**** BEGIN LOGGING AT Mon Feb 14 23:59:56 2005 Feb 15 00:00:42 s/is/in/ Feb 15 00:07:14 It looks like the NPE library is missing a component. It doesn't build because some macros are missing, but there are no errors about missing headers. Feb 15 00:07:28 I'm guessing it's an unexpected omission. Feb 15 00:07:38 nite all Feb 15 00:30:18 Bewoolie, yes, I wrote the 1.0.25 in .... wait a minute hes not here anymore. Feb 15 00:32:20 yep....not here. Feb 15 01:07:55 wow - there's a real crowd here in openjtag now ... Feb 15 01:08:14 yes...more than I'm used to! Feb 15 01:08:18 but nice Feb 15 01:08:30 prpplague, TekGecko: g'day Feb 15 05:25:50 dyoung: what kind of reset? a full system reset, or a tap logic reset with an additional nTRST ? Feb 15 07:19:00 ho ho ho Feb 15 07:19:02 morning all Feb 15 07:23:25 <[g2]> morning Feb 15 07:23:34 <[g2]> dyoung, thx for the e-mail.... Feb 15 07:48:54 qbot: seen beewoolie Feb 15 07:48:54 beewoolie <~beewoolie@florence.buici.com> was last seen on IRC in channel #openjtag, 7h 41m 16s ago, saying: 'nite all'. Feb 15 07:49:10 qbot: ~emulate abcs Feb 15 07:49:10 no idea, prpplague Feb 15 07:49:13 doh Feb 15 07:49:21 qbot: ~`tuesdays Feb 15 07:49:21 prpplague: are you smoking crack? Feb 15 07:49:23 qbot: ~tuesdays Feb 15 07:49:24 prpplague: I don't know, could you explain it? Feb 15 07:49:36 hmm qbot needs an education **** BEGIN LOGGING AT Tue Feb 15 09:32:54 2005 Feb 15 10:10:55 * prpplague needs a beer Feb 15 10:16:50 * Tiersten pours a beer into the floppy drive to send to prpplague Feb 15 10:17:05 Only 1.4MB worth. Sorry Feb 15 10:30:13 Tiersten: thanks Feb 15 10:48:03 hmm, apex is giving me a headache Feb 15 11:06:50 argh i need to insert some debuggin leds routines Feb 15 11:59:40 morning(barely) Feb 15 12:03:48 hi ka6sox-zzzz Feb 15 12:03:57 oops, not-ka6sox that is Feb 15 12:04:49 hiya jacques Feb 15 12:04:59 will you have time to work later on Gastro? Feb 15 12:08:49 don't see why not Feb 15 12:09:03 thanks..I'll let you know when he is up and on what IP. Feb 15 12:09:18 bbs Feb 15 12:09:21 ok Feb 15 12:41:22 where is beewoolie when you need him Feb 15 13:35:13 beewoolie: hey Feb 15 13:35:18 beewoolie: _you! Feb 15 13:35:37 beewoolie: apex is giving me a headache Feb 15 13:35:49 Take two aspirin... Feb 15 13:35:56 What is it? Feb 15 13:35:57 beewoolie: you got a few minutes to impart some knowledge? Feb 15 13:36:04 probably. Feb 15 13:36:15 beewoolie: relocate_apex Feb 15 13:36:23 beewoolie: seems apex is dying there Feb 15 13:36:32 beewoolie: doesn't return from the function call Feb 15 13:36:51 there is some new code to help with this, a little. Feb 15 13:36:52 beewoolie: the bootstrap seems to correct Feb 15 13:36:55 There is the debug_ll code. Feb 15 13:37:06 beewoolie: yea Feb 15 13:37:14 You can force it to output characters and hex to tell you what's going on. Feb 15 13:37:20 beewoolie: i created one since it didn't exist for the lh7952x Feb 15 13:37:32 I've posted one since then. Feb 15 13:37:40 I think. Feb 15 13:37:53 Anyway, usually, the problem is that sdram isn't really there. Feb 15 13:38:01 Do you have jtag? Feb 15 13:38:05 beewoolie: yea Feb 15 13:38:56 BDI? Feb 15 13:39:06 * beewoolie forgets easily Feb 15 13:39:07 ahh ok the 1.0.25 has the debug for the lh79520 Feb 15 13:39:15 beewoolie: trace32 Feb 15 13:39:27 If you are going to make patches to apex, I can recommend quilt. Feb 15 13:39:39 It will make it easier to follow updates to the baseline code. Feb 15 13:39:57 beewoolie: but i've not been able to make heads or tails of the why its dying in the relocation Feb 15 13:40:06 beewoolie: hmm, well, i prefer