Re: [Qemu-devel] [PATCH v6 06/24] memfd: add fallback for memfd

[Marc-André Lureau]

Hi

----- Original Message -----
> On Tue, Sep 29, 2015 at 06:34:36PM +0200, address@hidden wrote:
> > From: Marc-André Lureau <address@hidden>
> > 
> > Add an open/unlink/mmap fallback for system that do not support memfd.
> > This patch may require additional SELinux policies to work for enforced
> > systems, but should gracefully fail nonetheless.
> > 
> > Signed-off-by: Marc-André Lureau <address@hidden>
> 
> I'd rather just fail migration.

So we don't provide this compatibility code and migration should fail.

Would it be enough to check if memfd works at early runtime and add a migration 
blocker for vhost-user? Or is it possible to recover if migration fails when 
memfd fails to allocate? I would thing the former is better.

> 
> > ---
> >  util/memfd.c | 22 ++++++++++++++++++++--
> >  1 file changed, 20 insertions(+), 2 deletions(-)
> > 
> > diff --git a/util/memfd.c b/util/memfd.c
> > index 3168902..970b5b0 100644
> > --- a/util/memfd.c
> > +++ b/util/memfd.c
> > @@ -84,8 +84,26 @@ void *qemu_memfd_alloc(const char *name, size_t size,
> > unsigned int seals,
> >              return NULL;
> >          }
> >      } else {
> > -        perror("memfd");
> > -        return NULL;
> > +        const char *tmpdir = getenv("TMPDIR");
> > +        gchar *fname;
> > +
> > +        tmpdir = tmpdir ? tmpdir : "/tmp";
> > +
> > +        fname = g_strdup_printf("%s/memfd-XXXXXX", tmpdir);
> > +        mfd = mkstemp(fname);
> > +        unlink(fname);
> > +        g_free(fname);
> > +
> > +        if (mfd == -1) {
> > +            perror("mkstemp");
> > +            return NULL;
> > +        }
> > +
> > +        if (ftruncate(mfd, size) == -1) {
> > +            perror("ftruncate");
> > +            close(mfd);
> > +            return NULL;
> > +        }
> >      }
> >  
> >      ptr = mmap(0, size, PROT_READ | PROT_WRITE, MAP_SHARED, mfd, 0);
> > --
> > 2.4.3
>