On Tue, Jul 01, 2003 at 01:25:43PM +0200, Benjamin Herrenschmidt wrote: > > > but as benh pointed out i want to implement the standard networking > > interface as well, so we can do tftp, nfs etc. > > You probably need only a very small subset of it though. The big thing > beeing: do you want TCP or not ? If you don't, then you can do something > very simple that is plenty enough for tftp & nfs on an UDP basis. good point, udp only would probably be sufficient. > If you want TCP, it becomes more complex and we can't really rely on > later OF own TCP implementation. yup > > > - It's probably easiest to insist that the filesystems only read/write > > > blocks, and allow the upper layer to handle smaller chunks. > > > > no i think this would be problematic. > > The probleme here is that your low level native blk size can be > different than your filesystem requested block size, so you need some > level of adaptation in between how so? the filesystem drivers will just use read() with whatever size argument they need for a particular operation. -- Ethan Benson http://www.alaska.net/~erbenson/
Attachment:
pgp00003.pgp
Description: PGP signature