patch Feb 15 13:40:22 If you have a debugger, the most expedient way to discover the problem source it to step the code. Feb 15 13:40:30 quilt uses patch. It's a patch manager. Feb 15 13:40:42 beewoolie: yea thats what i've been doing Feb 15 13:40:51 It has a patch stack and can quickly keep track of the changes in groups. Feb 15 13:40:59 What instruction is failing? Feb 15 13:40:59 ahh Feb 15 13:41:18 beewoolie: the relocate_apex just stays in a loop Feb 15 13:41:21 beewoolie: never exits Feb 15 13:42:07 It's a simple loop. Look at the registers. It copies until one register is greater than the end point. Feb 15 13:43:37 just not used to the inline asm syntax Feb 15 13:44:00 guess i'll try moving forward with 1.0.25 Feb 15 13:44:51 Disassembled it with the debugger. Feb 15 13:45:34 beewoolie: you tested 1.0.24 with the lpd79520 board? Feb 15 13:46:46 Not specifically. I'm testing against the lpd79524. Feb 15 13:46:59 I'll go upstairs and try it. Feb 15 13:49:27 beewoolie: its gonna be hard to get our patchs done against apex with the vesion changing alot Feb 15 13:50:04 My intuition is that either you are changing too much, or you need a patch manager. Feb 15 13:50:34 Your changes ought to be localized into files that I don't have. Except for a few hooks. Feb 15 13:50:38 beewoolie: not really changing anything, just adding our platform Feb 15 13:50:52 Right. So, a patch manager will make that easy. Feb 15 13:51:01 beewoolie: if you say so Feb 15 13:51:03 :) Feb 15 13:51:35 Intesting. it doesn't boot on the 79520. It is getting past the relocator, though. Feb 15 13:53:08 it's running. I think I wonked one of the uart pieces. Feb 15 13:53:58 is that with the 1.0.24 or 25? Feb 15 13:54:03 1.0.25 Feb 15 13:54:41 beewoolie: are you getting any debug stuff from debug_ll ? Feb 15 13:54:48 yes. Feb 15 13:54:51 That part works. Feb 15 13:55:28 ok, i'll gerry-rig the lpd79520 config for my board for a quick test Feb 15 13:55:43 I think I'm initializing the wrong serial port. Feb 15 13:56:13 This is one of the areas that needs to be improved. The default serial port should be in the config file. Feb 15 13:56:40 beewoolie: yea, i searched for it earlier this morning Feb 15 13:58:31 When I get the kconfig part working, this will not be so hard to fix. Feb 15 13:59:54 Found the problem. Feb 15 14:00:07 Which serial port do you use Feb 15 14:00:09 ? Feb 15 14:00:20 uart1 Feb 15 14:00:28 OK. Then you don't need to change it. Feb 15 14:01:43 Do you want a new version, or do you want to make the change youself? Feb 15 14:03:22 The error is in the serial.c file. It should use UART instead of UART_PHYS in the __REG macros at the top of the file. Feb 15 14:04:03 It works on the 79524 because that board uses U0 for the console. Feb 15 14:04:14 beewoolie: naw. i can make that change Feb 15 14:04:23 beewoolie: let me have a quick look Feb 15 14:05:10 BTW, I added a make target...just for you...that builds every configuration. It lets me make sure I haven't broken someting. Feb 15 14:06:44 beewoolie: hehe Feb 15 14:06:56 it's caught me once already. Feb 15 14:07:18 beewoolie: hmm, well 1.0.25 doesn't get any uart data Feb 15 14:07:29 with the putc_ll Feb 15 14:07:36 There is another error. Feb 15 14:07:53 the init-lh79520 clears the wrong bit in the rcpc for U1 Feb 15 14:08:05 It is clearing U0 Feb 15 14:09:24 gotcha, changed Feb 15 14:17:22 beewoolie: still nothing on the uart Feb 15 14:17:25 * prpplague looks again Feb 15 14:17:59 Not even the debug output? Feb 15 14:18:03 debug_ll Feb 15 14:18:09 OK. here's how we do this. Feb 15 14:18:17 Display memory at 0xfffe1000 Feb 15 14:18:23 That's the base address of U1. Feb 15 14:19:03 Um, wait. Feb 15 14:19:59 It's 0xfffc0000 for U0 Feb 15 14:20:45 It's fffc1000 for u1 which is what we're using Feb 15 14:21:01 it should look like this: Feb 15 14:21:13 fffc1000: 08 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ........ ........ Feb 15 14:21:13 fffc1010: 00 00 00 00 00 00 00 00 38 00 00 00 00 00 00 00 ........ 8....... Feb 15 14:21:13 fffc1020: 00 00 00 00 08 00 00 00 00 00 00 00 70 00 00 00 ........ ....p... Feb 15 14:21:13 fffc1030: 01 03 00 00 12 00 00 00 00 00 00 00 00 00 00 00 ........ ........ Feb 15 14:21:55 it will take me a few to get setup to do that Feb 15 14:26:07 Once you've got that and it looks good. write 65 to 0xfffc1000 which should put a A on the serial output. If that doen't work, you know where to look. Feb 15 14:26:09 <[g2]> morning all Feb 15 14:26:12 [g2]: hi Feb 15 14:28:34 beewoolie: yea, well the hardware works Feb 15 14:28:53 beewoolie: i have it booting with blob and our custom bootloader Feb 15 14:29:10 I didn't say anything about hardware not working Feb 15 14:30:02 beewoolie: i shoved some led functions in Feb 15 14:30:02 beewoolie: i suspect something isn't going right for the relocation, the sdram probably isn't setup properly Feb 15 14:31:07 All of that is easy to verify with the debugger. Feb 15 14:31:35 I think it behooves you to get the debug_ll working first. It should be easy and it gives you something to stand on. Feb 15 14:33:37 beewoolie: yea testing now Feb 15 14:34:50 <[g2]> beewoolie, have any luck with the opie builds ? Feb 15 14:35:01 [g2]: get this. the NPE library doesn't build. It's got some undefined constants. Feb 15 14:35:21 I'm starting with scratchbox this morning. I think this might make it work. Feb 15 14:35:32 <[g2]> GL Feb 15 14:35:38 <[g2]> the NPE stuff is a furball Feb 15 14:35:51 I can tell. Feb 15 14:36:02 I'd load the firmware myself, but the code is really really bad. Feb 15 14:36:05 I mean bad bad. Feb 15 14:36:16 they have lines in there that say things like this: Feb 15 14:36:18 <[g2]> nod. kas11 *hated* it Feb 15 14:36:29 variable = variable + 10 ; // Add 10 to variable Feb 15 14:37:06 did kas11 get it to build? Feb 15 14:37:20 <[g2]> I think our time might be better spent making usb-serial work Feb 15 14:37:36 <[g2]> and leave that to the kernel Feb 15 14:37:51 I've had good luck with CSR. Feb 15 14:37:53 Except that the kernel won't start the NPE without the bootloader doing an init Feb 15 14:38:04 Lemme h ave a stab at 1.5 and get back to you later. Feb 15 14:38:27 1.4 makes without incident. Feb 15 14:38:33 dyoung: if you can get it to compile and link the library, I can put it into apex. Feb 15 14:38:36 I have the 1.4 epk if that would help. Feb 15 14:38:40 I'll take 1.4 for now. Feb 15 14:38:47 I have the epk, too. what is it? Feb 15 14:38:59 ecos package. Feb 15 14:39:04 I dont know the firmat offhand Feb 15 14:39:19 I don't have ecos and I don't want it. that's another big smelly mess Feb 15 14:39:27 Right. Feb 15 14:39:50 Lemme diddle for a a bit. I'll reportback later. Feb 15 14:40:08 I don't really care how I get access to the library. As long as I can link to it, it ought to be good. Feb 15 14:40:25 ok Feb 15 14:40:35 <[g2]> I'm gonna try booting directly to the hd from APEX Feb 15 14:41:00 <[g2]> well having the kernel boot directly from the HD via APEX Feb 15 14:41:01 With what driver? Feb 15 14:41:06 All that CSR Stuff is in a horrid mess on my disk, so its gonna take a bit to get it sorted out. Feb 15 14:41:11 Oh. That should be OK Feb 15 14:41:23 dyoung: thanks Feb 15 14:41:33 <[g2]> dyoung, ditto. Feb 15 