**** BEGIN LOGGING AT Mon May 05 02:59:57 2008 May 05 03:24:04 anyone know why I'd see the "not have eRcOmM" message? I have redboot modified for 16MB access, and using slugosbe image built. Using "slugimage" (with -M) verifies the trailer is there. May 05 04:42:43 rwhitby: when you return, if you could send me the 16MB-detected RedBoot that works for you. I May 05 04:42:51 I'll then modify wiki with proper instructions May 05 09:12:31 03rwhitby * r107 10slugimage/trunk/slugimage: Removed the unused Loader config partition May 05 13:44:55 03rwhitby * r109 10slugimage/trunk/slugimage: Added code to detect unpacking of apex images without a Ramdisk and to not include the header. May 05 13:48:56 03rwhitby * r110 10slugimage/trunk/tests/ (12 files in 5 dirs): Added unpack and repack testing for all test cases May 05 19:47:42 RobNC, did you get the 16MB slugimage built? May 05 20:10:24 So can the Slugs be upgraded to 16MB of flash too? May 05 20:10:48 never mind, found the wiki page May 05 23:11:26 ka6sox: sorry I was away. Got the slugimage built, need to confirm if I can disassemble and re-assemble using slugimage. May 05 23:11:52 joshin: yes called Obese Slug (or FatFlash) May 05 23:13:39 seems like "make openwrt" had problem, for some reason. says to re-run make with V=99 to see what's going on May 05 23:14:32 yeah something wrong with "make -C package/wpa_supplicant compile" May 05 23:16:03 ah. May 05 23:16:34 but note that's different than what I did earlier. Earlier I did "make slugosbe" and tried that image. May 05 23:17:01 right May 05 23:17:15 this one was a Obese Slug image of???? May 05 23:18:13 doing "make update" now... 21 revs since last night. Oh obese slug, or more correctly, FatFlash slug, is automatically built as part of the process for "make slugosbe" and le May 05 23:18:58 gonna try it once update is done May 05 23:21:24 okay so in /temp/deploy/images you have a 16flash version? May 05 23:23:41 crap same problem - 16 bytes too big. Must be a problem in either the build process or slugimage. May 05 23:23:57 yep - it's slugosbe-4.10-alpha-nslu2-16mb.bin May 05 23:25:20 rwhitby is your man for that stuff. May 05 23:26:16 ka6sox yeah I don't want to bug him too much. Unfortunately, he has probably forgotten more than I know about this! May 05 23:26:36 too true... for me too! May 05 23:30:16 Interesting thing is that I have a complicated setup. Oh and the "eRcOmM" message is the trailer in flash. My redboot seems like it has some of the mods necessary, but not all of them, to access 16MB. Rod claims to have it working via hacking RedBoot. My RedBoot does see all 16MB of the flash. May 05 23:30:25 But without the trailers in the right place, no dice. May 05 23:34:53 At least I confirmed my extra 3 USB ports are working. Problem must be with the Debian etch, it wouldn't recognize my external USB HDDs (self-powered). When I flashed with D-I, the kernel sees it but for some reason my debian install doesn't recognize those other ports. Weird... May 05 23:35:51 ya, that is wierd. May 05 23:37:37 I'd like to get away from debian b/c I can't help if that's too bulky b/c designed for desktops with gobs of memory. I'd like to try either openwrt (uclibc) or slugosbe, but for fatslug. BTW, I got upslug2 working but same issue with eRcOmM - so my redboot is broken it seems. May 05 23:37:54 (in regards to accessing 16MB and letting the 2nd stage apex work with that). May 05 23:38:35 (brb) May 05 23:39:24 we should ask rwhitby when he gets around about this...I think that your FIS may be the problem. May 05 23:45:53 yeah but I really have two problems: 1) redboot is not detecting trailer, 2) my build of slugosbe FatFlash, for some reason, cannot be disassembled and assembled back. Problem 2) is a fundamental issue on either the compilation process or the slugimage disassembly/reassembly. May 05 23:46:29 right..I think that there are exactly *2* of these in the world..... May 05 23:46:38 or should I say 10 May 05 23:46:49 pretty soon there will be 3... May 05 23:47:13 okay back laters...I have a 6hr drive now May 05 23:47:24 ugh, where does work have you now? May 05 23:48:56 ok good luck and have a good one May 06 00:04:28 RobNC: morning May 06 00:12:29 Hi, rwhitby. evening my time :-) May 06 00:12:37 ~ugt May 06 00:12:38 from memory, ugt is Universal Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html May 06 00:12:59 Did you try the latest slugimage? May 06 00:12:59 ah thanks! :-) didn't know that. May 06 00:13:15 after "make distclean; make slugosbe"? May 06 00:16:29 RobNC: nm - I have now reproduced your problem locally. May 06 00:16:41 yes, the disassembly of the image via slugimage still says it's 16-bytes too large to re-assemble. However, if you could send me the RedBoot you got to work, I'll update the wiki and do diffs to compare yours versus the one on the wiki. It seems that mine is partially modified, enough to see the 16MB flash but not enough to look for sercomm trailer at end of flash perhaps. that's why I get the "not have eRcOmM" messa May 06 00:17:08 Sorry about that - was hoping it was something I was doing. May 06 00:17:38 don't be sorry - finding bugs is a good thing May 06 00:17:51 :D May 06 00:18:18 heh so it seems I am truly in the bleeding edge :-) May 06 00:20:16 RobNC: 385d5a9b4263570885b83f13b3a904eb redboot.bin May 06 00:20:30 rwhitby: would it be easier to build a new redboot from sources than byte hack? Seems like RedBoot has to be built by makes from svn anyway, unless it's just some binary dump. May 06 00:21:27 we don't build redboot in