> -----Original Message----- [snip] > >> diff --git a/xen/common/ioreq.c b/xen/common/ioreq.c > >> index a319c88..72b5da0 100644 > >> --- a/xen/common/ioreq.c > >> +++ b/xen/common/ioreq.c > >> @@ -591,8 +591,8 @@ static void hvm_ioreq_server_deinit(struct > >> ioreq_server *s) > >> put_domain(s->emulator); > >> } > >> > >> -int hvm_create_ioreq_server(struct domain *d, int bufioreq_handling, > >> - ioservid_t *id) > >> +static int hvm_create_ioreq_server(struct domain *d, int > >> bufioreq_handling, > >> + ioservid_t *id) > > Would this not be a good opportunity to drop the 'hvm_' prefix (here and > > elsewhere)? > > It would be, I will drop. > > > May I ask, are you ok with that alternative approach proposed by Jan and > already implemented in current version (top level dm-op handling > arch-specific > and call into ioreq_server_dm_op() for otherwise unhandled ops)? >
Yes, I think per-arch hypercall handlers is the tidiest way to go. Paul > Initial discussion here: > https://lore.kernel.org/xen-devel/1606732298-22107-10-git-send-email-olekst...@gmail.com/ > > -- > Regards, > > Oleksandr Tyshchenko