out, ticket, memory_order_acquire);
 
 	/*
 	 * Acquire ticket on read-side in order to allow them
@@ -172,10 +172,10 @@ struct rte_pflock {
 	 * speculatively.
 	 */
 	w = RTE_PFLOCK_PRES | (ticket & RTE_PFLOCK_PHID);
-	ticket = __atomic_fetch_add(&pf->rd.in, w, __ATOMIC_RELAXED);
+	ticket = atomic_fetch_add_explicit(&pf->rd.in, w, memory_order_relaxed);
 
 	/* Wait for any pending readers to flush. */
-	rte_wait_until_equal_16(&pf->rd.out, ticket, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&pf->rd.out, ticket, memory_order_acquire);
 }
 
 /**
@@ -192,10 +192,10 @@ struct rte_pflock {
 rte_pflock_write_unlock(rte_pflock_t *pf)
 {
 	/* Migrate from write phase to read phase. */
-	__atomic_fetch_and(&pf->rd.in, RTE_PFLOCK_LSB, __ATOMIC_RELEASE);
+	atomic_fetch_and_explicit(&pf->rd.in, RTE_PFLOCK_LSB, memory_order_release);
 
 	/* Allow other writers to continue. */
-	__atomic_fetch_add(&pf->wr.out, 1, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->wr.out, 1, memory_order_release);
 }
 
 #ifdef __cplusplus
lib/eal/include/rte_ticketlock.h +4 -4
index 5db0d8a..5206b62 100644
@@ -31,7 +31,7 @@
 typedef union {
 	uint32_t tickets;
 	struct {
-		uint16_t current;
+		uint16_t _Atomic current;
 		uint16_t next;
 	} s;
 } rte_ticketlock_t;
@@ -63,7 +63,7 @@
 rte_ticketlock_lock(rte_ticketlock_t *tl)
 {
 	uint16_t me = __atomic_fetch_add(&tl->s.next, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&tl->s.current, me, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&tl->s.current, me, memory_order_acquire);
 }
 
 /**
@@ -75,8 +75,8 @@
 static inline void
 rte_ticketlock_unlock(rte_ticketlock_t *tl)
 {
-	uint16_t i = __atomic_load_n(&tl->s.current, __ATOMIC_RELAXED);
-	__atomic_store_n(&tl->s.current, i + 1, __ATOMIC_RELEASE);
+	uint16_t i = atomic_load_explicit(&tl->s.current, memory_order_relaxed);
+	atomic_store_explicit(&tl->s.current, i + 1, memory_order_release);
 }
 
 /**
lib/eal/loongarch/include/rte_pause.h +1 -1
index 4302e1b..65a68fc 100644
@@ -9,7 +9,7 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
+#include "rte_pause.h"
 
 #include "generic/rte_pause.h"
 
lib/eal/ppc/include/rte_pause.h +1 -1
index 16e47ce..98f3e06 100644
@@ -9,7 +9,7 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
+#include "rte_pause.h"
 
 #include "generic/rte_pause.h"
 
lib/eal/riscv/include/rte_pause.h +1 -1
index cb8e9ca..51b024a 100644
@@ -11,7 +11,7 @@
 extern "C" {
 #endif
 
-#include "rte_atomic.h"
+#include "rte_pause.h"
 
 #include "generic/rte_pause.h"
 
lib/ring/rte_ring_c11_pvt.h +22 -20

89 lines not on this page · raw message

lib/ring/rte_ring_core.h +2 -2
index 82b2370..cf3cb84 100644
@@ -66,8 +66,8 @@ enum rte_ring_sync_type {
  * but offset for *sync_type* and *tail* values should remain the same.
  */
 struct rte_ring_headtail {
-	volatile uint32_t head;      /**< prod/consumer head. */
-	volatile uint32_t tail;      /**< prod/consumer tail. */
+	uint32_t _Atomic head;      /**< prod/consumer head. */
+	uint32_t _Atomic tail;      /**< prod/consumer tail. */
 	RTE_STD_C11
 	union {
 		/** sync type of prod/cons */
lib/ring/rte_ring_generic_pvt.h +12 -7
index 5acb6e5..12a3ca8 100644
@@ -89,11 +89,14 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sp)
-			r->prod.head = *new_head, success = 1;
-		else
-			success = rte_atomic32_cmpset(&r->prod.head,
-					*old_head, *new_head);
+		if (is_sp) {
+			r->prod.head = *new_head;
+			success = 1;
+		} else
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->prod.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 	} while (unlikely(success == 0));
 	return n;
 }
@@ -162,8 +165,10 @@
 			rte_smp_rmb();
 			success = 1;
 		} else {
-			success = rte_atomic32_cmpset(&r->cons.head, *old_head,
-					*new_head);
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->cons.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 		}
 	} while (unlikely(success == 0));
 	return n;
lib/ring/rte_ring_peek_elem_pvt.h +1 -1
index bb0a7d5..6707e38 100644
@@ -59,7 +59,7 @@
 
 	pos = tail + num;
 	ht->head = pos;
-	__atomic_store_n(&ht->tail, pos, __ATOMIC_RELEASE);
+	atomic_store_explicit(&ht->tail, pos, memory_order_release);
 }
 
 /**
Signature · 1 line
1.8.3.1
#16 lore raw

PATCH v3 4/4 eal: adapt rte spinlock and rwlock APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_spinlock.h and rte_rwlock.h APIs to use standard C11 atomics. Update consumers of the spinlock and rwlock APIs for the API break.

---
 lib/eal/include/generic/rte_rwlock.h   | 46 ++++++++++++++++++----------------
 lib/eal/include/generic/rte_spinlock.h | 21 ++++++++--------
 lib/eal/x86/include/rte_spinlock.h     |  2 +-
 3 files changed, 36 insertions(+), 33 deletions(-)
lib/eal/include/generic/rte_rwlock.h +24 -22
index 9e083bb..b659c4c 100644
@@ -22,6 +22,8 @@
  *  https://locklessinc.com/articles/locks/
  */
 
+#include <stdatomic.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -57,7 +59,7 @@
 #define RTE_RWLOCK_READ	 0x4	/* Reader increment */
 
 typedef struct __rte_lockable {
-	int32_t cnt;
+	int32_t _Atomic cnt;
 } rte_rwlock_t;
 
 /**
@@ -92,21 +94,21 @@
 
 	while (1) {
 		/* Wait while writer is present or pending */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed)
 		       & RTE_RWLOCK_MASK)
 			rte_pause();
 
 		/* Try to get read lock */
-		x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-				       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+		x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				       memory_order_acquire) + RTE_RWLOCK_READ;
 
 		/* If no writer, then acquire was successful */
 		if (likely(!(x & RTE_RWLOCK_MASK)))
 			return;
 
 		/* Lost race with writer, backout the change. */
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELAXED);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_relaxed);
 	}
 }
 
@@ -127,20 +129,20 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 	/* fail if write lock is held or writer is pending */
 	if (x & RTE_RWLOCK_MASK)
 		return -EBUSY;
 
 	/* Try to get read lock */
-	x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-			       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+	x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+			       memory_order_acquire) + RTE_RWLOCK_READ;
 
 	/* Back out if writer raced in */
 	if (unlikely(x & RTE_RWLOCK_MASK)) {
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELEASE);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_release);
 
 		return -EBUSY;
 	}
@@ -158,7 +160,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ, memory_order_release);
 }
 
