**** BEGIN LOGGING AT Sun Mar 06 23:59:56 2005 Mar 07 01:23:53 morning kas11 beewoolie-away Mar 07 01:24:02 morning. Mar 07 01:24:03 cya from the mine in a few Mar 07 01:24:28 Hi ka6sox-away ;) Mar 07 01:24:37 k Mar 07 01:25:40 headed into the office. Mar 07 01:25:53 go get 'em Mar 07 01:27:09 <[g2]> morning Mar 07 01:35:15 [g2]: have any insights into the ehci driver? Mar 07 01:39:43 <[g2]> beewoolie-away, not yet Mar 07 01:39:49 <[g2]> I got a good sleep Mar 07 02:21:23 <[g2]> beewoolie-away, I did scratch down some notes at http://www.nslu2-linux.org/wiki/OpenSlug/EhciIssue Mar 07 02:26:07 [g2]: I added a couple of comments. Mar 07 02:47:36 did anyone look into nslu2_pci_map_irq yet?...dropping INTD and changing to mod 3 looks wrong..and I was the moron responsible. Mar 07 02:50:58 sorta depends on bus and slot...stuff a printk in before return to see what the inputs are and what gets mapped Mar 07 02:57:37 make than slot and pin Mar 07 03:03:08 <[g2]> beewoolie-away, thx for the note in the wiki Mar 07 03:03:16 beewoolie-away: lo Mar 07 03:03:29 prpplague: bi Mar 07 03:03:38 <[g2]> kas11, I'd like to check that in a bit Mar 07 03:04:11 beewoolie-away: whats cookin Mar 07 03:05:03 prpplague: little things. Some small bug fixes in apex. The ext2 driver is almost done. Mar 07 03:05:13 lovely Mar 07 03:05:23 beewoolie-away: still tweaking my mmc driver Mar 07 03:06:14 prpplague: I'm gonna have to implement some kind of driver chaining soon in order to make these fs drivers viable. Mar 07 03:10:33 beewoolie-away: yea, i see that there is considerable fat stuff in the drv-cf.c Mar 07 03:11:23 prpplague: I don't think any of it is used anymore. Mar 07 03:11:37 prpplague: the first implementation put all of the code in the cf driver, but I've pulled it all out. Mar 07 03:11:38 beewoolie-away: yea, i've yanking most of it Mar 07 03:11:51 beewoolie-away: i think alot of it was just commented out Mar 07 03:45:32 [g2-away] nobody cared comes from do_level_IRQ iirc...at certain places in PCI code, 4 slots and ints is assumed...somewhere this needs to be filtered for the NSLU2 since we only have 3 ints...and no physically defined slots Mar 07 03:49:58 kas11: I don't see how that can cause the nobody-cared message. Mar 07 03:50:39 kas11: My understanding is that the error comes from the fact that there is an interrupt triggered and detected by the kernel, but the irq handler has nothing to do. Mar 07 03:50:44 0,1 and 2 are valid indicies into the table...a bogus index is readily generated by slot or pin out of our bounds Mar 07 03:51:02 kas11: When I've seen these before, it is caused by a race in the driver. Mar 07 03:51:19 kas11: I still don't see how that matters. Mar 07 03:51:49 kas11: how is a bogus index generated? Mar 07 03:51:56 iic...I just saw something in the code that looked problematic...but may have nothing to do with that problem Mar 07 03:53:18 slot and pin seem to be constants of 4...that won't work since only 3 ints are available for the PCI bus on the nslu2 Mar 07 03:53:22 kas11: It seems that an interrupt is being asserted for irq26, the ehci vector, that doesn't match a valid event. Mar 07 03:53:44 kas11: Sure, but if nothing generates the fourth interrupt, it ought not matter. Mar 07 03:54:30 kas11: Unless, for some reason, that interrupt is mapping to the ehci vector *and* it is generating spurrious interrupts. Mar 07 03:54:52 but what if something is assigned to what would normally be INTD?...higher levels assume that would be valid...*shrug* Mar 07 03:55:57 If the intd pin has been reassigned, sure that could be an issue. That may account for the spurrious interrupt. I'm skeptical that it is an issue because we only see the nobody-cared messages when the USB driver is under load. Mar 07 03:56:44 makes me wonder if someone is trying for an extra irq for something under load Mar 07 03:57:02 What is your hypothesis? Mar 07 03:58:15 the code is wrong anyhow...may have nothing to do with anything but what would normally be valid a pin and/or slot will generate a bad mapping Mar 07 03:59:01 that code being specific to the NSLU2... Mar 07 04:00:02 someone needs to put a printk there and make sure that it isn't called later when the load occurs... Mar 07 04:01:04 simple enough to do if you are setup to build an openslug kernel...unfortunately, I'm not Mar 07 05:57:37 <[g2]-away> ka6sox-away, ping do you know where FatSlug is ? Mar 07 05:58:04 I'm packing up your now. Mar 07 05:58:13 (the other one) Mar 07 06:00:04 <[g2]-away> cool thx Mar 07 06:18:37 IIRC, INTD is reassigned to power. Mar 07 07:05:41 dyoung: I'm still not clear how this difference in the number of interrupt pins could have any effect on the kernel. Mar 07 07:06:32 dyoung: If the feature isn't present, and the PCI controller doesn't generate bogus interrupts for the unused pin, then there should be no issue. Mar 07 07:07:19 dyoung: BTW, I released a version of APEX with an ext2 filesystem driver. We're not quite all of the way there, but that is the last hard part. Mar 07 07:09:31 <[g2]> beewoolie-away, I think error just happend on flash during a "rm" command Mar 07 07:10:03 [g2]: have you had a chance to try bonnie? Mar 07 07:10:10 <[g2]> no Mar 07 07:10:19 <[g2]> :( Mar 07 07:10:34 beewoolie-away, I dont understand either. A lot of this PCI stuff is too much for my feeble brain. Mar 07 07:11:02 <[g2]> actually thing relating to the PCI is the suspend/resume stuff Mar 07 07:11:21 <[g2]> I tried running without CONFIG_PM but that seems busted Mar 07 07:12:22 <[g2]> beewoolie-away, how big is your flash stick ? Mar 07 07:12:39 The reason that I think bonnie is an interesting test is that we can determine if the throughput is reasonable. Mar 07 07:12:58 <[g2]> Ok it does occur on flash :) Mar 07 07:12:59 [g2]: if it is, we can assume, for the time being, that the lost interrupt is innocuous. Mar 07 07:13:13 [g2]: I see that you are getting errors on flash. Mar 07 07:13:23 <[g2]> yes Mar 07 07:13:29 [g2]: The largest flash I've got is 256MiB. Mar 07 07:13:30 <[g2]> not as often but yes Mar 07 07:13:42 <[g2]> is it empty ? Mar 07 07:13:51 [g2]: It probably could be. Mar 07 07:14:19 <[g2]> Ok it happens on flash and hd with ext2 and ext3 Mar 07 07:14:20 I'll tell you what, I'll see about putting openslug on one of my slugs. Mar 07 07:14:51 Let me pull the latest from oe and from your bk repo and see what happens. Mar 07 07:15:03 <[g2]> Ok that's what I'm running Mar 07 07:15:13 <[g2]> I pulled about an hour ago Mar 07 07:15:33 * beewoolie-away pulls Mar 07 07:16:48 something seems to be odd with the slug repo. Mar 07 07:17:05 hehe is it a little sluggish? Mar 07 07:17:12 * beewoolie grins Mar 07 07:17:30 I'm a little scared of trying to ^c. Mar 07 07:17:41 Ah, there it goes. Mar 07 07:19:00 ...it's in super-slug-slo-mo Mar 07 07:21:07 ...building Mar 07 07:21:53 <[g2]> dd if=/dev/zero of=zero bs=1M count=240 Mar 07 07:21:58 <[g2]> and rm zero Mar 07 07:22:01 <[g2]> both do it Mar 07 07:22:10 <[g2]> not always but often Mar 07 07:22:41 <[g2]> mount -t ext2 /dev/sdb1 /media/cf Mar 07 07:22:48 <[g2]> and cd /media/cf Mar 07 07:23:19 <[g2]> kids need food, Mr. Mom's gotta run for a bit Mar 07 08:26:46 [g2]-food: It isn't building. Mar 07 08:26:54 ERROR: Nothing provides unslung-able-image Mar 07 08:27:05 ERROR: dependency unslung-able-image (for unslung-image) not satisfied Mar 07 08:27:05 NOTE: no buildable providers for unslung-image Mar 07 08:27:31 2104.100u 343.872s 44:40.36 91.3% 0+0k 0+0io 1345pf+0w Mar 07 08:28:28 <[g2]-food> that's because you distro is probably openslug Mar 07 08:28:40 <[g2]-food> you should be bitbake openslug-image Mar 07 08:29:33 <[g2]-food> beewoolie, you are trying to build openslug right ? Mar 07 08:29:47 I think I got that wrong. Let me try again. Mar 07 08:30:50 <[g2]-food> :) Mar 07 08:44:42 <[g2]-food> is the NEC supposed to be level or edge triggered ? Mar 07 08:44:48 What is the fastest, reasonably priced machine we can buy these days? Mar 07 08:45:00 <[g2]-food> This is an interesting post http://lists.arm.linux.org.uk/pipermail/linux-arm-kernel/2004-December/025906.html Mar 07 08:45:20 <[g2]-food> beewoolie, can you be more specific with that question ? Mar 07 08:45:24 Good question. If it were wrong, I think it would have been *way* broken. Mar 07 08:46:18 I cannot know how it works without the datasheet. Mar 07 08:46:35 <[g2]-food> I think I've got a copy I'll check Mar 07 08:47:00 <[g2]-food> what kind of machine are you talking about a build machine ? Mar 07 08:47:10 You read my mind. Mar 07 08:48:22 <[g2]-food> You've got less than a 1GB memory right ? Mar 07 08:48:22 <[g2]-food> :) Mar 07 08:50:27 did anyone come up with a manual for the NEC? when I looked they wanted people to jump thru hoops Mar 07 08:50:43 <[g2]-food> I'll check if I've got one Mar 07 08:50:54 <[g2]-food> I think I dl'ed it yesterday or the day before Mar 07 08:51:53 Shoot me a copy dude. Mar 07 08:52:06 <[g2]-food> I've gotta find it Mar 07 08:52:16 <[g2]-food> I think I pulled it straight from the web site Mar 07 08:52:28 It didn't let me do so. Mar 07 08:52:50 not being in hoop-jumping mood, I said [insert rude words here] very much and did something else Mar 07 08:54:18 <[g2]-food> what's wrong with this link ? http://www.necel.com/usb/ja/dir1/S16265EJ4V0DS00.pdf Mar 07 08:54:44 <[g2]-food> it doesn't have a lock on it :) Mar 07 08:55:39 did y'all try changing the latency setting on the ehci-hcd driver?... Mar 07 08:55:47 While that is, literally, the datasheet. It doesn't explain the program interface. Mar 07 08:55:58 I was looking for the whole devel manual Mar 07 08:56:01 <[g2]-food> a while ago Mar 07 08:56:41 build of openslug took about an hour. Mar 07 08:56:45 agrees with beewoolie...and still objects to hoops Mar 07 09:00:40 So, now that I have built openslug. Can I use the linksys fw to install it? Mar 07 09:09:14 [g2]-food: Are you still eating? Mar 07 09:09:27 <[g2]-food> no Mar 07 09:09:41 I've got the redboot prompt. Mar 07 09:09:48 How do I put openslug on this thing? Mar 07 09:11:12 <[g2]> http://www.nslu2-linux.org/wiki/HowTo/RecoverFromABadFlash Mar 07 09:12:05 I understand the procedure. What I don't know is which file I am to use and whether or not I can pull it from the web interface. Mar 07 09:12:29 <[g2]> Do you have the linksys stock firmware on there ? Mar 07 09:12:39 Yep. Mar 07 09:12:47 <[g2]> then just flash upgrade Mar 07 09:12:53 <[g2]> admin -> upgrade Mar 07 09:13:10 The .img file? Mar 07 09:13:26 <[g2]> browse and select the tmp/deploy/images/...flash... Mar 07 09:13:43 -rw-rw-r-- 1 elf elf 8388608 2005-03-07 16:53 openslug-nslu2-20050307233923.flashdisk.img Mar 07 09:14:18 <[g2]> that's the lastest ? then yes Mar 07 09:14:45 This image is the size of the whole of flash. Are you sure? Mar 07 09:20:45 It seems to boot faster. Mar 07 09:22:17 I have 148 pages of nec users manual... Mar 07 09:22:23 Hmm. OK. It's booted. Mar 07 09:22:33 Dropbear seems to be running. Mar 07 09:23:30 Should we be talking about this on the nslu2-linux channel? Mar 07 09:23:41 I dunno? Mar 07 09:23:55 Should I upload S16336E someplace for those interested? Mar 07 09:24:09 Can it go on the wiki? Mar 07 09:24:25 I jumped thru hoopw Mar 07 09:24:58 s...waiting for NEC official ok Mar 07 09:26:17 [g2]: no vfat filesystem in openslug? Mar 07 09:26:27 <[g2]> probably not Mar 07 09:26:40 <[g2]> vfat doesn't support symlinks Mar 07 09:27:05 <[g2]> fisk and mke2fs are there :) Mar 07 09:27:08 And your point is? Mar 07 09:27:19 Folks are going to have vfat devices. Mar 07 09:27:21 <[g2]> it's useless for system partitions Mar 07 09:27:33 Not for storage, it is isn't. Mar 07 09:27:42 <[g2]> it's trivial to add Mar 07 09:27:55 #modprobe vfat... Mar 07 09:28:09 Not as trivial as you think. Mar 07 09:28:20 <[g2]> we've be working on the boot from ext3 on hd/flash stick and the ehci issue along with a couple dozen others Mar 07 09:30:19 <[g2]> is on the list http://www.nslu2-linux.org/wiki/OpenSlug/BetaRelease Mar 07 09:30:28 <[g2]> it's Mar 07 09:31:24 <[g2]> beewoolie, if we were going to talk about this in another channel #openslug would be the one Mar 07 09:32:10 <[g2]> kas11, YOUR VERY WELCOME THERE TOO ! Mar 07 10:14:18 <[g2]> ~seen desaxena_away Mar 07 10:14:19 [g2]: i haven't seen 'desaxena_away' Mar 07 10:14:19 [g2]: i haven't seen 'desaxena_away' Mar 07 10:14:33 <[g2]> ~seen deaxena_away Mar 07 10:14:33 [g2]: i haven't seen 'deaxena_away' Mar 07 10:14:34 [g2]: i haven't seen 'deaxena_away' Mar 07 10:14:38 <[g2]> ~seen dsaxena_away Mar 07 10:14:39 [g2]: i haven't seen 'dsaxena_away' Mar 07 10:14:39 dsaxena_away is currently on #nslu2-linux #elinux Mar 07 10:57:34 dyoung-web: I seem to have wedged my slug. It isn't bricked because I can get to redboot. It won't boot tho. Mar 07 10:57:42 == Executing boot script in 2.000 seconds - enter ^C to abort Mar 07 10:57:42 RedBoot> boot;exec 0x01d00000 Mar 07 10:57:42 not have eRcOmM Mar 07 10:57:42 Entry to download Mar 07 10:57:48 What's the deal? Mar 07 10:58:02 It isn't copying the kernel? Mar 07 10:59:55 There are zeroes at 0x50060000 which is not what I'd expect Mar 07 11:02:02 Have not ercom means redboot cant find the sercomm trailer.... Mar 07 11:04:33 dyoung-web: It looks like the procedure on the web site didn't work. Mar 07 11:05:04 My guess is that the flash wasn't being erased before being written. Mar 07 11:05:11 the easiest way for you to fix it at this point is to "upslug" it. Mar 07 11:05:12 When I erased it first, it seems to be OK. Mar 07 11:05:47 Control-C it, "upgrade" it, then "upslug" it. Mar 07 11:06:15 Or yeah you can tftp it and fis write it too. Mar 07 11:06:34 Does the wiki allow for or tags? Mar 07 11:07:03 The tftp procedure didn't work for me. I had to erase first. Mar 07 11:07:11 It came back this time. Mar 07 11:07:50 I think I'm going to go back to APEX until it is an acceptable replacement for redboot. The system as is gives me heebeejeebees. Mar 07 11:28:55 I agree. Mar 07 11:30:58 When you have time we should poke and push into this NPE initialization thing; and be done with it once and for all. Then I'll move to apex permanently. Mar 07 11:33:58 That's what I'm going to do next. Mar 07 11:41:11 dyoung-web: Did you ever get the library to build? Mar 07 11:41:14 I'm more than ready to help you do it. Mar 07 11:41:31 More? than ready. Mar 07 11:41:31 you know, since I have no fear of breaking it. Mar 07 11:41:48 Just lacking in time. Mar 07 11:41:49 give me a minute. Let me see how far it builds right now. Mar 07 11:41:52 I was pretty close. Mar 07 11:42:03 I got close, but I got distracted with a yard. Mar 07 11:42:49 It's furlongs and fortnights that get me most of the time. Mar 07 11:43:31 heh. Mar 07 11:44:12 The weather has been mostly wet since december, the last 2 weekends were the only nice weekends for a while; my grass got totally out of control. Mar 07 11:49:40 Hmm. Mar 07 11:49:57 I gotta get outta here or my a$$ will be grass. Mar 07 11:50:12 OK. I'm just looking at the library now. Mar 07 11:50:29 "Keeping the woman waiting is like asking for death". Mar 07 11:57:52 dyoung-web: I know what the problem is. The library assumes that some addresses are statically mapped which are *not* statically mapped anymore. **** BEGIN LOGGING AT Mon Mar 07 12:07:45 2005 Mar 07 13:40:39 dyoung-web: I've posted my progress with the access library. It's pretty sloppy work, IMHO. You can get my patches and scripts from my wiki. Mar 07 13:41:01 http://wiki.buici.com/twiki/bin/view/Main/IXPAccessLibrary Mar 07 13:41:12 I don't have complete instructions, yet. Mar 07 13:41:37 If you want to make a go of it, let me know how far you get. Mar 07 22:08:57 same crowd! :) Mar 07 22:09:45 Is the status of APEX available on a wiki or page somewhere? Mar 07 22:10:08 <[g2]> getting Mar 07 22:12:02 <[g2]> http://wiki.buici.com/twiki/bin/view/Main/ApexBootloader Mar 07 22:19:04 thanks. Mar 07 22:25:50 nice page. Mar 07 22:26:37 Did anybody actually flashed their slug with it yet? Mar 07 22:27:13 with apex? Mar 07 22:27:26 I don't care much for all the fluff like splash screen, netconsole, USB this and that. As long as it gives me serial access to command it's good. Mar 07 22:27:28 yep. Mar 07 22:28:06 It's a bootloader after all. Not a tool for the average user. Mar 07 22:28:10 VoodooZ_work: yea several ppl are testing it Mar 07 22:28:16 good. Mar 07 22:28:25 VoodooZ_work: but thats the good thing about apex, you can choose the options you want Mar 07 22:28:31 I'd like to try it out but I don't have jtag so I'm a bit paranoid. Mar 07 22:28:37 VoodooZ_work: just don't compile in the ones you don't want Mar 07 22:28:41 yes! that's a good thing. Mar 07 22:29:07 Who's using it to boot openslug right now? Mar 07 22:29:47 i could not say with any specifity, seems like it is ka6sox and [g2] Mar 07 22:30:07 ok. Mar 07 22:30:44 <[g2]> VoodooZ_work, you can run it from RAM to get familiar with it Mar 07 22:30:53 <[g2]> I wouldn't flash it just yet Mar 07 22:30:55 If It boots openslug's jffs2 correctly with basic serial I'd try it. Mar 07 22:31:05 ok. Mar 07 22:31:23 <[g2]> there are two issues (1 of which you probably don't care about) Mar 07 22:31:34 [g2]: would you say it's early alpha or close to beta? Mar 07 22:32:00 <[g2]> 1/ The NPE's aren't initialized so the ixp doesn't come up --- This is the one you probably don't care about Mar 07 22:32:21 <[g2]> closer to later alpha Mar 07 22:32:33 <[g2]> 2/ From RAM is was getting a checksum issue. Mar 07 22:32:49 looks good then. Mar 07 22:33:06 <[g2]> I think dyoung/bewoolie is the only one that has it flashed Mar 07 22:33:24 I guess I'll have to start monitoring this space more to keep up with the progress. Mar 07 22:33:31 <[g2]> I shoul be getting FatSlug latter this week and then I'll be able to flash and do more testing Mar 07 22:33:39 good! Mar 07 22:34:09 <[g2]> and bewoolie has a pretty good idea about the NPE isue Mar 07 22:34:11 <[g2]> issue Mar 07 22:34:14 keep up the great work [g2]. you rock! Mar 07 22:34:28 ~praise bewoolie Mar 07 22:34:28 All Hail bewoolie! Mar 07 22:34:30 <[g2]> thx, I'm just part of a GREAT community Mar 07 22:34:33 All hail bewoolie! Mar 07 22:35:33 Well, if the community is like one of those multi-seat bicycle, you're definitely at the steering end! Mar 07 22:36:14 and rwhitby too of course, you guys just share the steering every 8/12 hours! :) Mar 07 22:36:34 <[g2]> oh there are lots of key contributors Mar 07 22:36:50 <[g2]> it's really great Mar 07 22:37:49 yep. I started here with the goal of equipping my robot with a great CPU and ended up delaying the robot to keep having fun with the board instead! Mar 07 22:39:54 <[g2]> you'll really be ready for next years competition ! Mar 07 22:41:11 Yeah. I hope so but unfortunately a great CPU doesn't make for a great robot. I should be working on my AI instead but summer is coming so I'll be getting the motorcycle fever soon. :) Mar 07 22:42:08 I tried my first machine vision test yesterday on my slug but it looked kind of slow. I'll have to investigate more. Mar 07 23:00:44 <[g2]> there are only so many things we can help you with :) Mar 07 23:26:22 hey kas11! **** ENDING LOGGING AT Mon Mar 07 23:59:56 2005