**** BEGIN LOGGING AT Sun Feb 13 23:59:56 2005 Feb 14 07:23:27 morning Feb 14 07:29:39 <[g2]> morning ! Feb 14 07:30:31 today I will be in the office for a little while...but I need to "rescue" my car from the side of the mountian before a landslide burries it. Feb 14 07:32:41 its pressed up against the mountian and has one wheel in a ditch. Feb 14 07:34:51 <[g2]> doh! Feb 14 07:35:32 <[g2]> ka6sox-away, do you think you can send back the FatSlug soon ? Feb 14 07:36:30 yes Feb 14 07:36:47 <[g2]> great thx. Feb 14 07:36:57 today is a little bit stressful as I need to get the car out of there..otherwise it very well could be burried. Feb 14 07:37:28 we are expecting a soaker by the end of the week. Feb 14 07:37:39 <[g2]> well get it out!, unless you're wanting another car real bad Feb 14 07:38:09 uh...that would be the second car in 5 mo...the Insurance Co would probably cancel me. Feb 14 07:38:31 <[g2]> what happened to the first one ? Feb 14 07:38:45 Taco'd it on the freeway. Feb 14 07:38:53 3 car pileup Feb 14 07:38:53 <[g2]> OUCH Feb 14 07:39:00 yeah. Feb 14 07:41:42 my ride is here...gotta scoot I'll be on in about 30 minutes from the office...catch you there. Feb 14 07:42:12 <[g2]> nod Feb 14 07:46:07 morning all Feb 14 07:47:33 <[g2]> morning Feb 14 07:48:02 whats cookin? Feb 14 08:27:24 cooking with GAS Feb 14 08:32:31 gnu assembler? Feb 14 08:51:29 yes, how did you guess? Feb 14 08:57:32 no guess, every time I see GAS I think gnu assembler :-) Feb 14 11:12:09 <[g2]> g'day beewoolie Feb 14 11:12:22 gday Feb 14 11:12:31 did you get my mail about oe? Feb 14 11:12:51 <[g2]> the one from O'dark thirty about unslung ? Feb 14 11:13:10 Don't know that one. Feb 14 11:13:21 Oh, is odark midnight? Feb 14 11:13:37 unslung isn't building for me from OE anymore. Feb 14 11:13:49 <[g2]> do you get the TARGET_OS issue ? Feb 14 11:14:03 I set it to linux_uclib Feb 14 11:14:28 <[g2]> there were so many details in that e-mail I couldn't get throught them all :) Feb 14 11:14:40 <[g2]> actually, TARGET_OS = "linux" for Unslung Feb 14 11:14:50 I didn't get an email about. Let me try that... Feb 14 11:15:07 ...one of these days... Feb 14 11:15:14 <[g2]> beewoolie, I was referring to the one you sent me :) Feb 14 11:15:24 o. Feb 14 11:15:49 I only meant to say that it was working and then it wasn't. Feb 14 11:16:08 <[g2]> well that happens with this stuff for now. Feb 14 11:16:20 <[g2]> I'm sure will get around to fixing it sooner or later :) Feb 14 11:16:27 <[g2]> we'll Feb 14 11:16:32 That's the kind of attitude that has let Microsoft software get so big. Feb 14 11:17:35 BTW, I was helping my dad get one of these slugs going. We found that the instructions for changing the http server didn't work. I updated the wiki. I'm curious, though, if you agree with our solution. Feb 14 11:17:54 <[g2]> It's the old "When you're up to your a** in aligators, it's hard to remember you mission is to drain the swamp" kinda thing :) Feb 14 11:18:35 <[g2]> didn't notice the updated on the wiki Feb 14 11:18:38 [g2], thats exactly how I feel. Feb 14 11:18:38 hey, I resemble that..."looking at the pond out back" Feb 14 11:19:25 heh Feb 14 11:19:27 I know, too. I never want to go all the way to the beginning. fortunately, the kernel works pretty well. :-) Feb 14 11:19:29 <[g2]> ka6sox, any luck with the car Feb 14 11:20:05 headed up now...going to try to make it soon(tm) Feb 14 11:20:31 <[g2]> beewoolie, it's a goal of mine to get the nslu2 Openslug changes in the kernel accepted upstream Feb 14 11:20:58 I didn't know you'd made any. What sorts of changes have you done? Feb 14 11:21:05 <[g2]> so by like 2.6.15 or something ppl can just make ARCH=arm nslu2_defconfg and make zImage Feb 14 11:21:39 cya in a bit. Feb 14 11:21:39 Well, a new defconfig is easy. Are there driver-level changes? Feb 14 11:21:44 ka6sox: ta Feb 14 11:21:57 <[g2]> I haven't done any myself, but kas11 did some I/O changes and there is stuff IXP Feb 14 11:22:18 * kas11 tosses ka6sox a winch Feb 14 11:22:29 I recall that she wasn't willing to indent the code to kernel standards. Feb 14 11:22:29 <[g2]> ka6sox, GL Feb 14 11:22:58 More slides took the car away? Feb 14 11:23:18 <[g2]> beewoolie, that's OK, I'm willing to indent to the proper standards etc.... Feb 14 11:23:28 There's always...indent. Feb 14 11:23:35 <[g2]> once I learn what they are and how has the indenter :) Feb 14 11:23:43 she was willing to do a certain amount...but couldn't get OE to work so had nothing to backport changes too Feb 14 11:23:53 Let me know if you have anything to submit. rmk and I are real tight. ;-) Feb 