 /**
@@ -178,10 +180,10 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 	if (x < RTE_RWLOCK_WRITE &&
-	    __atomic_compare_exchange_n(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
-					1, __ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+	    atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
+					memory_order_acquire, memory_order_relaxed))
 		return 0;
 	else
 		return -EBUSY;
@@ -201,22 +203,22 @@
 	int32_t x;
 
 	while (1) {
-		x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+		x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 		/* No readers or writers? */
 		if (likely(x < RTE_RWLOCK_WRITE)) {
 			/* Turn off RTE_RWLOCK_WAIT, turn on RTE_RWLOCK_WRITE */
-			if (__atomic_compare_exchange_n(&rwl->cnt, &x, RTE_RWLOCK_WRITE, 1,
-							__ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+			if (atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, RTE_RWLOCK_WRITE,
+							memory_order_acquire, memory_order_relaxed))
 				return;
 		}
 
 		/* Turn on writer wait bit */
 		if (!(x & RTE_RWLOCK_WAIT))
-			__atomic_fetch_or(&rwl->cnt, RTE_RWLOCK_WAIT, __ATOMIC_RELAXED);
+			atomic_fetch_or_explicit(&rwl->cnt, RTE_RWLOCK_WAIT, memory_order_relaxed);
 
 		/* Wait until no readers before trying again */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) > RTE_RWLOCK_WAIT)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) > RTE_RWLOCK_WAIT)
 			rte_pause();
 
 	}
@@ -233,7 +235,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_WRITE, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_WRITE, memory_order_release);
 }
 
 /**
@@ -247,7 +249,7 @@
 static inline int
 rte_rwlock_write_is_locked(rte_rwlock_t *rwl)
 {
-	if (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) & RTE_RWLOCK_WRITE)
+	if (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) & RTE_RWLOCK_WRITE)
 		return 1;
 
 	return 0;
lib/eal/include/generic/rte_spinlock.h +11 -10
index c50ebaa..d92432d 100644
@@ -17,6 +17,8 @@
  * All locks must be initialised before use, and only initialised once.
  */
 
+#include <stdatomic.h>
+
 #include <rte_lcore.h>
 #ifdef RTE_FORCE_INTRINSICS
 #include <rte_common.h>
@@ -28,7 +30,7 @@
  * The rte_spinlock_t type.
  */
 typedef struct __rte_lockable {
-	volatile int locked; /**< lock status 0 = unlocked, 1 = locked */
+	int _Atomic locked; /**< lock status 0 = unlocked, 1 = locked */
 } rte_spinlock_t;
 
 /**
@@ -65,10 +67,10 @@
 {
 	int exp = 0;
 
-	while (!__atomic_compare_exchange_n(&sl->locked, &exp, 1, 0,
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED)) {
-		rte_wait_until_equal_32((volatile uint32_t *)&sl->locked,
-			       0, __ATOMIC_RELAXED);
+	while (!atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed)) {
+		rte_wait_until_equal_32((uint32_t _Atomic *)&sl->locked,
+			       0, memory_order_relaxed);
 		exp = 0;
 	}
 }
@@ -89,7 +91,7 @@
 rte_spinlock_unlock(rte_spinlock_t *sl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_store_n(&sl->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&sl->locked, 0, memory_order_release);
 }
 #endif
 
@@ -112,9 +114,8 @@
 	__rte_no_thread_safety_analysis
 {
 	int exp = 0;
-	return __atomic_compare_exchange_n(&sl->locked, &exp, 1,
-				0, /* disallow spurious failure */
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED);
+	return atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed);
 }
 #endif
 
@@ -128,7 +129,7 @@
  */
 static inline int rte_spinlock_is_locked (rte_spinlock_t *sl)
 {
-	return __atomic_load_n(&sl->locked, __ATOMIC_ACQUIRE);
+	return atomic_load_explicit(&sl->locked, memory_order_acquire);
 }
 
 /**
lib/eal/x86/include/rte_spinlock.h +1 -1
index 0b20ddf..3e7f8ac 100644
@@ -78,7 +78,7 @@ static inline int rte_tm_supported(void)
 }
 
 static inline int
-rte_try_tm(volatile int *lock)
+rte_try_tm(int _Atomic *lock)
 {
 	int i, retries;
 
Signature · 1 line
1.8.3.1
#18 lore raw

PATCH v4 1/4 build: require minimum c11 compiler

In reply to Tyler Retzlaff

Update top level meson.build to require c_std=c11 and remove use of -std=gnu99.

---
 drivers/net/failsafe/meson.build  | 1 -
 lib/eal/common/eal_common_errno.c | 1 +
 meson.build                       | 1 +
 3 files changed, 2 insertions(+), 1 deletion(-)
drivers/net/failsafe/meson.build +0 -1
index 6013e13..c1d3610 100644
@@ -7,7 +7,6 @@ if is_windows
     subdir_done()
 endif
 
-cflags += '-std=gnu99'
 cflags += '-D_DEFAULT_SOURCE'
 cflags += '-D_XOPEN_SOURCE=700'
 cflags += '-pedantic'
lib/eal/common/eal_common_errno.c +1 -0
index ef8f782..d4e2c13 100644
@@ -4,6 +4,7 @@
 
 /* Use XSI-compliant portable version of strerror_r() */
 #undef _GNU_SOURCE
+#define _POSIX_C_SOURCE 200112L
 
 #include <stdio.h>
 #include <string.h>
meson.build +1 -0
index 39cb738..70b54f0 100644
@@ -9,6 +9,7 @@ project('DPDK', 'c',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c11',
             'default_library=static',
             'warning_level=2',
         ],
Signature · 1 line
1.8.3.1
#19 lore raw

PATCH v4 2/4 devtools: forbid use of GCC atomic builtins

In reply to Tyler Retzlaff

Refrain from using compiler __atomic_xxx builtins DPDK now uses standard C11 atomics atomic_xxx.

---
 devtools/checkpatches.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
devtools/checkpatches.sh +8 -0
index 43f5e36..a32f02e 100755
@@ -102,6 +102,14 @@ check_forbidden_additions() { # <patch>
 		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
 		"$1" || res=1
 
+	# refrain from using compiler __atomic_xxx builtins
+	awk -v FOLDERS="lib drivers app examples" \
+		-v EXPRESSIONS="__atomic_.*\\\(" \
+		-v RET_ON_FAIL=1 \
+		-v MESSAGE='Using __atomic_xxx builtins' \
+		-f $(dirname $(readlink -f $0))/check-forbidden-tokens.awk \
+		"$1" || res=1
+
 	# refrain from using compiler __atomic_thread_fence()
 	# It should be avoided on x86 for SMP case.
 	awk -v FOLDERS="lib drivers app examples" \
Signature · 1 line
1.8.3.1
#17 lore raw

v4 cover eal: update public API to use stdatomic atomics

In reply to Tyler Retzlaff

Update EAL public API to use standard C11 atomics. In accordance with technical board discussion February 2023.
http://mails.dpdk.org/archives/dev/2023-February/263516.html

This initial series makes no attempt to adapt every use of builtin atomics from GCC it intends only to adapt the public API of the EAL presented as a compatibility surface to applications.

Further series will be submitted to adapt the rest of DPDK to use standard atomics that are 'internal' and not do not straddle the EAL API.

The choice has been made to break these two changes apart to allow more precise analysis of any potential performance impact this change or subsequent changes may introduce.

Additionally, it is not possible to break the commits up per driver/lib as both the provider and consumer of the APIs need to be changed in the same commit to maintain non-broken build and function.

