On Thu, Jul 17, 2008 at 10:02 PM, Magnus Lundin <[EMAIL PROTECTED]> wrote: >> On Thu, Jul 17, 2008 at 8:41 PM, Spen <[EMAIL PROTECTED]> wrote: >>>> >>>> Any objections? >>>> >>> >>> Personally i would prefer to keep the existing behaviour. >> >> I guess you mean that you are speaking on behalf of what users >> would want? > > That is a VERY good argument ! > >> The problem is that the behaviour is *unexpected*. > Why "unexpected"?
Would an end user expect the following sequence to work from GDB using a tested configuration script for his target? reset load I'd have to say yes. If "reset" does not execute the init sequence, then the answer is "possibly". >> If there is a reset init script, then why isn't it run by default as it >> is for other tools? OpenOCD is a bit off the beaten track here having >> this config option. > If there is a reset init script and a reset_init command, of course the > init script should be run. If I do a reset, there are many possible > scenarios where I want to keep as much processor state as possible, so > reset halt is a very reasonable bahiviour unless a different default reset > behaviour is specified. I don't know what to make of the above. You are an expert on these things, so you understand all the different types of resets and you have all of them available to you. I'm trying to understand what this means for less seasoned users. >> Minimally I'd like to require all scripts in the target library to >> have reset_init as the reset behaviour. > Please, let us try to make core OpenOCD as stable and versatile as > possible, and not demand default behaviour. Please refrain from the > "require". It is perfectly OK to write a target library with reset_init as > reset bahaviour. It is a very different thing to "require" rthis > behaviour. When I say "target library" I'm referring to scripts committed to src/target/target in OpenOCD. You can do anything you want with your own configuration scripts. The scripts in src/target/target (should) follow the rules laid out in src/target/target/readme.txt With the new tcl stuff, there will be all the configurability experts could ever want. -- Øyvind Harboe http://www.zylin.com/zy1000.html ARM7 ARM9 XScale Cortex JTAG debugger and flash programmer _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
