* Simon Horman [2010-09-25 21:34:02 +0900]: > On Fri, Sep 24, 2010 at 10:36:09AM +0200, Sergio Gelato wrote: > > The main problem is that perdition/io.c:io_pipe() and its caller > > perdition/perdition.c:perdition_log_close() use int counters > > while the corresponding arguments in vanessa_socket_pipe_func() > > are declared size_t. I'd worry about stack corruption on platforms > > where sizeof(size_t) > sizeof(int). > > > > Suggested fix: declare those counters size_t, and (for cosmetic purposes) > > cast them to unsigned long before formatting them with %lu instead of %d. > > Thanks, I'll fix that. > > Do you think it warrants an update to the testing (= already frozen squeeze) > package?
I do: amd64 has sizeof(size_t)==8 and sizeof(int)==4. -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
