On Tue, Aug 16, 2005 at 01:39:40PM +0100, Andrew de Quincey wrote: > On Tuesday 16 Aug 2005 13:27, Jeremy Hall wrote: > > Hello, > > > > What I wanted is: > > > > 1: number of blocks corrected after FEC > > > > 2: number of blocks uncorrectable > > > > I'm guessing I can't get that with either of these numbers. > > Yup. Trying to think of a way round the problem. > > Johannes suggested having the frontend thread poll the values regularly (e.g. > once a second), and cache the values internally. Then the IOCTLs called by > the user application return those internal cached values. > > Johannes, why is blocking in the IOCTLs to monitor the BER etc bad? Is it > because the API is defined not to? Well, it is expected that it blocks for the duration of an i2c transfer, but not for arbitrary time. Ideally it shouldn't block at all. Johannes