Finally, the first patch in this series is a duplication of Bruce Richardson series https://patchwork.dpdk.org/project/dpdk/list/?series=29048 and is only being provided to enable the CI run on this series. It is expected that Bruce's series when committed will allow the build: require minimum c11 compiler here to be dropped from this series.

v4:
* fix previous cut & paste mistake including rte_pause.h when rte_atomic.h was intended in loongarch, ppc, riscv rte_pause.h (sorry was tired)

v3:
* restore incorrectly dropped architecture specific #include "rte_pause.h" for loongarch, ppc, riscv
* convert parameters to stdatomic in RTE_WAIT_UNTIL_MASKED macro usage (missed previously)
* lib/ring missed some conversion to stdatomic on struct rte_ring _Atomic fields (picked up by clang which is more strict than gcc)
* drivers/mlx5 use stdatomic operations on struct rte_ring _Atomic fields (picked up by clang which is more strict than gcc)

v2:
* remove volatile qualification from variables, structure fields and parameters that have been qualified _Atomic
* restore removal of #include <rte_atomic.h> in rte_ring_core.h the ring library still uses other rte_atomic APIs and types that are not replaced by the standard or compiler intrinsics
* add missing adaption to stdatomic in ring/rte_ring_c11_pvt.h

Tyler Retzlaff (4):
build: require minimum c11 compiler
devtools: forbid use of GCC atomic builtins
eal: adapt rte pause APIs to use C11 atomics
eal: adapt rte spinlock and rwlock APIs to use C11 atomics

 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 devtools/checkpatches.sh                 |   8 ++
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/failsafe/meson.build         |   1 -
 drivers/net/mlx5/mlx5_hws_cnt.h          |   4 +-
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   6 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_common_errno.c        |   1 +
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  51 ++++++-------
 lib/eal/include/generic/rte_rwlock.h     |  46 ++++++------
 lib/eal/include/generic/rte_spinlock.h   |  21 +++---
 lib/eal/include/rte_mcslock.h            |  14 ++--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/eal/x86/include/rte_spinlock.h       |   2 +-
 lib/ring/rte_ring_c11_pvt.h              |  42 ++++++-----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 meson.build                              |   1 +
 36 files changed, 337 insertions(+), 278 deletions(-)
Signature · 1 line
1.8.3.1
#21 lore raw

PATCH v4 4/4 eal: adapt rte spinlock and rwlock APIs to use C11 atomics

Adapt rte_spinlock.h and rte_rwlock.h APIs to use standard C11 atomics. Update consumers of the spinlock and rwlock APIs for the API break.

---
 lib/eal/include/generic/rte_rwlock.h   | 46 ++++++++++++++++++----------------
 lib/eal/include/generic/rte_spinlock.h | 21 ++++++++--------
 lib/eal/x86/include/rte_spinlock.h     |  2 +-
 3 files changed, 36 insertions(+), 33 deletions(-)
lib/eal/include/generic/rte_rwlock.h +24 -22
index 9e083bb..b659c4c 100644
@@ -22,6 +22,8 @@
  *  https://locklessinc.com/articles/locks/
  */
 
+#include <stdatomic.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -57,7 +59,7 @@
 #define RTE_RWLOCK_READ	 0x4	/* Reader increment */
 
 typedef struct __rte_lockable {
-	int32_t cnt;
+	int32_t _Atomic cnt;
 } rte_rwlock_t;
 
 /**
@@ -92,21 +94,21 @@
 
 	while (1) {
 		/* Wait while writer is present or pending */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed)
 		       & RTE_RWLOCK_MASK)
 			rte_pause();
 
 		/* Try to get read lock */
-		x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-				       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+		x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				       memory_order_acquire) + RTE_RWLOCK_READ;
 
 		/* If no writer, then acquire was successful */
 		if (likely(!(x & RTE_RWLOCK_MASK)))
 			return;
 
 		/* Lost race with writer, backout the change. */
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELAXED);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_relaxed);
 	}
 }
 
@@ -127,20 +129,20 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 	/* fail if write lock is held or writer is pending */
 	if (x & RTE_RWLOCK_MASK)
 		return -EBUSY;
 
 	/* Try to get read lock */
-	x = __atomic_fetch_add(&rwl->cnt, RTE_RWLOCK_READ,
-			       __ATOMIC_ACQUIRE) + RTE_RWLOCK_READ;
+	x = atomic_fetch_add_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+			       memory_order_acquire) + RTE_RWLOCK_READ;
 
 	/* Back out if writer raced in */
 	if (unlikely(x & RTE_RWLOCK_MASK)) {
-		__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ,
-				   __ATOMIC_RELEASE);
+		atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ,
+				   memory_order_release);
 
 		return -EBUSY;
 	}
@@ -158,7 +160,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_READ, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_READ, memory_order_release);
 }
 
 /**
@@ -178,10 +180,10 @@
 {
 	int32_t x;
 
-	x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+	x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 	if (x < RTE_RWLOCK_WRITE &&
-	    __atomic_compare_exchange_n(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
-					1, __ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+	    atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, x + RTE_RWLOCK_WRITE,
+					memory_order_acquire, memory_order_relaxed))
 		return 0;
 	else
 		return -EBUSY;
@@ -201,22 +203,22 @@
 	int32_t x;
 
 	while (1) {
-		x = __atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED);
+		x = atomic_load_explicit(&rwl->cnt, memory_order_relaxed);
 
 		/* No readers or writers? */
 		if (likely(x < RTE_RWLOCK_WRITE)) {
 			/* Turn off RTE_RWLOCK_WAIT, turn on RTE_RWLOCK_WRITE */
-			if (__atomic_compare_exchange_n(&rwl->cnt, &x, RTE_RWLOCK_WRITE, 1,
-							__ATOMIC_ACQUIRE, __ATOMIC_RELAXED))
+			if (atomic_compare_exchange_weak_explicit(&rwl->cnt, &x, RTE_RWLOCK_WRITE,
+							memory_order_acquire, memory_order_relaxed))
 				return;
 		}
 
 		/* Turn on writer wait bit */
 		if (!(x & RTE_RWLOCK_WAIT))
-			__atomic_fetch_or(&rwl->cnt, RTE_RWLOCK_WAIT, __ATOMIC_RELAXED);
+			atomic_fetch_or_explicit(&rwl->cnt, RTE_RWLOCK_WAIT, memory_order_relaxed);
 
 		/* Wait until no readers before trying again */
-		while (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) > RTE_RWLOCK_WAIT)
+		while (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) > RTE_RWLOCK_WAIT)
 			rte_pause();
 
 	}
@@ -233,7 +235,7 @@
 	__rte_unlock_function(rwl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_fetch_sub(&rwl->cnt, RTE_RWLOCK_WRITE, __ATOMIC_RELEASE);
+	atomic_fetch_sub_explicit(&rwl->cnt, RTE_RWLOCK_WRITE, memory_order_release);
 }
 
 /**
@@ -247,7 +249,7 @@
 static inline int
 rte_rwlock_write_is_locked(rte_rwlock_t *rwl)
 {
-	if (__atomic_load_n(&rwl->cnt, __ATOMIC_RELAXED) & RTE_RWLOCK_WRITE)
+	if (atomic_load_explicit(&rwl->cnt, memory_order_relaxed) & RTE_RWLOCK_WRITE)
 		return 1;
 
 	return 0;
lib/eal/include/generic/rte_spinlock.h +11 -10
index c50ebaa..d92432d 100644
@@ -17,6 +17,8 @@
  * All locks must be initialised before use, and only initialised once.
  */
 
+#include <stdatomic.h>
+
 #include <rte_lcore.h>
 #ifdef RTE_FORCE_INTRINSICS
 #include <rte_common.h>
@@ -28,7 +30,7 @@
  * The rte_spinlock_t type.
  */
 typedef struct __rte_lockable {
-	volatile int locked; /**< lock status 0 = unlocked, 1 = locked */
+	int _Atomic locked; /**< lock status 0 = unlocked, 1 = locked */
 } rte_spinlock_t;
 