14 11:24:07 <[g2]> thx Feb 14 11:24:29 <[g2]> lennert actually offered to help me with the processs Feb 14 11:25:06 <[g2]> I was going to take him up on his offer if you didn't have some burning flame under your backside to get it done Feb 14 11:25:29 <[g2]> and if it's not terribly onerous to be a maintainer Feb 14 11:27:08 I would certainly clean up x1205-rtc if I had anyone to test it...I hate testing corner cases Feb 14 11:28:36 <[g2]> kas11, I'll be happy to test some. There are tons of things that need testing Feb 14 11:28:55 <[g2]> I think it's all a work-in-progress and many ppl have appreciated the work you did Feb 14 11:29:23 * [g2] is getting ready to use it right now, voodooz put up some stuff on the wiki just this morning Feb 14 11:30:12 I'm perfectly happy letting someone else post the patches. Feb 14 11:30:27 understood...is why I didn't push it...I've since added things that got absolutely no testing...so it all sorta died a natural death Feb 14 11:32:39 beewoolie: morning Feb 14 11:32:49 beewoolie: got apex sorta booting Feb 14 11:33:08 prpplague: awesome Feb 14 11:33:34 anyone got experience reverse engineering contents of a rom? Feb 14 11:33:42 (for arm) Feb 14 11:33:43 disarm Feb 14 11:33:56 * prpplague googles Feb 14 11:33:59 I don't think it understands BE Feb 14 11:36:32 beewoolie: hmm, i'm not finding an app called disarm Feb 14 11:36:35 beewoolie: url? Feb 14 11:36:42 * beewoolie looks Feb 14 11:37:06 <[g2]> kas11, it hasn't died, it's kinda hibernating :) Feb 14 11:37:11 beewoolie: we will probably go with apex now that the 2.95 toolchain issue is resolved Feb 14 11:37:34 It wasn't asm related at all. It had to do with the attributes. Feb 14 11:37:35 [g2] I meant locally ;) Feb 14 11:37:43 kas11: i have trouble with oe as well Feb 14 11:39:12 prpplague: http://homepage.ntlworld.com/g.mccaughan/g/software.html Feb 14 11:40:03 beewoolie: thanks Feb 14 11:40:08 beewoolie: you used it before? Feb 14 11:40:11 prpplague: some. Feb 14 11:40:33 prpplague: I've mused as to whether I could get binutils to do the dissassemly. Feb 14 11:41:19 beewoolie: i've never done any dissassembly Feb 14 11:41:24 beewoolie: so this is kinda new Feb 14 11:48:52 <[g2]> beewoolie, I'm open for any advice from you also on submitting changes for the kernel, so please to be bashful :) Feb 14 11:49:07 <[g2]> s/to be/don't be/ Feb 14 11:50:17 there are lots of pitfalls. The easiest way to start is to let me look at a patch you want included. You can also go to the arm linux pages and read Russell's developer info pages. The bottom line is "make it easy for me (RMK) to maximize the likelihood that the patch will be applied." Feb 14 12:06:10 <[g2]> beewoolie, nod. Feb 14 12:06:45 <[g2]> I'll check out the page, all the changes should be pretty trivial because they'll all apply to the nslu2 branch Feb 14 12:07:05 <[g2]> little likelyhood of messing other stuff up Feb 14 12:29:05 [g2]: We've talked about using the nslu2 archid in the bootloader. Because the kernel as shipped by Linksys ignores the bootloader Feb 14 12:29:21 ...we can set the 'right' id in APEX and RedBoot now. Feb 14 12:29:30 when the right kernel comes along...booyeah! Feb 14 12:29:55 [g2]: btw, setting the OS to linux seems to have fixed the problem. Feb 14 13:09:04 <[g2]> beewoolie, the original Make file for the 2.6.7 kernel / then 2.6.9 set the machine Id properly Feb 14 13:09:48 <[g2]> If the linksys kernel really ignores that then we can just switch the OpenSlug kernel over Feb 14 13:10:03 <[g2]> hmmm, maybe that's not the caes Feb 14 13:15:24 <[g2]> beewoolie, The OpenSlug kernel should be built with both the IXP425 and NSLU2 config support initially. Feb 14 13:15:55 <[g2]> At some point we should probably drop support for the IXP425 Feb 14 13:31:58 [g2]: I think you're missing the point. The kernel doesn't read the archid given by the bootloader. The kernel still needs one, though. They just override it in the head.S. Feb 14 13:32:59 I believe you can build kernels with both id's no problem. Feb 14 13:33:17 <[g2]> beewoolie, I think you're missing the point, OpenSlug's kernel should to boot from both APEX and Redboot Feb 14 13:33:36 <[g2]> Or are you saying that Redboot doesn't pass the id Feb 14 13:33:39 NP. Feb 14 13:33:54 I'm not sure if redboot passes a kernel id. I've have to read the code. Feb 14 13:34:01 it is possible they didn't even bother. Feb 14 13:34:18 <[g2]> If they don't bother, then we're fine Feb 14 13:34:25 Remember, that was a long time ago. I don't think anyone worried about these ids back in the 2.4 days. Anyone except for