On 10/1/2021 8:24 PM, Eelco Chaudron wrote:
See comments below.
On 15 Sep 2021, at 14:43, Chris Mi wrote:
Process sFlow offload packet in handler thread if handler id is 0.
Signed-off-by: Chris Mi <c...@nvidia.com>
Reviewed-by: Eli Britstein <el...@nvidia.com>
---
ofproto/ofproto-dpif-upcall.c | 63 +++++++++++++++++++++++++++++++++++
1 file changed, 63 insertions(+)
diff --git a/ofproto/ofproto-dpif-upcall.c b/ofproto/ofproto-dpif-upcall.c
index 1c9c720f0..4a36a45bb 100644
--- a/ofproto/ofproto-dpif-upcall.c
+++ b/ofproto/ofproto-dpif-upcall.c
@@ -22,6 +22,7 @@
#include "connmgr.h"
#include "coverage.h"
#include "cmap.h"
+#include "lib/dpif-offload-provider.h"
#include "lib/dpif-provider.h"
#include "dpif.h"
#include "openvswitch/dynamic-string.h"
@@ -779,6 +780,57 @@ udpif_get_n_flows(struct udpif *udpif)
return flow_count;
}
+static void
+process_offload_sflow(struct udpif *udpif, struct dpif_offload_sflow *sflow)
+{
+ const struct dpif_sflow_attr *attr = sflow->attr;
+ const struct user_action_cookie *cookie;
+ struct dpif_sflow *dpif_sflow;
+ struct ofproto_dpif *ofproto;
+ struct upcall upcall;
+ uint32_t iifindex;
+ struct flow flow;
+
+ if (!attr) {
+ VLOG_WARN_RL(&rl, "%s: sFlow upcall is missing its attribute",
+ __func__);
Here we should remove the function in the log message. A developer can easily
find it, and it might confuse the end-user.
+ return;
+ }
+
+ cookie = nl_attr_get(attr->userdata);
Here we need to also check that the length of the attr is at least
sizeof(struct user_action_cookie)
Done.
+ if (!cookie) {
+ VLOG_WARN_RL(&rl, "%s: user action cookie is missing", __func__);
Remove function name, and make it sflow specific;
“sFlow user action cookie is missing”
Done.
+ return;
+ }
+ ofproto = ofproto_dpif_lookup_by_uuid(&cookie->ofproto_uuid);
+ if (!ofproto) {
+ VLOG_WARN_RL(&rl, "%s: sFlow upcall can't find ofproto dpif for UUID "
+ UUID_FMT, __func__, UUID_ARGS(&cookie->ofproto_uuid));
Please remove function name.
+ return;
+ }
+ dpif_sflow = ofproto->sflow;
+ if (!dpif_sflow) {
+ VLOG_WARN_RL(&rl, "%s: sFlow upcall is missing dpif information",
+ __func__);
Please remove function name.
Done.
+ return;
+ }
+
+ memset(&flow, 0, sizeof flow);
+ if (attr->tunnel) {
+ memcpy(&flow.tunnel, attr->tunnel, sizeof flow.tunnel);
+ }
+ iifindex = sflow->iifindex;
+ flow.in_port.odp_port = netdev_ifindex_to_odp_port(iifindex);
+ memset(&upcall, 0, sizeof upcall);
+ upcall.flow = &flow;
+ upcall.cookie = *cookie;
+ upcall.packet = &sflow->packet;
+ upcall.sflow = dpif_sflow;
+ upcall.ufid = &attr->ufid;
+ upcall.type = SFLOW_UPCALL;
+ process_upcall(udpif, &upcall, NULL, NULL);
+}
+
/* The upcall handler thread tries to read a batch of UPCALL_MAX_BATCH
* upcalls from dpif, processes the batch and installs corresponding flows
* in dpif. */
@@ -795,6 +847,17 @@ udpif_upcall_handler(void *arg)
dpif_recv_wait(udpif->dpif, handler->handler_id);
latch_wait(&udpif->exit_latch);
}
+ /* Only handler id 0 thread process sFlow offload packet. */
+ if (handler->handler_id == 0) {
+ struct dpif_offload_sflow sflow;
+ int err;
+
+ dpif_offload_sflow_recv_wait(udpif->dpif);
+ err = dpif_offload_sflow_recv(udpif->dpif, &sflow);
+ if (!err) {
+ process_offload_sflow(udpif, &sflow);
+ }
Here we only read and process a single sflow upcall for each poll block. Should
we not do the same as the upcall handling? It's reading max 64 entries before
continuation?
Done.
+ }
poll_block();
}
--
2.27.0
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