 /**
@@ -65,10 +67,10 @@
 {
 	int exp = 0;
 
-	while (!__atomic_compare_exchange_n(&sl->locked, &exp, 1, 0,
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED)) {
-		rte_wait_until_equal_32((volatile uint32_t *)&sl->locked,
-			       0, __ATOMIC_RELAXED);
+	while (!atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed)) {
+		rte_wait_until_equal_32((uint32_t _Atomic *)&sl->locked,
+			       0, memory_order_relaxed);
 		exp = 0;
 	}
 }
@@ -89,7 +91,7 @@
 rte_spinlock_unlock(rte_spinlock_t *sl)
 	__rte_no_thread_safety_analysis
 {
-	__atomic_store_n(&sl->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&sl->locked, 0, memory_order_release);
 }
 #endif
 
@@ -112,9 +114,8 @@
 	__rte_no_thread_safety_analysis
 {
 	int exp = 0;
-	return __atomic_compare_exchange_n(&sl->locked, &exp, 1,
-				0, /* disallow spurious failure */
-				__ATOMIC_ACQUIRE, __ATOMIC_RELAXED);
+	return atomic_compare_exchange_strong_explicit(&sl->locked, &exp, 1,
+				memory_order_acquire, memory_order_relaxed);
 }
 #endif
 
@@ -128,7 +129,7 @@
  */
 static inline int rte_spinlock_is_locked (rte_spinlock_t *sl)
 {
-	return __atomic_load_n(&sl->locked, __ATOMIC_ACQUIRE);
+	return atomic_load_explicit(&sl->locked, memory_order_acquire);
 }
 
 /**
lib/eal/x86/include/rte_spinlock.h +1 -1
index 0b20ddf..3e7f8ac 100644
@@ -78,7 +78,7 @@ static inline int rte_tm_supported(void)
 }
 
 static inline int
-rte_try_tm(volatile int *lock)
+rte_try_tm(int _Atomic *lock)
 {
 	int i, retries;
 
Signature · 1 line
1.8.3.1
#20 lore raw

PATCH v4 3/4 eal: adapt rte pause APIs to use C11 atomics

In reply to Tyler Retzlaff

Adapt rte_pause.h APIs to use standard C11 atomics. Update consumers of the pause APIs for the API break.

---
 app/test-bbdev/test_bbdev_perf.c         | 123 ++++++++++++++++++++-----------
 app/test/test_func_reentrancy.c          |   8 +-
 app/test/test_mcslock.c                  |  12 +--
 app/test/test_mempool_perf.c             |   8 +-
 app/test/test_pflock.c                   |  12 +--
 app/test/test_pmd_perf.c                 |  10 +--
 app/test/test_ring_perf.c                |   8 +-
 app/test/test_rwlock.c                   |   8 +-
 app/test/test_spinlock.c                 |   8 +-
 app/test/test_stack_perf.c               |  12 +--
 app/test/test_ticketlock.c               |   8 +-
 app/test/test_timer.c                    |  16 ++--
 drivers/event/opdl/opdl_ring.c           |  47 ++++++------
 drivers/net/mlx5/mlx5_hws_cnt.h          |   4 +-
 drivers/net/thunderx/nicvf_rxtx.c        |   5 +-
 drivers/net/thunderx/nicvf_struct.h      |   2 +-
 lib/bpf/bpf_pkt.c                        |   6 +-
 lib/distributor/distributor_private.h    |   2 +-
 lib/distributor/rte_distributor_single.c |  44 +++++------
 lib/eal/arm/include/rte_pause_64.h       |  28 +++----
 lib/eal/common/eal_memcfg.h              |   2 +-
 lib/eal/include/generic/rte_pause.h      |  51 ++++++-------
 lib/eal/include/rte_mcslock.h            |  14 ++--
 lib/eal/include/rte_pflock.h             |  22 +++---
 lib/eal/include/rte_ticketlock.h         |   8 +-
 lib/ring/rte_ring_c11_pvt.h              |  42 ++++++-----
 lib/ring/rte_ring_core.h                 |   4 +-
 lib/ring/rte_ring_generic_pvt.h          |  19 +++--
 lib/ring/rte_ring_peek_elem_pvt.h        |   2 +-
 29 files changed, 291 insertions(+), 244 deletions(-)
app/test-bbdev/test_bbdev_perf.c +78 -45
index 276bbf0..c3a3a28 100644
@@ -143,7 +143,7 @@ struct test_op_params {
 	uint16_t num_to_process;
 	uint16_t num_lcores;
 	int vector_mask;
-	uint16_t sync;
+	uint16_t _Atomic sync;
 	struct test_buffers q_bufs[RTE_MAX_NUMA_NODES][MAX_QUEUES];
 };
 
@@ -158,7 +158,7 @@ struct thread_params {
 	uint8_t iter_count;
 	double iter_average;
 	double bler;
-	uint16_t nb_dequeued;
+	uint16_t _Atomic nb_dequeued;
 	int16_t processing_status;
 	uint16_t burst_sz;
 	struct test_op_params *op_params;
@@ -3021,27 +3021,32 @@ typedef int (test_case_function)(struct active_device *ad,
 	if (test_vector.op_type == RTE_BBDEV_OP_TURBO_DEC)
 		deq = rte_bbdev_dequeue_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_DEC)
 		deq = rte_bbdev_dequeue_ldpc_dec_ops(dev_id, queue_id,
 				&tp->dec_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_LDPC_ENC)
 		deq = rte_bbdev_dequeue_ldpc_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else if (test_vector.op_type == RTE_BBDEV_OP_FFT)
 		deq = rte_bbdev_dequeue_fft_ops(dev_id, queue_id,
 				&tp->fft_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 	else /*RTE_BBDEV_OP_TURBO_ENC*/
 		deq = rte_bbdev_dequeue_enc_ops(dev_id, queue_id,
 				&tp->enc_ops[
-					__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED)],
+					atomic_load_explicit(&tp->nb_dequeued,
+						memory_order_relaxed)],
 				burst_sz);
 
 	if (deq < burst_sz) {
@@ -3052,8 +3057,9 @@ typedef int (test_case_function)(struct active_device *ad,
 		return;
 	}
 
-	if (__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) + deq < num_ops) {
-		__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	if (atomic_load_explicit(&tp->nb_dequeued, memory_order_relaxed) + deq < num_ops) {
+		atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+			memory_order_relaxed);
 		return;
 	}
 
@@ -3126,7 +3132,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	tp->mbps += (((double)(num_ops * tb_len_bits)) / 1000000.0) /
 			((double)total_time / (double)rte_get_tsc_hz());
 
-	__atomic_fetch_add(&tp->nb_dequeued, deq, __ATOMIC_RELAXED);
+	atomic_fetch_add_explicit(&tp->nb_dequeued, deq,
+		memory_order_relaxed);
 }
 
 static int
@@ -3165,9 +3172,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3222,10 +3230,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3262,9 +3272,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops,
 				num_to_process);
@@ -3313,10 +3324,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3352,9 +3365,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3399,10 +3413,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3439,9 +3455,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3488,10 +3505,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3528,9 +3547,10 @@ typedef int (test_case_function)(struct active_device *ad,
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
 	__atomic_store_n(&tp->processing_status, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&tp->nb_dequeued, 0, memory_order_relaxed);
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops,
 			num_to_process);
@@ -3575,10 +3595,12 @@ typedef int (test_case_function)(struct active_device *ad,
 			/* Wait until processing of previous batch is
 			 * completed
 			 */
-			rte_wait_until_equal_16(&tp->nb_dequeued, enqueued, __ATOMIC_RELAXED);
+			rte_wait_until_equal_16(&tp->nb_dequeued,
+				enqueued, memory_order_relaxed);
 		}
 		if (j != TEST_REPETITIONS - 1)
-			__atomic_store_n(&tp->nb_dequeued, 0, __ATOMIC_RELAXED);
+			atomic_store_explicit(&tp->nb_dequeued, 0,
+				memory_order_relaxed);
 	}
 
 	return TEST_SUCCESS;
@@ -3613,7 +3635,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3732,7 +3755,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3867,7 +3891,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -3990,7 +4015,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_dec_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4121,7 +4147,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4222,7 +4249,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_enc_op_alloc_bulk(tp->op_params->mp, ops_enq,
 			num_ops);
@@ -4323,7 +4351,8 @@ typedef int (test_case_function)(struct active_device *ad,
 
 	bufs = &tp->op_params->q_bufs[GET_SOCKET(info.socket_id)][queue_id];
 
