> On Tue, Sep 05, 2000 at 09:18:04AM -0600, Davida, Joe wrote:
> > Is what Bill Jones is saying correct?
> > I am just starting on XFS, and dont
> > have enough familiarity with it's
> > architecture.
> The LVM trick he proposes will not work on current Linux
> or Linux/XFS (at least not without changes to LVM). So you
> have a 2TB limit. 1TB may be safer due to possible
> driver signedness bugs.
I think the safest comment to make at the moment is that the path of least
resistence to getting above the 2 Tbyte limit is probably a kiobuf enabled
version of LVM, this path too has 2 Tbyte limits in it at the moment.
ll_rw_block and generic_make_request are the places where the across the
volume limit is getting imposed. If we could make it to a remapping request
function without having to map a disk address through a 32 bit sector number
then it should be doable.
The other end of the pipe is the XFS inode number, and I am still not
convinced that mkfs options can tune the filesystem to get above the
2 Tbyte boundary, a mount option which restricts inode placement to
within the first 2 Tbytes of allocation groups should not be too hard.