[dpdk-dev] [PATCH 1/1] net/af_xdp: shared UMEM support

Tahhan, Maryam maryam.tahhan at intel.com
Thu Sep 10 13:55:51 CEST 2020



> -----Original Message-----
> From: Loftus, Ciara <ciara.loftus at intel.com>
> Sent: Monday 7 September 2020 17:16
> To: dev at dpdk.org
> Cc: Loftus, Ciara <ciara.loftus at intel.com>
> Subject: [PATCH 1/1] net/af_xdp: shared UMEM support
> 
> Kernel v5.10 will introduce the ability to efficiently share a UMEM between
> AF_XDP sockets bound to different queue ids on the same or different
> devices. This patch integrates that functionality into the AF_XDP PMD.
> 
> A PMD will attempt to share a UMEM with others if the shared_umem=1
> vdev arg is set. UMEMs can only be shared across PMDs with the same
> mempool, up to a limited number of PMDs goverened by the size of the
> given mempool.
> Sharing UMEMs is not supported for non-zero-copy (aligned) mode.
> 
> The benefit of sharing UMEM across PMDs is a saving in memory due to not
> having to register the UMEM multiple times. Throughput was measured to
> remain within 2% of the default mode (not sharing UMEM).
> 
> A version of libbpf >= v0.2.0 is required and the appropriate pkg-config file
> for libbpf must be installed such that meson can determine the version.
> 
> Signed-off-by: Ciara Loftus <ciara.loftus at intel.com>

<snip>

> 
> +/* List which tracks PMDs to facilitate sharing UMEMs across them. */
> +struct internal_list {
> +	TAILQ_ENTRY(internal_list) next;
> +	struct rte_eth_dev *eth_dev;
> +};
> +
> +TAILQ_HEAD(internal_list_head, internal_list); static struct
> +internal_list_head internal_list =
> +	TAILQ_HEAD_INITIALIZER(internal_list);
> +
> +static pthread_mutex_t internal_list_lock = PTHREAD_MUTEX_INITIALIZER;

[Tahhan, Maryam] do multiple threads typically initialize and ethdev/invoke the underlying driver? 
Most apps I've seen initialize the ports one after the other in the starting thread - so if there's not multiple threads doing initialization - we may want to consider removing this mutex...
Or maybe do you see something potentially removing a port while a port is being added?

<snip>


More information about the dev mailing list