rmk that is. Feb 14 13:34:36 Well, sort of. Feb 14 13:35:00 If the 2.6 kernels are ignoring the passed id, then we still have an issue. Feb 14 13:35:29 If my suspicion is correct, the head.S is setting the arch id, even in the 2.6 kernels. Feb 14 13:37:00 <[g2]> we should check the code and the bootloader, we may be able to immediately switch to the nslu2 id and modify patch the kernel for the NSLU2 Feb 14 13:37:48 Right. I'm bulding unslung Feb 14 14:09:12 beewoolie: the splash screen code is working? Feb 14 14:09:20 beewoolie: for apex Feb 14 14:11:29 beewoolie: i _hate_ that darn cpld on the lpd boards Feb 14 14:17:06 Is that a question? Feb 14 14:18:14 Their CPLD is not my favorite either. Feb 14 14:19:20 beewoolie: yea, i just noticed the code for the splash screen, is that working on your board? Feb 14 14:20:46 yeah. It's not pretty, but it does work. Feb 14 14:21:02 It needs to be recoded to use some compression. Feb 14 14:21:18 beewoolie: yea that would be nice Feb 14 14:21:36 Moreover, it isn't wasn't designed as a splash-screen module. It was added to test the LCD controller. Feb 14 14:21:39 beewoolie: i'm using the lpd79520 stuff as a base for doing my stuff Feb 14 14:21:42 Spash screen is a side effect. Feb 14 14:21:49 beewoolie: having to go back and remove the cpld stuff Feb 14 14:21:55 Does it get to the command prompt? Feb 14 14:22:14 beewoolie: no not yet, i have to make a couple changes for our serial port Feb 14 14:22:16 The idea would be to make a new mach_ and change the init code. Feb 14 14:22:39 Likewise, the serial port should be simply changing the base address. Feb 14 14:22:46 beewoolie: we have some custom hardware Feb 14 14:22:52 I've been reworking the code to make this all easier. Feb 14 14:23:11 How does your hardware change the uart operation? Feb 14 14:23:35 beewoolie: we have the uart multiplexed on an external connector Feb 14 14:24:01 beewoolie: we have to enable some gpio's to determine the function Feb 14 14:25:04 beewoolie: when the handheld is not in the cradle the rs-232 transceiver is powered down to conserv power Feb 14 14:25:54 beewoolie: yea, i've already defined a new mach_ Feb 14 14:26:07 beewoolie: just using the lpd79520 as a base Feb 14 14:26:13 beewoolie: as a guide Feb 14 14:39:00 prpplague: None of that should effect the bootloader, IMHO. You can set the right GPIO as part of the init code and then set it again when the loader jumps to the kernel. Feb 14 14:39:29 beewoolie: agreed, just haven't done that yet Feb 14 14:40:08 I'm planning to add a mode where apex doesn't even have a command line. It should make the loader really really small. Feb 14 14:40:24 I need to implement the rest of kconfig, though, for it to be viable. Feb 14 14:40:45 beewoolie: yea, that would be nice Feb 14 14:41:07 beewoolie: i've got to add the mmc support , cs8900 and usb Feb 14 14:44:35 [g2]: Guess what? And the answer is *not* chicken butt. Feb 14 14:44:45 +#ifdef CONFIG_ARCH_NSLU2 Feb 14 14:44:45 + mov r7, #(MACH_TYPE_NSLU2 & 0xff) Feb 14 14:44:45 + orr r7, r7, #(MACH_TYPE_NSLU2 & 0xff00) Feb 14 14:44:45 +#endif Feb 14 14:45:50 In other words, your openslug kernel is forcing the arch id. I'm not sure why the author of this patch did it this way. It's actually more bytes than the more conventional method. Feb 14 14:45:57 e.g. mov r7, #MACH_TYPE_COTULLA_IDP Feb 14 14:47:01 ...for most IDs. In the nslu2 case, it's exactly the same number of bytes. Feb 14 14:47:37 Some have said that it is more cache efficient to load all of the bytes through the code stream. I think that readability is more important. Feb 14 14:47:42 <[g2]> beewoolie, I think that was a first attempt at the NSLU2 specific changes Feb 14 14:48:13 <[g2]> depends on how important speed is Feb 14 14:48:23 At this point in the code, it is immeasurable. Feb 14 14:48:34 <[g2]> nod. Feb 14 14:48:37 as I recall, you have a problem when the id doesn't fit in 8 bits Feb 14 14:48:54 You just have to change the instruction coding. Feb 14 14:50:47 mov r7, =MACH_TYPE_NSLU2 ; Load indirect. Feb 14 14:51:10 The assembler does the pc offset magic. Feb 14 14:51:55 For loads over 16 bits it is a clear code-space win Feb 14 14:52:36 [g2]: I wrote just to point out that the nslu2 kernel, like the linksys kernel, is ignoring the bootloader's arch id Feb 14 14:53:06 beewoolie: hmm, i'm not seeing a config option for the splash screen Feb 14 14:53:44 prpplague: yeah. There isn't one proper. I put the splash screen in the lcd controller test code. Like I said, it's not really finished. Feb 14 14:54:19 beewoolie: ahh ok Feb 14 14:54:24 prpplague: