Paolo Bonzini <pbonz...@redhat.com> writes: > On 22/01/2015 15:55, Markus Armbruster wrote: >> == Look like a bug == >> >> blockdev-nbd.c:35: leaked_handle: Handle variable "fd" going out of >> scope leaks the handle. > > It's a false positive. > > After nbd_client_new calls nbd_send_negotiate, either it returns or > client escapes via QTAILQ_INSERT_TAIL (either in nbd_client_new or in > nbd_handle_export_name). > > So I think it's the same as below. > > >> == Look like false positive == >> >> The ones in qemu-char.c should be suppressed by our model of >> g_io_channel_unix_new(). Can't see how it screwed that up. > > It seems okay to me too, but these are exactly the false positive that > the g_malloc model is supposed to avoid...
Hmm, I forgot to model g_realloc_n(). And I think I can improve the modelling of "can / can't return null". Let me tinker some more... [...]
