christos@ wrote: > In article <[email protected]>, > Izumi Tsutsui <[email protected]> wrote: > > > >The problem is getdisksize() was introduced after 5.0 > >and it assumed DIOCGPART always returned valid values, > >but no one confirmed if the assumption was correct. > > What do you think the drivers called before to get the > size? getdisksize() was introduced just to centralize > checking both ioctls(). I.e. there is no behavioral > change in the absence of wedges.
Hmm, as you say md(4) would have failed in msdosfs_mountfs() even before getdisksize() was introduced and the only problem was in disklabel support addition to md(4). Anyway KASSERT() (or sanity check) should also be centralized to getdisksize()? --- Izumi Tsutsui