I'm preoccupied with getting a demo together. Once I figure out how to build opie, I should be able to get on to other things. Feb 14 14:54:31 beewoolie: np Feb 14 14:54:56 prpplague: I hacked in the logo screen with the hope that it will be part of this demo. Feb 14 14:56:05 <[g2]-away> beewoolie, I'd like to change that for OpenSlug Feb 14 14:57:01 beewoolie: the logo screen is one of our requirments Feb 14 14:58:02 beewoolie: hehe man that is one ugly jtag dongle Feb 14 14:58:41 prpplague: show me the money! Feb 14 14:58:56 beewoolie: hehe Feb 14 14:59:26 * beewoolie thinks prplague might be getting the hang of it Feb 14 15:19:48 beewoolie: how come you didn't get you some nice protoboards to make your jtag? Feb 14 15:20:31 prpplague: I got what I could find. Perfboard works fine. Protoboards, IIRC, are not a permanent solution. Feb 14 15:21:27 beewoolie: http://www.beldynsys.com/ Feb 14 15:21:47 beewoolie: well worth the investment if you tinker around with electronics Feb 14 15:22:01 We've been over this before. Feb 14 15:22:20 When I built mine, I couldn't find anything reasonably priced. Feb 14 15:22:21 beewoolie: oh, didn't realize that Feb 14 15:22:29 beewoolie: hehe Feb 14 15:22:39 IIRC, these products aren't cheap. Feb 14 15:23:20 beewoolie: you can get about 200 of the small boards for about $75 Feb 14 15:23:36 What am I going to do with 200 of them? Sell them on ebay? Feb 14 15:23:45 beewoolie: you can get them individually from www.sparkfun.com Feb 14 15:24:08 As I said, we've been down this road before. Feb 14 15:24:12 hehe Feb 14 15:24:21 beewoolie: just nagging Feb 14 15:24:36 * prpplague likes his projects to be pretty as well as functional Feb 14 15:26:19 I think it's pretty. Feb 14 15:27:08 hehe Feb 14 15:27:37 beewoolie: i'm just nagging you, anything that works is pretty in my book Feb 14 17:40:20 <[g2]> beewoolie, do you use any special tools for managing patchsets ? Feb 14 17:41:20 [g2]: I just started using quilt for some tasks. Feb 14 17:41:31 <[g2]> in OE :) Feb 14 17:42:48 <[g2]> I'm looking for something that will take to unified diffs and split out the components and compare them Feb 14 17:43:15 <[g2]> short of doing that myself Feb 14 17:47:49 * [g2] starts playing with quilt Feb 14 17:48:05 <[g2]> now I'm quilting with my friends.... go figure Feb 14 17:48:18 tee hee Feb 14 17:51:50 [g2]: What are you trying to do? Feb 14 17:52:22 <[g2]> I think I need to apply two patch sets and quilt diff thme Feb 14 17:52:25 <[g2]> the Feb 14 17:52:26 <[g2]> m Feb 14 17:52:28 <[g2]> :) Feb 14 17:52:40 You want the differences between two patches? Feb 14 17:52:49 <[g2]> nod Feb 14 17:52:59 Interestng. Feb 14 17:53:15 I think that that is probably the best way. Feb 14 17:53:24 <[g2]> well there are 187 patches in one patchset, and 189 in the second Feb 14 17:53:33 <[g2]> I'm guessing 3 files were added Feb 14 17:53:33 I'm guessing that what you really needs is what is the same in the two patches. Feb 14 17:53:45 I have a script that breaks patches into separate files. Feb 14 17:53:57 You could use it to compare the individual patch files. Feb 14 17:54:10 It breaks the patch into one file per file patched. Feb 14 17:54:19 <[g2]> that's what I was going to do, but I wondered if there's a tool out there Feb 14 17:54:48 <[g2]> In my 60 read of the quilt man page it looks like it could do it Feb 14 17:55:04 Well, it doesn't really do what you want. What you need is something that will go file by file, apply the patches and then diff the two versions. sounds pretty useful. In fact, it would be handy for it to produce a patch of the differences. Feb 14 17:55:17 It wouldn't be that hard to write, really. Feb 14 17:55:57 <[g2]> no not hard at all, but practical and no need to invent the wheel if it's already done Feb 14 17:56:01 Suck in the two patches. Break them into files. Create the set of all files modified. One by one, produce patched files... Feb 14 17:56:27 You can see if patcher does it. My understanding is that quilt is the top of the food-chain where that sort of program lives. Feb 14 17:56:37 marina Feb 14 17:57:13 <[g2]> with quilt I think it might be do able Feb 14 17:57:38 <[g2]> I think one could a add a patch and apply the patch then push Feb 14 17:57:46 <[g2]> add the second apply Feb 14 17:57:51 <[g2]> then diff the two trees Feb 14 17:58:08 <[g2]> but I don't know yet Feb 14 17:58:16 I don't think it has the right stuff. Feb 14 17:58:25 It doesn't like changing the order of patches. Feb 14 17:58:41 <[g2]> ka6sox-away, marina ? Feb 14 17:58:44 I think either you'll have to apply the