arm11 code is broken in 2046->2047 ref: https://lists.berlios.de/pipermail/openocd-development/2009-September/010644.html
I've looked at this and it is a pretty hellish problem to figure out from a code inspection point of view. Anyone taking the time to look at 2046 in detail will conclude that it is blind luck that 2046 worked and that there is no way of figuring out what it did really did in all cases by code inspection.... For your own sanity I don't recommend trying to decipher 2046 vs. 2047... Short story: it involves a global variable for the end state in convoluted ways... Oh... and I committed 2047... :-| Moving on.... Here are a couple of things that need to be sorted out for it to be humanly possible to figure out what the arm11 code does and expect: - "tms_sequence long" will pass through TAP_DRPAUSE going from TAP_DRSHIFT to TAP_IDLE. - "tms_sequence short" will *skip* TAP_DRPAUSE going from TAP_DRSHIFT to TAP_IDLE. I've got a request out to test arm11 w/"tms_sequence long". If "tms_sequence long" works w/svn head, then we must fix the arm11 code to have much more explict handling of going passing through any required code. Perhaps everything that is needed is for the arm11 code to go through TAP_DRPAUSE when moving from TAP_DRSHIFT to TAP_IDLE? If the problem does not turn out to be "tms_sequence long/short", then my next preferred step would be to try if getting rid of all ARM11_TAP_DEFAULT, does the trick. However, here I'm out of my depth, because the code offers no solid leads on what the expected end state really is... -- Øyvind Harboe Embedded software and hardware consulting services http://www.zylin.com _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
