Comment - In jlink_execute_reset(jtag_command_t *cmd)
jlink_tap_execute(); Should be removed and be superflous, actually both of them should be removed. This is because the jtag command queue MUST be clean at this moment, controlled from jtag_add_reset. Otherwise we get tap sequences sent while reset is active and also different behaviour on synchronous interfaces (Zylin and maybe parport) and usb interfaces. The low lewel jlink_reset is synchrous and contains the necessary usb writes, as is the ft2232 reset. This work is part of the next small patch, coming soon, but in essence, the drivers should do NOTHING more for reset than toggle the hw lines. The rest is up to jtag_add_reset and gobal sequencing from target reset handlers. We can see with hw traces of the actual jtag hw communications how openocd sends scan sequences while both srts and trst are in reset, which is of course the route of some reset problems. Have a nice day Magnus _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
