Re: qemu-user viability (was Re: [SECURITY] [DSA 5983-1] qemu security update)
Please, everyone, tell me.
Suppose we had suid-root /bin/sh for 15 years. We noticed this
and removed the suid bit from it, finally - because this way everyone's
system was trivially vulnerable to a trivial local root - there isn't
even "exploit" necessary, just run /bin/sh and be root.
Now you're asking to return things back "because real life is not
a theoretical university thesis". Should we keep /bin/sh suid-root
because some users setups broke when we revoked suid bit from a
binary which never, ever, supposed to be suid?
Sure you can rebuild your /bin/sh to make it suid (I dunno why do
you want to rebuild it when there's chmod, but this is a different
question). You're free to do it, that's your system. This does
not mean everyone else system should be trivially owned like we
had.
And yes, as I mentioned before, some setups might break - the ones
which relied on suid/sgid bits. If you had setup which relied on
/bin/sh being suid-root, it wont work anymore. And it is not me
who should tell you how to change your setup, because I don't
know your setup. /bin/sh is not supposed to be suid-root, and
there's no way it will be kept suid-root.
What are we talking about? I don't understand.
Thanks,
/mjt
Reply to: