Re: [PATCH v2 5/6] tools/oxenstored: Rework Domain evtchn handling to use port_pair

From: Christian Lindig <christian.lindig@citrix.com>
To: Andrew Cooper <Andrew.Cooper3@citrix.com>
Cc: Xen-devel <xen-devel@lists.xenproject.org>,
	David Scott <dave@recoil.org>,
	Edwin Torok <edvin.torok@citrix.com>,
	Rob Hoes <Rob.Hoes@citrix.com>
Subject: Re: [PATCH v2 5/6] tools/oxenstored: Rework Domain evtchn handling to use port_pair
Date: Thu, 1 Dec 2022 11:59:02 +0000	[thread overview]
Message-ID: <B01EB9B0-2E59-460E-9F1B-04F2406C788B@citrix.com> (raw)
In-Reply-To: <20221130165455.31125-6-andrew.cooper3@citrix.com>

> On 30 Nov 2022, at 16:54, Andrew Cooper <Andrew.Cooper3@citrix.com> wrote:
> 
> Inter-domain event channels are always a pair of local and remote ports.
> Right now the handling is asymmetric, caused by the fact that the evtchn is
> bound after the associated Domain object is constructed.
> 
> First, move binding of the event channel into the Domain.make() constructor.
> This means the local port no longer needs to be an option.  It also removes
> the final callers of Domain.bind_interdomain.
> 
> Next, introduce a new port_pair type to encapsulate the fact that these two
> should be updated together, and replace the previous port and remote_port
> fields.  This refactoring also changes the Domain.get_port interface (removing
> an option) so take the opportunity to name it get_local_port instead.
> 
> Also, this fixes a use-after-free risk with Domain.close.  Once the evtchn has
> been unbound, the same local port number can be reused for a different
> purpose, so explicitly invalidate the ports to prevent their accidental misuse
> in the future.
> 
> This also cleans up some of the debugging, to always print a port pair.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> ---
> CC: Christian Lindig <christian.lindig@citrix.com>
> CC: David Scott <dave@recoil.org>
> CC: Edwin Torok <edvin.torok@citrix.com>
> CC: Rob Hoes <Rob.Hoes@citrix.com>

Acked-by: Christian Lindig <christian.lindig@citrix.com>

> 
> v2:
> * New
> ---
> tools/ocaml/xenstored/connections.ml |  9 +----
> tools/ocaml/xenstored/domain.ml      | 75 ++++++++++++++++++------------------
> tools/ocaml/xenstored/domains.ml     |  2 -
> 3 files changed, 39 insertions(+), 47 deletions(-)
> 
> diff --git a/tools/ocaml/xenstored/connections.ml b/tools/ocaml/xenstored/connections.ml
> index 7d68c583b43a..a80ae0bed2ce 100644
> --- a/tools/ocaml/xenstored/connections.ml
> +++ b/tools/ocaml/xenstored/connections.ml
> @@ -48,9 +48,7 @@ let add_domain cons dom =
> 	let xbcon = Xenbus.Xb.open_mmap ~capacity (Domain.get_interface dom) (fun () -> Domain.notify dom) in
> 	let con = Connection.create xbcon (Some dom) in
> 	Hashtbl.add cons.domains (Domain.get_id dom) con;
> -	match Domain.get_port dom with
> -	| Some p -> Hashtbl.add cons.ports p con;
> -	| None -> ()
> +	Hashtbl.add cons.ports (Domain.get_local_port dom) con

I would prefer Hashtbl.replace. Hashtbl.add shadows an existing binding which becomes visible again after Hashtabl.remove. When we are sure that we only have one binding per key, we should use replace instead of add. 

