I think the best way is borrowing constructor and destructor concepts from C++. For each internal data structure foo, maintain a foo_create function as the constructor(many data structures are already done), and a foo_destroy function as the destructor. Then we avoid duplicate and complicate clean work at each failure exit in foo_create. <div class="gmail_quote">2010/11/23 Kristian Høgsberg <<a href="mailto:krh@bitplanet.net">krh@bitplanet.net</a>> <blockquote style="BORDER-LEFT: #ccc 1px solid; MARGIN: 0px 0px 0px 0.8ex; PADDING-LEFT: 1ex" class="gmail_quote"> <div class="im">On Wed, Nov 17, 2010 at 10:16 PM, 林昊翔 <<a href="mailto:linhaoxiang@gmail.com">linhaoxiang@gmail.com</a>> wrote: > The scenario is: > struct A *foo() > { > A *a; > > a = malloc(sizeof(A)); > if (a == NULL) > return NULL; > > a->b = malloc(sizeof(B)); > if (a->b == NULL) > return NULL; > > return a; > } </div>You're right, there's a bunch of those in the code. Could you please create your patches by committing the changes and then either attach the output of git format-patch -1 (where -1 means "one commit", use whatever number of patches you have) or even better use git send-email -1 --to=<a href="mailto:wayland-devel@lists.freedesktop.org">wayland-devel@lists.freedesktop.org</a>. Thanks, Kristian </blockquote></div>