On Wed, Feb 25, 2009 at 5:42 PM, Laurent Gauch <laurent.ga...@amontec.com> wrote: >> >> >/ Maybe the check should be put back anyway, even when we know that >> />/ it takes 7 or less steps in case someone changes the code later. >> />/ >> />/ >> /No doubt that check is needed, or >> >> a function rewrite which changes tms to bit a bit vector, and then >> "bits" (which BTW should be called bitcount to make it clear this >> parameter is not the bit vector) can be any number and then the >> function could be used to send the tap machine on an arbitrary nearly >> endless journey, including looping. >> >> As written, it is a mine field. >> >> Since I am a dummy, I have experience with "APIs for Dummys". >> >> Dick >> > > Actually, you have all functions in OpenOCD JTAG layer to do any TAP > State transition. > > Just adjust _end_state before sending new _scan. > > end_state should only be STABLE states as : IDLE RESET DRPAUSE IRPAUSE. > > DO NOT add any _tms functions. Why ? Because it is a bit dangerous to > let user calling the JTAG HAL with their own _tms. > > If you let users going in UNSTABLE states, we will have a lot of new > troubles in OpenOCD. > > > As I tell you before, there are enough functions for passing from DRSCAN > to a new DRSCAN without passing in IDLE state: > > > fix _end_state to DRPAUSE > > add your _scan > > add _state_move to DRPAUSE > > fix _end_state to IDLE > > add your new _scan > > Before adding new functions to JTAG HAL, please make sure to understand > the actual ones !
Code show us code ! Greetings _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
