On Sun, Aug 8, 2010 at 8:50 PM, David Brownell <davi...@pacbell.net> wrote: > I don't recall problems in this space... quite > the opposite (arm9).
It's subtle, but you may have seen and dismissed it. I know I have ;-) > If "reset run" leaves the target in reset or halt > instead of running, that's a bug to fix > by ensuring the target is running. The bug is that deassert does not update target->state, but relies on poll to be invoked on a timer at some laterpoint. This is doesn't work with poll off and is unpredictable with polling on. The actual target state is fine. E.g. tcl proc's can fail whereas the same commands in telnet can succeed: Polling on: reset run halt => works reset run ; halt => fails as there is no time for polling to be invoked on timer. I've explained the responsibilities of deassert a bit more in the .h file now. I pondered "fixing" this in the reset procedure, but I figured the real bug was that the deassert implementation needed to leave the target and OpenOCD's structures in a well defined state. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
