**** BEGIN LOGGING AT Fri Oct 13 02:59:57 2006 Oct 13 06:43:53 03oleo * r4224 10optware/trunk/ (3 files in 2 dirs): transmission: bug fixes in scripts Oct 13 11:34:46 mtn: connecting to monotone.nslu2-linux.org Oct 13 11:34:46 mtn: network error: failed to connect: Connection refused Oct 13 11:35:04 Anybody else getting this? Oct 13 11:35:29 lemme see Oct 13 11:35:38 The box is pingable at least. Oct 13 11:35:56 And has quite a few ports open:22/tcp open ssh Oct 13 11:35:56 25/tcp filtered smtp Oct 13 11:35:56 80/tcp open http Oct 13 11:35:56 111/tcp open rpcbind Oct 13 11:35:56 113/tcp open auth Oct 13 11:35:57 135/tcp filtered msrpc Oct 13 11:35:59 139/tcp filtered netbios-ssn Oct 13 11:36:01 443/tcp open https Oct 13 11:36:03 445/tcp filtered microsoft-ds Oct 13 11:36:07 828/tcp open unknown Oct 13 11:36:09 1984/tcp open bigbrother Oct 13 11:36:11 2401/tcp open cvspserver Oct 13 11:36:11 Its in the middle if a sync atm Oct 13 11:36:13 4444/tcp filtered krb524 Oct 13 11:36:13 VoodooZ: gimme a quick mtn ommand line to test, haven't got the URLs handy Oct 13 11:36:18 yikes! port 135/139 open! that's bad! Oct 13 11:36:32 I don't know. I just did make update under head. Oct 13 11:36:35 let me see... Oct 13 11:37:33 mtn update -b org.nslu2-linux.dev??? Oct 13 11:37:50 I'm just fishing ths out of the Makefile though so.... Oct 13 11:38:09 I do not have anything local against nslu2-linux.dev at this time. Oct 13 11:38:16 or use the command 'status' Oct 13 11:39:07 try now Oct 13 11:39:39 mtn: error: I/O failure while talking to peer monotone.nslu2-linux.org, disconnecting Oct 13 11:40:15 and it doubt it's my connection as I'm connecting from work to my home box which worked all week Oct 13 11:40:23 different error at least. :) Oct 13 11:40:31 I know exactly what the cause is. Oct 13 11:40:41 we're working to fix the problem. Oct 13 11:41:07 basically when a sync is being run, and a client connects to the server, the sync kills the serve. Oct 13 11:41:11 Ok, thanks dyoung_ Oct 13 11:41:13 and then the sync hangs Oct 13 11:41:19 Just wanted to let you guys know. Oct 13 11:42:00 okay the sync finshed Oct 13 11:42:02 try now? Oct 13 11:43:45 looks better. Oct 13 11:43:47 thanks. Oct 13 11:44:01 np. Oct 13 11:44:12 we'll get that resolved pretty soon hopefully. Oct 13 11:44:14 the #oe guys told me somebody got my pocketsphinx package in mtn. should I expect it to show up inour head too. Oct 13 11:44:40 what? I thought you just resolved it? it's updating... Oct 13 11:44:45 Yeah it syncs every 5min or so. Oct 13 11:44:51 Yeah, its a recurring problem though Oct 13 11:45:06 I see. Oct 13 11:45:18 the permanent solution is being crafted this week. Oct 13 11:45:42 (the solution is to unbundle the servingi and syncing features from each other. Oct 13 11:45:45 ) Oct 13 11:46:39 VoodooZ: I've been meaning to ask you if things work for you under 64-bit these days? Oct 13 11:48:41 ACtually, I changed company in April so I no longer have access to the nice SMP boxes. Oct 13 11:49:09 I'm planning on upgrading my home box soon though so I'll be getting an AMD X2 probably. Oct 13 11:50:02 hehe Oct 13 11:50:08 I just got a X2 Oct 13 11:50:12 I'm happy. Oct 13 11:50:17 I'm running it in 32bit tho Oct 13 11:50:28 since