Messages in this thread | | | Date | Thu, 17 Jun 1999 00:14:27 -0400 | Subject | Re: file size limit | From | tytso@mit ... |
| |
From: "Stephen C. Tweedie" <sct@redhat.com> Date: Wed, 16 Jun 1999 14:12:28 +0100 (BST)
> If we move to extents, we can start to do nifty stuff: there is a > 32bit start block number, and a 32bit length field in an extent. That > length field is WAY over dimensioned. You can grab a few bits there > for "move to another meta-group". Meta-group being a group of 4G > blocks.
The plan is to use 64 bits for both keys and locations in the extent map. The extra bits in the location will let us span over multiple devices if we want to.
The plan was to use 64 bit block and inode pointers as part of doing the B-tree directory and extent extensions to ext2/ext3. 32 bits would be reserved for inode and block number, 16 bits would be reserved as an index into device/volume table, and 16 bits would be reserved for future use.
I was not planning on extents to use 32 bit length field, because as you point out, that is a vast overprovisioning of that field. I was thinking more like 16 bits, maximum.
- Ted
- 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/
| |