Messages in this thread | | | From | "Peter T. Breuer" <> | Subject | Re: Total And Complete Wackyness | Date | Fri, 11 Feb 2000 20:35:10 +0100 (MET) |
| |
"A month of sundays ago jmcmullan@linuxcare.com wrote:" > Guus Sliepen <guus@warande3094.warande.uu.nl> wrote: > > [re: layering a protocol over a block device] > > > > This MAY sound like fiction, but this is already a problem with raid (just > > mount a raid over /dev/hda1..4, you still can access or mount hda1..hda4 > > seperately). > > > Hmz, MIT magic cookies ;)? > > Vhat are deez, ``Majeek cookees''? > > Seriously, what is the MIT magic cookies approach? Do you have a > terse explanation, or a good URL? I'd love to hear it.
It leaves a lot to the imagination ...
My yoke device already fixes some of the problem with leaving devices accessible under a raid mirror, by simply not requiring any overlying device. It hooks together the raw devices, hijacking their request_fn's in the kernel, and doing other bits of cuckoo-type things. (uh, probably available via www.freshmeat.net; /yoke).
But I don't see why part of the action of the ordinary raid device couldn't be to do the same thing in reverse .. disable access via the underlying devices. I might suggest a patch in a while (soon as I can get hold of a recent kernel source). It should probably stash the ops structure already registerered for the device and substitute its own. To check .. if raid depends on the strcuture being available just where it used to be ...
Peter
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |