> > > > Regarding 5.7.12.11 in Unipro v1.8, PA rejects sebsequent requests > > > > following the first request from upper layer or remote. > > > > In this situation, PA responds w/ BUSY in cases when they come > > > > from upper layer and does nothing for the requests. So HCI doesn't > > > > receive ind, a.k.a. indicator for its requests and an interrupt, > > > > IS.UPMS isn't generated. > > > > > > > > When LINERESET occurs, the error handler issues PMC which is > > > > recognized as a request for PA. If a host's PA gets or raises > > > > LINERESET, and a request for PMC, this could be a concurrent > > > > situation mentioned above. And I found that situation w/ my > > environment. > > > Can you please elaborate on how this concurrency can happen? > > > My understanding is that both line reset indication and uic command > > > are protected by host_lock? > > > > Yes and one thing I have to correct on the clause: 5.7.12.11 -> > > 5.7.12.1.1 > > > > And I’m talking about the situation w/ some requests w/ PACP. > OK. Thanks. > > However, Please note that Clause 5.7.12.1.1 "Concurrency Resolution" of > the unipro spec, is dealing with local-peer concurrency. Not 2 concurrent > local requests. > As such, I think you need to explain that this is not a host issue. > > Thanks, > Avri I talked with the experts on this again. They said the situation also includes 'two local requests case' because they see 'a local request' mentioned in the spec could be also PA_INIT.req form the local's upper layer. -- "A local request shall be rejected when the PA Layer is processing a local request or a peer request 2679 from the peer Device." Could you explain more why you don't think it's the case? Thanks. Kiwoong Kim
