common/cnxk: fix atomic load in batch ops

In roc batch alloc wait code, __ATOMIC_RELAXED is changed to __ATOMIC_ACQUIRE in order to avoid potential out of order loads. Fixes: 50d08d3 ("common/cnxk: fix batch alloc completion poll logic") Cc: [email protected] Signed-off-by: Nawal Kishor <[email protected]>
DPDK · Jan 22, 2025 · ccde23a · ccde23a
1 parent e931818
commit ccde23a
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 1 deletion.
diff --git a/.mailmap b/.mailmap
@@ -1077,6 +1077,7 @@ Natanael Copa <[email protected]>
 Nathan Brown <[email protected]>
 Nathan Law <[email protected]>
 Nathan Skrzypczak <[email protected]>
+Nawal Kishor <[email protected]>
 Neel Patel <[email protected]> <[email protected]>
 Neil Horman <[email protected]>
 Nelio Laranjeiro <[email protected]>

diff --git a/drivers/common/cnxk/roc_npa.h b/drivers/common/cnxk/roc_npa.h
@@ -255,7 +255,7 @@ roc_npa_batch_alloc_wait(uint64_t *cache_line, unsigned int wait_us)
 	/* Batch alloc status code is updated in bits [5:6] of the first word
 	 * of the 128 byte cache line.
 	 */
-	while (((__atomic_load_n(cache_line, __ATOMIC_RELAXED) >> 5) & 0x3) ==
+	while (((__atomic_load_n(cache_line, __ATOMIC_ACQUIRE) >> 5) & 0x3) ==
 	       ALLOC_CCODE_INVAL)
 		if (wait_us && (plt_tsc_cycles() - start) >= ticks)
 			break;