I have vivid memories of your 64-bit issues. Oct 13 11:50:32 nice. Oct 13 11:50:55 With Intel finally getting their act together I'm tempted do get a Core2 but I'm sticking with AMD thru the "hard" times. Oct 13 11:51:07 I did a compile comparison for the optware toolchain X2 4200 versus Dual Xeon PE1800 3.0Ghz. Oct 13 11:51:39 The X2 4200 wins by 3 min. 28min on the Xeon 3Ghz and 25min on the X2 Oct 13 11:51:43 wow! Oct 13 11:51:49 What was your -j command? Oct 13 11:51:57 I didnt use one. ;-) Oct 13 11:52:12 really? Oct 13 11:52:15 Yep Oct 13 11:52:24 I had to do it 3 times just to make sure I wasnt hallucinating Oct 13 11:52:36 so just the default build with no PARALELL shit? Oct 13 11:52:42 Righto. Oct 13 11:53:03 Nice! You should try it -j 3 or more for fun. Oct 13 11:53:15 IT might a nice diff on the opterons Oct 13 11:53:26 Hmm, thats a interesting experiemtn too. Oct 13 11:53:32 I'll fire one off right now. Oct 13 11:53:50 3, 4,5 & 6 should be sufficient I would think. Oct 13 11:54:06 although I was testing on a quad system so... Oct 13 11:54:29 Do you know the proper Argument for setting it up? Oct 13 11:54:51 It's PARALELL something in local.conf Oct 13 11:55:37 dyoung_, what's the PE1800 based on? not the new Core2? Oct 13 11:55:37 Oh, I was testing just building the optware toolchain, outside of oe. I didnt quite get that far yet on this machine. Oct 13 11:55:58 oh i see. Oct 13 11:56:00 The PE1800 is dual Xeon 3.0Ghz; those were "HT" CPU's Oct 13 11:56:22 ha! that's probably why. they were groslly inefficient Oct 13 11:56:40 The new quad cores are going to kick as at highly parallel stuff though Oct 13 11:56:41 I just tried with -j 4 Oct 13 11:56:51 so? Oct 13 11:56:59 dunno, ask me in 20min or so. ;-) Oct 13 11:57:03 hehehe. Oct 13 11:57:23 I was going to say. wow! I need another coffeee! :) Oct 13 11:57:47 dyoung_, btw, do you know if there's any progress on getting glibc 2.4/2.5 running on the slug? Oct 13 11:58:00 it would be nice to have NPTL instead of pthreads. Oct 13 11:58:30 I know it sounds crazy for a slug but efficiency is important. Oct 13 11:58:30 Hmm, I think I saw nail mutter some choice words; but I didnt really hear anything as far as progress. Oct 13 11:58:59 yeah. I talked to him quickly yesterday and it wasn't looking good. Oct 13 11:59:24 koen seemed to be having some success with 2.5 but that's a totally different machine. Oct 13 12:00:08 I have submitted squashfs-lzma to the nslu2-linux ml. What's currently in use on the slugs? Oct 13 12:02:36 in the flash? Oct 13 12:03:57 dyoung_: yes Oct 13 12:04:29 jffs2 Oct 13 12:04:43 likewise, nice! Is it really demanding processor wise? Oct 13 12:05:06 VoodooZ: I intend to compare it speedwise against JFFS2 next week. Oct 13 12:05:25 good! Let us know or update the wiki as this would be real useful. Oct 13 12:05:37 by simply doing a "cp -ra / /dev/null" or something Oct 13 12:05:42 Do you have approx compression ratios? Oct 13 12:07:29 squashfs could be used as a live backup for workstation too. I like it. Oct 13 12:08:05 I just wonder how big of a size difference it makes in average. Oct 13 12:13:32 VoodooZ: hold on, I will build both jffs2 and squash-lzma to compare Oct 13 12:13:41 cool Oct 13 12:13:59 It could