Johann F. wrote: >>> I can't make out the important changes from the gratuitous ones. Can >>> you please revert all the u8 -> uint8_t changes as well as whitespace >>> changes? >>> >>> -- >>> Rick Altherr >>> kc8...@kc8apf.net >>> >> Its been a week now. I can boil it down and find the essence of the >> changes. I was going to get back to this code in the next couple of >> weeks anyway. There could be some useful changes in this mess. If they >> are there, I will find them. >> >> >> Dick >> >> > > There are my changes again. I noticed that i forget to handle > runtest_requires_tck in new function xs_shift. > > Johann >
Johann, I am feeling no love for this patch. I do not see where any flow is different than the original code, other than you are multiplying xruntest by 16 on successive attempts. This multiply by 16 is not part of the Xylinx docs or algorithm for the intended chip, let alone any other FPGA or arbitrary SVF recipient. This xruntest value can be set to a higher number in your toolset which generates the xsvf file? I'll give it another look, I have an FPGA to program here. Maybe I am missing something in the patch, but I went over it line by line. And yes, we lost the runtest_requires_tck support in your patch also. In the next several days, I will try my fpga hardware and see if there is any general problem persisting. Troubleshooting things like this are made far easier by my new tap_set_state() logging functionality. However, there are still tap state transitions that are not being logged, and I think this is a needed enhancement that I would like to correct as soon as I have time to. All state transitions should go through tap_set_state(). When logging is turned off, there is no perceptible overhead, since it is a compile time off/on option. Dick _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
