[dpdk-dev] [PATCH v6 04/10] vhost: add two new messages to support a shared buffer

[Yu, Jin]

> -----Original Message-----
> From: Maxime Coquelin [mailto:maxime.coquelin at redhat.com]
> Sent: Saturday, September 7, 2019 1:23 AM
> To: Yu, Jin <jin.yu at intel.com>; dev at dpdk.org
> Cc: Liu, Changpeng <changpeng.liu at intel.com>; Bie, Tiwei
> <tiwei.bie at intel.com>; Wang, Zhihong <zhihong.wang at intel.com>; Lin Li
> <lilin24 at baidu.com>; Xun Ni <nixun at baidu.com>; Yu Zhang
> <zhangyu31 at baidu.com>
> Subject: Re: [PATCH v6 04/10] vhost: add two new messages to support a
> shared buffer
> 
> 
> 
> On 8/29/19 4:12 PM, JinYu wrote:
> > This patch introduces two new messages VHOST_USER_GET_INFLIGHT_FD
> and
> > VHOST_USER_SET_INFLIGHT_FD to support transferring a shared buffer
> > between qemu and backend.
> >
> > Signed-off-by: Lin Li <lilin24 at baidu.com>
> > Signed-off-by: Xun Ni <nixun at baidu.com>
> > Signed-off-by: Yu Zhang <zhangyu31 at baidu.com>
> > Signed-off-by: Jin Yu <jin.yu at intel.com>
> > ---
> >  lib/librte_vhost/vhost.h      |   7 +
> >  lib/librte_vhost/vhost_user.c | 242
> ++++++++++++++++++++++++++++++++++
> >  lib/librte_vhost/vhost_user.h |   4 +-
> >  3 files changed, 252 insertions(+), 1 deletion(-)
> >
> > diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h index
> > 884befa85..39f645b97 100644
> > --- a/lib/librte_vhost/vhost.h
> > +++ b/lib/librte_vhost/vhost.h
> > @@ -286,6 +286,12 @@ struct guest_page {
> >  	uint64_t size;
> >  };
> >
> > +struct inflight_mem_info {
> > +	int		fd;
> > +	void		*addr;
> > +	uint64_t	size;
> > +};
> > +
> >  /**
> >   * Device structure contains all configuration information relating
> >   * to the device.
> > @@ -303,6 +309,7 @@ struct virtio_net {
> >  	uint32_t		nr_vring;
> >  	int			dequeue_zero_copy;
> >  	struct vhost_virtqueue	*virtqueue[VHOST_MAX_QUEUE_PAIRS * 2];
> > +	struct inflight_mem_info inflight_info;
> 
> Could it be a pointer and allocated only if the feature has been negotiated?
Ok, will fix it.
> 
> >  #define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)
> >  	char			ifname[IF_NAME_SZ];
> >  	uint64_t		log_size;
> > diff --git a/lib/librte_vhost/vhost_user.c
> > b/lib/librte_vhost/vhost_user.c index c9e29ece8..647929234 100644
> > --- a/lib/librte_vhost/vhost_user.c
> > +++ b/lib/librte_vhost/vhost_user.c
> > @@ -49,6 +49,15 @@
> >  #define VIRTIO_MIN_MTU 68
> >  #define VIRTIO_MAX_MTU 65535
> >
> > +#define INFLIGHT_ALIGNMENT	64
> > +#define INFLIGHT_VERSION	0xabcd
> > +#define VIRTQUEUE_MAX_SIZE	1024
> > +
> > +#define CLOEXEC		0x0001U
> > +
> > +#define ALIGN_DOWN(n, m) ((n) / (m) * (m)) #define ALIGN_UP(n, m)
> > +ALIGN_DOWN((n) + (m) - 1, (m))
> > +
> >  static const char *vhost_message_str[VHOST_USER_MAX] = {
> >  	[VHOST_USER_NONE] = "VHOST_USER_NONE",
> >  	[VHOST_USER_GET_FEATURES] = "VHOST_USER_GET_FEATURES",
> @@ -78,6
> > +87,8 @@ static const char *vhost_message_str[VHOST_USER_MAX] = {
> >  	[VHOST_USER_POSTCOPY_ADVISE]  =
> "VHOST_USER_POSTCOPY_ADVISE",
> >  	[VHOST_USER_POSTCOPY_LISTEN]  =
> "VHOST_USER_POSTCOPY_LISTEN",
> >  	[VHOST_USER_POSTCOPY_END]  = "VHOST_USER_POSTCOPY_END",
> > +	[VHOST_USER_GET_INFLIGHT_FD] =
> "VHOST_USER_GET_INFLIGHT_FD",
> > +	[VHOST_USER_SET_INFLIGHT_FD] =
> "VHOST_USER_SET_INFLIGHT_FD",
> >  };
> >
> >  static int send_vhost_reply(int sockfd, struct VhostUserMsg *msg); @@
> > -160,6 +171,16 @@ vhost_backend_cleanup(struct virtio_net *dev)
> >  		dev->log_addr = 0;
> >  	}
> >
> > +	if (dev->inflight_info.addr) {
> > +		munmap(dev->inflight_info.addr, dev->inflight_info.size);
> > +		dev->inflight_info.addr = NULL;
> > +	}
> > +
> > +	if (dev->inflight_info.fd > 0) {
> > +		close(dev->inflight_info.fd);
> > +		dev->inflight_info.fd = -1;
> > +	}
> > +
> >  	if (dev->slave_req_fd >= 0) {
> >  		close(dev->slave_req_fd);
> >  		dev->slave_req_fd = -1;
> > @@ -1165,6 +1186,225 @@ virtio_is_ready(struct virtio_net *dev)
> >  	return 1;
> >  }
> >
> > +static int
> > +mem_create(const char *name, unsigned int flags) { #ifdef
> > +__NR_memfd_create
> > +	return syscall(__NR_memfd_create, name, flags); #else
> > +	return -1;
> > +#endif
> > +}
> 
> Maybe you could use memfd_create, as done in:
> lib/librte_eal/linux/eal/eal_memalloc.c
> 
Ok, I will see.