On 1/23/2016 1:38 AM, Richardson, Bruce wrote: > On Thu, Jan 21, 2016 at 09:02:41AM +0000, Van Haaren, Harry wrote: >>> From: Qiu, Michael >>> Sent: Thursday, January 21, 2016 6:14 AM >>> To: Van Haaren, Harry <harry.van.haaren at intel.com>; david.marchand at >>> 6wind.com >>> Cc: dev at dpdk.org >>> Subject: Re: [dpdk-dev] [PATCH] eal: add function to check if primary proc >>> alive >>> <snip> >>> As we could start up many primaries, how does your secondary process >>> work with them? >> When a primary process initializes, the location of the config file is >> important. The default is /var/run/.rte_config >> >> To run multiple primary processes, the --file-prefix= option is used to >> specific a custom location for the config file. Eg: --file-prefix=testing >> /var/run/.testing_config >> >> The rte_eal_check_primary_alive(const char*) function takes a char* >> parameter - this is the location of the config file that the secondary >> process will wait for. Setting it to the correct value will make this >> secondary process wait for the corresponding primary process. >> >> Regards, -Harry > Since a given secondary process only works with a single primary process, I'm > not > sure why the user should want or need to pass in this parameter. What's the > use > case for a secondary process wanting to know about a different primary > process? > The details of what the config file is should largely be hidden from the user > IMHO.
So using the prefix, and get the file name inside the API(--file-prefix=xxx then the config file /var/run/.xxx_config), if no perfix, then could be /var/run/.rte_config. Just a suggestion. Maybe there are better solutions . Thanks, Michael > If you want to allow a secondary to query an arbitrary primary process can you > still allow a NULL string to query the default primary based on the passed in > file-prefix parameter (if any)? > > /Bruce >
