* [PATCH] net/zxdh: avoid use of rte_memcpy
@ 2025-09-26 18:10 Stephen Hemminger
0 siblings, 0 replies; only message in thread
From: Stephen Hemminger @ 2025-09-26 18:10 UTC (permalink / raw)
To: dev; +Cc: Stephen Hemminger, Junlong Wang, Lijie Shan
The regular library memcpy has more error checking and is as
fast or faster for small fixed sizes. Use struct assignment
for even more type checking.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
---
drivers/net/zxdh/zxdh_flow.c | 53 ++++++++++++++++--------------------
1 file changed, 23 insertions(+), 30 deletions(-)
diff --git a/drivers/net/zxdh/zxdh_flow.c b/drivers/net/zxdh/zxdh_flow.c
index 7a2e20ac39..1167f53641 100644
--- a/drivers/net/zxdh/zxdh_flow.c
+++ b/drivers/net/zxdh/zxdh_flow.c
@@ -94,9 +94,9 @@ static void
zxdh_adjust_flow_op_rsp_memory_layout(void *old_data,
size_t old_size, void *new_data)
{
- rte_memcpy(new_data, old_data, sizeof(struct zxdh_flow));
+ memcpy(new_data, old_data, sizeof(struct zxdh_flow));
memset((char *)new_data + sizeof(struct zxdh_flow), 0, 4);
- rte_memcpy((char *)new_data + sizeof(struct zxdh_flow) + 4,
+ memcpy((char *)new_data + sizeof(struct zxdh_flow) + 4,
(char *)old_data + sizeof(struct zxdh_flow),
old_size - sizeof(struct zxdh_flow));
}
@@ -1328,14 +1328,11 @@ static int fd_flow_parse_pattern(struct rte_eth_dev *dev, const struct rte_flow_
ipv4_mask->hdr.type_of_service;
key->frag_flag = (ipv4_spec->hdr.fragment_offset != 0) ? 1 : 0;
key_mask->frag_flag = (ipv4_mask->hdr.fragment_offset != 0) ? 1 : 0;
- rte_memcpy((uint32_t *)key->src_ip + 3,
- &ipv4_spec->hdr.src_addr, 4);
- rte_memcpy((uint32_t *)key->dst_ip + 3,
- &ipv4_spec->hdr.dst_addr, 4);
- rte_memcpy((uint32_t *)key_mask->src_ip + 3,
- &ipv4_mask->hdr.src_addr, 4);
- rte_memcpy((uint32_t *)key_mask->dst_ip + 3,
- &ipv4_mask->hdr.dst_addr, 4);
+
+ memcpy((uint32_t *)key->src_ip + 3, &ipv4_spec->hdr.src_addr, 4);
+ memcpy((uint32_t *)key->dst_ip + 3, &ipv4_spec->hdr.dst_addr, 4);
+ memcpy((uint32_t *)key_mask->src_ip + 3, &ipv4_mask->hdr.src_addr, 4);
+ memcpy((uint32_t *)key_mask->dst_ip + 3, &ipv4_mask->hdr.dst_addr, 4);
}
break;
case RTE_FLOW_ITEM_TYPE_IPV6:
@@ -1364,14 +1361,10 @@ static int fd_flow_parse_pattern(struct rte_eth_dev *dev, const struct rte_flow_
key->nw_proto = ipv6_spec->hdr.proto;
key_mask->nw_proto = ipv6_mask->hdr.proto;
- rte_memcpy(key->src_ip,
- &ipv6_spec->hdr.src_addr, 16);
- rte_memcpy(key->dst_ip,
- &ipv6_spec->hdr.dst_addr, 16);
- rte_memcpy(key_mask->src_ip,
- &ipv6_mask->hdr.src_addr, 16);
- rte_memcpy(key_mask->dst_ip,
- &ipv6_mask->hdr.dst_addr, 16);
+ memcpy(key->src_ip, &ipv6_spec->hdr.src_addr, 16);
+ memcpy(key->dst_ip, &ipv6_spec->hdr.dst_addr, 16);
+ memcpy(key_mask->src_ip, &ipv6_mask->hdr.src_addr, 16);
+ memcpy(key_mask->dst_ip, &ipv6_mask->hdr.dst_addr, 16);
}
break;
case RTE_FLOW_ITEM_TYPE_TCP:
@@ -1477,8 +1470,8 @@ static int fd_flow_parse_pattern(struct rte_eth_dev *dev, const struct rte_flow_
"Invalid vxlan mask");
return -rte_errno;
}
- rte_memcpy(key->vni, vxlan_spec->vni, 3);
- rte_memcpy(key_mask->vni, vxlan_mask->vni, 3);
+ memcpy(key->vni, vxlan_spec->vni, 3);
+ memcpy(key_mask->vni, vxlan_mask->vni, 3);
break;
}
case RTE_FLOW_ACTION_TYPE_VOID:
@@ -1573,10 +1566,10 @@ fd_flow_parse_vxlan_encap(struct rte_eth_dev *dev __rte_unused,
switch (items->type) {
case RTE_FLOW_ITEM_TYPE_ETH:
item_eth = items->spec;
