I've been trying to run a real-time process to feed a D/A chip using the spidev driver but am finding the call to /dev/spidevX.Y is responsible for "random" very long latencies that do not occur if I call the "driver" to toggle a GPIO pin instead.
My most recent attempts have been with Robert Nelson's 3.9-rc6 kernel for the Pandaboard ES and he has suggested I contact the OMAP and SPI maintainers, hence I am sending this message. Details are here: http://eewiki.net/display/linuxonarm/PandaBoard_Comments?focusedCommentId=13599653&#comment-13599653 A sample test program that will illustrate the issue can be downloaded here: http://eewiki.net/download/attachments/12059008/swave_spidev.c?version=1&modificationDate=1365778735227&api=v2 Robert has run this test code (after changing the GPIO and spidev name strings for the Beagleboard) and has found the xM is much closer to usable, but what is curious is that the worst case and number of "missed samples" are made worse when running as root. Hoping you have a solution. --wally. ------------------------------------------------------------------------------ Try New Relic Now & We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, & servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr _______________________________________________ spi-devel-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/spi-devel-general
