> Take for example: > > fields[0].tap = etm_reg->jtag_info->tap; > fields[0].num_bits = 32; > fields[0].out_value = reg->value; > fields[0].in_value = NULL; > fields[0].check_value = NULL; > fields[0].check_mask = NULL; > > fields[1].tap = etm_reg->jtag_info->tap; > fields[1].num_bits = 7; > fields[1].out_value = malloc(1);
:-) That malloc() is just a silly anachronism in OpenOCD. We should be mostly rid of them now... > buf_set_u32(fields[1].out_value, 0, 7, reg_addr); > fields[1].in_value = NULL; > fields[1].check_value = NULL; > fields[1].check_mask = NULL; > > fields[2].tap = etm_reg->jtag_info->tap; > fields[2].num_bits = 1; > fields[2].out_value = malloc(1); > buf_set_u32(fields[2].out_value, 0, 0, 0); > fields[2].in_value = NULL; > fields[2].check_value = NULL; > fields[2].check_mask = NULL; I think everybody will agree that this is one of the least elegant parts of OpenOCD. I've worked on improving this part of OpenOCD. The check_value/mask no longer have to be set. Also, I believe that switching to 32 bit instead of 8 bit word size would help(get rid of lots of buf_set_u32()'s). > OUT_FIELD(fields[0], 32, reg->value); > OUT_FIELD(fields[1], 7, reg_addr); > OUT_FIELD(fields[2], 1, 0); > > it would be macros you might have to look up once when you deal with > openocd for the first time, but that is a tiny price to pay for > immediately seeing the relevant parts of the action. Macros/helper fn's could improve things, but for this particular case, I believe the root cause of the problem should be attacked. Currently the worst part of the fields.xxx mess is that 8 bit instead of 32 bit words are used... -- Øyvind Harboe 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
