fix: optimize DropReason eBPF map lookups (#730)

# Description

For all eBPF program in the DropReason plugin (except `inet_csk_accept`
which has issue we need to investigate
#715):
- only make ebpf map calls when necessary
- omit setting some packet fields to 0 right after `memset` is called on
the entire struct

**Details**:
Previously, we did a map lookup regardless of whether the input `retVal`
indicated a drop. Now, only for drops.
We also skip a map delete when there wasn't a earlier kprobe that saved
the corresponding PID.

## Checklist

- [x] I have read the [contributing
documentation](https://retina.sh/docs/contributing).
- [x] I signed and signed-off the commits (`git commit -S -s ...`). See
[this
documentation](https://docs.github.com/en/authentication/managing-commit-signature-verification/about-commit-signature-verification)
on signing commits.
- [x] I have correctly attributed the author(s) of the code.
- [x] I have tested the changes locally.
- [x] I have followed the project's style guidelines.
- [x] I have updated the documentation, if necessary.
- [x] I have added tests, if applicable.

Signed-off-by: Igor Klemenski <igor.klemenski@microsoft.com>

pkg/plugin/dropreason/_cprog/drop_reason.c

@@ -249,9 +249,6 @@ int BPF_KPROBE(nf_hook_slow, struct sk_buff *skb, struct nf_hook_state *state)
 
     struct packet p;
     __builtin_memset(&p, 0, sizeof(p));
-
-    p.in_filtermap = false;
-    p.skb_len = 0;
     get_packet_from_skb(&p, skb);
 
     __u64 pid_tgid = bpf_get_current_pid_tgid();
@@ -271,26 +268,25 @@ int BPF_KRETPROBE(nf_hook_slow_ret, int retVal)
 {
     __u64 pid_tgid = bpf_get_current_pid_tgid();
     __u32 pid = pid_tgid >> 32;
-    struct packet *p = bpf_map_lookup_elem(&drop_pids, &pid);
-    bpf_map_delete_elem(&drop_pids, &pid);
 
-    if (!p)
+    if (retVal >= 0)
     {
+        bpf_map_delete_elem(&drop_pids, &pid);
         return 0;
     }
 
-    if (retVal >= 0)
+    struct packet *p = bpf_map_lookup_elem(&drop_pids, &pid);
+    if (!p)
     {
         return 0;
     }
 
+    bpf_map_delete_elem(&drop_pids, &pid);
+
     update_metrics_map(ctx, IPTABLE_RULE_DROP, 0, p);
     return 0;
 }
 
-// static __always_inline int
-// exit_tcp_connect(struct pt_regs *ctx, int ret)
-
 /*
 This function checks the return value of tcp_v4_connect and
  update the metrics map accordingly.
@@ -309,9 +305,6 @@ int BPF_KRETPROBE(tcp_v4_connect_ret, int retVal)
     struct packet p;
     __builtin_memset(&p, 0, sizeof(p));
 
-    p.in_filtermap = false;
-    p.skb_len = 0;
-
     update_metrics_map(ctx, TCP_CONNECT_BASIC, retVal, &p);
     return 0;
 }
@@ -387,9 +380,6 @@ int BPF_KPROBE(nf_nat_inet_fn, void *priv, struct sk_buff *skb, const struct nf_
 
     struct packet p;
     __builtin_memset(&p, 0, sizeof(p));
-
-    p.in_filtermap = false;
-    p.skb_len = 0;
     get_packet_from_skb(&p, skb);
 
     __u64 pid_tgid = bpf_get_current_pid_tgid();
@@ -403,17 +393,21 @@ int BPF_KRETPROBE(nf_nat_inet_fn_ret, int retVal)
 {
     __u64 pid_tgid = bpf_get_current_pid_tgid();
     __u32 pid = pid_tgid >> 32;
-    struct packet *p = bpf_map_lookup_elem(&natdrop_pids, &pid);
-    bpf_map_delete_elem(&natdrop_pids, &pid);
 
-    if (!p)
+    if (retVal != NF_DROP)
+    {
+        bpf_map_delete_elem(&natdrop_pids, &pid);
         return 0;
+    }
 
-    if (retVal != NF_DROP)
+    struct packet *p = bpf_map_lookup_elem(&natdrop_pids, &pid);
+    if (!p)
     {
         return 0;
     }
 
+    bpf_map_delete_elem(&natdrop_pids, &pid);
+
     update_metrics_map(ctx, IPTABLE_NAT_DROP, 0, p);
     return 0;
 }
@@ -432,9 +426,6 @@ int BPF_KPROBE(nf_conntrack_confirm, struct sk_buff *skb)
 
     struct packet p;
     __builtin_memset(&p, 0, sizeof(p));
-
-    p.in_filtermap = false;
-    p.skb_len = 0;
     get_packet_from_skb(&p, skb);
 
     __u64 pid_tgid = bpf_get_current_pid_tgid();
@@ -448,17 +439,21 @@ int BPF_KRETPROBE(nf_conntrack_confirm_ret, int retVal)
 {
     __u64 pid_tgid = bpf_get_current_pid_tgid();
     __u32 pid = pid_tgid >> 32;
+
+    if (retVal != NF_DROP)
+    {
+        bpf_map_delete_elem(&natdrop_pids, &pid);
+        return 0;
+    }
+
     struct packet *p = bpf_map_lookup_elem(&natdrop_pids, &pid);
-    bpf_map_delete_elem(&natdrop_pids, &pid);
-
     if (!p)
-        return 0;
-
-    if (retVal != NF_DROP)
     {
         return 0;
     }
 
+    bpf_map_delete_elem(&natdrop_pids, &pid);
+
     update_metrics_map(ctx, CONNTRACK_ADD_DROP, retVal, p);
     return 0;
 }