- rte_memcpy(&dh_flow->encap0.dst_mac1, item_eth->dst.addr_bytes, 2);
- rte_memcpy(&dh_flow->encap1.src_mac1, item_eth->src.addr_bytes, 2);
- rte_memcpy(&dh_flow->encap0.dst_mac2, &item_eth->dst.addr_bytes[2], 4);
- rte_memcpy(&dh_flow->encap1.src_mac2, &item_eth->src.addr_bytes[2], 4);
+ memcpy(&dh_flow->encap0.dst_mac1, item_eth->dst.addr_bytes, 2);
+ memcpy(&dh_flow->encap1.src_mac1, item_eth->src.addr_bytes, 2);
+ memcpy(&dh_flow->encap0.dst_mac2, &item_eth->dst.addr_bytes[2], 4);
+ memcpy(&dh_flow->encap1.src_mac2, &item_eth->src.addr_bytes[2], 4);
dh_flow->encap0.dst_mac1 = rte_bswap16(dh_flow->encap0.dst_mac1);
dh_flow->encap1.src_mac1 = rte_bswap16(dh_flow->encap1.src_mac1);
dh_flow->encap0.dst_mac2 = rte_bswap32(dh_flow->encap0.dst_mac2);
@@ -1592,9 +1585,9 @@ fd_flow_parse_vxlan_encap(struct rte_eth_dev *dev __rte_unused,
dh_flow->encap0.tos = item_ipv4->hdr.type_of_service;
dh_flow->encap0.ttl = item_ipv4->hdr.time_to_live;
addr = rte_bswap32(item_ipv4->hdr.src_addr);
- rte_memcpy((uint32_t *)dh_flow->encap1.sip.ip_addr + 3, &addr, 4);
+ memcpy((uint32_t *)dh_flow->encap1.sip.ip_addr + 3, &addr, 4);
addr = rte_bswap32(item_ipv4->hdr.dst_addr);
- rte_memcpy((uint32_t *)dh_flow->encap0.dip.ip_addr + 3, &addr, 4);
+ memcpy((uint32_t *)dh_flow->encap0.dip.ip_addr + 3, &addr, 4);
break;
case RTE_FLOW_ITEM_TYPE_IPV6:
item_ipv6 = items->spec;
@@ -1603,7 +1596,7 @@ fd_flow_parse_vxlan_encap(struct rte_eth_dev *dev __rte_unused,
(item_ipv6->hdr.vtc_flow & RTE_IPV6_HDR_TC_MASK) >>
RTE_IPV6_HDR_TC_SHIFT;
dh_flow->encap0.ttl = item_ipv6->hdr.hop_limits;
- rte_memcpy(dh_flow->encap1.sip.ip_addr, &item_ipv6->hdr.src_addr, 16);
+ memcpy(dh_flow->encap1.sip.ip_addr, &item_ipv6->hdr.src_addr, 16);
dh_flow->encap1.sip.ip_addr[0] =
rte_bswap32(dh_flow->encap1.sip.ip_addr[0]);
dh_flow->encap1.sip.ip_addr[1] =
@@ -1612,7 +1605,7 @@ fd_flow_parse_vxlan_encap(struct rte_eth_dev *dev __rte_unused,
rte_bswap32(dh_flow->encap1.sip.ip_addr[2]);
dh_flow->encap1.sip.ip_addr[3] =
rte_bswap32(dh_flow->encap1.sip.ip_addr[3]);
- rte_memcpy(dh_flow->encap0.dip.ip_addr, &item_ipv6->hdr.dst_addr, 16);
+ memcpy(dh_flow->encap0.dip.ip_addr, &item_ipv6->hdr.dst_addr, 16);
dh_flow->encap0.dip.ip_addr[0] =
rte_bswap32(dh_flow->encap0.dip.ip_addr[0]);
dh_flow->encap0.dip.ip_addr[1] =
@@ -1884,7 +1877,7 @@ vf_flow_msg_process(enum zxdh_msg_type msg_type, struct rte_eth_dev *dev,
struct zxdh_flow_op_rsp *flow_rsp = (struct zxdh_flow_op_rsp *)flow_op_rsp;
dh_flow->hash_search_index = hw->hash_search_index;
- rte_memcpy(&flow_msg->dh_flow, dh_flow, sizeof(struct zxdh_flow));
+ flow_msg->dh_flow = *dh_flow;
zxdh_msg_head_build(hw, msg_type, &msg_info);
ret = zxdh_vf_send_msg_to_pf(dev, &msg_info, sizeof(struct zxdh_msg_info),
@@ -1909,7 +1902,7 @@ vf_flow_msg_process(enum zxdh_msg_type msg_type, struct rte_eth_dev *dev,
if (msg_type == ZXDH_FLOW_HW_ADD)
dh_flow->flowentry.hw_idx = flow_rsp->dh_flow.flowentry.hw_idx;
if (count)
- rte_memcpy((void *)count, &flow_rsp->count, sizeof(flow_rsp->count));
+ memcpy((void *)count, &flow_rsp->count, sizeof(flow_rsp->count));
return ret;
}
--
2.47.3
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2025-09-26 18:10 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-09-26 18:10 [PATCH] net/zxdh: avoid use of rte_memcpy Stephen Hemminger
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).