regression from hide_ipv6_interfaces

[aviramha]

User started crashing on macOS with the hook enabled.
Running a Java Bootspring app.
mirrord.json

{
  "target": {
    "namespace": "stage",
    "path": "rollout/C"
  },
  "feature": {
    "fs": true,
    "env": {
      "override": {
        "JAVA_OPTS": ""
      }
    },
    "network": {
      "incoming": {
        "ignore_ports": [9119, 8080]
      },
      "dns": true
    }
  }
}

Crash report (from Console):

Thread 2 Crashed:
0   libsystem_kernel.dylib        	       0x192cd15f0 __pthread_kill + 8
1   libsystem_pthread.dylib       	       0x192d09c20 pthread_kill + 288
2   libsystem_c.dylib             	       0x192c16a30 abort + 180
3   libsystem_malloc.dylib        	       0x192b26dc4 malloc_vreport + 896
4   libsystem_malloc.dylib        	       0x192b2a430 malloc_report + 64
5   libsystem_malloc.dylib        	       0x192b44494 find_zone_and_free + 528
6   13673373759347336826-libmirrord_layer_arm64.dylib	       0x10560f258 mirrord_layer::socket::ops::getifaddrs::ha84c6adc12fd5c9b + 1084
7   13673373759347336826-libmirrord_layer_arm64.dylib	       0x10562e0b8 mirrord_layer::socket::hooks::getifaddrs_detour::hf7a35b7d241ec18a + 128
8   ???                           	       0x1049a890c ???
9   libnet.dylib                  	       0x1054c93f8 enumInterfaces + 48
10  libnet.dylib                  	       0x1054ca264 Java_java_net_NetworkInterface_getAll + 36
11  ???                           	       0x112c67670 ???
12  ???                           	       0x112c62150 ???
13  ???                           	       0x112c62150 ???
14  ???                           	       0x112c62150 ???
15  ???                           	       0x112c5c140 ???

related #2842 #2895 #2849

[linear]

MBE-512 regression from hide_ipv6_interfaces

[aviramha]

Looking into the code, I believe we have a major implementation issue.
We get the interfaces from FN_GETIFADDRS which mallocs it for us
We create a new list, that isn't malloc'd on its own, then we free it (free in middle of block, that shouldn't work in general)
and when it does work - it means we're using after free because it actually frees our block as well!

we need to change the implementation so we'll use malloc to create our own, clean list then free the original one.