On 05.07.2022 17:26, Jan Beulich wrote: > On 22.06.2022 16:38, Rahul Singh wrote: >> --- a/xen/include/xen/event.h >> +++ b/xen/include/xen/event.h >> @@ -73,12 +73,12 @@ int evtchn_allocate_port(struct domain *d, unsigned int >> port); >> >> /* Allocate a new event channel */ >> int __must_check evtchn_alloc_unbound(evtchn_alloc_unbound_t *alloc, >> - evtchn_port_t port); >> + evtchn_port_t port, bool is_static); >> >> /* Bind an event channel port to interdomain */ >> int __must_check evtchn_bind_interdomain(evtchn_bind_interdomain_t *bind, >> struct domain *ld, >> - evtchn_port_t port); >> + evtchn_port_t port, bool >> is_static); > > Didn't even pay attention to this the first time through: You're > again touching functions you did alter already in earlier patches, > and with them their pre-existing call sites. This is not only > unnecessary code churn but also makes it harder to follow where a > change came from when (perhaps much later) using "git blame" or > alike. Please bring these functions into their intended shape in > a single step (each).
One more thing: Especially "bind" now has quite a few parameters for which without dom0less (i.e. particularly on x86) only a single value would ever be passed. Without LTO the compiler could still deal with this if the function remained static in all non-dom0less cases. Please consider whether you want to do so, or whether you want to find another solution to address this concern. Jan