-	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	rte_wait_until_equal_16(&tp->op_params->sync, SYNC_START,
+		memory_order_relaxed);
 
 	ret = rte_bbdev_fft_op_alloc_bulk(tp->op_params->mp, ops_enq, num_ops);
 	TEST_ASSERT_SUCCESS(ret, "Allocation failed for %d ops", num_ops);
@@ -4519,7 +4548,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	else
 		return TEST_SKIPPED;
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4542,7 +4572,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = bler_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4641,7 +4672,8 @@ typedef int (test_case_function)(struct active_device *ad,
 			throughput_function = throughput_pmd_lcore_enc;
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_WAIT, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_WAIT,
+		memory_order_relaxed);
 
 	/* Main core is set at first entry */
 	t_params[0].dev_id = ad->dev_id;
@@ -4664,7 +4696,8 @@ typedef int (test_case_function)(struct active_device *ad,
 				&t_params[used_cores++], lcore_id);
 	}
 
-	__atomic_store_n(&op_params->sync, SYNC_START, __ATOMIC_RELAXED);
+	atomic_store_explicit(&op_params->sync, SYNC_START,
+		memory_order_relaxed);
 	ret = throughput_function(&t_params[0]);
 
 	/* Main core is always used */
@@ -4694,8 +4727,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	 * Wait for main lcore operations.
 	 */
 	tp = &t_params[0];
-	while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-		op_params->num_to_process) &&
+	while ((atomic_load_explicit(&tp->nb_dequeued,
+		memory_order_relaxed) < op_params->num_to_process) &&
 		(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 		TEST_FAILED))
 		rte_pause();
@@ -4708,8 +4741,8 @@ typedef int (test_case_function)(struct active_device *ad,
 	for (used_cores = 1; used_cores < num_lcores; used_cores++) {
 		tp = &t_params[used_cores];
 
-		while ((__atomic_load_n(&tp->nb_dequeued, __ATOMIC_RELAXED) <
-			op_params->num_to_process) &&
+		while ((atomic_load_explicit(&tp->nb_dequeued,
+			memory_order_relaxed) < op_params->num_to_process) &&
 			(__atomic_load_n(&tp->processing_status, __ATOMIC_RELAXED) !=
 			TEST_FAILED))
 			rte_pause();
app/test/test_func_reentrancy.c +4 -4
index ae9de6f..833ceaf 100644
@@ -54,11 +54,11 @@
 #define MAX_LCORES	(rte_memzone_max_get() / (MAX_ITER_MULTI * 4U))
 
 static uint32_t obj_count;
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 #define WAIT_SYNCHRO_FOR_WORKERS()   do { \
 	if (lcore_self != rte_get_main_lcore())                  \
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED); \
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed); \
 } while(0)
 
 /*
@@ -438,7 +438,7 @@ struct test_case test_cases[] = {
 		return -1;
 
 	__atomic_store_n(&obj_count, 0, __ATOMIC_RELAXED);
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	cores = RTE_MIN(rte_lcore_count(), MAX_LCORES);
 	RTE_LCORE_FOREACH_WORKER(lcore_id) {
@@ -448,7 +448,7 @@ struct test_case test_cases[] = {
 		rte_eal_remote_launch(pt_case->func, pt_case->arg, lcore_id);
 	}
 
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	if (pt_case->func(pt_case->arg) < 0)
 		ret = -1;
app/test/test_mcslock.c +6 -6
index 52e45e7..3c26c69 100644
@@ -42,7 +42,7 @@
 
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_mcslock_per_core(__rte_unused void *arg)
@@ -75,7 +75,7 @@
 	rte_mcslock_t ml_perf_me;
 
 	/* wait synchro */
-	rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -100,14 +100,14 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
 	memset(time_count, 0, sizeof(time_count));
 
 	printf("\nTest with lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	lock = 1;
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
@@ -116,11 +116,11 @@
 
 	printf("\nTest with lock on %u cores...\n", (rte_lcore_count()));
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_mempool_perf.c +4 -4
index ce7c624..06f9fc1 100644
@@ -88,7 +88,7 @@
 static int use_external_cache;
 static unsigned external_cache_size = RTE_MEMPOOL_CACHE_MAX_SIZE;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 /* number of objects in one bulk operation (get or put) */
 static unsigned n_get_bulk;
@@ -188,7 +188,7 @@ struct mempool_test_stats {
 
 	/* wait synchro for workers */
 	if (lcore_id != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	start_cycles = rte_get_timer_cycles();
 
@@ -233,7 +233,7 @@ struct mempool_test_stats {
 	int ret;
 	unsigned cores_save = cores;
 
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 
 	/* reset stats */
 	memset(stats, 0, sizeof(stats));
@@ -258,7 +258,7 @@ struct mempool_test_stats {
 	}
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 
 	ret = per_lcore_mempool_test(mp);
 
app/test/test_pflock.c +6 -6
index 38da6bc..28addf8 100644
@@ -31,7 +31,7 @@
 
 static rte_pflock_t sl;
 static rte_pflock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_pflock_per_core(__rte_unused void *arg)
@@ -69,7 +69,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -99,7 +99,7 @@
 	const unsigned int lcore = rte_lcore_id();
 
 	printf("\nTest with no lock on single core...\n");
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -107,7 +107,7 @@
 
 	printf("\nTest with phase-fair lock on single core...\n");
 	lock = 1;
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 	printf("Core [%u] Cost Time = %"PRIu64" us\n",
 			lcore, time_count[lcore]);
@@ -116,12 +116,12 @@
 	printf("\nPhase-fair test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_pmd_perf.c +5 -5
index 3ef590c..1562bbb 100644
@@ -537,7 +537,7 @@ enum {
 	return 0;
 }
 
-static uint64_t start;
+static uint64_t _Atomic start;
 
 static inline int
 poll_burst(void *args)
@@ -575,7 +575,7 @@ enum {
 		num[portid] = pkt_per_port;
 	}
 
-	rte_wait_until_equal_64(&start, 1, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_64(&start, 1, memory_order_acquire);
 
 	cur_tsc = rte_rdtsc();
 	while (total) {
@@ -629,9 +629,9 @@ enum {
 
 	/* only when polling first */
 	if (flags == SC_BURST_POLL_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 1, memory_order_relaxed);
 	else
-		__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&start, 0, memory_order_relaxed);
 
 	/* start polling thread
 	 * if in POLL_FIRST mode, poll once launched;
@@ -655,7 +655,7 @@ enum {
 
 	/* only when polling second  */
 	if (flags == SC_BURST_XMIT_FIRST)
-		__atomic_store_n(&start, 1, __ATOMIC_RELEASE);
+		atomic_store_explicit(&start, 1, memory_order_release);
 
