From: Nik M. <add...@gm...> - 2006-03-09 22:29:47
|
This may end up as a double post on the archives, due to sending from the wrong email account. On 3/9/06, Nik Martin <nm...@ra...> wrote: > On 3/9/06, Dave Hylands <dhy...@gm...> wrote: > > Hi Nik, > > > > On 3/9/06, Nik Martin <add...@gm...> wrote: > > > My problem was the KERNELDIR version !=3D 2.6.15 > > > Nik > > > > > > Also, It's building the proper dir in /proc, but not in /dev. adding > > > debug statements now to try to figure out why > > > > You need to manually create the entry in /dev (that's the udev support > > stuff I was talking about). If you want to make the changes yourself, > > then take a look at the differences between the latest and previous > > versions of robostix_drv.c: > > I just re-read and noticed the /udev reference. I'm new to kernel hacking= . > > > <http://websvn.gumstix.com/diff.php?repname=3DBuildroot&path=3D%2Fbranc= hes%2Fprojects%2Frobostix%2Fgumstix%2Frobostix_drv.c&rev=3D0&sc=3D0> > > > > If memory serves me correctly, the name passed into > > class_device_create will be the name of the /dev entry created. > > > > > Also, since the main app is still just looping looking for events, I'= m > > > not sure what I'm gaining in regards to performance. I'm assured to > > > not miss an event but the event will be caught at the speed of the > > > main loop. I need to a fairly responsive 5 button keypad UI, so I > > > need to figure out how to fire software interrupts in the main app > > > based on interrups in the char driver, to eliminate a time critical > > > main loop. > > > > The main app should just be issuing a non-blocking read call. The read > > will return when a button is pressed. If you need to do other stuff as > > well, then launch a thread which does the blocking and have it signal > > the main thread. > > > > It's also possible to implement support for the select call, but that > > isn't currently in the sample driver (although it's pretty easy to > > add). > > > > What all does you app do? > I have to read a flow meter which outputs a square wave whose frequency is relative to the rate of flow. It's pretty slow, like 28-30 Hz, but the accuracy is important. This just needs to happen in the background, and I reference the reading every few seconds to do some calculations. The other IO is a just a set of buttons that act as a keyboard to an ncurses UI we're writing for those 2.7" sony LCD's from Goldmine Electronics and Earth LCD. Another process ( or maybe just a thread in my main app) will be running a neural network, training on the data the main app is collecting, and outputting a set of predictive control statements that will be controlling varion loads in a home. Its a DOE funded project, but I'm hoping to get some commercial appeal out of the nifty board we're building to support the project with. > > Nik > > > > > -- > > Dave Hylands > > Vancouver, BC, Canada > > http://www.DaveHylands.com/ > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by xPML, a groundbreaking scripting lang= uage > > that extends applications into web and mobile media. Attend the live we= bcast > > and join the prime developer group breaking into this new coding territ= ory! > > http://sel.as-us.falkag.net/sel?cmdlnk&kid=110944&bid$1720&dat=121642 > > _______________________________________________ > > gumstix-users mailing list > > gum...@li... > > https://lists.sourceforge.net/lists/listinfo/gumstix-users > > > |