definitely help with our space woes. Oct 13 12:14:59 -rw-r--r-- 1 leon leon 8257536 2006-10-13 14:09 clock-image-ixp4xx-20061013120720.rootfs.jffs2 Oct 13 12:15:01 -rwx------ 1 leon leon 5312512 2006-10-13 14:09 clock-image-ixp4xx-20061013120720.rootfs.squashfs Oct 13 12:15:03 -rw-r--r-- 1 leon leon 4587520 2006-10-13 12:52 subtitle-image-ixp4xx-20061013105212.rootfs.jffs2 Oct 13 12:15:04 -rwx------ 1 leon leon 3932160 2006-10-13 12:52 subtitle-image-ixp4xx-20061013105212.rootfs.squashfs Oct 13 12:15:28 the first image has more data (besides just executables and libraries). there is the biggest win. Oct 13 12:16:32 Unless of course the slugos determine to be unpractical. I don't know that much. Oct 13 12:16:49 VoodooZ: the squashfs in the listing above uses actually LZMA compression Oct 13 12:17:05 * VoodooZ is talking with his boss... Oct 13 12:17:07 VoodooZ: I wonder if someone uses LZMA with JFFS2 Oct 13 12:22:26 not sure. Oct 13 12:22:33 I've been out of the loop in the last year. Oct 13 12:22:50 just started again. what version of slugos are you using? Oct 13 12:23:36 I guess I could finding the rootfs script to see if there's compression in there. Oct 13 12:23:55 How is that lzma done normally? seperately or is it part of jffs2? Oct 13 12:29:34 It must not pass the -j option along Oct 13 12:29:54 is there a way to gobalise that? Oct 13 12:41:39 euh? Oct 13 13:02:23 dyoung_: You mean: how do I enable parallel builds with bitbake invoked makes? Oct 13 13:03:01 dyoung_: if so, put PARALLEL_MAKE = "-j4" in local.conf Oct 13 13:03:38 VoodooZ: squashfs with lzma would compress fixed-size blocks of files using LZMA Oct 13 13:04:06 i see. Oct 13 13:04:08 VoodooZ: same for JFFS-LZMA (but I have seen no working patch for that yet). Oct 13 13:04:35 I wasn't even aware we were using compression. Oct 13 13:04:44 that's how little I know! :) Oct 13 13:04:45 VoodooZ: it is also possible to compress existing initrd images with LZMA, but then the whole IMAGE is compressed, not the filesystem file content itself Oct 13 13:05:31 VoodooZ: well, I am not really a slug(os) user, I just hang around here because I have a custom board that "looks like" the slug or IXDP425, so the software effort has 99% overlap Oct 13 13:06:11 VoodooZ: I commit useful stuff back to openembedded (like squashfs-lzma) and come around here to ask whether you guys actually use it or not (it helps to get feedback) :-) Oct 13 13:06:13 good. I guess the guys did a good job. Overlap is good. Oct 13 13:06:21 definitely. Oct 13 13:06:59 Make sure the big guns know about it though as they are the ones making the big architectural decisions. (rwhitby, Nail, dyoung, [g2], etc...) Oct 13 13:08:17 * [g2] puffs up and pounds on chest ..... ooohh ooh oooh Oct 13 13:08:41 shhh - one of the big boys is listening in :-) Oct 13 13:08:59 <[g2]> no I just saw the blinking window Oct 13 13:09:49 JFFS2 with LZMA compression is nice, kernel compressed with LZMA is nice, but I have not seen working patches yet (in the kernel case, only for MIPS and x86). Oct 13 13:10:07 * [g2] glances up... likewise I've been running squahsfs for a month or so Oct 13 13:10:19 <[g2]> actually a squashed debian Oct 13 13:10:22 So I'll probably stick with squashfs-lzma with a JFFS2 overlay using something as unionfs Oct 13 13:11:12 <[g2]> likewise I was mounting it from a tiny <1MB jffs2 Oct 13 13:11:15 [g2]: I am running squashfs-lzma as rootfs, but I mean a writable filesystem. Oct 13 13:11:35 [g2]: JFFS2 with LZMA compression. Oct 13 13:11:40 <[g2]> likewise yeah you can just bind mount/unionfs overtop Oct 13 13:12:03 <[g2]> overtop of the squashfs that is Oct 13 13:12:52 [g2]: yep, but further savings from that would be a lzma-compressed kernel, and a jffs2 that used lzma (either through mkjffs2 and optionally even run-time). Oct 13 13:13:08 [g2]: gaining 10% Oct 13 13:13:22 <[g2]> is that compared to squashfs 3.0 ? Oct 13 13:13:28 <[g2]> or jffs2 Oct 13 13:13:54 10% against squashfs 3.0 Oct 13 13:13:57 <[g2]> likewise btw I think those are great idea and certainly things to be investigated made working Oct 13 13:14:03 more against jffs2, see here: Oct 13 13:14:06 -rw-r--r-- 1 leon leon 8257536 2006-10-13 14:09 clock-image-ixp4xx-20061013120720.rootfs.jffs2 Oct 13 13:14:07 -rwx------ 1 leon leon 5312512 2006-10-13 14:09 clock-image-ixp4xx-20061013120720.rootfs.squashfs Oct 13 13:14:09 -rw-r--r-- 1 leon leon 4587520 2006-10-13 12:52 subtitle-image-ixp4xx-20061013105212.rootfs.jffs2 Oct 13 13:14:11 -rwx------ 1 leon leon 3932160 2006-10-13 12:52 subtitle-image-ixp4xx-20061013105212.rootfs.squashfs Oct 13 13:14:26 [g2]: the squashfs images are LZMA compressed here. Oct 13 13:15:46 [g2]: I yesterday added lzma compression to squashfs 3.1r2, I will concentrate on other work first. Oct 13 13:16:09 <[g2]> likewise I'd think if it's not already there, supporting LMZA on the kernel image would be a decent win Oct 13 13:16:33 <[g2]> as it would allow larger kernel in a fixed amount of space (as in the nslu2 case) Oct 13 13:17:00 <[g2]> iirc the OpenWRT guys have been doing this for a while Oct 13 13:17:08 [g2]: that's MIPS right? Oct 13 13:17:28 <[g2]> yeah, mainly, but they run a lots of arches Oct 13 13:17:34 [g2]: yes, and i386 patches are all around, but no ARM yet. Oct 13 13:17:58 <[g2]> dunno haven't checked, there's an IXP port Oct 13 13:18:36 <[g2]> IXP port for OpenWRT, not LMZA in the kernel decompressor (but haven't check that either) Oct 13 13:19:36 <[g2]> likewise just pinged someone in #madwifi Oct 13 13:19:54 <[g2]> asking about LMZA for ARM/XScale Oct 13 13:20:21 [g2]: Jeremy from bitsum.com was doing JFFS2 LZMA support in July on freewrt. That should be arch independent (fs) code. Oct 13 13:22:05 [g2]: I know, but I didn't log madwifi, and you said someone asked for it there? Oct 13 13:22:32 <[g2]> yeah Kaloz Oct 13 13:22:53 <[g2]> I think he was in OpenWRT before I was Oct 13 13:23:31 <[g2]> I think he's one of the guys working on getting the Gateway box running OpenWRT Oct 13 13:23:45 [g2]: hmm, I should run some IRC live topic grepper, any recommendations for such a tool? Oct 13 13:23:57 s/topic/chatter Oct 13 13:24:11 <[g2]> I think that's was riker does :) Oct 13 13:24:30 <[g2]> ~bot Oct 13 13:24:31 I ain't no stinkin' bot. I am a finely tuned and hand crafted tool. Oh wait... I guess I am a bot (that you should not abuse). Oct 13 13:24:54 <[g2]> the bot is probably logging 50 channels Oct 13 13:25:18 <[g2]> but IMHO I think it's just better to