Subject: armtool patches Date: 20-01-2004 18:57 From: Petr Novak To: Erwin Authried CC: Stas Khirman , Bruno Lopes F. Cabral Hi Erwin [...] While working with the armtool, I have done some small changes. Let me comment on them first, you can have a look at the sources later. First, while for most of your work it is fine that the CPU is stopped except the x command, I wanted to be able to for instance place a breakpoint, get the register dump, and continue at the full speed. First, I thought that the instruction counting is wrong, but after all the only "problem" was that the TAP controller was not left in the run/idle mode after the armtool has returned to command shell. Next, I have added some very primitive breakpoint functionality, wich has helped me to trace function call parameters and similar tasks while analyzing the standard firmware supplied with the SMC box. I have made the tool more correct in operations it needs to wait for the CPU to finish performing the operation. The original version did simply a short busy wait and has assumed that the operation has finished properly. I have used a more strict approach waiting for the status bit, at the obvious cost of the tool being substantially slower. However, with the hardware and tools I was not used to, I have preferred this stricter approach to the "I know it has to be that way" one. I have also tried to add more JTAG stuff, but I have to admit that once uClinux started to work, I did not really come back to finish the low-level stuff, as I had to finish my project where the SMC was just a tool and not a final objective. [...] The patch is attached, not all changes are really relevant, I was getting used to using the tool I may have added some debug print and other stuff not really doing any useful work. For the Intel StrataFlash erase/programming, I have used the same approach as you did - to load the data into SDRAM with armtool with a small code to perform the actual flash write. As the write to the Intel StrataFlash chips is much simpler than with other makes of flash, the program is simpler than yours for other flash architectures. There is a small caveat that the SMC hardware has a gate for flash write on one of the GPIO pins, so the pin has to have the correct value for any flash write to be effective. If you want to check anything in the changes or if you have some idea how to extend the tool, please drop me a line. Cheers, Petr Novak vice-chairman DOBNET, rural broadband non-profit WISP we operate in the lower Berounka region (Berounka river, Brdy mountains, Cesky Kras caves, under the Karlstejn castle, a famous golf course, a most wonderfull spot only 40km south of Prague) Czechia [...]