On Wed, Feb 13, 2019 at 3:08 AM Tony Chuang <yhchuang@xxxxxxxxxxx> wrote: > > On Behalf Of Tony Chuang > > > From: Brian Norris [mailto:briannorris@xxxxxxxxxxxx] > > > Do you have some kind of memory mapping/ordering issue or something? I > > > wouldn't think you should expect to just drop packets on the floor so > > > often like this. > > > > > > > Looks like the cause is that the DMA might not have done when the interrupt > > arrived. Need to do more test to figure it out. > > > I tested and checked again. I found that I didn't enable DMA sync for > the TRX path. After some work it can be resolved if I turn DMA sync on. > So I will include this in the PATCH v5. Great! I'd also highly recommend dropping that loop. It just pretends to do something that it doesn't really do. Leaving a single iteration as a sort of debugging check is OK, if it's truly a bug any time we fail it. Regards, Brian
