tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Vnode scope implementation

Am 21.08.2009 um 12:19 schrieb YAMAMOTO Takashi:



Sorry for the delay, I was a bit busy.

I don't see any reason to hold with implementing the vnode scope just
because we can't control operations on a remote host. Therefore, I'll
give it a few more days (just in case) before moving forward.

what will you do for nfs etc, by "moving forward"?
do nothing and give up mandating the use of vnode scope?

I think it's quite clear that in the case of networked file systems it is always the serving machine that has "the last word" on security decisions and restrictions. So while a nfs client could tighten restrictions, it can not loosen them. I think this is a fact we have to live with.




Elad Efrat wrote:
On Tue, Aug 4, 2009 at 2:07 AM, YAMAMOTO Takashi< > wrote:

but IIUC it can prevent what the
remote file-system would allow
sometimes it can, but in general it can't.

1. a client sends a request to a server.
2. the server decided to allow the operation, and actually process it,
and return the result to the client.

ie. you don't have a chance to pass "fs_decision" to kauth.

Ugh, the above was an obvious typo on my part. I agree, we don't have
a chance to pass fs_decision to kauth(9) because the remote server
will already perform the operation.

Should we enforce that
limitation on all file-systems, or make remote file-systems an
exception? Veriexec sets a precedent of the latter, which I think
makes sense. Do you have something else in mind?
i'm not sure if "remote file-systems or not" is a good classification

How else would you classify it? and you still don't answer the
question of whether you have a solution that will not be considered
"broken". ;)



Home | Main Index | Thread Index | Old Index