[libvirt] [PATCH 1/2] virt-aa-helper: add rules for shmem devices

[Christian Ehrhardt]

On Tue, Nov 19, 2019 at 10:25 PM Jamie Strandboge <jamie at canonical.com> wrote:
>
> On Tue, 22 Oct 2019, Christian Ehrhardt wrote:
>
> > Shared memory devices need qemu to be able to access certain paths
> > either for the shared memory directly (mostly ivshmem-plain) or for a
> > socket (mostly ivshmem-doorbell).
> >
> > Add logic to virt-aa-helper to render those apparmor rules based
> > on the domain configuration.
> >
> > https://bugzilla.redhat.com/show_bug.cgi?id=1761645
> >
> > Signed-off-by: Christian Ehrhardt <christian.ehrhardt at canonical.com>
> > ---
> >  src/security/virt-aa-helper.c | 35 +++++++++++++++++++++++++++++++++++
> >  1 file changed, 35 insertions(+)
> >
> > diff --git a/src/security/virt-aa-helper.c b/src/security/virt-aa-helper.c
> > index 7d7262ca39..8c261f0010 100644
> > --- a/src/security/virt-aa-helper.c
> > +++ b/src/security/virt-aa-helper.c
> > @@ -958,6 +958,7 @@ get_files(vahControl * ctl)
> >      int rc = -1;
> >      size_t i;
> >      char *uuid;
> > +    char *mem_path = NULL;
> >      char uuidstr[VIR_UUID_STRING_BUFLEN];
> >      bool needsVfio = false, needsvhost = false, needsgl = false;
> >
> > @@ -1224,6 +1225,39 @@ get_files(vahControl * ctl)
> >          }
> >      }
> >
> > +    for (i = 0; i < ctl->def->nshmems; i++) {
> > +        if (ctl->def->shmems[i]) {
> > +            virDomainShmemDef *shmem = ctl->def->shmems[i];
> > +            /* server path can be on any type and overwrites defaults */
> > +            if (shmem->server.enabled &&
> > +                shmem->server.chr.data.nix.path) {
> > +                    if (vah_add_file(&buf, shmem->server.chr.data.nix.path,
> > +                            "rw") != 0)
> > +                        goto cleanup;
>
> I'll let others comment on the code changes, but this apparmor rule
> looks ok.
>
> > +            } else {
>
> That said, I wonder about the logic here since up above we have:
>
>   if (shmem->server.enabled && shmem->server.chr.data.nix.path)
>
> but here we just have 'else'. What happens if server.enabled is false
> and server.chr.data.nix.path is set or vice versa? Does this 'else'
> clause correctly handle those scenarios?

Yes if either of the above isn't fulfilled it will fall back to use
the default paths.
So a single else without other checks should be correct.
The following switch then differs the default paths used base on the model.

>
> > +                switch (shmem->model) {
> > +                case VIR_DOMAIN_SHMEM_MODEL_IVSHMEM_PLAIN:
> > +                    /* until exposed, recreate qemuBuildShmemBackendMemProps */
> > +                    if (virAsprintf(&mem_path, "/dev/shm/%s", shmem->name) < 0)
> > +                        goto cleanup;
> > +                    break;
> > +                case VIR_DOMAIN_SHMEM_MODEL_IVSHMEM_DOORBELL:
> > +                case VIR_DOMAIN_SHMEM_MODEL_IVSHMEM:
> > +                     /* until exposed, recreate qemuDomainPrepareShmemChardev */
> > +                    if (virAsprintf(&mem_path, "/var/lib/libvirt/shmem-%s-sock",
> > +                            shmem->name) < 0)
> > +                        goto cleanup;
> > +                    break;
> > +                }
> > +                if (mem_path != NULL) {
> > +                    if (vah_add_file(&buf, mem_path, "rw") != 0)
> > +                        goto cleanup;
> > +                }
> > +            }
> > +        }
> > +    }
> > +
>
>
> --
> Jamie Strandboge             | http://www.canonical.com



-- 
Christian Ehrhardt
Staff Engineer, Ubuntu Server
Canonical Ltd