**** BEGIN LOGGING AT Mon Aug 17 02:59:56 2009
Aug 17 19:28:32 <Circuitsoft> Hello. I'm trying to figure out how to connect OpenOCD to a chip, but the datasheet has too much detail.
Aug 17 19:40:39 <roxfan> you need to find at least tck, tdi, tdo and tms
Aug 17 19:41:06 <roxfan> trst and srst can be necessary, depending on the core
Aug 17 19:51:36 <Circuitsoft> It's connected, but I'm having issues with configuration.
Aug 17 19:51:39 <Circuitsoft> It's an ARM7tdmi
Aug 17 19:52:04 <Circuitsoft> I know that it has three scan chains. 0 is full boundary of core, 1 is bus subset, 2 is ICE.
Aug 17 19:52:10 <Circuitsoft> Is that any standard tdmi?
Aug 17 19:52:45 <roxfan> no idea
Aug 17 19:53:13 <roxfan> but i think you should be to at least get the idcode
Aug 17 19:53:48 <Circuitsoft> All I know so far is that the full IR len is 12.
Aug 17 19:54:08 <Circuitsoft> urjtag started doing an exhaustive search finding the DRs of all possible instructions.
Aug 17 19:54:19 <Circuitsoft> Didn't find anything useful.
Aug 17 19:54:27 <roxfan> did it show you the id?
Aug 17 19:54:32 <Circuitsoft> no.
Aug 17 19:54:37 <roxfan> weird
Aug 17 19:55:06 <roxfan> um, why pm?
Aug 17 19:55:10 <Circuitsoft> The part is an sta8058.
Aug 17 19:55:16 <Circuitsoft> Afraid of flood-ban
Aug 17 19:55:28 <roxfan> for three lines?
Aug 17 19:55:33 <roxfan> use a pastebin then...
Aug 17 19:55:35 <Circuitsoft> I think it's happened before.
Aug 17 19:55:36 <Circuitsoft> ok.
Aug 17 19:56:24 <roxfan> either way, 7e1e1e1f looks fishy...
Aug 17 19:56:43 <Circuitsoft> http://pastebin.ca/1532910
Aug 17 19:57:02 <Circuitsoft> I don't actually know anything about what the idcodes should be.
Aug 17 19:57:51 <roxfan> well, you can try replacing fc3c3c3f by 7e1e1e1f
Aug 17 19:58:24 <Circuitsoft> I'm not even sure if the flash exists on the scan chain.
Aug 17 19:58:32 <Circuitsoft> It's a multi-chip-module.
Aug 17 19:58:35 <roxfan> unlikely
Aug 17 19:58:53 <roxfan> wait
Aug 17 19:59:00 <roxfan> you mean boundary scan?
Aug 17 19:59:18 <Circuitsoft> I thought I had to declare it in the chain even if I'm not using it directly.
Aug 17 19:59:22 <Circuitsoft> Error: number of discovered devices in JTAG chain (2) does not match (enabled) configuration (1), total taps: 1
Aug 17 19:59:42 <roxfan> hmm so it found only one device
Aug 17 19:59:58 <Circuitsoft> Uh, vice-versa, I got rid of the flash newtap line.
Aug 17 20:00:10 <roxfan> ah
Aug 17 20:00:36 <roxfan> well it's unlikely than the flash will be in jtag tap chain
Aug 17 20:01:01 <Circuitsoft> ok
Aug 17 20:01:06 <roxfan> usually you either access it via boundary scan or using cpu's debug mode
Aug 17 20:02:53 <Circuitsoft> Datasheet quote: "In this case, the user can store its own "loader" in the internal RAM through JTAG, in a way suitable for production programming equipment. The detailed procedure will be released and described in a separate document."
Aug 17 20:03:16 <roxfan> yeah that's what openocd does
Aug 17 20:03:31 <Circuitsoft> And that's what I'm trying to configure it to do.
Aug 17 20:03:41 <roxfan> it uploads a small code that runs in the core and sends results via jtag
Aug 17 20:04:05 <roxfan> anyway, first you need to make openocd recognize the devices
Aug 17 20:04:42 <Circuitsoft> Info : JTAG tap: sta8058.cpu tap/device found: 0x7e1e1e1f (mfg: 0x70f, part: 0xe1e1, ver: 0x7)
Aug 17 20:04:43 <Circuitsoft> Info : JTAG Tap/device matched
Aug 17 20:04:45 <Circuitsoft> Warn : Tap/Device does not have IDCODE
Aug 17 20:04:51 <roxfan> o.O
Aug 17 20:04:58 <roxfan> never seen this one...
Aug 17 20:05:19 <Circuitsoft> So, perhaps that's not actually an ARM core, and I should be targeting the next device in the chain?
Aug 17 20:06:15 <roxfan> no idea really...
Aug 17 20:06:20 <Circuitsoft> Well, the next item in the chain has a tapid of 0x00000000
Aug 17 20:07:25 <Circuitsoft> I can't imagine that's an arm7tdmi
Aug 17 20:54:22 <Circuitsoft> Well, I've figured something out.
Aug 17 20:54:28 <Circuitsoft> I have my reset_config wrong.
Aug 17 20:54:33 <Circuitsoft> Now, I just need to figure out what it should be.
Aug 17 21:31:22 <Circuitsoft> roxfan: Is it possible to autodetect the reset_config?
Aug 17 21:31:52 <roxfan> umm don't think so
Aug 17 21:32:06 <Circuitsoft> Where can I find a list of all possible flags to try?
Aug 17 21:32:16 <roxfan> in the doc?
Aug 17 21:32:26 <Circuitsoft> Doesn't appear to be there.
Aug 17 21:32:50 <Circuitsoft> http://openocd.berlios.de/doc/html/Reset-Configuration.html makes no mention of a complete list of values.
Aug 17 21:34:38 <roxfan> Possible values are none (the default), trst_only, srst_only and trst_and_srst. 
Aug 17 21:34:41 <roxfan> etc
Aug 17 21:36:10 <Circuitsoft> srst_only and trst_only both fail to reset the device.
Aug 17 21:36:18 <Circuitsoft> OpenOCD can only connect immediately after a power cycle.
Aug 17 21:36:35 <Circuitsoft> Given that, is there any chance of it working without needing a power cycle?
Aug 17 21:36:39 <roxfan> are you sure those signals are connected?
Aug 17 21:37:05 <Circuitsoft> Reasonably. I'll have to check the schematic, but I think so.
Aug 17 21:37:45 <Circuitsoft> If neither of them works, though, will I always have to power-cycle the device before it'll connect?
Aug 17 21:41:00 <roxfan> well, i think usually openocd should be able to break the running target
**** ENDING LOGGING AT Tue Aug 18 02:59:56 2009