patches to two sets of files, or write a script to do it Feb 14 17:58:58 what is marina? google came up bust except for some cute pictures Feb 14 17:59:24 <[g2]> I'll bet it's a MAC thingy Feb 14 17:59:37 Ah. That would explain why I don't see it. Feb 14 18:01:31 [g2], Marina is my wife. Feb 14 18:01:58 I'm not finding anything about a patch diff program. Feb 14 18:02:07 I might have to write one. It is something I've needed before. Feb 14 18:03:38 <[g2]> beewoolie, hmmmm Feb 14 18:05:41 <[g2]> beewoolie, thx Feb 14 18:06:22 beewoolie: I've heard that there is a LE version for NPE available from Intel. Feb 14 18:06:37 Huh! That would be really nice. Feb 14 18:06:59 1.5 is what I read. Feb 14 18:07:11 I'll look. I haven't had a chance to look into why the ethernet interface isn't starting in openslug. I'm guessing that the bootloader needs to do an init. Feb 14 18:07:41 jacques knows all about it. Feb 14 18:07:58 <[g2]> jacques, was talking about that earlier 1.5 Feb 14 18:08:50 I'm just trying to get things moving along in the bootloader world! Feb 14 18:09:21 <[g2]> ok, so back in bootloader world Feb 14 18:09:37 <[g2]> beewoolie, is flashing built in to the latest APEX ? Feb 14 18:09:45 Yeah. Feb 14 18:09:53 [g2], I made it to the office for exactly 20 minutes Feb 14 18:09:55 <[g2]> has it been tested by someone ? Feb 14 18:09:56 There is a nor-flash driver Feb 14 18:10:02 I've used it. Feb 14 18:10:23 Don't go using it without a JTAG. ;-) Feb 14 18:10:40 <[g2]> That's exactly what I was going to do Feb 14 18:10:46 <[g2]> and without serial Feb 14 18:11:55 <[g2]> I mean if I build an APEX and then have mabye you or dyoung flash it and boot a version of OpenSlug, then I'm willing to risk it Feb 14 18:12:44 <[g2]> I'll copy out my Redboot/sysconf parts first Feb 14 18:12:51 v1.2 Feb 14 18:12:51 - Created separate integration patches for big and little endian kernels Feb 14 18:12:56 From the intel site. Feb 14 18:13:21 okay cool Feb 14 18:13:45 Sounds good. I'm using the same CFI driver on several platforms. Feb 14 18:13:54 Feb 14 18:14:03 so, I'm confident that that code works. Feb 14 18:14:18 There are lots of other pieces, though, that I'm not so sure about. Feb 14 18:15:00 Aside from the hassle factor, is there a good reason for us to go to an LE kernel? Feb 14 18:15:22 <[g2]> eventually we should be able to run both LE and BE Feb 14 18:15:34 <[g2]> I'd stick with BE for now Feb 14 18:16:19 <[g2]> I mean, I'm really toying with an Intel IXP free platform Feb 14 18:17:11 <[g2]> now that usbnet is fixed, I been booting directly with the usb-to-ethernet adapter Feb 14 18:17:32 <[g2]> what I really need is a usb2.0-to-gigabit adapter Feb 14 18:18:06 <[g2]> Although I could use the usb host-to-host adapter and forward from the other host box Feb 14 18:18:30 <[g2]> that'll be a good "crush" test Feb 14 18:20:10 Can you get enough speed through USB2 to make the gig adapter worth it? Feb 14 18:20:28 <[g2]> well usb2.0 tops out at 480Mhz Feb 14 18:20:45 <[g2]> I'm $29 gigabit cards to >600Mhz Feb 14 18:21:01 <[g2]> through my $70 gigabit switch Feb 14 18:21:14 Hz != Mib Feb 14 18:21:32 MHz != Mib Feb 14 18:22:02 <[g2]> nod on the 480-to-600, but the 600 was bits per second Feb 14 18:22:15 <[g2]> as measured with ttcp Feb 14 18:22:32 <[g2]> via tcp Feb 14 18:22:37 I got patches from the Intel site, but no libraries. Feb 14 18:22:51 Need to look some more. Feb 14 18:23:16 <[g2]> So the ethernet can surely crush the 100mhz which the slug could handle in UDP Feb 14 18:23:22 <[g2]> via the ttcp test Feb 14 18:29:22 It looks like we need to install the microcode for the NPE in the bootloader. If that's all, this should not be too bad. Feb 14 18:30:06 I felt a break in the force. Feb 14 18:30:21 it was like someone wants me to JTAG write something... Feb 14 18:30:31 <[g2]> :) Feb 14 18:30:33 Like a whole planet crying out? Feb 14 18:31:08 <[g2]> dyoung-web, have you tried replacing Redboot from APEX via the APEX cmd line ? Feb 14 18:31:26 Replacing with apex? Feb 14 18:31:32 Yes, thats how I got it in there in the first place. Feb 14 18:31:58 <[g2]> I said replacing Redboot from APEX Feb 14 18:32:11 [g2]: You can load APEX into SDRAM, run it, and then use it to write itself to flash. Feb 14 18:32:18 Yeah, I jknow but there are several permutations of what you asked. Feb 14 18:32:41 <[g2]> and once it's in flash, can it reflash Redboot over itself ? Feb 14 18:32:46 RedBoot -> Apex; Apex -> Apex; Redboot->RedBoot, etc. Feb 14 18:32:48 <[g2]> that was my question Feb 14 18:33:00 You can. I've never tried. Feb 14 18:33:11 <[g2]> right