On Thu, Aug 26, 2021 at 12:08 PM Alexei Starovoitov <alexei.starovoitov@xxxxxxxxx> wrote: > > On Thu, Aug 26, 2021 at 11:18 AM sunyucong@xxxxxxxxx > <sunyucong@xxxxxxxxx> wrote: > > > > Reporting back: I tried a select() based approach, (as attached below) > > but unfortunately it doesn't seem to work. During testing, I am > > always getting full timeout errors as the socket never seems to become > > ready to read(). My guess is that this has something to do with the > > sockets being created through sockpair() , but I am unable to confirm. > > > > On the other hand, the previous patch approach works perfectly fine, I > > would still like to request to apply that instead. > > Ok. Applied your earlier patch, but it's a short term workaround. > select() should work for af_unix. > I suspect something got broken with the redirect. > Cong, Jiang, > could you please take a look ? Like I mentioned before, I suspect there is some delay in one of the queues on the way or there is a worker wakeup latency. I will try adding some tracepoints to see if I can capture it. Thanks.