14:57:08 beewoolie: btw, is there a reason you have the -g as one of the compile flags? Feb 15 14:57:42 <[g2]> I think that's to generate the debug info :) Feb 15 14:58:24 [g2] is a smart one. yeah. it's handy to have debug info. You can ignore the 2.95.3 errors about it. Feb 15 14:59:32 <[g2]> I'd guess the BDI can load the map and do symbolic debugging right ? Feb 15 14:59:51 <[g2]> OpenJTAG should be able to be able to do the same Feb 15 15:00:33 No BDI doesn't debug Feb 15 15:00:43 It can talk to GDB which needs the debug info. Feb 15 15:01:07 Well, scratchbox fails the first test. Feb 15 15:05:41 <[g2]> beewoolie, did you see that it looks like siddy nailed the IXP driver issue Feb 15 15:05:46 * TekGecko runs in a circle Feb 15 15:06:21 <[g2]> hey TekGecko, where did you come from ? :) Feb 15 15:07:22 :) Iwas here last night Feb 15 15:07:28 .. but you signed off before I said anything Feb 15 15:07:44 <[g2]> I needed sleep :) Feb 15 15:08:34 :) Feb 15 15:09:05 Anywho, going to start playing w/ some jtag stuff and I saw this channel. Feb 15 15:09:11 <[g2]> Feb 14 21:40:35 Er. Feb 15 15:09:11 <[g2]> Feb 14 21:40:48 .. he left. Feb 15 15:09:11 <[g2]> Feb 14 21:40:48 Heh Feb 15 15:09:19 Yep Feb 15 15:09:40 <[g2]> I saw that part and welcome to jtag playland Feb 15 15:10:08 <[g2]> but that you still didn't mention how you ambled across this channel Feb 15 15:10:35 .. saw that prpplague was in here as well as in #pixterdev, which Im in as well. Feb 15 15:10:35 * [g2] is interested in how the community grows Feb 15 15:10:46 <[g2]> nod. Feb 15 15:10:54 Working on hacking the JuiceBox, and was talking w/ prpplague the other night Feb 15 15:11:06 <[g2]> cool Feb 15 15:35:43 Blender meal? Feb 15 15:35:53 [g2]: Blender meal? Feb 15 15:36:08 <[g2]> pizza Feb 15 15:36:13 [g2]: Wait, I know. Ensure. Feb 15 15:36:45 * [g2] doesn't have to forage for food :) Feb 15 15:41:38 Heh Feb 15 16:00:17 <[g2]> beewoolie, this is a kernel error or a APEX message ? Feb 15 16:00:35 <[g2]> Uncompressing Linux........................................................... Feb 15 16:00:35 <[g2]> crc error Feb 15 16:00:35 <[g2]> -- System halted Feb 15 16:00:46 That is from the kernel decompressor. Feb 15 16:00:56 It's saying that the data it's decompressing is bad. Feb 15 16:01:14 Usually means that not all of the kernel image was copied to RAM correctly. Feb 15 16:01:38 You can copy more bytes than the length of the compressed kernel. But not fewer. Feb 15 16:01:55 <[g2]> hmmm Feb 15 16:02:11 Hasn't it worked for you? Feb 15 16:02:34 <[g2]> booting directly to something always gives that error Feb 15 16:11:26 hi there. Feb 15 16:18:00 [g2]: I don't think it's because you're booting directly. I'd guess we've got a memory conflict. Feb 15 16:18:30 [g2]: Can you show me where things are going in memory? Where is the kernel being copied? what is the command line. Feb 15 16:18:36 <[g2]> beewoolie, it's interesting that it happens for me and not dyoung Feb 15 16:18:54 Suggests operator differences. Feb 15 16:19:06 <[g2]> between flash and ram Feb 15 16:19:12 <[g2]> or dyoung and me :) Feb 15 16:19:14 beewoolie: you get my message about the iocon_uartmux? Feb 15 16:19:26 did not. Feb 15 16:19:31 what is up? Feb 15 16:20:31 beewoolie: on the 79520 the uart1 isn't available at reset Feb 15 16:20:38 beewoolie: you need to set the mux pins Feb 15 16:21:00 beewoolie: IOCON_UARTMUX = 0x0c Feb 15 16:21:13 beewoolie: i gotta run, be back in awhie Feb 15 16:21:21 prpplague: k Feb 15 16:21:27 [g2]: Who is siddy? Feb 15 16:21:40 <[g2]> an OpenSlugger Feb 15 16:21:45 <[g2]> from nslu2-linux Feb 15 16:21:52 beewoolie: just add that at the beginning of the debug_ll section in init-lh79520.c Feb 15 16:22:00 prpplague: K. I'll check on it. Feb 15 16:22:22 [g2]: what did he do? I saw that there was a comment on the IXP drivers from you. Feb 15 16:22:51 <[g2]> http://www.nslu2-linux.org/wiki/OpenSlug/StabilizeEthernetDriver Feb 15 16:23:18 <[g2]> He noticed the IXP driver doesn't have to be run in a polled mode and came up with a fix Feb 15 16:24:10 Ah. There may be more to that story. The 'new' ethernet driver interface is a hybrid between polling and interrupt driven. Feb 15 16:24:27 I spoke with one of the netguys about using this interface in embedded systems. Feb 15 16:26:00 <[g2]> which interface ? Feb 15 16:26:11 <[g2]> the ixp intf ? Feb 15 16:26:34 He seemd to feel that there isn't a benefit without SMP Feb 15 16:27:11 <[g2]> you mean the hybrid api interface ? Feb 15 16:27:17 The 'interface' is a programming interface. Yeah. API Feb 15 16:27:55 <[g2]> I mean for an $80 box this its alot of fun Feb 15 16:28:04 I considered adding it to my lh79524 driver. Feb 15 16:28:59 I still might. Feb 15 16:29:06 <[g2]> I'm kinda curious how much leg room the processor has, I'd be surprised if it's very tuned Feb 15 16:29:12 It requires some real changes to the driver structure. Feb 15 16:31:21 <[g2]> around 7-8 years ago I was debugging the performance problems on moto 860 for ATM Feb 15 16:32:17 <[g2]> I just put a big logic analyzer on the bus and captured several M instrucion executions at a shot it was like 25ms Feb 15 16:32:28 <[g2]> worth of CPU time Feb 15 16:33:14 <[g2]> I just used the map file from the linker and a perl script and we had plots/graphs of where all the time was spent Feb 15 16:33:50 <[g2]> It took about 3 days to setup and 1 1/2 days later we had a 10X throughput improvement Feb 15 16:34:31 <[g2]> have you ever played with the kernel profiling ? Feb 15 16:41:33 [g2]: Haven't. I don't even have a good way to measure tcp throughput. There's this tool, ttcp, that is supposed to work. It just crashes when I use it. Feb 15 16:42:00 <[g2]> I've used ttcp several times, even on the slug Feb 15 16:45:11 <[g2]> beewoolie, are you trying to run it on the slug ? Feb 15 16:45:23 No. I was testing throughput from one of the LPD boards. Feb 15 16:45:40 <[g2]> It's worked fine on several targets Feb 15 16:50:08 It was year ago that I last tried it. Feb 15 16:51:13 <[g2]> well it works pretty well Feb 15 16:51:31 <[g2]> I've run it on mips, x86 and xscale Feb 15 17:12:32 brain fart. Feb 15 17:12:45 so I got the 1.5 files. Feb 15 17:13:01 how much of it do we need for Apex Integration? Feb 15 17:13:29 This is building stuff for Linux. Should I look at the epk instead? And if so, what to do wit it? Feb 15 17:13:39 because it makes a million .o files. Feb 15 17:13:51 no .a or any other kind of library as far as I can tell Feb 15 17:16:44 dyoung-web: I need to call the IxDl functions to initialize the NPE. Feb 15 17:16:54 I think. Feb 15 17:17:01 If the epk work, then so be it. Feb 15 17:23:54 IxD... Okay lemme poke around some and see. Feb 15 17:24:01 what tyurns up. Feb 15 17:24:55 dyoung-web: The other thing, which I haven't done, is looked at redboot to see what it calls. I'd push the micrcode into the NPE myself, but I'm being a little lazy. Feb 15 17:26:39 thats availiable as a separated c file now. Feb 15 17:27:08 Right, but the data is structured and their code is poop. Feb 15 17:27:30 IxNpeDl ? Feb 15 17:27:38 It is far easier to link to the library and let it push to code into the NPE than it is to reproduce the functionality. Feb 15 17:27:42 I think so, yeah. Feb 