save the TAX right ? Feb 14 18:33:12 APEX can write anything from memory to flash. Feb 14 18:33:19 Okay so you want to take a Apex Enabled Slug and write redboot back into it? Feb 14 18:33:22 <[g2]> that's my point Feb 14 18:33:42 That should work... Feb 14 18:33:48 With APEX? ---> xr 0x1000 # XReceive redboot to 0x1000 Feb 14 18:33:51 <[g2]> since APEX runs from memory, we should put a copy in the sysconf partition Feb 14 18:34:00 erase nor:0+256k Feb 14 18:34:13 copy 0x1000+256k nor:0 Feb 14 18:34:15 reboot Feb 14 18:34:29 copy mem:0x1000 256k nor:0 Feb 14 18:34:33 nod Feb 14 18:34:34 I've done that. It works, but it doesn't test everything. Feb 14 18:34:41 no. Feb 14 18:34:42 no Feb 14 18:34:42 Those commands are familiar to me these days. ;-) Feb 14 18:34:42 no Feb 14 18:34:53 That command won't work. Feb 14 18:35:09 The form is driver:start+length Feb 14 18:35:40 (sorryr, this client cant type the plus char) Feb 14 18:35:56 2 2=4 Feb 14 18:36:01 4-4=0 Feb 14 18:36:04 2^2=4 Feb 14 18:36:06 Its just the Plus. Feb 14 18:36:06 ;-) Feb 14 18:36:18 <[g2]> & Feb 14 18:36:24 & Feb 14 18:36:35 I think its so I cant go [g2]plusplus Feb 14 18:36:43 ~karma [g2] Feb 14 18:36:43 [g2] has neutral karma Feb 14 18:39:18 <[g2]> beewoolie, can we start APEX from within linux ? Feb 14 18:39:36 <[g2]> or from the linux cmd lind Feb 14 18:39:39 <[g2]> line Feb 14 18:39:41 How would that work? Feb 14 18:39:51 dyoung-web:twang Feb 14 18:40:04 I suppose we could write a driver to do it. You'd have to be really clever. Feb 14 18:40:43 <[g2]> Well assuming there's no watchdog, you'd disable interrupt and overwrite low-memory and soft reset Feb 14 18:40:48 You'd have to pull the CPU out of MMU mode in a way that the current instruction was still pointing to a valid piece of memory. I suppose it could be done. Feb 14 18:41:28 Hmm. There would have to be several pages mapped to low memory. I suppose it is possible. Feb 14 18:41:47 Is this an attempt to bootstrap? Feb 14 18:41:49 <[g2]> 4-5 right :) Feb 14 18:42:00 <[g2]> 16-20K Feb 14 18:42:13 <[g2]> depending on page size Feb 14 18:42:20 <[g2]> which I'm assuming is 4K Feb 14 18:42:22 The thing is that those pages have to be 1:1 mapped. Feb 14 18:42:55 <[g2]> well I'm sure the kernels got 0 mapped for the vectors :) Feb 14 18:42:58 I think that the kernel moves the vectors to 0xffff...., Feb 14 18:43:20 That way, address zero can be a bonafide fault. Feb 14 18:43:33 <[g2]> I was reading about that in the ARM info today Feb 14 18:43:57 <[g2]> that the vectors were up high Feb 14 18:44:06 So, what is the profit here? Feb 14 18:44:26 Oh, wait. Is this a way to test that the loader can load a particular kernel? Feb 14 18:44:39 I think you're better off using the serial point. Feb 14 18:44:44 s/point/port Feb 14 18:44:59 <[g2]> I was thinking easy switch between Redboot and APEX Feb 14 18:45:07 <[g2]> switching Feb 14 18:45:16 You can write it to flash from Linux, no? Feb 14 18:45:22 <[g2]> put a copy in the sysconf at the end and its always there Feb 14 18:45:31 Hmm. Feb 14 18:45:48 I don't think so. They clobber that segment any time they want to. No? Feb 14 18:46:13 <[g2]> Linksys does... OpenSlug doesn't right now Feb 14 18:46:29 what is the goal here? Feb 14 18:46:49 <[g2]> bootloader recovery from a running program Feb 14 18:47:10 I don't follow. Feb 14 18:47:11 <[g2]> we really don't *need* it Feb 14 18:47:39 <[g2]> mtd-utils works from the cmd line and are built into OpenSlug now Feb 14 18:47:54 <[g2]> and both Redboot and APEX can write from the cmdline Feb 14 18:48:19 <[g2]> We just need to teach APEX usb-serial on the NSLU2 Feb 14 18:48:42 Can do. Feb 14 18:48:58 For the sake of a USB console, no? Feb 14 18:49:06 Hey, can we use a USB hub on the slug/ Feb 14 18:49:19 <[g2]> Then ppl can have a solderless bootloader with serial Feb 14 18:49:32 That would be a good thing. Feb 14 18:49:36 <[g2]> or a serial connection to the bootloader without solder Feb 14 18:49:44 Did you get my comments about the bootloader UI using the two buttons? Feb 14 18:49:54 <[g2]> no Feb 14 18:50:08 <[g2]> what was your comment ? Feb 14 18:50:09 I updated the wiki. It's kinda like this. Feb 14 18:50:13 up up down down up down down Feb 14 18:50:23 There will be a special version for dyong. Feb 14 18:50:41 For Super Pursuit Mode? Feb 14 18:50:44 Cool. ;-) Feb 14 18:50:46 MY version will watch for the reset button being held while the unit is power-ed on. Feb 14 18:51:07 The LEDs will show that the loader is waiting for a mode choice. Feb 14 18:51:29 The on button changes the mode and the LEDS reflect that