Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752233AbdHMAm2 (ORCPT ); Sat, 12 Aug 2017 20:42:28 -0400 Received: from aserp1040.oracle.com ([141.146.126.69]:34117 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752149AbdHMAm1 (ORCPT ); Sat, 12 Aug 2017 20:42:27 -0400 Subject: Re: [PATCH v3 04/13] xen/pvcalls: implement socket command and handle events To: Stefano Stabellini , xen-devel@lists.xen.org Cc: linux-kernel@vger.kernel.org, jgross@suse.com, Stefano Stabellini References: <1501541855-7354-1-git-send-email-sstabellini@kernel.org> <1501541855-7354-4-git-send-email-sstabellini@kernel.org> From: Boris Ostrovsky Message-ID: Date: Sat, 12 Aug 2017 20:42:05 -0400 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.2.1 MIME-Version: 1.0 In-Reply-To: <1501541855-7354-4-git-send-email-sstabellini@kernel.org> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit X-Source-IP: aserv0021.oracle.com [141.146.126.233] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 7021 Lines: 220 On 07/31/2017 06:57 PM, Stefano Stabellini wrote: > Send a PVCALLS_SOCKET command to the backend, use the masked > req_prod_pvt as req_id. This way, req_id is guaranteed to be between 0 > and PVCALLS_NR_REQ_PER_RING. We already have a slot in the rsp array > ready for the response, and there cannot be two outstanding responses > with the same req_id. > > Wait for the response by waiting on the inflight_req waitqueue and > check for the req_id field in rsp[req_id]. Use atomic accesses and > barriers to read the field. Note that the barriers are simple smp > barriers (as opposed to virt barriers) because they are for internal > frontend synchronization, not frontend<->backend communication. > > Once a response is received, clear the corresponding rsp slot by setting > req_id to PVCALLS_INVALID_ID. Note that PVCALLS_INVALID_ID is invalid > only from the frontend point of view. It is not part of the PVCalls > protocol. > > pvcalls_front_event_handler is in charge of copying responses from the > ring to the appropriate rsp slot. It is done by copying the body of the > response first, then by copying req_id atomically. After the copies, > wake up anybody waiting on waitqueue. > > pvcallss_lock protects accesses to the ring. > > Create a new struct sock_mapping and convert the pointer into an > uint64_t and use it as id for the new socket to pass to the backend. The > struct will be fully initialized later on connect or bind. In this patch > the struct sock_mapping is empty, the fields will be added by the next > patch. > > sock->sk->sk_send_head is not used for ip sockets: reuse the field to > store a pointer to the struct sock_mapping corresponding to the socket. > This way, we can easily get the struct sock_mapping from the struct > socket. > > Signed-off-by: Stefano Stabellini > CC: boris.ostrovsky@oracle.com > CC: jgross@suse.com > --- > drivers/xen/pvcalls-front.c | 119 ++++++++++++++++++++++++++++++++++++++++++++ > drivers/xen/pvcalls-front.h | 8 +++ > 2 files changed, 127 insertions(+) > create mode 100644 drivers/xen/pvcalls-front.h > > diff --git a/drivers/xen/pvcalls-front.c b/drivers/xen/pvcalls-front.c > index 2afe36d..7c4a7cb 100644 > --- a/drivers/xen/pvcalls-front.c > +++ b/drivers/xen/pvcalls-front.c > @@ -20,6 +20,8 @@ > #include > #include > > +#include "pvcalls-front.h" > + > #define PVCALLS_INVALID_ID UINT_MAX > #define PVCALLS_RING_ORDER XENBUS_MAX_RING_GRANT_ORDER > #define PVCALLS_NR_REQ_PER_RING __CONST_RING_SIZE(xen_pvcalls, XEN_PAGE_SIZE) > @@ -38,11 +40,128 @@ struct pvcalls_bedata { > }; > static struct xenbus_device *pvcalls_front_dev; > > +struct sock_mapping { > + bool active_socket; > + struct list_head list; > + struct socket *sock; > +}; > + > +static inline int get_request(struct pvcalls_bedata *bedata, int *req_id) > +{ > + *req_id = bedata->ring.req_prod_pvt & (RING_SIZE(&bedata->ring) - 1); > + if (RING_FULL(&bedata->ring) || > + READ_ONCE(bedata->rsp[*req_id].req_id) != PVCALLS_INVALID_ID) > + return -EAGAIN; > + return 0; > +} > + > static irqreturn_t pvcalls_front_event_handler(int irq, void *dev_id) > { > + struct xenbus_device *dev = dev_id; > + struct pvcalls_bedata *bedata; > + struct xen_pvcalls_response *rsp; > + uint8_t *src, *dst; > + int req_id = 0, more = 0, done = 0; > + > + if (dev == NULL) > + return IRQ_HANDLED; > + > + bedata = dev_get_drvdata(&dev->dev); > + if (bedata == NULL) > + return IRQ_HANDLED; > + > +again: > + while (RING_HAS_UNCONSUMED_RESPONSES(&bedata->ring)) { > + rsp = RING_GET_RESPONSE(&bedata->ring, bedata->ring.rsp_cons); > + > + req_id = rsp->req_id; > + dst = (uint8_t *)&bedata->rsp[req_id] + sizeof(rsp->req_id); > + src = (uint8_t *)rsp + sizeof(rsp->req_id); > + memcpy(dst, src, sizeof(*rsp) - sizeof(rsp->req_id)); > + /* > + * First copy the rest of the data, then req_id. It is > + * paired with the barrier when accessing bedata->rsp. > + */ > + smp_wmb(); > + WRITE_ONCE(bedata->rsp[req_id].req_id, rsp->req_id); > + > + done = 1; > + bedata->ring.rsp_cons++; > + } > + > + RING_FINAL_CHECK_FOR_RESPONSES(&bedata->ring, more); > + if (more) > + goto again; > + if (done) > + wake_up(&bedata->inflight_req); > return IRQ_HANDLED; > } > > +int pvcalls_front_socket(struct socket *sock) > +{ > + struct pvcalls_bedata *bedata; > + struct sock_mapping *map = NULL; > + struct xen_pvcalls_request *req; > + int notify, req_id, ret; > + > + if (!pvcalls_front_dev) > + return -EACCES; > + /* > + * PVCalls only supports domain AF_INET, > + * type SOCK_STREAM and protocol 0 sockets for now. > + * > + * Check socket type here, AF_INET and protocol checks are done > + * by the caller. > + */ > + if (sock->type != SOCK_STREAM) > + return -ENOTSUPP; > + > + bedata = dev_get_drvdata(&pvcalls_front_dev->dev); > + > + map = kzalloc(sizeof(*map), GFP_KERNEL); > + if (map == NULL) > + return -ENOMEM; > + /* > + * sock->sk->sk_send_head is not used for ip sockets: reuse the > + * field to store a pointer to the struct sock_mapping > + * corresponding to the socket. This way, we can easily get the > + * struct sock_mapping from the struct socket. Is this a safe assumption? Is it conceivable that at some point in the future sk_send_head might become used? > + */ > + WRITE_ONCE(sock->sk->sk_send_head, (void *)map); > + > + spin_lock(&bedata->pvcallss_lock); > + list_add_tail(&map->list, &bedata->socket_mappings); > + > + ret = get_request(bedata, &req_id); > + if (ret < 0) { > + spin_unlock(&bedata->pvcallss_lock); > + return ret; What is happening to the mapping you've linked above? Should it be unlinked and freed? (or defer its creation until we've got a request slot?) > + } > + req = RING_GET_REQUEST(&bedata->ring, req_id); > + req->req_id = req_id; > + req->cmd = PVCALLS_SOCKET; > + req->u.socket.id = (uint64_t) map; > + req->u.socket.domain = AF_INET; > + req->u.socket.type = SOCK_STREAM; > + req->u.socket.protocol = 0; IPPROTO_IP? -boris > + > + bedata->ring.req_prod_pvt++; > + RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&bedata->ring, notify); > + spin_unlock(&bedata->pvcallss_lock); > + if (notify) > + notify_remote_via_irq(bedata->irq); > + > + wait_event(bedata->inflight_req, > + READ_ONCE(bedata->rsp[req_id].req_id) == req_id); > + > + ret = bedata->rsp[req_id].ret; > + /* read ret, then set this rsp slot to be reused */ > + smp_mb(); > + WRITE_ONCE(bedata->rsp[req_id].req_id, PVCALLS_INVALID_ID); > + > + return ret; > +} > + > static const struct xenbus_device_id pvcalls_front_ids[] = { > { "pvcalls" }, > { "" } > diff --git a/drivers/xen/pvcalls-front.h b/drivers/xen/pvcalls-front.h > new file mode 100644 > index 0000000..b7dabed > --- /dev/null > +++ b/drivers/xen/pvcalls-front.h > @@ -0,0 +1,8 @@ > +#ifndef __PVCALLS_FRONT_H__ > +#define __PVCALLS_FRONT_H__ > + > +#include > + > +int pvcalls_front_socket(struct socket *sock); > + > +#endif >