Re: [RFC][PATCH] loginuid through procfs (+ a question)
On Thursday 13 January 2005 15:45, Stephen Smalley wrote:
David Woodhouse pointed out that no value in the uid space is
reserved
for such use.
I see places in the kernel setting -1 for uid on init, though. For example,
look at security/keys/keyctl.c
You could introduce a separate flag indicating whether the loginuid
is
set at all, and export both the flag and the uid value, with the latter
simply being 0 if the flag indicates that the loginuid has not been set.
If uid_t is 32 bits, I doubt anyone will have 4 billion users. -1 would work.
We can also check for that value being set through theinterface and flag it
as an error. This should probably be discussed on lkml I think. Its a system
wide policy used in different places. more people than us probably have the
same issue.
-Steve Grubb