Hi Maxime, > -----Original Message----- > From: Maxime Coquelin <[email protected]> > Sent: Thursday, October 13, 2022 4:02 PM > To: Ding, Xuan <[email protected]>; Xia, Chenbo <[email protected]> > Cc: [email protected]; Hu, Jiayu <[email protected]>; He, Xingguang > <[email protected]>; Yang, YvonneX <[email protected]>; > Jiang, Cheng1 <[email protected]>; Wang, YuanX > <[email protected]>; Ma, WenwuX <[email protected]> > Subject: Re: [PATCH v4 1/2] vhost: introduce DMA vchannel unconfiguration > > > > On 10/13/22 08:40, [email protected] wrote: > > From: Xuan Ding <[email protected]> > > > > Add a new API rte_vhost_async_dma_unconfigure() to unconfigure DMA > > vchannels in vhost async data path. Lock protection are also added to > > protect DMA vchannels configuration and unconfiguration from > > concurrent calls. > > > > Signed-off-by: Xuan Ding <[email protected]> > > --- > > doc/guides/prog_guide/vhost_lib.rst | 6 +++ > > doc/guides/rel_notes/release_22_11.rst | 4 ++ > > lib/vhost/rte_vhost_async.h | 18 +++++++ > > lib/vhost/version.map | 3 ++ > > lib/vhost/vhost.c | 69 ++++++++++++++++++++++++-- > > 5 files changed, 95 insertions(+), 5 deletions(-) > > > > diff --git a/doc/guides/prog_guide/vhost_lib.rst > > b/doc/guides/prog_guide/vhost_lib.rst > > index bad4d819e1..d3cef978d0 100644 > > --- a/doc/guides/prog_guide/vhost_lib.rst > > +++ b/doc/guides/prog_guide/vhost_lib.rst > > @@ -323,6 +323,12 @@ The following is an overview of some key Vhost > API functions: > > Get device type of vDPA device, such as VDPA_DEVICE_TYPE_NET, > > VDPA_DEVICE_TYPE_BLK. > > > > +* ``rte_vhost_async_dma_unconfigure(dma_id, vchan_id)`` > > + > > + Clean DMA vChannels finished to use. This function needs to be > > + called after the deregistration of async DMA vchannel has been > > + finished. > > + > > Vhost-user Implementations > > -------------------------- > > > > diff --git a/doc/guides/rel_notes/release_22_11.rst > > b/doc/guides/rel_notes/release_22_11.rst > > index 2da8bc9661..3be150122c 100644 > > --- a/doc/guides/rel_notes/release_22_11.rst > > +++ b/doc/guides/rel_notes/release_22_11.rst > > @@ -236,6 +236,10 @@ New Features > > > > strings $dpdk_binary_or_driver | sed -n 's/^PMD_INFO_STRING= //p' > > > > +* **Added DMA vChannel unconfiguration for async vhost.** > > + > > + * Added support to unconfigure DMA vChannels that have been > unregistered. > > + > > > > Removed Items > > ------------- > > diff --git a/lib/vhost/rte_vhost_async.h b/lib/vhost/rte_vhost_async.h > > index 1db2a10124..6ee4f7258d 100644 > > --- a/lib/vhost/rte_vhost_async.h > > +++ b/lib/vhost/rte_vhost_async.h > > @@ -266,6 +266,24 @@ rte_vhost_async_try_dequeue_burst(int vid, > uint16_t queue_id, > > struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t > count, > > int *nr_inflight, int16_t dma_id, uint16_t vchan_id); > > > > +/** > > + * @warning > > + * @b EXPERIMENTAL: this API may change, or be removed, without prior > notice. > > + * > > + * Unconfigure DMA vChannels in asynchronous data path. > > + * This function should be called after the DMA vChannel has been > unregistered. > > + * > > + * @param dma_id > > + * the identifier of DMA device > > + * @param vchan_id > > + * the identifier of virtual DMA channel > > + * @return > > + * 0 on success, and -1 on failure > > + */ > > +__rte_experimental > > +int > > +rte_vhost_async_dma_unconfigure(int16_t dma_id, uint16_t vchan_id); > > + > > #ifdef __cplusplus > > } > > #endif > > diff --git a/lib/vhost/version.map b/lib/vhost/version.map index > > 7a00b65740..0b61870870 100644 > > --- a/lib/vhost/version.map > > +++ b/lib/vhost/version.map > > @@ -94,6 +94,9 @@ EXPERIMENTAL { > > rte_vhost_async_try_dequeue_burst; > > rte_vhost_driver_get_vdpa_dev_type; > > rte_vhost_clear_queue; > > + > > + # added in 22.11 > > + rte_vhost_async_dma_unconfigure; > > }; > > > > INTERNAL { > > diff --git a/lib/vhost/vhost.c b/lib/vhost/vhost.c index > > 8740aa2788..9fbc56229a 100644 > > --- a/lib/vhost/vhost.c > > +++ b/lib/vhost/vhost.c > > @@ -23,6 +23,7 @@ > > > > struct virtio_net *vhost_devices[RTE_MAX_VHOST_DEVICE]; > > pthread_mutex_t vhost_dev_lock = PTHREAD_MUTEX_INITIALIZER; > > +static rte_spinlock_t vhost_dma_lock = RTE_SPINLOCK_INITIALIZER; > > I think a mutex would be more appropiate.
Thanks for your suggestion. Maybe mutex is more efficient, I will switch to mutex in next version. Regards, Xuan > > Thanks, > Maxime
