Hi all,

I've set this patch under review on Patchwork. This is the first time I push for open source project. Please correct me if this is not the right procedure.

I couldn't find a stable way to reproduce the issue, but it does happen time to time. Not seeing the same issue after applying this patch on my environment.

Any comments on the patch?

Thanks!
Vincent

On Fri, 24 Jan 2020 at 20:43, Vincent Hou <vincent.houyi@gmail.com> wrote:
map_newlink() may abort when interface list changed between netlink
request for getting interfaces and getting addresses. This commit is
ported from the same change from glibc commit.

Signed-off-by: Vincent Hou <vincent.houyi@gmail.com>
---

Hi all,

App code making a getaddrinfo() call may get a occasional abort from
map_newlink(). The reason of the abort is getifaddrs() uses an out-dated
interface list.

In getaddrinfo() call, it will call a function getifaddrs() who retrieves
all interfaces and addresses from kernel via netlink, returning an array
of interfaces and addresses. It will then call a function map_newlink()
to build a mapping between the interfaces’ index into the returning array
to the interface id in kernel. Then getifaddrs() will bind the
relationship between interface and the address using the mapping from
previous step.

The interfaces and addresses are retrieved from two separate netlink
requests. It retrieves interface lists then followed by address list.
Between two requests, kernel may make change to interfaces and addresses.
If a new interface and address is added between the requests, the
newly-added interface doesn’t present in the interface list, but the
new-added address does. So the address will point to an non-existent
interface index.

This issue is fixed by glibc and this change ports the commit from glibc
(with some minor changes).

Thanks,
Vincent Hou
 libc/inet/ifaddrs.c | 53 +++++++++++++++++++++++++++++++++++----------
 1 file changed, 42 insertions(+), 11 deletions(-)

diff --git a/libc/inet/ifaddrs.c b/libc/inet/ifaddrs.c
index 0c9310651..72771d35a 100644
--- a/libc/inet/ifaddrs.c
+++ b/libc/inet/ifaddrs.c
@@ -339,17 +339,19 @@ map_newlink (int idx, struct ifaddrs_storage *ifas, int *map, int max)
       else if (map[i] == idx)
        return i;
     }
-  /* This should never be reached. If this will be reached, we have
-     a very big problem.  */
-  abort ();
+
+  /* This means interfaces changed inbetween the reading of the
+     RTM_GETLINK and RTM_GETADDR information.  We have to repeat
+     everything.  */
+  return -1;
 }


 /* Create a linked list of `struct ifaddrs' structures, one for each
    network interface on the host machine.  If successful, store the
    list in *IFAP and return 0.  On errors, return -1 and set `errno'.  */
-int
-getifaddrs (struct ifaddrs **ifap)
+static int
+getifaddrs_internal (struct ifaddrs **ifap)
 {
   struct netlink_handle nh = { 0, 0, 0, NULL, NULL };
   struct netlink_res *nlp;
@@ -496,6 +498,13 @@ getifaddrs (struct ifaddrs **ifap)
                 kernel.  */
              ifa_index = map_newlink (ifim->ifi_index - 1, ifas,
                                       map_newlink_data, newlink);
+             if (__builtin_expect (ifa_index == -1, 0))
+               {
+               try_again:
+                 result = -EAGAIN;
+                 free (ifas);
+                 goto exit_free;
+               }
              ifas[ifa_index].ifa.ifa_flags = ifim->ifi_flags;

              while (RTA_OK (rta, rtasize))
@@ -580,9 +589,11 @@ getifaddrs (struct ifaddrs **ifap)
                 that we have holes in the interface part of the list,
                 but we always have already the interface for this address.  */
              ifa_index = newlink + newaddr_idx;
-             ifas[ifa_index].ifa.ifa_flags
-               = ifas[map_newlink (ifam->ifa_index - 1, ifas,
-                                   map_newlink_data, newlink)].ifa.ifa_flags;
+             int idx = map_newlink (ifam->ifa_index - 1, ifas,
+                                    map_newlink_data, newlink);
+             if (__builtin_expect (idx == -1, 0))
+               goto try_again;
+             ifas[ifa_index].ifa.ifa_flags = ifas[idx].ifa.ifa_flags;
              if (ifa_index > 0)
                ifas[ifa_index - 1].ifa.ifa_next = &ifas[ifa_index].ifa;
              ++newaddr_idx;
@@ -768,9 +779,13 @@ getifaddrs (struct ifaddrs **ifap)
              /* If we didn't get the interface name with the
                 address, use the name from the interface entry.  */
              if (ifas[ifa_index].ifa.ifa_name == NULL)
-               ifas[ifa_index].ifa.ifa_name
-                 = ifas[map_newlink (ifam->ifa_index - 1, ifas,
-                                     map_newlink_data, newlink)].ifa.ifa_name;
+               {
+                 int idx = map_newlink (ifam->ifa_index - 1, ifas,
+                                        map_newlink_data, newlink);
+                 if (__builtin_expect (idx == -1, 0))
+                   goto try_again;
+                 ifas[ifa_index].ifa.ifa_name = ifas[idx].ifa.ifa_name;
+               }

              /* Calculate the netmask.  */
              if (ifas[ifa_index].ifa.ifa_addr
@@ -850,6 +865,22 @@ getifaddrs (struct ifaddrs **ifap)

   return result;
 }
+
+
+/* Create a linked list of `struct ifaddrs' structures, one for each
+   network interface on the host machine.  If successful, store the
+   list in *IFAP and return 0.  On errors, return -1 and set `errno'.  */
+int
+getifaddrs (struct ifaddrs **ifap)
+{
+  int res;
+
+  do
+    res = getifaddrs_internal (ifap);
+  while (res == -EAGAIN);
+
+  return res;
+}
 libc_hidden_def(getifaddrs)

 void
--
2.18.0