know the ppl and the projects Oct 13 13:26:16 <[g2]> likewise have you played with > 64MB config btw on the IXP ? Oct 13 13:26:33 [g2]: Only on the IXDP425, yes Oct 13 13:26:39 <[g2]> that's fine Oct 13 13:26:47 <[g2]> it's perfect actually :) Oct 13 13:26:58 <[g2]> did you run into any issues with that ? Oct 13 13:27:00 [g2]: I am notably bad at remembering names just from IRC. Oct 13 13:27:23 <[g2]> room for growth or an xchat plugin :) Oct 13 13:27:45 * VoodooZ dreams of having more RAM on his slugbot Oct 13 13:27:47 [g2]: no issues that I can remember, are you looking for any? Oct 13 13:28:08 <[g2]> no they are looking for me :) Oct 13 13:28:37 <[g2]> which kernel are you on btw if you can say Oct 13 13:28:46 [g2]: haha, well I did some work on the SDRAM controller configurations overall several board revisions and with non-Intel certified SDRAM configs. Oct 13 13:28:58 s/overall/over Oct 13 13:29:25 [g2]: I am using OE .dev 2.6.17 ixp4xx-kernel but with most of the patches disabled. Oct 13 13:29:27 <[g2]> yeah I remember you were asking what the timings were Oct 13 13:30:04 [g2]; yup, but we use a x32 SDRAM, so it took me a bit to get the bits right (and to understand their meaning). Oct 13 13:30:26 [g2]: The hardware design guidelines helped me more than the IXP4xx developer guide, btw. Oct 13 13:30:31 <[g2]> memory controllers are usually that way the first time around :) Oct 13 13:30:46 [g2]: what issue do you have? Oct 13 13:31:22 <[g2]> I see some instability, but I've only got one board Oct 13 13:32:09 [g2]: I currently have instability as well, and we think it is due to unmatched SDRAM trace lengths. Oct 13 13:32:12 <[g2]> likewise do you have a MACH_ID and are you planning on pushing the config/changes upstream ? Oct 13 13:32:31 [g2]: no (but will) resp. yes Oct 13 13:32:52 <[g2]> no repository ? Oct 13 13:34:10 [g2]: I will open some epositories in february at the company I work for (I do have an employer besides my freelance work). Oct 13 13:34:35 <[g2]> Feb '07 ? Oct 13 13:34:37 [g2]: I first need to make sure non-GPL code does not leak away. Oct 13 13:35:19 <[g2]> that make sense Oct 13 13:35:20 [g2]: yes, but I can send you anything you want right now. I have a modified 1.92 redboot, and a small patch for our machine in the kernel. Oct 13 13:35:41 <[g2]> thx I'm running 2.02/3 Oct 13 13:36:13 <[g2]> I can load -m disk hda1:/boot/zImage Oct 13 13:36:23 <[g2]> where hda1 is ext2 on CF Oct 13 13:37:04 <[g2]> but we should probably want to coordinate upstream patches some Oct 13 13:37:18 [g2]: nice. where are the instabilities? in redboot, linux? spontaneous hangs? does redboot show the GDB stub output? Oct 13 13:38:02 <[g2]> last I checked, if mem=64M is passed in the system is 100% stable Oct 13 13:38:35 <[g2]> with mem=128M, the system is fine in Linux under light load, but under heavy load it locks up Oct 13 13:38:39 <[g2]> like bonnie++ Oct 13 13:39:21 [g2]: do you have a hardware debugger (I have a BDI2000 here)? Oct 13 13:39:23 <[g2]> so I know the HW is all fine (except for possibly 1 address line) Oct 13 13:39:38 [g2]: did you check for memory aliasing (using redboot)? Oct 13 13:40:04 <[g2]> I think so, but I it wouldn't hurt to check again Oct 13 13:40:21 [g2]: x16 