Re: [Libguestfs] virt-v2v valgrind errors in libosinfo

On Tuesday, 14 April 2020 11:53:30 CEST Richard W.M. Jones wrote: I think this might have to do with the glib allocator (libosinfo is based on glib, so it allocate using it), that allocates in chunks by default to avoid fragmentation, IIRC. Do you still get the same issues if you export G_SLICE=always-malloc when running valgrind? This is correct according to the way we use OsinfoOs so far, i.e. only by using what OsinfoDb has, because OsinfoDb holds references on the OsinfoOs objects. I talked with Fabiano (main libosinfo developer) about this, and sadly OsinfoDb has also references to other objects (like the devices) inside each OsinfoOs, so it is not possible to get the ownership of all the OsinfoOs (g_object_ref) and then get rid of the OsinfoDb (g_object_unref). Hm it is never freed? Wouldn't that result in actual leaks, since OsinfoDb_t_finalize (g_object_unref'ing the OsinfoDb) wouldn't be called? I thought about this, and according to knowledge this would be needed only if we want osinfo_os objects alive even when the osinfo_db gets "out of scope". Considering that the osinfo_db is kept basically statically this should be fine. Oops yes. "Db" and "Os" sadly look too similar... -- Pino Toscano