> 
> let select ?(only_if = (fun _ -> true)) cons =
> 	Hashtbl.fold (fun _ con (ins, outs) ->
> @@ -97,10 +95,7 @@ let del_domain cons id =
> 		let con = find_domain cons id in
> 		Hashtbl.remove cons.domains id;
> 		(match Connection.get_domain con with
> -		 | Some d ->
> -		   (match Domain.get_port d with
> -		    | Some p -> Hashtbl.remove cons.ports p
> -		    | None -> ())
> +		 | Some d -> Hashtbl.remove cons.ports (Domain.get_local_port d)
> 		 | None -> ());
> 		del_watches cons con;
> 		Connection.close con
> diff --git a/tools/ocaml/xenstored/domain.ml b/tools/ocaml/xenstored/domain.ml
> index d59a9401e211..ecdd65f3209a 100644
> --- a/tools/ocaml/xenstored/domain.ml
> +++ b/tools/ocaml/xenstored/domain.ml
> @@ -19,14 +19,31 @@ open Printf
> let debug fmt = Logging.debug "domain" fmt
> let warn  fmt = Logging.warn  "domain" fmt
> 
> +(* An event channel port pair.  The remote port, and the local port it is
> +   bound to. *)
> +type port_pair =
> +{
> +	local: Xeneventchn.t;
> +	remote: int;
> +}
> +
> +(* Sentinal port_pair with both set to EVTCHN_INVALID *)
> +let invalid_ports =
> +{
> +	local = Xeneventchn.of_int 0;
> +	remote = 0
> +}
> +
> +let string_of_port_pair p =
> +	sprintf "(l %d, r %d)" (Xeneventchn.to_int p.local) p.remote
> +
> type t =
> {
> 	id: Xenctrl.domid;
> 	mfn: nativeint;
> 	interface: Xenmmap.mmap_interface;
> 	eventchn: Event.t;
> -	mutable remote_port: int;
> -	mutable port: Xeneventchn.t option;
> +	mutable ports: port_pair;
> 	mutable bad_client: bool;
> 	mutable io_credit: int; (* the rounds of ring process left to do, default is 0,
> 	                           usually set to 1 when there is work detected, could
> @@ -41,8 +58,8 @@ let is_dom0 d = d.id = 0
> let get_id domain = domain.id
> let get_interface d = d.interface
> let get_mfn d = d.mfn
> -let get_remote_port d = d.remote_port
> -let get_port d = d.port
> +let get_remote_port d = d.ports.remote
> +let get_local_port d = d.ports.local
> 
> let is_bad_domain domain = domain.bad_client
> let mark_as_bad domain = domain.bad_client <- true
> @@ -56,54 +73,36 @@ let is_paused_for_conflict dom = dom.conflict_credit <= 0.0
> 
> let is_free_to_conflict = is_dom0
> 
> -let string_of_port = function
> -	| None -> "None"
> -	| Some x -> string_of_int (Xeneventchn.to_int x)
> -
> let dump d chan =
> -	fprintf chan "dom,%d,%nd,%d\n" d.id d.mfn d.remote_port
> +	fprintf chan "dom,%d,%nd,%d\n" d.id d.mfn d.ports.remote
> 
> let rebind_evtchn d remote_port =
> -	begin match d.port with
> -	| None -> ()
> -	| Some p -> Event.unbind d.eventchn p
> -	end;
> +	Event.unbind d.eventchn d.ports.local;
> 	let local = Event.bind_interdomain d.eventchn d.id remote_port in
> -	debug "domain %d rebind (l %s, r %d) => (l %d, r %d)"
> -	      d.id (string_of_port d.port) d.remote_port
> -	      (Xeneventchn.to_int local) remote_port;
> -	d.remote_port <- remote_port;
> -	d.port <- Some (local)
> +	let ports = { local; remote = remote_port } in
> +	debug "domain %d rebind %s => %s"
> +	      d.id (string_of_port_pair d.ports) (string_of_port_pair ports);
> +	d.ports <- ports
> 
> let notify dom =
> -	match dom.port with
> -	| None -> warn "domain %d: attempt to notify on unknown port" dom.id
> -	| Some port -> Event.notify dom.eventchn port
> -
> -let bind_interdomain dom =
> -	begin match dom.port with
> -	| None -> ()
> -	| Some port -> Event.unbind dom.eventchn port
> -	end;
> -	dom.port <- Some (Event.bind_interdomain dom.eventchn dom.id dom.remote_port);
> -	debug "bound domain %d remote port %d to local port %s" dom.id dom.remote_port (string_of_port dom.port)
> -
> +	Event.notify dom.eventchn dom.ports.local
> 
> let close dom =
> -	debug "domain %d unbound port %s" dom.id (string_of_port dom.port);
> -	begin match dom.port with
> -	| None -> ()
> -	| Some port -> Event.unbind dom.eventchn port
> -	end;
> +	debug "domain %d unbind %s" dom.id (string_of_port_pair dom.ports);
> +	Event.unbind dom.eventchn dom.ports.local;
> +	dom.ports <- invalid_ports;
> 	Xenmmap.unmap dom.interface
> 
> -let make id mfn remote_port interface eventchn = {
> +let make id mfn remote_port interface eventchn =
> +	let local = Event.bind_interdomain eventchn id remote_port in
> +	let ports = { local; remote = remote_port } in
> +	debug "domain %d bind %s" id (string_of_port_pair ports);
> +{
> 	id = id;
> 	mfn = mfn;
> -	remote_port = remote_port;
> +	ports;
> 	interface = interface;
> 	eventchn = eventchn;
> -	port = None;
> 	bad_client = false;
> 	io_credit = 0;
> 	conflict_credit = !Define.conflict_burst_limit;
> diff --git a/tools/ocaml/xenstored/domains.ml b/tools/ocaml/xenstored/domains.ml
> index 26018ac0dd3d..2ab0c5f4d8d0 100644
> --- a/tools/ocaml/xenstored/domains.ml
> +++ b/tools/ocaml/xenstored/domains.ml
> @@ -126,7 +126,6 @@ let create doms domid mfn remote_port =
> 	let interface = Xenctrl.map_foreign_range xc domid (Xenmmap.getpagesize()) mfn in
> 	let dom = Domain.make domid mfn remote_port interface doms.eventchn in
> 	Hashtbl.add doms.table domid dom;
> -	Domain.bind_interdomain dom;
> 	dom
> 
> let xenstored_kva = ref ""
> @@ -144,7 +143,6 @@ let create0 doms =
> 
> 	let dom = Domain.make 0 Nativeint.zero remote_port interface doms.eventchn in
> 	Hashtbl.add doms.table 0 dom;
> -	Domain.bind_interdomain dom;
> 	Domain.notify dom;
> 	dom
> 
> -- 
> 2.11.0
> 