Re: [PATCH v2 1/1] python/machine: Fix AF_UNIX path too long on macOS

[Peter Delevoryas]

On Mon, Jul 18, 2022 at 09:56:17AM +0100, Daniel P. Berrangé wrote:
> On Sat, Jul 16, 2022 at 10:34:34AM -0700, Peter Delevoryas wrote:
> > On macOS, private $TMPDIR's are the default. These $TMPDIR's are
> > generated from a user's unix UID and UUID [1], which can create a
> > relatively long path:
> > 
> >     /var/folders/d7/rz20f6hd709c1ty8f6_6y_z40000gn/T/
> > 
> > QEMU's avocado tests create a temporary directory prefixed by
> > "avo_qemu_sock_", and create QMP sockets within _that_ as well.
> > The QMP socket is unnecessarily long, because a temporary directory
> > is created for every QEMUMachine object.
> > 
> >     /avo_qemu_sock_uh3w_dgc/qemu-37331-10bacf110-monitor.sock
> 
> 
> Looking at this again, I realize my suggestion for dealing with the
> second part of the path was mistaken.
> 
> The "qemu-37331-10bacf110-monitor.sock" part is combining two
> pieces.
> 
> First the result of
> 
>   f"qemu-{os.getpid()}-{id(self):02x}"
> 
> is
> 
>   qemu-37331-10bacf110
> 
> and the code later than appends '-monitor.sock'
> 
> So...
> 
> > 
> > The path limit for unix sockets on macOS is 104: [2]
> > 
> >     /*
> >      * [XSI] Definitions for UNIX IPC domain.
> >      */
> >     struct  sockaddr_un {
> 
> >         unsigned char   sun_len;        /* sockaddr len including null */
> >         sa_family_t     sun_family;     /* [XSI] AF_UNIX */
> >         char            sun_path[104];  /* [XSI] path name (gag) */
> >     };
> > 
> > This results in avocado tests failing on macOS because the QMP unix
> > socket can't be created, because the path is too long:
> > 
> >     ERROR| Failed to establish connection: OSError: AF_UNIX path too long
> > 
> > This change reduces the size of both paths, and removes the unique
> > identification information from the socket name, since it seems to be
> > unnecessary.
> > 
> > This commit produces paths like the following:
> > 
> >     pdel@pdel-mbp:/var/folders/d7/rz20f6hd709c1ty8f6_6y_z40000gn/T
> >     $ tree qemu*
> >     qemu_oc7h7f3u
> >     ├── qmp-console.sock
> >     └── qmp-monitor.sock
> > 
> > [1] 
> > https://apple.stackexchange.com/questions/353832/why-is-mac-osx-temp-directory-in-weird-path
> > [2] 
> > /Library/Developer/CommandLineTools/SDKs/MacOSX12.3.sdk/usr/include/sys/un.h
> > 
> > Signed-off-by: Peter Delevoryas <peter@pjd.dev>
> > ---
> >  python/qemu/machine/machine.py         | 2 +-
> >  tests/avocado/avocado_qemu/__init__.py | 2 +-
> >  2 files changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/python/qemu/machine/machine.py b/python/qemu/machine/machine.py
> > index 37191f433b..b1823966b3 100644
> > --- a/python/qemu/machine/machine.py
> > +++ b/python/qemu/machine/machine.py
> > @@ -157,7 +157,7 @@ def __init__(self,
> >          self._wrapper = wrapper
> >          self._qmp_timer = qmp_timer
> >  
> > -        self._name = name or f"qemu-{os.getpid()}-{id(self):02x}"
> > +        self._name = name or "qmp"
> 
> ...my suggestion here was wrong.
> 
> We don't need the os.getpid() unoiqueness because the tmpdir already
> ensures that is safe, but keeping 'id(self)' is a good idea, if the
> test case creates multiple machines concurrently. Bearing in mind we
> later append '-monitor.sock' we don't need 'qmp' in the self._name.

Yeah you're right, it's a good idea to make the QEMUMachine name unique.

When I made v2, I looked around to see if we were creating 2
QEMUMachine's concurrently. I didn't see anything, but looking again, I
see this in tests/migration/guestperf/engine.py:

    if self._dst_host != "localhost":
        dstmonaddr = ("localhost", 9001)
    else:
        dstmonaddr = "/var/tmp/qemu-dst-%d-monitor.sock" % os.getpid()
    srcmonaddr = "/var/tmp/qemu-src-%d-monitor.sock" % os.getpid()

    src = QEMUMachine(self._binary,
                      args=self._get_src_args(hardware),
                      wrapper=self._get_src_wrapper(hardware),
                      name="qemu-src-%d" % os.getpid(),
                      monitor_address=srcmonaddr)

    dst = QEMUMachine(self._binary,
                      args=self._get_dst_args(hardware, uri),
                      wrapper=self._get_dst_wrapper(hardware),
                      name="qemu-dst-%d" % os.getpid(),
                      monitor_address=dstmonaddr)

This test case would probably still pass with this change, because the
monitor_address is specified, but I think you're right, it's a good idea
to make the default name different for each QEMUMachine instance to
avoid breaking anything.

> 
> So on reflection I think I should have suggested using:
> 
>     self._name = name or f"{id(self):02x}"

+1, yeah I'll change it to that

> 
> And *in addition*, a few lines later change:
> 
>             self._monitor_address = os.path.join(
>                 self.sock_dir, f"{self._name}-monitor.sock"
>             )
> 
> To
> 
>             self._monitor_address = os.path.join(
>                 self.sock_dir, f"{self._name}.qmp"
>             )

Oh ok, yeah that looks good too. Thanks for the detailed review and
suggestions!

Peter

> 
> 
> > diff --git a/tests/avocado/avocado_qemu/__init__.py 
> > b/tests/avocado/avocado_qemu/__init__.py
> > index ed4853c805..43b8c8848c 100644
> > --- a/tests/avocado/avocado_qemu/__init__.py
> > +++ b/tests/avocado/avocado_qemu/__init__.py
> > @@ -296,7 +296,7 @@ def require_accelerator(self, accelerator):
> >                          "available" % accelerator)
> >  
> >      def _new_vm(self, name, *args):
> > -        self._sd = tempfile.TemporaryDirectory(prefix="avo_qemu_sock_")
> > +        self._sd = tempfile.TemporaryDirectory(prefix="qemu_")
> 
> This bit is fine.

+1

> 
> >          vm = QEMUMachine(self.qemu_bin, base_temp_dir=self.workdir,
> >                           sock_dir=self._sd.name, log_dir=self.logdir)
> >          self.log.debug('QEMUMachine "%s" created', name)
> 
> With regards,
> Daniel
> -- 
> |: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
> |: https://libvirt.org         -o-            https://fstop138.berrange.com :|
> |: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
>