SDRAM, or x32? And how many? Oct 13 13:40:25 <[g2]> x16 Oct 13 13:40:31 <[g2]> two chips Oct 13 13:40:38 <[g2]> 32x16 iirc Oct 13 13:41:06 2 chips of 32Mx16bits makes 128 Mbytes, yes. Oct 13 13:41:27 <[g2]> :) Oct 13 13:42:00 [g2]: sanity check :-) Oct 13 13:42:12 <[g2]> that's fine :) Oct 13 13:43:17 [g2]: what's your SDRAM_CONFIG set to (let me just figure that out myself, hold on while I open the PDFs) Oct 13 13:43:38 * NAiL looks at squashfs patches Oct 13 13:47:37 NAiL: thanks Oct 13 13:48:28 <[g2]> likewise I was just starting to look at the issue again Oct 13 13:48:49 <[g2]> it turns out my 128MB box is my Internet web server/trac server Oct 13 13:49:00 <[g2]> uptime ~180 days Oct 13 13:49:43 <[g2]> I'm actually switching network service providers this weekend Oct 13 13:49:54 <[g2]> so that box will get freed up soon Oct 13 13:51:44 <[g2]> likewise I've got 2 256MB boxes, but those are chip stacked Oct 13 13:52:32 [g2]: SDR_CONFIG[5:0] = 011100 Oct 13 13:52:52 [g2]: 128 MB LOFT, right? Oct 13 13:52:56 <[g2]> yeah Oct 13 13:53:02 <[g2]> a test box Oct 13 13:53:18 [g2]: the SDR_CONFIG matches? Oct 13 13:54:16 [g2]: My best guess for the instability would be power dip (under stress load), or critical timing due to unmatched lengths to SDRAMs. Oct 13 13:54:51 <[g2]> it's not power as the thing is wall plugged Oct 13 13:55:02 likewise: patches look good by me Oct 13 13:55:12 Can't look into merging them until after the weekend though Oct 13 13:55:21 so buzz me next week ;) Oct 13 13:55:24 <[g2]> it's not unmatched lengths as there's only 1 extra line between the 64 and 128 config Oct 13 13:55:50 <[g2]> so it could only be 1 line if it was taht (the high-order address) Oct 13 13:56:00 <[g2]> s/taht/that/ Oct 13 13:56:00 [g2] meant: so it could only be 1 line if it was that (the high-order address) Oct 13 13:57:19 NAiL: ok, if I think about it :-) Oct 13 13:58:15 [g2]: did you check that the SDR_CONFIG matches your SDRAM parts #ROW, #COLUMNS? Oct 13 13:58:42 <[g2]> likewise last time I looked into yes Oct 13 13:58:59 <[g2]> it's been like a year since I last checked Oct 13 14:02:31 * [g2] trying test on 256MB Oct 13 14:02:47 <[g2]> likewise thx for the input btw Oct 13 14:04:08 [g2]: I'll let you know what our instability arises from. If I cool the IXP4xx from 48 Celsius to below 35, everything works reliably.... Oct 13 14:04:52 <[g2]> that sounds like a solder joint issue Oct 13 14:06:37 [g2]: yes, I'll probably have to run it through an X-ray scan soon. Oct 13 14:07:14 <[g2]> that'd be cool Oct 13 14:09:27 have to go to a nearby shop, bbs Oct 13 14:51:17 03bzhou * r4225 10optware/trunk/make/py-usb.mk: py-usb: upstream upgrade to 0.3.5 Oct 13 19:31:22 03blaster8 * r468 10kernel/trunk/patches/2.6.19/ (4 files): Bump to 2.6.19-rc2, clean ixp_npe driver, fixup led-cpu-trigger patch Oct 13 19:42:04 03blaster8 * r469 10kernel/trunk/patches/2.6.19/defconfig: Update defconfig for 2.6.19-rc2 Oct 13 22:38:58 03blaster8 * r470 10kernel/trunk/ (4 files in 2 dirs): Update post-2.6.19-rc1 ixp_npe cleanups and docs Oct 13 23:38:22 is there some way I can see what the load on my slug is? i can't find a package containing top and uptime complains of a missing file **** ENDING LOGGING AT Sat Oct 14 02:59:57 2006