mode. Say, the disk lights are 0 and 1. Feb 14 18:51:36 There are four states of those two lights. Feb 14 18:51:59 So, we've got normal boot, failsafe boot, upgrade mode, and something else. Feb 14 18:52:11 Press reset again and away it goes. Feb 14 18:52:26 No messing with timed button presses. Feb 14 18:52:26 <[g2]> actually 8 Feb 14 18:52:37 If you using three lights? Feb 14 18:52:46 <[g2]> Red/Green on the Ready, and the two disk lights Feb 14 18:52:56 <[g2]> 12 Feb 14 18:52:57 There are four leds. 16 states. Feb 14 18:53:08 I don't want to make it too gawdawful complex. Feb 14 18:53:15 Do we need more than four options? Feb 14 18:53:15 <[g2]> I think ethernet is driven by hw Feb 14 18:53:34 Right. There are five leds on that edge. Feb 14 18:53:38 One is a like light. Feb 14 18:53:43 s/like/link/ Feb 14 18:53:55 Two for disk lights. One green ready and one red. Feb 14 18:54:07 dyoung knows all about this now Feb 14 18:54:22 I like that UI better thanmine. Feb 14 18:54:28 <[g2]> I think it's a wonderful I idea Feb 14 18:54:41 4 states is more than enough i think? Feb 14 18:54:42 Trouble is, the ready and red led are difficult to distinguish when the system is closed. Feb 14 18:54:55 <[g2]> If we can, I use the power-on button for choice toggling Feb 14 18:55:12 So, can we make a failsafe boot mode that allows the user to get telnet access and recover a corrupt filesystem? Feb 14 18:55:24 That's what I said. Feb 14 18:55:40 Hold reset to on power-up to enable the mode. Feb 14 18:55:45 Press power to choose. Feb 14 18:55:48 <[g2]> nod. Feb 14 18:55:49 N times Feb 14 18:55:53 Press reset to go. Feb 14 18:56:08 <[g2]> cycles through, or just hold-the power-on Feb 14 18:56:14 If the user idles for 10 seconds, we do a normal boot. Feb 14 18:56:17 <[g2]> for 2 seconds Feb 14 18:56:22 We could to that. Feb 14 18:56:31 the main issue with the failsafe mode is the size of the ixp400/ixp425_eth drivers Feb 14 18:56:32 I'd rather use reset so it is a little difficult to do. Feb 14 18:56:32 <[g2]> I don't like the idle Feb 14 18:56:40 <[g2]> unless it's like 1 minute Feb 14 18:56:45 <[g2]> even then Feb 14 18:56:46 Can't we recover the jffs without ethernet? Feb 14 18:57:12 thats what I was thinking Feb 14 18:57:13 Whatever. I think it is easy to choose a boot mode in 10 or 30 seconds. Feb 14 18:57:36 is pilfer the ixp400/ixp425_eth from the jffs2... Feb 14 18:57:42 Better yet, the failsafe mode gives then a USB serial console. Feb 14 18:57:44 it should be doable. Its already been done actually. Feb 14 18:57:53 <[g2]> I think reset to get in is fine, but but not to get out of the mode Feb 14 18:58:03 Why? Feb 14 18:58:13 <[g2]> cause it's a pain Feb 14 18:58:22 This is a failsafe mode. Feb 14 18:58:22 <[g2]> and not real benefit Feb 14 18:58:41 <[g2]> holding a button for 3-5 seconds is no mistake Feb 14 18:58:49 Well, that's an implementation detail that isn't necessary to hash out right now. Feb 14 18:58:54 We'll send it to the focus groups. Feb 14 18:58:56 <[g2]> nod. Feb 14 18:59:02 <[g2]> LOL Feb 14 18:59:23 I had a 'situation' this weekend where I lost control of the slug. Feb 14 18:59:31 Luckily, we had the parts to add a serial port. Feb 14 18:59:36 "lost control...." Feb 14 18:59:41 We'd been messing with the http servers. Feb 14 18:59:43 <[g2]> temporarialy Feb 14 18:59:47 is that like losing control of a car? Feb 14 18:59:55 Trouble is that the firmware was clobbering some of our changes. Feb 14 18:59:58 you slug goes slithering away on its own? Feb 14 18:59:59 <[g2]> a controlled skid of sorts :) Feb 14 19:00:05 icy patch. Feb 14 19:00:14 Leaving a trail of SlugSlime? Feb 14 19:00:16 ;-) Feb 14 19:00:18 busy changing CDs at the time Feb 14 19:00:20 <[g2]> was the ol'man impressed :) Feb 14 19:00:27 Hell yeah! Feb 14 19:00:43 You can go to his site. It's nothing much right now, but it is running on a slug. Feb 14 19:00:56 <[g2]> I'd love to Feb 14 19:01:19 http://www.dickshouse.net Actually, it's less than impressive. Feb 14 19:01:25 But it *is* running on a slug. Feb 14 19:01:41 he needs to block port 8080 Feb 14 19:01:43 <[g2]> Have you seen polkadotninja ? Feb 14 19:01:51 is it on a slug? Feb 14 19:01:57 <[g2]> nod Feb 14 19:02:10 the other guy too.,. bytered Feb 14 19:02:26 Nice. Feb 14 19:02:39 <[g2]> call him that ADMIN is up on the web ! Feb 14 19:02:53 Back to my story. The bog didn't come back with the telnet server, so we couln't get back to fix the errors. Feb 14 19:03:01 I know. Feb 14 19:03:20 <[g2]> DOH! Feb 14 19:03:21 You can only really mess is up with the password...which might be the default. Feb 