On Thu, Mar 26, 2020 at 11:08:40PM +0100, Vlastimil Babka wrote: > On 3/26/20 9:24 PM, Kees Cook wrote: > I didn't want to modify param for the sake of error prints, but perhaps > the replacements won't confuse system admin too much? Ah, fair enough. Should be fine to do it against "path" then. Ignore that bit from me. ;) > >> + filp_close(file, NULL); > > > > Please check the return value of filp_close() and treat that as an error > > for this function too. > > Well I could print it, but not much else? The unmount will probably fail > in that case? Maybe? This is just a nit of mine from tracking horrible bugs that turned out to be unreported 'close' failures. :) > But I guess the "mount on first applicable argument" approach would work > with this scheme as well: > > struct vfsmount *proc_mnt = NULL; > parse_args(..., &proc_mnt, ...) Yes please! That would be perfect. (And yeah, it's a sensible optimization to do it "as needed"; I hadn't thought of that.) -- Kees Cook
