Re: INFO: task hung in nbd_ioctl

From: Richard W.M. Jones
Date: Thu Oct 17 2019 - 12:49:12 EST


On Thu, Oct 17, 2019 at 09:36:34AM -0700, Eric Biggers wrote:
> On Thu, Oct 17, 2019 at 05:28:29PM +0100, Richard W.M. Jones wrote:
> > On Thu, Oct 17, 2019 at 10:47:59AM -0500, Mike Christie wrote:
> > > On 10/17/2019 09:03 AM, Richard W.M. Jones wrote:
> > > > On Tue, Oct 01, 2019 at 04:19:25PM -0500, Mike Christie wrote:
> > > >> Hey Josef and nbd list,
> > > >>
> > > >> I had a question about if there are any socket family restrictions for nbd?
> > > >
> > > > In normal circumstances, in userspace, the NBD protocol would only be
> > > > used over AF_UNIX or AF_INET/AF_INET6.
> > > >
> > > > There's a bit of confusion because netlink is used by nbd-client to
> > > > configure the NBD device, setting things like block size and timeouts
> > > > (instead of ioctl which is deprecated). I think you don't mean this
> > > > use of netlink?
> > >
> > > I didn't. It looks like it is just a bad test.
> > >
> > > For the automated test in this thread the test created a AF_NETLINK
> > > socket and passed it into the NBD_SET_SOCK ioctl. That is what got used
> > > for the NBD_DO_IT ioctl.
> > >
> > > I was not sure if the test creator picked any old socket and it just
> > > happened to pick one nbd never supported, or it was trying to simulate
> > > sockets that did not support the shutdown method.
> > >
> > > I attached the automated test that got run (test.c).
> >
> > I'd say it sounds like a bad test, but I'm not familiar with syzkaller
> > nor how / from where it generates these tests. Did someone report a
> > bug and then syzkaller wrote this test?
>
> It's an automatically generated fuzz test.
>
> There's rarely any such thing as a "bad" fuzz test. If userspace
> can do something that causes the kernel to crash or hang, it's a
> kernel bug, with very few exceptions (e.g. like writing to
> /dev/mem).
>
> If there are cases that aren't supported, like sockets that don't
> support a certain function or whatever, then the code needs to check
> for those cases and return an error, not hang the kernel.

Oh I see. In that case I agree, although I believe this is a
root-only API and root has a lot of ways to crash the kernel, but sure
it could be fixed to restrict sockets to one of:

- AF_LOCAL or AF_UNIX
- AF_INET or AF_INET6
- AF_INET*_SDP (? no idea what this is, but it's used by nbd-client)

Here are some ways NBD is used in real code:

libnbd$ git grep AF_
fuzzing/libnbd-fuzz-wrapper.c: if (socketpair (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0, sv) == -1) {
generator/states-connect-socket-activation.c: s = socket (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
generator/states-connect-socket-activation.c: addr.sun_family = AF_UNIX;
generator/states-connect.c: fd = socket (AF_UNIX, SOCK_STREAM|SOCK_NONBLOCK|SOCK_CLOEXEC, 0);
generator/states-connect.c: struct sockaddr_un sun = { .sun_family = AF_UNIX };
generator/states-connect.c: if (socketpair (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0, sv) == -1) {


nbdkit$ git grep AF_
plugins/info/info.c: case AF_INET:
plugins/info/info.c: if (inet_ntop (AF_INET, &addr->sin_addr,
plugins/info/info.c: case AF_INET6:
plugins/info/info.c: if (inet_ntop (AF_INET6, &addr6->sin6_addr,
plugins/info/info.c: case AF_UNIX:
plugins/nbd/nbd-standalone.c: struct sockaddr_un sock = { .sun_family = AF_UNIX };
plugins/nbd/nbd-standalone.c: fd = socket (AF_UNIX, SOCK_STREAM, 0);
server/sockets.c: sock = socket (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
server/sockets.c: sock = set_cloexec (socket (AF_UNIX, SOCK_STREAM, 0));
server/sockets.c: addr.sun_family = AF_UNIX;
tests/test-layers.c: if (socketpair (AF_LOCAL, SOCK_STREAM, 0, sfd) == -1) {
tests/test-socket-activation.c: sock = socket (AF_UNIX, SOCK_STREAM /* NB do not use SOCK_CLOEXEC */, 0);
tests/test-socket-activation.c: addr.sun_family = AF_UNIX;
tests/test-socket-activation.c: sock = socket (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
tests/web-server.c: listen_sock = socket (AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
tests/web-server.c: addr.sun_family = AF_UNIX;

nbd$ git grep AF_
gznbd/gznbd.c: if(socketpair(AF_UNIX, SOCK_STREAM, 0, pr)){
nbd-client.c: if (ai->ai_family == AF_INET)
nbd-client.c: ai->ai_family = AF_INET_SDP;
nbd-client.c: else (ai->ai_family == AF_INET6)
nbd-client.c: ai->ai_family = AF_INET6_SDP;
nbd-client.c: un_addr.sun_family = AF_UNIX;
nbd-client.c: if ((sock = socket(AF_UNIX, SOCK_STREAM, 0)) == -1) {
nbd-client.c: if (socketpair(AF_UNIX, SOCK_STREAM, 0, plainfd) < 0)
nbd-server.c: if(netaddr.ss_family == AF_UNIX) {
nbd-server.c: client->clientaddr.ss_family = AF_UNIX;
nbd-server.c: if(client->clientaddr.ss_family == AF_UNIX) {
nbd-server.c: assert((ai->ai_family == AF_INET) || (ai->ai_family == AF_INET6));
nbd-server.c: if(ai->ai_family == AF_INET) {
nbd-server.c: } else if(ai->ai_family == AF_INET6) {
nbd-server.c: socketpair(AF_UNIX, SOCK_STREAM, 0, sockets);
nbd-server.c: sa.sun_family = AF_UNIX;
nbd-server.c: sock = socket(AF_UNIX, SOCK_STREAM, 0);
nbdsrv.c: int addrlen = addr->sa_family == AF_INET ? 4 : 16;
nbdsrv.c: assert(addr->sa_family == AF_INET || addr->sa_family == AF_INET6);
nbdsrv.c: case AF_INET:
nbdsrv.c: case AF_INET6:
tests/code/trim.c: socketpair(AF_UNIX, SOCK_STREAM, AF_UNIX, spair);
tests/run/nbd-tester-client.c: if (socketpair(AF_UNIX, SOCK_STREAM, 0, plainfd) < 0) {
tests/run/nbd-tester-client.c: if ((sock = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
tests/run/nbd-tester-client.c: addr.sun_family = AF_UNIX;
tests/run/nbd-tester-client.c: addr.sin_family = AF_INET;
tests/run/nbd-tester-client.c: if (socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == -1) {


qemu-nbd is a bit hard to grep like this, but it only supports
Unix domain sockets or TCP/IP.


Rich.

--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
Fedora Windows cross-compiler. Compile Windows programs, test, and
build Windows installers. Over 100 libraries supported.
http://fedoraproject.org/wiki/MinGW