On Monday, September 23, 2013 9:23 AM, Ian Abbott wrote: > On 2013-09-21 00:33, H Hartley Sweeten wrote: >> Tidy up this driver a bit. >> >> Unfortunately not a great reduction in the lines of code but it did >> uncover a couple bugs in the 8112 board and async command support. >> >> The async command support still needs to be looked over. >> 1) The analog input sample read in the interrupt routine is not >> returned to the user. >> 2) The scan_end_src and/or stop_src appears to be broken. As noted >> in the interrupt routine, nothing sets 'ntrig' which is used to >> cause the COMEDI_CB_EOA event. [snip] > Looks good. > > I can take a look at the async command support once this series has been > committed if you like. Basically, comedi_buf_put() can be used in the > interrupt handler to write the sample to the comedi buffer. If it > returns 0 (unsuccessful), the interrupt routine should set the > COMEDI_CB_OVERFLOW event flag (it can also set COMEDI_CB_ERROR, but that > is already set by comedi_buf_put()). If comedi_buf_put() returns 1 > (successful), then since the "scan length" is always 1 (a single sample) > in this driver, it can set the COMEDI_CB_EOS event flag. > pcl711_ai_cmd() should set devpriv->ntrig to the number of scans (or the > number of conversions) if cmd->stop_src == TRIG_COUNT. The interrupt > handler should only decrement devpriv->ntrig if s->async->cmd.stop_src > == TRIG_COUNT. (Also, devpriv->ntrig should be an unsigned int.) > > Reviewed-by: Ian Abbott <abbotti@xxxxxxxxx> Thanks! I'm hoping this series, and the adl_pci6208 cleanup, get applied by Greg before Stephen Rothwell freezes linux-next on Friday. Once this is in if you can fix up the command support that would be great! Regards, Hartley _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
