On Fri, 31 Aug 2018 10:25:47 +0200 Gaëtan Rivet <gaetan.ri...@6wind.com> wrote:
> Hi Stephen, > > On Thu, Aug 30, 2018 at 03:35:11PM -0700, Stephen Hemminger wrote: > > From: Stephen Hemminger <sthem...@microsoft.com> > > > > Implement callback functionality on link state changes. > > This is not really driven off of interrupt file descriptor like most other > > PMD's. Instead, it happens when a link state change message arrives > > in the common ring buffer. > > > > Does this mean that the lsc event will be processed in a dataplane > thread? Looking at the _rte_eth_dev_callback_process() call, it seems > so. > > Shouldn't this be executed in the context of the eal-intr-thread > instead? This thread is marked control and should be configured with the > proper afinity, unless dataplane threads. > > Maybe I missed something, it's just to double-check that this will > behave nicely with applications relying on the eal-intr-thread afinity. > > -- > Gaëtan Rivet > 6WIND There is no EAL API to take the event and propogate it over to the eal interrupt thread. The interrupt thread is buried in the internals of EAL.