14 19:03:25 doh! Feb 14 19:03:28 <[g2]> I'm in Feb 14 19:03:48 You're a real hacker. Feb 14 19:04:15 Are these intel libraries really libraries? Feb 14 19:04:18 <[g2]> ppl google for web cam Feb 14 19:04:26 <[g2]> with the default pw Feb 14 19:04:26 Or is it just the microcode that is a closed library? Feb 14 19:04:36 I've seen that webcam thing. Feb 14 19:04:39 It's kinda cool. Feb 14 19:05:11 webcam or web cam? Feb 14 19:05:22 webmam Feb 14 19:05:25 I dont see anything cool. Feb 14 19:05:51 You have to open your eyes first Feb 14 19:06:20 Do we know anyone familiar with the NPE libraries? Feb 14 19:08:01 theyre ugly as hell Feb 14 19:08:13 I must be blind. Feb 14 19:08:26 http://www.boingboing.net/2005/01/10/still_more_random_go.html Feb 14 19:08:45 <[g2]> can you guys bop over to #nslu2 for a minute ? Feb 14 19:10:43 no. Feb 14 19:10:52 I dont have access to that channel from here. Feb 14 19:12:01 Unles you mean nslu2-linux Feb 14 19:12:17 dyoung-web: #nslu2 Feb 14 19:12:41 nope, cant do it. Feb 14 19:12:47 lemme find another client. Feb 14 19:23:18 dyoung-web: What part of nslu2 didn't work for you? Feb 14 19:23:44 I'm on now using cgiirc; and it only has 3 channels programmed into it Feb 14 19:23:55 [g2]: That search doesn't seem to find that many slugs Feb 14 19:24:17 <[g2]> well there are only 2K downloads of the binaries Feb 14 19:24:33 Soon we'll own the world! Feb 14 19:24:45 <[g2]> but there are lots of standard units out there Feb 14 19:25:17 With the "Firmware Version" string, there are only three matches. Feb 14 19:25:41 One is linksys and one is a duplicate of the first hit. Feb 14 19:25:53 <[g2]> which string ? Feb 14 19:26:02 Hahahahahahahahahahaha Feb 14 19:26:03 user_pass.cgi Feb 14 19:26:06 Do this. Feb 14 19:26:10 In google. Feb 14 19:26:34 "Firmware Version" "Network Storage Link for USB 2.0 Disks" Feb 14 19:26:42 Select the third link. Feb 14 19:27:39 we can u pgrade him to unslung. Feb 14 19:27:48 rofl Feb 14 19:27:56 Ooh or dyoung's New Secret ZombieWare. Feb 14 19:28:35 It looks like there's no password at all. Feb 14 19:28:47 I wonder if its a nslu2 though Feb 14 19:28:57 It says it is. Feb 14 19:28:58 try go to http://www.linksysdata.com/ui/ Feb 14 19:29:06 <[g2]> ~whois linksysdata.com Feb 14 19:29:06 [g2]: bugger all, i dunno Feb 14 19:29:40 Or it's a compendium of every UI. Feb 14 19:30:01 Administrative Contact: Cisco Linksys. Feb 14 19:30:09 heh Feb 14 19:30:23 <[g2]> I went to the update page Feb 14 19:30:32 me too Feb 14 19:30:40 <[g2]> did you browse ? Feb 14 19:30:48 IOW, there aren't that many folks on the net...yet. Feb 14 19:30:51 yeah, but I dont have a firmware image handy. Feb 14 19:31:02 doo ood Feb 14 19:31:16 <[g2]> I didn't browse Feb 14 19:31:36 There weren't files. It also says that apache is serving the pages. Feb 14 19:31:38 <[g2]> we should probably send e-mail to the web admin or /. Feb 14 19:31:54 <[g2]> maybe a honeypot Feb 14 19:32:06 * ka6sox-away must be on /ignore :) Feb 14 19:32:13 <[g2]> for h4x0r5 like us Feb 14 19:32:25 hehe Feb 14 19:32:34 <[g2]> doo ood Feb 14 19:32:38 Nah I think a legit page for linksys/cisco to test drive the interfaces. Feb 14 19:32:48 So, do we know anyone familiar with the NPE libraries? Feb 14 19:32:53 <[g2]> why on the web ? Feb 14 19:33:03 for the focus groups ;-) Feb 14 19:33:21 <[g2]> there listening in this channel and have already started Feb 14 19:33:26 For support engineers. Feb 14 19:33:32 bewoolie, how familiar is famliar? Feb 14 19:33:50 <[g2]> beewoolie, there's a binary that needs to get downloaded and some stuff to be enabled in the chip Feb 14 19:34:08 <[g2]> and we may be violating Intel's license Feb 14 19:34:12 AFAICT, the only 'binary' part is the microcode. Feb 14 19:34:23 <[g2]> right Feb 14 19:34:27 There isn't really a library per-se. There is no link library. Feb 14 19:35:02 <[g2]> the bootload would need to dl the microcode for it to work Feb 14 19:35:06 ka6sox-away Feb 14 19:35:09 OK. Feb 14 19:35:21 dyoung-web Feb 14 19:35:25 I'm fine with the microcode. Feb 14 19:35:34 you were calling earlier? Feb 14 19:36:00 Is the microcode sensitive to BE/LE? Feb 14 19:36:06 It doesn't seem so. Feb 14 19:36:14 yeah...took most of the day to retrieve my Auto from the side of the mountian. Feb 14 19:36:22 It seems like it must be the C code that has some sort of endian dependency Feb 14 19:36:29 tonight its going to rain..heavily Feb 14 19:36:56 It's dangerous where you live. Feb 14 19:37:12 yes. Feb 14 19:37:19 steep and dangerous. Feb 14 19:37:23 <[g2]> got your car back ? Feb 14 19:37:29 <[g2]> BE Feb 14 19:37:30