On 03/08/2020 19:21, Oleksandr Tyshchenko wrote: > diff --git a/xen/common/Makefile b/xen/common/Makefile > index 06881d0..f6fc3f8 100644 > --- a/xen/common/Makefile > +++ b/xen/common/Makefile > @@ -70,6 +70,7 @@ extra-y := symbols-dummy.o > > obj-$(CONFIG_COVERAGE) += coverage/ > obj-y += sched/ > +obj-$(CONFIG_IOREQ_SERVER) += hvm/ > obj-$(CONFIG_UBSAN) += ubsan/ > > obj-$(CONFIG_NEEDS_LIBELF) += libelf/ > diff --git a/xen/common/hvm/Makefile b/xen/common/hvm/Makefile > new file mode 100644 > index 0000000..326215d > --- /dev/null > +++ b/xen/common/hvm/Makefile > @@ -0,0 +1 @@ > +obj-y += ioreq.o > diff --git a/xen/common/hvm/ioreq.c b/xen/common/hvm/ioreq.c > new file mode 100644 > index 0000000..7e1fa23 > --- /dev/null > +++ b/xen/common/hvm/ioreq.c > <snip>
HVM is an internal detail of arch specific code. It should not escape into common code. >From x86's point of view, there is nothing conceptually wrong with having an IOREQ server for PV guests, although it is very unlikely at this point that adding support would be a good use of time. Please make this into a proper top-level common set of functionality. ~Andrew
