Re: [Qemu-devel] [PATCH] monitor: fix use after free

[Luiz Capitulino <lcapitulino@redhat.com>]

On Mon, 18 Aug 2014 22:03:31 +0200
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Mon, Aug 18, 2014 at 02:05:46PM -0400, Luiz Capitulino wrote:
> > On Sun, 17 Aug 2014 11:45:17 +0200
> > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > 
> > > The function monitor_fdset_dup_fd_find_remove() references member of
> > > 'mon_fdset' which - when remove flag is set - may be freed in function
> > > monitor_fdset_cleanup().
> > > remove is set by monitor_fdset_dup_fd_remove which in practice
> > > does not need the returned value, so make it void,
> > > and return -1 from monitor_fdset_dup_fd_find_remove.
> > > 
> > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > Reported-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
> > 
> > I've applied this one to the qmp tree, but I have one comment below.
> > 
> > > ---
> > >  include/monitor/monitor.h | 2 +-
> > >  monitor.c                 | 8 +++++---
> > >  stubs/fdset-remove-fd.c   | 3 +--
> > >  3 files changed, 7 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/include/monitor/monitor.h b/include/monitor/monitor.h
> > > index 3d6929d..78a5fc8 100644
> > > --- a/include/monitor/monitor.h
> > > +++ b/include/monitor/monitor.h
> > > @@ -64,7 +64,7 @@ AddfdInfo *monitor_fdset_add_fd(int fd, bool has_fdset_id, int64_t fdset_id,
> > >                                  Error **errp);
> > >  int monitor_fdset_get_fd(int64_t fdset_id, int flags);
> > >  int monitor_fdset_dup_fd_add(int64_t fdset_id, int dup_fd);
> > > -int monitor_fdset_dup_fd_remove(int dup_fd);
> > > +void monitor_fdset_dup_fd_remove(int dup_fd);
> > >  int monitor_fdset_dup_fd_find(int dup_fd);
> > >  
> > >  #endif /* !MONITOR_H */
> > > diff --git a/monitor.c b/monitor.c
> > > index cdbaa60..ba1908f 100644
> > > --- a/monitor.c
> > > +++ b/monitor.c
> > > @@ -2542,8 +2542,10 @@ static int monitor_fdset_dup_fd_find_remove(int dup_fd, bool remove)
> > >                      if (QLIST_EMPTY(&mon_fdset->dup_fds)) {
> > >                          monitor_fdset_cleanup(mon_fdset);
> > >                      }
> > > +                    return -1;
> > 
> > Returning -1 here looks wrong. A better fix is probably to split this
> > function into two functions as it's doing two unrelated things currently.
> > 
> > I've applied the fix anyway because the only user of remove=true doesn't
> > care about the return value and also because the bad semantic is less worse
> > than the use after free...
> 
> 
> I agree generally, it's just that I don't understand this function.
> For example, doesn't it leak memory? When
> monitor_fdset_dup_fd_find_remove drops a MonFdsetFd entry from list, it does
> not seem to free it.

It seems so. Corey, are you still around? Can you take a look at this?

> 
> 
> 
> > > +                } else {
> > > +                    return mon_fdset->id;
> > >                  }
> > > -                return mon_fdset->id;
> > >              }
> > >          }
> > >      }
> > > @@ -2555,9 +2557,9 @@ int monitor_fdset_dup_fd_find(int dup_fd)
> > >      return monitor_fdset_dup_fd_find_remove(dup_fd, false);
> > >  }
> > >  
> > > -int monitor_fdset_dup_fd_remove(int dup_fd)
> > > +void monitor_fdset_dup_fd_remove(int dup_fd)
> > >  {
> > > -    return monitor_fdset_dup_fd_find_remove(dup_fd, true);
> > > +    monitor_fdset_dup_fd_find_remove(dup_fd, true);
> > >  }
> > >  
> > >  int monitor_handle_fd_param(Monitor *mon, const char *fdname)
> > > diff --git a/stubs/fdset-remove-fd.c b/stubs/fdset-remove-fd.c
> > > index b3886d9..7f6d61e 100644
> > > --- a/stubs/fdset-remove-fd.c
> > > +++ b/stubs/fdset-remove-fd.c
> > > @@ -1,7 +1,6 @@
> > >  #include "qemu-common.h"
> > >  #include "monitor/monitor.h"
> > >  
> > > -int monitor_fdset_dup_fd_remove(int dupfd)
> > > +void monitor_fdset_dup_fd_remove(int dupfd)
> > >  {
> > > -    return -1;
> > >  }
>