The error handling in ksmbd_server_init() uses "one function to free
everything style" which is impossible to audit and leads to several
canonical bugs.  When we free something that wasn't allocated it may be
uninitialized, an error pointer, freed in a different function or we
try freeing "foo->bar" when "foo" is a NULL pointer.  And since the
code is impossible to audit then it leads to memory leaks.

In the ksmbd_server_init() function, every goto will lead to a crash
because we have not allocated the work queue but we call
ksmbd_workqueue_destroy() which tries to flush a NULL work queue.
Another bug is if ksmbd_init_buffer_pools() fails then it leads to a
double free because we free "work_cache" twice.  A third type of bug is
that we forgot to call ksmbd_release_inode_hash() so that is a resource
leak.

A better way to write error handling is for every function to clean up
after itself and never leave things partially allocated.  Then we can
use "free the last successfully allocated resource" style.  That way
when someone is reading the code they can just track the last resource
in their head and verify that the goto matches what they expect.

In this patch I modified ksmbd_ipc_init() to clean up after itself and
then I converted ksmbd_server_init() to use gotos to clean up.

Fixes: cabcebc31de4 ("cifsd: introduce SMB3 kernel server")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>