**** BEGIN LOGGING AT Tue May 18 02:59:57 2010 May 18 14:22:58 hi, I am trying to gdb my vmlinux, using openocd-0.4.0 on opensuse-11.2. All seem to work fine http://gist.github.com/405052 , but why the transfer rate is so low? May 18 14:42:49 ferar: you may need to adjust jtag_khz May 18 14:43:06 ok, thanks May 18 14:46:14 Weaselweb: however , In my configuration I used the jtag_rclk 3000, http://gist.github.com/405081. Shouldn't that automatically adjust to the right frequency? May 18 14:47:34 on my at91 I can use 6000 KHz after PLL and MCK has been setup, but adaptive clocking needs a supported ICE interface May 18 14:49:58 What are the operating systems? May 18 14:50:35 I'm wondering about ethernet issues to localhost. May 18 14:50:53 I'm running openocd on linux May 18 14:51:11 and gdb of course too May 18 14:52:19 I'm also wondering what the JTAG device is. May 18 14:53:15 Its being operated from within a VM? May 18 14:54:29 for me the interface is j-link, and OS is opensuse no VM May 18 14:54:46 the board is samsung s3c2450 May 18 14:57:20 Weaselweb: is it necessary to set PLL and MCK? or I can simply let u-boot does that, then halt after u-boot loads! then from gdb load vmlinux? May 18 14:57:40 I have no idea about j-link interfaces. May 18 14:57:47 me neither May 18 14:58:12 ferar: you can let u-boot do the initializing stuff and start with a high jtag_khz May 18 14:58:29 ok, thanks May 18 18:29:55 when I connect via gdb, and do: (gdb) info all-registers I get different output compared to that of telnet output: > arm reg what could be the issue? May 18 18:46:43 different how? May 18 19:34:44 roxfan: http://gist.github.com/405429, http://gist.github.com/405430 May 18 19:37:37 the values look the same to me May 18 19:38:18 the formatting is different, but what did you expect? May 18 19:38:23 the last two instructions are different May 18 19:38:35 ok, formating May 18 19:38:49 that's disassembly, not registers May 18 19:38:59 yep May 18 19:39:13 R13 is SP, and push x is the same as STMDB SP!, x May 18 19:41:23 anyone might know why this happens: http://gist.github.com/405434, and then the board gets out of halt and starts running !? May 18 19:43:13 the daemon output is http://gist.github.com/405439 May 18 20:21:32 ferar: what i mapped to that address? RAM, flash? May 18 20:22:52 Weaselweb: not sure really, I think SRAM May 18 20:23:19 note that MMU is on at this stage May 18 20:23:27 oh, ok May 18 20:23:41 then it depends what it is mapped at this stage May 18 20:23:43 and I have a 1 Gig of NAND visible to openocd May 18 20:24:21 you can't obviously set a software breakpoint into NAND (while you even can't execute directly from NAND either) May 18 20:26:44 well, I let u-boot load, then halt the target, start gdb, then load the vmlinux (48MB) to DRAM (SRAM is only 64K) May 18 20:27:23 u-boot initialized PLL, MMU, NAND becomes visible May 18 20:28:15 If I want to debug linux itself after it completly booted, I start u-boot and load the uImage from flash and attach then via gdb, so i dont need to load the vmlinux to SDRAM via ICE May 18 20:29:07 ok, well the problem I'm facing is that zImage does not even uncompress! May 18 20:29:32 I see u-boot's message: "Starting Kernel" and that's it! May 18 20:32:56 you can set a hardware breakpoint to your entry point, let u-boot load your image and jump to it, your debugger should stop right now and you can start debugging May 18 20:34:36 thanks, ok let me try this May 18 21:59:02 Weaselweb: I still can't figure out why my gdb can not set beakpoint and always get this Error: Unable to set 32 bit software breakpoint at address 00008058 - check that memory is read/writable ! May 18 21:59:40 anyone can tell me what the heck I'm doing wrong May 18 22:29:13 it's probably in the flash May 18 22:30:27 try "gdb_breakpoint_override hard" May 18 22:39:45 roxfan: thanks, that was it **** ENDING LOGGING AT Wed May 19 02:59:57 2010