 	/* wait for polling finished */
 	diff_tsc = rte_eal_wait_lcore(lcore);
app/test/test_ring_perf.c +4 -4
index 3972fd9..f0f5ef1 100644
@@ -320,7 +320,7 @@ struct thread_params {
 	return 0;
 }
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 static uint64_t queue_count[RTE_MAX_LCORE];
 
 #define TIME_MS 100
@@ -342,7 +342,7 @@ struct thread_params {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (time_diff < hz * TIME_MS / 1000) {
@@ -397,12 +397,12 @@ struct thread_params {
 		param.r = r;
 
 		/* clear synchro and start workers */
-		__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 		if (rte_eal_mp_remote_launch(lcore_f, &param, SKIP_MAIN) < 0)
 			return -1;
 
 		/* start synchro and launch test on main */
-		__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+		atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 		lcore_f(&param);
 
 		rte_eal_mp_wait_lcore();
app/test/test_rwlock.c +4 -4
index 4ae0bf8..dfbd0d6 100644
@@ -35,7 +35,7 @@
 
 static rte_rwlock_t sl;
 static rte_rwlock_t sl_tab[RTE_MAX_LCORE];
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 enum {
 	LC_TYPE_RDLOCK,
@@ -101,7 +101,7 @@ struct try_rwlock_lcore {
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcount < MAX_LOOP) {
@@ -134,12 +134,12 @@ struct try_rwlock_lcore {
 	printf("\nRwlock Perf Test on %u cores...\n", rte_lcore_count());
 
 	/* clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	if (rte_eal_mp_remote_launch(load_loop_fn, NULL, SKIP_MAIN) < 0)
 		return -1;
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(NULL);
 
 	rte_eal_mp_wait_lcore();
app/test/test_spinlock.c +4 -4
index 3f59372..38724a1 100644
@@ -48,7 +48,7 @@
 static rte_spinlock_recursive_t slr;
 static unsigned count = 0;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_spinlock_per_core(__rte_unused void *arg)
@@ -110,7 +110,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_get_timer_cycles();
 	while (lcount < MAX_LOOP) {
@@ -149,11 +149,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_stack_perf.c +6 -6
index 1eae00a..67510d6 100644
@@ -23,7 +23,7 @@
  */
 static volatile unsigned int bulk_sizes[] = {8, MAX_BURST};
 
-static uint32_t lcore_barrier;
+static uint32_t _Atomic lcore_barrier;
 
 struct lcore_pair {
 	unsigned int c1;
@@ -143,8 +143,8 @@ struct thread_args {
 	s = args->s;
 	size = args->sz;
 
-	__atomic_fetch_sub(&lcore_barrier, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_32(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_fetch_sub_explicit(&lcore_barrier, 1, memory_order_relaxed);
+	rte_wait_until_equal_32(&lcore_barrier, 0, memory_order_relaxed);
 
 	uint64_t start = rte_rdtsc();
 
@@ -173,7 +173,7 @@ struct thread_args {
 	unsigned int i;
 
 	for (i = 0; i < RTE_DIM(bulk_sizes); i++) {
-		__atomic_store_n(&lcore_barrier, 2, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, 2, memory_order_relaxed);
 
 		args[0].sz = args[1].sz = bulk_sizes[i];
 		args[0].s = args[1].s = s;
@@ -206,7 +206,7 @@ struct thread_args {
 		int cnt = 0;
 		double avg;
 
-		__atomic_store_n(&lcore_barrier, n, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_barrier, n, memory_order_relaxed);
 
 		RTE_LCORE_FOREACH_WORKER(lcore_id) {
 			if (++cnt >= n)
@@ -300,7 +300,7 @@ struct thread_args {
 	struct lcore_pair cores;
 	struct rte_stack *s;
 
-	__atomic_store_n(&lcore_barrier, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&lcore_barrier, 0, memory_order_relaxed);
 
 	s = rte_stack_create(STACK_NAME, STACK_SIZE, rte_socket_id(), flags);
 	if (s == NULL) {
app/test/test_ticketlock.c +4 -4
index 242c136..f12d1e5 100644
@@ -48,7 +48,7 @@
 static rte_ticketlock_recursive_t tlr;
 static unsigned int count;
 
-static uint32_t synchro;
+static uint32_t _Atomic synchro;
 
 static int
 test_ticketlock_per_core(__rte_unused void *arg)
@@ -111,7 +111,7 @@
 
 	/* wait synchro for workers */
 	if (lcore != rte_get_main_lcore())
-		rte_wait_until_equal_32(&synchro, 1, __ATOMIC_RELAXED);
+		rte_wait_until_equal_32(&synchro, 1, memory_order_relaxed);
 
 	begin = rte_rdtsc_precise();
 	while (lcore_count[lcore] < MAX_LOOP) {
@@ -153,11 +153,11 @@
 	printf("\nTest with lock on %u cores...\n", rte_lcore_count());
 
 	/* Clear synchro and start workers */
-	__atomic_store_n(&synchro, 0, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 0, memory_order_relaxed);
 	rte_eal_mp_remote_launch(load_loop_fn, &lock, SKIP_MAIN);
 
 	/* start synchro and launch test on main */
-	__atomic_store_n(&synchro, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&synchro, 1, memory_order_relaxed);
 	load_loop_fn(&lock);
 
 	rte_eal_mp_wait_lcore();
app/test/test_timer.c +8 -8
index 0c36dc9..cf89a19 100644
@@ -202,7 +202,7 @@ struct mytimerinfo {
 
 /* Need to synchronize worker lcores through multiple steps. */
 enum { WORKER_WAITING = 1, WORKER_RUN_SIGNAL, WORKER_RUNNING, WORKER_FINISHED };
-static uint16_t lcore_state[RTE_MAX_LCORE];
+static uint16_t _Atomic lcore_state[RTE_MAX_LCORE];
 
 static void
 main_init_workers(void)
@@ -210,7 +210,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_WAITING, __ATOMIC_RELAXED);
+		atomic_store_explicit(&lcore_state[i], WORKER_WAITING, memory_order_relaxed);
 	}
 }
 
@@ -220,10 +220,10 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		__atomic_store_n(&lcore_state[i], WORKER_RUN_SIGNAL, __ATOMIC_RELEASE);
+		atomic_store_explicit(&lcore_state[i], WORKER_RUN_SIGNAL, memory_order_release);
 	}
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_RUNNING, memory_order_acquire);
 	}
 }
 
@@ -233,7 +233,7 @@ struct mytimerinfo {
 	unsigned i;
 
 	RTE_LCORE_FOREACH_WORKER(i) {
-		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, __ATOMIC_ACQUIRE);
+		rte_wait_until_equal_16(&lcore_state[i], WORKER_FINISHED, memory_order_acquire);
 	}
 }
 
@@ -242,8 +242,8 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, __ATOMIC_ACQUIRE);
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_RUNNING, __ATOMIC_RELEASE);
+	rte_wait_until_equal_16(&lcore_state[lcore_id], WORKER_RUN_SIGNAL, memory_order_acquire);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_RUNNING, memory_order_release);
 }
 
 static void
@@ -251,7 +251,7 @@ struct mytimerinfo {
 {
 	unsigned lcore_id = rte_lcore_id();
 
-	__atomic_store_n(&lcore_state[lcore_id], WORKER_FINISHED, __ATOMIC_RELEASE);
+	atomic_store_explicit(&lcore_state[lcore_id], WORKER_FINISHED, memory_order_release);
 }
 
 
drivers/event/opdl/opdl_ring.c +26 -21

120 lines not on this page · raw message

drivers/net/mlx5/mlx5_hws_cnt.h +2 -2
index f462665..99977a6 100644
@@ -386,7 +386,7 @@ struct mlx5_hws_age_param {
 