15 17:28:01 theres also a IxNpeDlImageMgr Feb 15 17:28:18 I haven't reviewed RedBoot to see what it does. Feb 15 17:28:25 in any case, I have the 1.0 versions of these files built. Feb 15 17:28:48 I accidentally blew away the 1.4 built stuff, so I gotta read the notes on how to get back to that point. Feb 15 17:29:24 My goal would be to have instructions of the like: download zip files. Put them in this directory. Type make. Feb 15 17:29:50 nod Feb 15 17:30:25 I mean, those are the instructions for Apex users. We'll write all of the glue to make it work properly. If we only build some of the files, then so be it. Feb 15 17:30:33 The other issue I was having is that the NPE stuff doesnt like to build with my crosstool 3.3.4 . I never worked otu why. Feb 15 17:30:47 NOw that I know what we're looking for I can poke at it more. Feb 15 17:30:51 I found that they use a switch that isn't supported. Feb 15 17:31:08 -mcopy-byte-short or something. Feb 15 17:31:17 I ended up having to use their binary compiler to make it all go. Feb 15 17:31:21 -mshort-bytes Feb 15 17:31:22 their make files are so ugly. Feb 15 17:31:29 it makes me nauseated look at them. Feb 15 17:31:30 I got it to work. Feb 15 17:31:38 It's just that there are some missing constants. Feb 15 17:31:42 And these are dumb ones. Feb 15 17:31:59 IXP_NPE_PHYS like stuff. Feb 15 17:35:48 Aha. Feb 15 17:36:03 Okay, if you want sometihg for testing, I have a libextras.a that has it all. Feb 15 17:53:38 if youre interested http://thestuffguy.com/libextras.a Feb 15 17:59:51 What is libextras? Feb 15 18:08:26 "extras" to redboot, so that includes the ixp access stuff. Feb 15 18:25:38 I'll eventually work out how to make it build just the bits we want/need. Feb 15 18:29:15 I know what I said earlier, but I don't think I want to depend on redboot to make this work. Feb 15 18:32:59 beewoolie: you gonna be up late? Feb 15 18:33:15 Don't know. Feeling kinda tired already and it's 6:30p Feb 15 18:33:31 I made the change you suggested. I cannot explain how it worked for me. Feb 15 18:33:44 Neither do I Feb 15 18:33:48 beewoolie: ahh ok Feb 15 18:34:00 beewoolie: i'll be testing furture tomorrow with some more suggestions Feb 15 18:34:05 I've offered this in case you want to play today while I figure out how to make it go standalone. Feb 15 18:34:13 dyoung-web: Are you sure this library is legit? The section on NPE is awfully small. Feb 15 18:34:18 beewoolie: btw, is there a reason for the -g compiled flag? Feb 15 18:34:28 Its supposed to be. Feb 15 18:34:36 I might be wrong... Feb 15 18:34:39 I thought we talked about -g already. At least, [g2] and i did. Feb 15 18:34:47 1.2M Feb 15 18:34:52 beewoolie: sorry if you did, i didn't see it Feb 15 18:34:59 The redboot it makes boots. Feb 15 18:35:02 -g adds symbols. It helps with debug. You can ignore the compiler warnings. Feb 15 18:35:21 dyoung-web: It's just that the NPE section of the library seems realy small. Feb 15 18:35:34 beewoolie: yea i figured as much Feb 15 18:35:38 Also, It looks like it needs more symbols. Feb 15 18:35:50 00000000 *UND* 00000000 ixNpeDlNpeExecutionStart Feb 15 18:37:47 drat Feb 15 18:38:16 was hoping for a quick fix, but guess not. ;-) Feb 15 18:38:28 Strip the file and it's about 200K Feb 15 18:43:16 ttfn Feb 15 18:43:18 I can try make you a .a file... that symbol exists in a object file. Feb 15 18:43:28 okay Feb 15 18:43:38 I need remember how that works. RTFM or something.... Feb 15 18:51:54 armv5b-softfloat-linux-ar r test.a */*.o Feb 15 18:52:11 http://thestuffguy.com/test.a Feb 15 19:22:51 ttfn **** ENDING LOGGING AT Tue Feb 15 23:59:56 2005