On 16/08/17 10:52, Jan Beulich wrote: >>>> On 15.08.17 at 17:04, <[email protected]> wrote: >> On 15/08/17 14:49, Jan Beulich wrote: >>> @@ -2608,7 +2610,7 @@ static long gnttab_copy( >>> { >>> if ( i && hypercall_preempt_check() ) >>> { >>> - rc = i; >>> + rc = count - i; >> Somewhere, probably as a comment for gnttab_copy(), we should have a >> comment explaining why the return value is different from all other ops, >> which will also go somewhat to explaining the "rc = count - rc;" logic. > Sure. > >> I think it would also be wise to have an early exit in >> do_grant_table_op() for passing a count of 0. As far as I can tell, we >> will get all the way into the subop handler before discovering a count of 0. > Well, that would collide with {get,set}_version which don't currently > honor count (and hence existing callers may be passing zero here). > Otherwise I would agree with what you propose.
Lovely :( We've also got a number of other issues, like the fact that count, being unsigned int, gets silently truncated in the non-compat case. ~Andrew _______________________________________________ Xen-devel mailing list [email protected] https://lists.xen.org/xen-devel
