"Pattan, Reshma" <reshma.pattan at intel.com> writes: > Hi, > >> -----Original Message----- >> From: Aaron Conole [mailto:aconole at redhat.com] >> Sent: Friday, June 10, 2016 7:48 PM >> To: Pattan, Reshma <reshma.pattan at intel.com> >> Cc: dev at dpdk.org >> Subject: Re: [dpdk-dev] [PATCH v8 5/8] lib/librte_pdump: add new library for >> packet capturing support >> >> Hi Reshma, >> >> Reshma Pattan <reshma.pattan at intel.com> writes: >> >> > Added new library for packet capturing support. >> > >> > Added public api rte_pdump_init, applications should call this as part >> > of their application setup to have packet capturing framework ready. >> > >> > Added public api rte_pdump_uninit to uninitialize the packet capturing >> > framework. >> > >> > Added public apis rte_pdump_enable and rte_pdump_disable to enable and >> > disable packet capturing on specific port and queue. >> > >> > Added public apis rte_pdump_enable_by_deviceid and >> > rte_pdump_disable_by_deviceid to enable and disable packet capturing >> > on a specific device (pci address or name) and queue. >> > >> > Added public api rte_pdump_set_socket_dir to set the server socket >> > path. >> >> Thanks for this, it is quite useful! I am wondering, should the >> same API work for >> a client socket as well? The code becomes a bit easier to maintain, >> and the API >> behaves whether executed from client or server. >> Thoughts? > > In this patch, server socket path is added as argument to > rte_pdump_init() , so server socket path must be passed while calling > rte_pdump_init() API. > And rte_pdump_set_socket_dir() is added for clients , as client need > to know server socket path for contacting server, so application > should pass server socket path for clients using this API. > > Could you please clarify which of the below option you are looking to have? > a)If you want to have client and server sockets under same non default > path this can be done using same API. This just needs a tiny change in > the code. > > b)But if you want to have aserver and client sockets under different > paths, this can done using either of the below approaches. > b1)use same rte_pdump_set_socket_dir() API, but add a new argument to > specify if the path is for server or client socket.
This is probably the better option. I think it would result in the least surprise to an end developer, anyway. Thanks, Aaron > (or) > b2)have two separate APIs to set client and server socket paths. > > Which one do you prefer? > > Konstantin, any comments from your side, please add. > > Thanks, > Reshma > >> >> Thanks, >> Aaron >>