 	MLX5_ASSERT(r->prod.sync_type == RTE_RING_SYNC_ST);
 	MLX5_ASSERT(r->cons.sync_type == RTE_RING_SYNC_ST);
-	current_head = __atomic_load_n(&r->prod.head, __ATOMIC_RELAXED);
+	current_head = atomic_load_explicit(&r->prod.head, memory_order_relaxed);
 	MLX5_ASSERT(n <= r->capacity);
 	MLX5_ASSERT(n <= rte_ring_count(r));
 	revert2head = current_head - n;
@@ -394,7 +394,7 @@ struct mlx5_hws_age_param {
 	__rte_ring_get_elem_addr(r, revert2head, sizeof(cnt_id_t), n,
 			&zcd->ptr1, &zcd->n1, &zcd->ptr2);
 	/* Update tail */
-	__atomic_store_n(&r->prod.tail, revert2head, __ATOMIC_RELEASE);
+	atomic_store_explicit(&r->prod.tail, revert2head, memory_order_release);
 	return n;
 }
 
drivers/net/thunderx/nicvf_rxtx.c +3 -2
index defa551..0db9505 100644
@@ -385,9 +385,10 @@
 		ltail++;
 	}
 
-	rte_wait_until_equal_32(&rbdr->tail, next_tail, __ATOMIC_RELAXED);
+	rte_wait_until_equal_32(&rbdr->tail, next_tail,
+		memory_order_relaxed);
 
-	__atomic_store_n(&rbdr->tail, ltail, __ATOMIC_RELEASE);
+	atomic_store_explicit(&rbdr->tail, ltail, memory_order_release);
 	nicvf_addr_write(door, to_fill);
 	return to_fill;
 }
drivers/net/thunderx/nicvf_struct.h +1 -1
index 13cf8fe..38c72b6 100644
@@ -20,7 +20,7 @@ struct nicvf_rbdr {
 	struct rbdr_entry_t *desc;
 	nicvf_iova_addr_t phys;
 	uint32_t buffsz;
-	uint32_t tail;
+	uint32_t _Atomic tail;
 	uint32_t next_tail;
 	uint32_t head;
 	uint32_t qlen_mask;
lib/bpf/bpf_pkt.c +3 -3
index ffd2db7..d0d50cd 100644
@@ -25,7 +25,7 @@
 
 struct bpf_eth_cbi {
 	/* used by both data & control path */
-	uint32_t use;    /*usage counter */
+	uint32_t _Atomic use;    /*usage counter */
 	const struct rte_eth_rxtx_callback *cb;  /* callback handle */
 	struct rte_bpf *bpf;
 	struct rte_bpf_jit jit;
@@ -110,8 +110,8 @@ struct bpf_eth_cbh {
 
 	/* in use, busy wait till current RX/TX iteration is finished */
 	if ((puse & BPF_ETH_CBI_INUSE) != 0) {
-		RTE_WAIT_UNTIL_MASKED((uint32_t *)(uintptr_t)&cbi->use,
-			UINT32_MAX, !=, puse, __ATOMIC_RELAXED);
+		RTE_WAIT_UNTIL_MASKED(&cbi->use,
+			UINT32_MAX, !=, puse, memory_order_relaxed);
 	}
 }
 
lib/distributor/distributor_private.h +1 -1
index 7101f63..3b43d3d 100644
@@ -52,7 +52,7 @@
  * Only 64-bits of the memory is actually used though.
  */
 union rte_distributor_buffer_single {
-	volatile int64_t bufptr64;
+	int64_t _Atomic bufptr64;
 	char pad[RTE_CACHE_LINE_SIZE*3];
 } __rte_cache_aligned;
 
lib/distributor/rte_distributor_single.c +22 -22

117 lines not on this page · raw message

lib/eal/arm/include/rte_pause_64.h +14 -14
index 5f70e97..96ad050 100644
@@ -148,13 +148,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_16(volatile uint16_t *addr, uint16_t expected,
-		int memorder)
+rte_wait_until_equal_16(volatile uint16_t _Atomic *addr, uint16_t expected,
+		memory_order memorder)
 {
 	uint16_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_16(addr, value, memorder)
 	if (value != expected) {
@@ -167,13 +167,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_32(volatile uint32_t *addr, uint32_t expected,
-		int memorder)
+rte_wait_until_equal_32(volatile uint32_t _Atomic *addr, uint32_t expected,
+		memory_order memorder)
 {
 	uint32_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_32(addr, value, memorder)
 	if (value != expected) {
@@ -186,13 +186,13 @@ static inline void rte_pause(void)
 }
 
 static __rte_always_inline void
-rte_wait_until_equal_64(volatile uint64_t *addr, uint64_t expected,
-		int memorder)
+rte_wait_until_equal_64(volatile uint64_t _Atomic *addr, uint64_t expected,
+		memory_order memorder)
 {
 	uint64_t value;
 
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&
-		memorder != __ATOMIC_RELAXED);
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&
+		memorder != memory_order_relaxed);
 
 	__RTE_ARM_LOAD_EXC_64(addr, value, memorder)
 	if (value != expected) {
@@ -206,8 +206,8 @@ static inline void rte_pause(void)
 
 #define RTE_WAIT_UNTIL_MASKED(addr, mask, cond, expected, memorder) do {  \
 	RTE_BUILD_BUG_ON(!__builtin_constant_p(memorder));                \
-	RTE_BUILD_BUG_ON(memorder != __ATOMIC_ACQUIRE &&                  \
-		memorder != __ATOMIC_RELAXED);                            \
+	RTE_BUILD_BUG_ON(memorder != memory_order_acquire &&              \
+		memorder != memory_order_relaxed);                        \
 	const uint32_t size = sizeof(*(addr)) << 3;                       \
 	typeof(*(addr)) expected_value = (expected);                      \
 	typeof(*(addr)) value;                                            \
lib/eal/common/eal_memcfg.h +1 -1
index 8889ba0..0decc29 100644
@@ -18,7 +18,7 @@
  * Memory configuration shared across multiple processes.
  */
 struct rte_mem_config {
-	volatile uint32_t magic;   /**< Magic number - sanity check. */
+	uint32_t _Atomic magic;   /**< Magic number - sanity check. */
 	uint32_t version;
 	/**< Prevent secondary processes using different DPDK versions. */
 
lib/eal/include/generic/rte_pause.h +26 -25

116 lines not on this page · raw message

lib/eal/include/rte_mcslock.h +7 -7
index a805cb2..da25be4 100644
@@ -33,7 +33,7 @@
  */
 typedef struct rte_mcslock {
 	struct rte_mcslock *next;
-	int locked; /* 1 if the queue locked, 0 otherwise */
+	uint32_t _Atomic locked; /* 1 if the queue locked, 0 otherwise */
 } rte_mcslock_t;
 
 /**
@@ -53,7 +53,7 @@
 	rte_mcslock_t *prev;
 
 	/* Init me node */
-	__atomic_store_n(&me->locked, 1, __ATOMIC_RELAXED);
+	atomic_store_explicit(&me->locked, 1, memory_order_relaxed);
 	__atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
 
 	/* If the queue is empty, the exchange operation is enough to acquire
@@ -88,7 +88,7 @@
 	 * to spin on me->locked until the previous lock holder resets
 	 * the me->locked using mcslock_unlock().
 	 */
-	rte_wait_until_equal_32((uint32_t *)&me->locked, 0, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_32(&me->locked, 0, memory_order_acquire);
 }
 
