On Wed, Oct 21, 2009 at 10:58 AM, David Brownell <[email protected]> wrote: > On Wednesday 21 October 2009, Řyvind Harboe wrote: >> >> Are there any shortest paths that are ambiguous? >> > >> > I don't understand what you mean by "ambiguous"; >> > at any TCK edge, TMS will have only one value so >> > only one state will be entered. >> >> There are many ways to get from A to B. The "shortest" >> from A to B is, presumably unambiguous so one can >> then use endstate + pathmove to get a predictable behavior. > > So you're asking if there are cases where there > are two or more paths A --> B of the same length, > where there is no shorter path?
Yes. This is an assumption that it's easy to make when writing some target support.... Look at the arm11 code, there are some such assumptions in there somewhere but they are not documented or known. > Don't think there could be. But again, we're not > anywhere near there. > > It suffices to specify start and end state, given > requirements like IR-Pause --> DR-Pause needing > to pass through IR-Update but not Run. Sometimes there is a requirement that a state is *not* passed through.... if that state was on one out of two shortest paths, we could be in trouble... Typically the end state for a scan is an important case to have an unambiguous result for... -- Øyvind Harboe http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
