Date: Wed, 10 Jan 1996 00:31:41 +1100 From: Bruce Evans <bde@zeta.org.au> To: bde@zeta.org.au, msmith@atrad.adelaide.edu.au Cc: hackers@FreeBSD.ORG, phk@critter.tfs.com Subject: Re: Add new slice to running system, comments? Message-ID: <199601091331.AAA22987@godzilla.zeta.org.au>
next in thread | raw e-mail | index | archive | help
>> Fix ufs to have decent performance for swapping on a vn device first :-). >In this context, vn devices are barely a bandaid. Only if they don't have decent performance. The vn driver could handle contiguous files (non-growing) with about 0.001% overhead compared with raw devices. It just needs to read the block map of the device early and replace reading it later by a `+'. After reading the code :-), I think the vn devices already have decent performance for the contiguous && swapping case. There isn't much overhead except for a VOP_BMAP() call for each block. VOP_READ() and VOP_WRITE() aren't used for the swapping case (they would be extremely slow for msdosfs and merely OK for ufs). The original vn driver avoided VOP_READ() and VOP_WRITE() in all cases, but John rewrote it. Contiguous files are necessary so that the disk doesn't have to seek for between virtually adjacent blocks. A large block size may also be necessary. I now think you should use the vn device. >> Slices are supposed to be disjoint. There are enough complications for >Supposed phooey. I have a new use where nesting them would be very >desirable. Is there anything other than the design philosophy that >will melt down if slices overlap? Try it and see :-). There's nothing to stop overlapping slices except fdisks that refuse to create them. Other OS's might by confused by overlapping slices in the MBR. >> overlapping partitions inside slices. Since the windows swap file is in >> a DOS slice, it seems best to create a label for this slice and describe >> the swap file as the b partition in the label. You will need a label >> anyway since swapping only works on the b partition. >Ah. This conflicts with what PHK said, so I guess I'm going to have to >play with this. I was wrong. >Ok, let's stretch things a bit further. >I lay down a slab of contiguous clusters inside a FAT filesystem, and assign >them to a file, so nothing else will stuff with them. >I want to put some filesystems in there, and possibly some more swap >(because Windows swapfiles are usually too small), and I also want to be able >to get at a hypothetical Windows swapfile _and_ the surrounding FAT filesystem. Use vn devices and fix the vm (?) problems that required non-swapping accesses to go through the file system. I'm not sure how to handle booting from one of these slices. You could put the slice in the MBR only if there is a free entry. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199601091331.AAA22987>