 /**
@@ -120,14 +120,14 @@
 		/* More nodes added to the queue by other CPUs.
 		 * Wait until the next pointer is set.
 		 */
-		uintptr_t *next;
-		next = (uintptr_t *)&me->next;
+		uintptr_t _Atomic *next;
+		next = (uintptr_t _Atomic *)&me->next;
 		RTE_WAIT_UNTIL_MASKED(next, UINTPTR_MAX, !=, 0,
-			__ATOMIC_RELAXED);
+			memory_order_relaxed);
 	}
 
 	/* Pass lock to next waiter. */
-	__atomic_store_n(&me->next->locked, 0, __ATOMIC_RELEASE);
+	atomic_store_explicit(&me->next->locked, 0, memory_order_release);
 }
 
 /**
lib/eal/include/rte_pflock.h +11 -11
index a3f7291..1c32512 100644
@@ -40,8 +40,8 @@
  */
 struct rte_pflock {
 	struct {
-		uint16_t in;
-		uint16_t out;
+		uint16_t _Atomic in;
+		uint16_t _Atomic out;
 	} rd, wr;
 };
 typedef struct rte_pflock rte_pflock_t;
@@ -116,14 +116,14 @@ struct rte_pflock {
 	 * If no writer is present, then the operation has completed
 	 * successfully.
 	 */
-	w = __atomic_fetch_add(&pf->rd.in, RTE_PFLOCK_RINC, __ATOMIC_ACQUIRE)
+	w = atomic_fetch_add_explicit(&pf->rd.in, RTE_PFLOCK_RINC, memory_order_acquire)
 		& RTE_PFLOCK_WBITS;
 	if (w == 0)
 		return;
 
 	/* Wait for current write phase to complete. */
 	RTE_WAIT_UNTIL_MASKED(&pf->rd.in, RTE_PFLOCK_WBITS, !=, w,
-		__ATOMIC_ACQUIRE);
+		memory_order_acquire);
 }
 
 /**
@@ -139,7 +139,7 @@ struct rte_pflock {
 static inline void
 rte_pflock_read_unlock(rte_pflock_t *pf)
 {
-	__atomic_fetch_add(&pf->rd.out, RTE_PFLOCK_RINC, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->rd.out, RTE_PFLOCK_RINC, memory_order_release);
 }
 
 /**
@@ -160,8 +160,8 @@ struct rte_pflock {
 	/* Acquire ownership of write-phase.
 	 * This is same as rte_ticketlock_lock().
 	 */
-	ticket = __atomic_fetch_add(&pf->wr.in, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&pf->wr.out, ticket, __ATOMIC_ACQUIRE);
+	ticket = atomic_fetch_add_explicit(&pf->wr.in, 1, memory_order_relaxed);
+	rte_wait_until_equal_16(&pf->wr.out, ticket, memory_order_acquire);
 
 	/*
 	 * Acquire ticket on read-side in order to allow them
@@ -172,10 +172,10 @@ struct rte_pflock {
 	 * speculatively.
 	 */
 	w = RTE_PFLOCK_PRES | (ticket & RTE_PFLOCK_PHID);
-	ticket = __atomic_fetch_add(&pf->rd.in, w, __ATOMIC_RELAXED);
+	ticket = atomic_fetch_add_explicit(&pf->rd.in, w, memory_order_relaxed);
 
 	/* Wait for any pending readers to flush. */
-	rte_wait_until_equal_16(&pf->rd.out, ticket, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&pf->rd.out, ticket, memory_order_acquire);
 }
 
 /**
@@ -192,10 +192,10 @@ struct rte_pflock {
 rte_pflock_write_unlock(rte_pflock_t *pf)
 {
 	/* Migrate from write phase to read phase. */
-	__atomic_fetch_and(&pf->rd.in, RTE_PFLOCK_LSB, __ATOMIC_RELEASE);
+	atomic_fetch_and_explicit(&pf->rd.in, RTE_PFLOCK_LSB, memory_order_release);
 
 	/* Allow other writers to continue. */
-	__atomic_fetch_add(&pf->wr.out, 1, __ATOMIC_RELEASE);
+	atomic_fetch_add_explicit(&pf->wr.out, 1, memory_order_release);
 }
 
 #ifdef __cplusplus
lib/eal/include/rte_ticketlock.h +4 -4
index 5db0d8a..5206b62 100644
@@ -31,7 +31,7 @@
 typedef union {
 	uint32_t tickets;
 	struct {
-		uint16_t current;
+		uint16_t _Atomic current;
 		uint16_t next;
 	} s;
 } rte_ticketlock_t;
@@ -63,7 +63,7 @@
 rte_ticketlock_lock(rte_ticketlock_t *tl)
 {
 	uint16_t me = __atomic_fetch_add(&tl->s.next, 1, __ATOMIC_RELAXED);
-	rte_wait_until_equal_16(&tl->s.current, me, __ATOMIC_ACQUIRE);
+	rte_wait_until_equal_16(&tl->s.current, me, memory_order_acquire);
 }
 
 /**
@@ -75,8 +75,8 @@
 static inline void
 rte_ticketlock_unlock(rte_ticketlock_t *tl)
 {
-	uint16_t i = __atomic_load_n(&tl->s.current, __ATOMIC_RELAXED);
-	__atomic_store_n(&tl->s.current, i + 1, __ATOMIC_RELEASE);
+	uint16_t i = atomic_load_explicit(&tl->s.current, memory_order_relaxed);
+	atomic_store_explicit(&tl->s.current, i + 1, memory_order_release);
 }
 
 /**
lib/ring/rte_ring_c11_pvt.h +22 -20

89 lines not on this page · raw message

lib/ring/rte_ring_core.h +2 -2
index 82b2370..cf3cb84 100644
@@ -66,8 +66,8 @@ enum rte_ring_sync_type {
  * but offset for *sync_type* and *tail* values should remain the same.
  */
 struct rte_ring_headtail {
-	volatile uint32_t head;      /**< prod/consumer head. */
-	volatile uint32_t tail;      /**< prod/consumer tail. */
+	uint32_t _Atomic head;      /**< prod/consumer head. */
+	uint32_t _Atomic tail;      /**< prod/consumer tail. */
 	RTE_STD_C11
 	union {
 		/** sync type of prod/cons */
lib/ring/rte_ring_generic_pvt.h +12 -7
index 5acb6e5..12a3ca8 100644
@@ -89,11 +89,14 @@
 			return 0;
 
 		*new_head = *old_head + n;
-		if (is_sp)
-			r->prod.head = *new_head, success = 1;
-		else
-			success = rte_atomic32_cmpset(&r->prod.head,
-					*old_head, *new_head);
+		if (is_sp) {
+			r->prod.head = *new_head;
+			success = 1;
+		} else
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->prod.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 	} while (unlikely(success == 0));
 	return n;
 }
@@ -162,8 +165,10 @@
 			rte_smp_rmb();
 			success = 1;
 		} else {
-			success = rte_atomic32_cmpset(&r->cons.head, *old_head,
-					*new_head);
+			/* NOTE: review for potential ordering optimization */
+			success = atomic_compare_exchange_strong_explicit(&r->cons.head,
+					old_head, *new_head,
+					memory_order_seq_cst, memory_order_seq_cst);
 		}
 	} while (unlikely(success == 0));
 	return n;
lib/ring/rte_ring_peek_elem_pvt.h +1 -1
index bb0a7d5..6707e38 100644
@@ -59,7 +59,7 @@
 
 	pos = tail + num;
 	ht->head = pos;
-	__atomic_store_n(&ht->tail, pos, __ATOMIC_RELEASE);
+	atomic_store_explicit(&ht->tail, pos, memory_order_release);
 }
 
 /**
Signature · 1 line
1.8.3.1