On Thu, 2006-03-30 at 09:30 -0500, Stephen Smalley wrote:
> +int selinux_ctxid_to_string(u32 ctxid, char **ctx, u32 *ctxlen,
gfp_t gfp_mask);
On second look, I think you can drop the gfp_mask because we have to use
GFP_ATOMIC in context_struct_to_string regardless of what the caller is
using, due to taking the policy rdlock (and looking later in the patch,
you don't pass it down any further, so it serves no purpose). Also,
since you explicitly identify this as a _to_string interface, I think
you can drop the ctxlen argument; the returned string is NUL-terminated
anyway - unless the caller needs the length for some purpose.
Actually, I take the last point back - keep the ctxlen argument please.
Catherine Zhang's patch for Unix datagram SO_PASSSEC support requires
the length information, and she needs this interface too.
--
Stephen Smalley
National Security Agency