On Fri, Dec 02, 2011 at 08:48:50AM +0100, Paolo Bonzini wrote: > On 12/02/2011 08:42 AM, Stefan Weil wrote: > >>> > >>>- if (sharing_fds[0] == -1) > >>>+ if (sharing_fds[0] == -1) { > >>>+ g_free(sharing_fds); > >>> return 1; > >>>+ } > >>> > >>> if (device) { > >>> int ret; > >>Zhihui, > >> > >>Kernel should free all memory used by the process after it exits. So > >>there's no memory leak even without explicit free. > > > >That's correct. Nevertheless fixing this helps to find other more important > >memory leaks with static or dynamic code analyzers like cppcheck or > >valgrind. > > > >Reviewed-by: Stefan Weil <s...@weilnetz.de> > > Please hold off, I'm about to introduce large changes to qemu-nbd > for asynchronous I/O (including getting rid of sharing_fds in favor > of the main loop) and and I'd rather avoid gratuitous conflicts.
Okay, let's wait for Paolo's changes. Stefan