our build process, cause it requires the Intel proprietary microcode May 06 00:21:37 that's why it's a binary modification May 06 00:22:24 Ah I thought that was only for NPE, didn't realize redboot wasn't built without a plugin-type module (ala nvidia and others) May 06 00:23:54 Not to stray too far off topic, but I believe the latest kernels have open-source version of IXP eth driver. Is it now possible to build redboot w/opensrc driver or even apex with opensrc driver? May 06 00:25:13 both are possible - it just needs someone to do the work (which is not trivial) May 06 00:25:30 oh I see, can't even imagine. May 06 00:26:27 BTW, we have a svn repo where we can build redboot (it's called sluggo), but we haven't tested it on a real device yet. May 06 00:27:57 With redboot-16MB.bin I'll xdiff and update wiki. May 06 00:29:18 Dumb question but can I modify RedBoot from within RedBoot? I.e., does RedBoot copy itself to RAM and then run from RAM? May 06 00:30:02 RobNC: I have a paper copy here of the mods I made - 9 different places May 06 00:30:15 yes, you should be able to install a new redboot from inside redboot May 06 00:30:25 flash offsets of my modifications are: May 06 00:31:25 cf54, cfb0, d030, d048, doe8, d104, d110, d11c, d128, d5b0, d934, d944 May 06 00:31:43 let me know if you see diffs other than at those locations. May 06 00:31:59 once you've got the wiki page up, I'll check it and annotate it with comments May 06 00:32:03 ok, will do, compare to normal 8MB flash redboot image May 06 00:32:38 I made some mods initially but it seems that it needs polishing. May 06 00:32:47 I did it by looking at the redboot source code we have, then working out what needed to change in the C code, then finding those bits in the disassembly, then modifying the binary. May 06 00:33:27 wow great job! May 06 00:34:11 It all comes down to changing FLASH_SIZE in the source code, and working through all the places that is used. May 06 00:35:26 Being able to redbuild RedBoot from source helped, cause I didn't have to disassemble from binary, I could just compile C to assembly to find the right places. May 06 00:36:31 Anyway, I might have time this arvo to look at slugimage. In the meantime, you might be able to just cut a block off the end of Flashdisk and try that. May 06 00:37:38 that at least helps. Thanks for your efforts, indeed! Didn't know if flashdisk was sectorized and size as prefix. May 06 00:38:43 Flashdisk doesn't have a size prefix. Ramdisks do, but slugimage adds that. May 06 00:38:59 You can look at the start of the unpacked file to see if the size has been mistakenly put there. May 06 00:39:21 So is "Flashdisk" same as "Ramdisk"? May 06 00:39:36 no May 06 00:39:54 ramdisk always starts at a specific location, and has a size header. May 06 00:40:12 flashdisk (jffs2) cannot start at that location, it needs to start at the next block, so it can start from the start of the block May 06 00:40:35 Of course, just realized that. I meant ramfs. May 06 00:41:20 Debian images have kernel and ramdisk. SlugOS and OpenWrt images have kernel and flashdisk. May 06 00:41:36 i.e., the filesystem uncompressed into ram. Ah ok that makes sense now - thanks! May 06 00:43:52 I'll try out mods and get back to you. Thanks again for help with RedBoot FatFlash mods. I'll also have to look at slugimage to see diffs between 8MB and 16MB. You'd think that only flashdisk would be larger, since 32-bit addressing and no extra space by anything else. May 06 00:50:45 yeah, the bug is in slugimage - the flashdisk is the same for both May 06 00:50:57 (it's only built once and packed into an image twice) May 06 00:53:45 I see. perhaps process of packing twice has an error. Might matter for devices that can do 2x, 4x, 8x 8MB flash. BTW, I wonder if 32MB flash would work for NSLU2 (missing high address bit). It seems my company's software folks are greedy and want more than they need. :-) May 06 00:56:47 no, slugimage is called twice, once for 8mb once for 16mb - the same flashdisk image is used for each May 06 00:57:16 it's definitely a slugimge bug May 06 01:20:00 rwhitby: just did xxd and doing diff shows huge differences between my redboot and yours, as well as downloaded 4.8-beta-slugosbe slugimage disassembled RedBoot. If I concentrate on just the differences shown above, looks like the addresses are off. Are you using 16-bit addressing or 8-bit? One difference I happened to see (in text window of xxd) is eCos: Feb 3 2004 and the other is eCos: July 4, 2007. May 06 01:20:52 The Feb 3 2004 is what is found in the downloaded "slugosbe-4.8 beta" RedBoot disassembly with slugimage. May 06 01:23:29 interesting May 06 01:23:46 we get the redboot for the built images from linksys firmware images May 06 01:24:01 Weird, hmm, even the built version I made has the Feb 3 2004 date. May 06 01:24:01 where did you get the July 4 one from ? May 06 01:24:26 July 4 is yours. Look at location 0x26a0 thereabouts May 06 01:25:46 What made me look for this is the fact that it seems there are thousands of lines differences (when converting both your 16mb redboot to hex, and the one I made, as well as the one from the 4.8 download page). May 06 01:26:46 back later ... May 06 01:26:59 ok thanks rwhitby. ttyl May 06 02:35:21 Anyone here ever built u-boot for NSLU2? Looks like it builds, had to replace references to IPL_ixp400NpeLibrary-2_1.zip with IPL_ixp400NpeLibrary-2_4.zip however... May 06 02:56:52 hmm, u-boot loads, just overwrote redboot (note: I have JTAG - this is a *MUST*!!) **** ENDING LOGGING AT Tue May 06 02:59:57 2008