In article <[email protected]> you wrote:
> They wouldn't have to be marked: capabilities are inherited by
> default, with my patch (as is the Unix tradition: euid=0 or {r,s}uid=0
> are preserved upon execve()), normal processes have CAP_FORK and just
> pass it on if you don't do something special to remove it.
The Problem with that is, that a program can be started with some priveldges
without knowing it. Traditional programs only check for uid=0 and in that
case refuse to do some things. A program might not expect to be able to do a
priveldged action with not being uid=0.
I do think those programs are broken, but it is a argument, why the
inheritance should be off by default (unless the euid is always 0 if any
root privelddge is enabled)
> Yes. In general, processes have all "regular" capabilities, and they
> are inherited normally.
I like that, i think it was once called user capabilities. I wonder if we
need to make those distinctions at all.
BTW: some of the priveldges could also be avoided by adding access control
to some kernel objects. I really like the windows approach here, and it
would be possible to have the permissions bound to special files (which
would also allow ACLs). So for example instead of haing CAP_NET_ADMIN the
kernel could check "can the process open /dev/objects/netadmin for write".
Gruss
Bernd
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
