**** BEGIN LOGGING AT Wed May 19 02:59:57 2010 May 19 14:56:44 hi, having some errors with gdb again: http://gist.github.com/406384 May 19 14:56:55 anyone can give a hint pls May 19 14:58:46 ferar: i had a similar problem on my at91. seemingly I can't use a hardware breakpoint and single step afterwards May 19 14:58:53 try removing the breakpoint after halting May 19 14:59:08 you mean thbreak May 19 14:59:17 I was looking into that, thanks May 19 15:10:10 Aloha May 19 15:10:49 Is there a driver for Xilinx USB-JTAG 03fd:0008 preferrably on OS X? May 19 15:12:19 i guess you should check the xilinx website for os x drivers May 19 15:14:05 I see May 19 15:14:11 I thought I need a driver from openocd May 19 15:14:55 openocd needs support for this xliinx device, but the device still needs some driver for USB access May 19 15:15:23 e.g. my amontec tiny jtag key has severed by usbserial on linux, but openocd still needs support for the tinykey itself May 19 15:15:37 currently I'm using windows in a vmware to develop, but i hoped I could use openocd on os x to debug May 19 15:16:18 As Xilinx doesn't support anything except Linux and Windows, I guess they don't have OS X drivers just for the jtag May 19 15:18:03 are there openocd binaries for windows? May 19 15:18:20 does it need cygwin etc.? May 19 15:19:08 i dont use openocd on windows, but afaik there are binaries used without cygwin May 19 15:28:23 lilalinux: http://www.freddiechopin.info/ May 19 15:41:40 thx May 19 15:48:27 KotH: I installed it, but can't find the executable May 19 15:49:00 ah, got it May 19 15:50:08 it installs some unixish tools too May 19 15:50:26 think of openocd on windows as if you'd run it on linux May 19 15:50:32 hm, why does it try to open network connections? May 19 15:50:39 it's not a typical windows app and doesnt behave like one May 19 15:50:45 because openocd does May 19 15:50:45 k May 19 15:50:52 read the docs on the control ports May 19 15:50:58 ok May 19 15:59:54 lilalinux, JTAG to program or do other things? May 19 16:00:35 They used to have open source programming tools. May 19 16:04:20 ali_as: programming works just fine with ise May 19 16:04:31 I wanted to do some more stuff with jtag May 19 16:04:36 just experimenting May 19 16:05:03 however, I don't want to flash my precious Xilinx USB-JTAG cable May 19 16:05:57 as suggested here: http://www.ixo.de/info/usb_jtag/#xilinx May 19 16:13:04 uhm.. May 19 16:13:29 openocds purpose is mostly in programming and debugging CPUs May 19 16:13:44 dunno how well it works for FPGA stuff, but i dont think it's that good May 19 16:14:16 and if you dont want to fry your xilinx cable, get an amontec jtag-key tiny May 19 16:14:36 it's cheap, and oss xilinx tools support it May 19 16:20:50 jtag-key tiny is made by olimex May 19 16:21:22 amontec do jtagkey and jtagkey2 May 19 16:22:16 Olimex is cheapest. Amontec slightly fancier. Both work, but produce different bugs/issues IME May 19 16:22:54 openOCD works OK on CPLD/FPGA programming but there are almost no existing configs - you have to work it out May 19 16:38:31 Did openocd give up on supporting non ARM CPUs? May 19 16:42:30 hi, I'm having this problem of debugging a simple test program that toggles the scroll-lock on a s3c2450 based board. http://gist.github.com/406510 May 19 16:43:20 where this: 0x8064 <_clear_gpa7>: pop {r0, pc} came from gnu-as added it or something? May 19 16:47:54 It looks like the start of your program is somewhere else. May 19 16:48:26 I usually place a software breakpoint as the first instruction in my test program. May 19 16:48:51 That way whereever the linker puts my code I'm instantly at the start. May 19 16:49:01 I did place a hb at _start, and it did stop there May 19 16:49:05 Not sure how this is done on ARM though, it's been a while. May 19 16:49:35 I mean a physical instruction. On x86 this would be int3. May 19 16:49:39 then as I "si" and get to 0x8064 <_clear_gpa7>: May 19 16:50:02 If pc is being poped then the execution is going somewhere else. May 19 16:50:08 the pc changes (there is this pop thing!) May 19 16:50:12 Yes. May 19 16:50:53 well, someone tell from where the hec this pop came! nothing like that in the original code May 19 16:51:57 It's a normal way to jump to other code. May 19 16:52:06 Can you single step to see where it goes? May 19 16:52:07 humm May 19 16:52:36 yep I'm "si" all the time and goes to garbage, May 19 16:52:56 for I reset halt and nothing is loaded May 19 16:52:58 Also, not sure I'd use _start. May 19 16:53:37 Try putting your executable into a dissasembler. May 19 16:53:55 Check the labels, see where the code is... May 19 16:53:59 ok let me check that quickly May 19 16:54:27 bb an hour or so. May 19 16:58:27 here is the disassembley output http://gist.github.com/406536 May 19 16:58:36 all seems fine! May 19 17:46:12 What does your memory map look like? May 19 18:02:29 ali_as: http://gist.github.com/406618 May 19 18:03:58 Not what I mean, where is the flash, where is the ram, is address translation involved? May 19 18:05:29 no MMU is off May 19 18:06:03 MMU: disabled, D-Cache: disabled, I-Cache: enabled May 19 18:06:18 Ok, s3c2450? May 19 18:06:22 yep May 19 18:09:53 Is this some sort of gps or satnav? May 19 18:10:46 no, it is a simple handheld like May 19 18:10:55 with gprs May 19 18:11:29 you meant memory map (samsung documentation) May 19 18:11:49 Yes. May 19 18:12:14 where I can past pic? May 19 18:12:20 on the web May 19 18:13:20 found it just a sec May 19 18:15:49 http://picpaste.com/memmap.jpeg May 19 18:16:53 ali_as: http://picpaste.com/pics/memmap.1274293012.jpeg May 19 18:18:13 Ok, how is elf being loaded, that seemsto be a rom area. May 19 18:18:26 what is troubling me, is that the original code is not the same as what gdb is loading! I used arm-linux-as / arm-linux-ld to get the binary May 19 18:19:31 well first launch gdb as: gdb arm_test.elf May 19 18:19:33 Yes. Either your code is somewhere else or you ae trying to load it into rom. May 19 18:19:57 then : load May 19 18:20:42 and I get: http://gist.github.com/406637 May 19 18:21:00 Can you use assemble the code for the ram area? May 19 18:22:26 you mean 0x3800_0000 May 19 18:22:36 starting from there? May 19 18:24:00 Yes or sram at 0x4. May 19 18:24:23 Might be the ORG diective in the assembler. May 19 18:24:43 ali_as: but in my openocd I have the following configuration: $_TARGETNAME configure -work-area-virt 0x40002000 -work-area-phys 0x40002000 -work-area-size 0xc000 -work-area-backup 0 May 19 18:25:25 My understanding is the elf contains the address it expects to load to. May 19 18:27:22 well, I'm not setting any address that's for sure May 19 18:27:37 That may well be why it's defaulting to 0. May 19 18:28:38 fine, how can I set that? May 19 18:29:04 The directive I'm familier with is ORG. Might be .ORG 0xblah. May 19 18:29:23 I've not done ARM assembler in a few years. May 19 18:31:03 many thanks **** ENDING LOGGING AT Thu May 20 03:00:00 2010