Hi.. the reason why there is a userspace example is that it was one of the early codes available on the acme site... i made the patch for the kernel driver many months later. that is why there are 2 ways of doing it
you are correct, the in kernel space things go faster and cannot be interrupted that easily unless there is a HW int John On 29.11.2006 at , you wrote: > Hi Geert, > > I understood what acme wrote, but that's why I was wondering that > they are using such a solution instead of the existing device driver > solution. > As far as I know from previous UNIX, not LINUX nor embedded LINUX > drivers is that they were running in kernel mode and thus were not > interrupted during their runtime (at least from user applications). > That's why in my opinion the device driver solution is far more > sophisticated than the userspace solution. Although I haven't looked > at the implementation of it. > > What I can say from my own experiences with the PCF8591 in > conjunction with the 'new' device driver is that it seems that the > driver is working properly. > What I seem to have is a problem with the analog inputs. Their > behaviour is strongly dependent on the connected input. Open or > grounded inputs seem to disturb the normal operation. > > So im still investigating and searching for a permanent solution. > > So long, Klaus > > > > >
