**** BEGIN LOGGING AT Tue Mar 22 23:59:56 2005 Mar 23 00:40:46 03ingeba * 10unslung/sources/proftpd/control: Moved to makefile Mar 23 00:41:56 Morning Mar 23 00:41:56 03ingeba * 10unslung/make/proftpd.mk: Made scripts 755 and did control file generation Mar 23 01:08:36 ka6sox-away: I have registered all CVS developers in slugbug - in theory (with some alias) we are 131 developers on this project, many of whom seem to be out to lunch! I'll try to do all the packages soon. Mar 23 01:10:27 anyone around? Mar 23 01:11:04 Nah Mar 23 01:11:37 in particular rwhitby, ByronT, or ka6sox Mar 23 01:11:49 Nope - none of them I believe Mar 23 01:11:53 or anyone interested or concerned with licensing issue. Mar 23 01:11:55 drat. Mar 23 01:12:06 Okay, youre here though. ;-) Mar 23 01:12:15 so... Did w3cam work? Mar 23 01:12:52 My goal for all that stuff i pushed this weekend was to simplify your usbcam page. I hope I didnt step on your toes. Mar 23 01:13:00 You didn't at all Mar 23 01:13:09 I was working in parallel on the bits to make openslug do similar things. Mar 23 01:13:10 I have no money to buy a proper cam anyway Mar 23 01:13:31 I thought you had a OV511 lying around? Mar 23 01:13:35 I do Mar 23 01:13:40 But is sucks Mar 23 01:13:47 It is from '98 Mar 23 01:14:01 Ah, but enough to test. ;-) Mar 23 01:14:25 Yesterday and today I've been testing Siddy's traffic shaping on Unslung Mar 23 01:14:37 And it works well Mar 23 01:14:40 Cool. Mar 23 01:15:04 If only we could get USB support ship shape on this box, it would be easier for me to test webcam Mar 23 01:15:13 As it is, I can only have two devices connected. Mar 23 01:15:36 I have only one slug and it is "production", so testing time is a bit limited Mar 23 01:16:04 The usb driver issue is serious Mar 23 01:16:11 Bummers. No biggie though, I figure someone will eventually test it and/or complain then we'll know. Mar 23 01:16:23 Whihc USB issue is giving you problems? Mar 23 01:16:29 tell me: It is in OE, so a IPK is generated? Mar 23 01:16:40 For which part? Mar 23 01:17:01 Oh, the " usb.c: not enough configurations" Mar 23 01:17:10 The drivers Mar 23 01:17:27 Yes, the drivers come out (in that particular case) as kernel-module-ov511.ipk Mar 23 01:17:53 And these drivers are patched then generated as a part of unslung-kernel Mar 23 01:18:07 but rwhitby recently pushed a null update-modules script so you can ipkg install it without any complain from ipkg. Mar 23 01:18:40 - "null update-modules" ? Mar 23 01:18:44 Yes. They will be in tmp/deploy/ipk and are manually scped to the "oe" feed. Mar 23 01:19:11 #!/bin/sh; exit 0 Mar 23 01:19:39 so when the postinstall runs, it just happily does it without grumbling about ldconfig, etc, etc Mar 23 01:19:44 Ah Mar 23 01:20:19 lemmem find the exact place. this is in unslung-4.5 Mar 23 01:20:53 I'm motivated to set up iptables2 proper as ipk and with kernel modules as ipk, we can have easy install of traffic shaping Mar 23 01:20:57 Yes Mar 23 01:23:03 the bits are in packages/meta/unslung-image.bb Mar 23 01:23:16 echo "#!/bin/sh" > ${IMAGE_ROOTFS}/usr/sbin/update-modules Mar 23 01:23:16 echo "exit 0" >> ${IMAGE_ROOTFS}/usr/sbin/update-modules Mar 23 01:23:16 chmod ugo+x ${IMAGE_ROOTFS}/usr/sbin/update-modules Mar 23 01:23:20 and theh same for depmod Mar 23 01:23:30 and a rm of the postinst script Mar 23 01:23:40 update-modules.postinst Mar 23 01:23:47 so thats enough to trick it into working. Mar 23 01:24:07 So you can play theh same trick with the traffic shaping stuff. Mar 23 01:24:26 I will have to take a closer look at BB - haven't checked in anything in that system yet Mar 23 01:24:41 s/BB/BK/ Mar 23 01:24:51 is tc part of it, or a separate thing? If separate, it'll need to be a unslung package from cvs Mar 23 01:25:29 Tc is part of it - so it will be an Unslung package in CVS. The modules, however, will be fetched from the OE feed Mar 23 01:25:48 Good Plan Mar 23 01:25:57 Haven't seen Siddy around lately, have you? Mar 23 01:26:08 On #openslug maybe Mar 23 01:26:13 As long as the defconfig says foo=m, it'll get into the ipk directory and can be scp-ed to the feed. Mar 23 01:26:17 ~seen siddy Mar 23 01:26:18 siddy <~sid@slug.6bone.siddy.org> was last seen on IRC in channel #nslu2-linux, 5d 21h 50m 6s ago, saying: ':)'. Mar 23 01:26:29 Guess hes on holiday. Mar 23 01:26:35 Good for him Mar 23 01:27:07 Will these modules automatically be put into the feed? Mar 23 01:27:30 Are you running openslug? Mar 23 01:27:34 one of these days they will be. that structure doesnt exist ye, i dont think. Mar 23 01:27:45 Right now I have openslug on the development slug. Mar 23 01:28:16 but I returned to unslung compilling for a while. Mar 23 01:28:18 What's the usb status there? Still the problem with lockups? What about # of devices connected? Mar 23 01:28:59 usb statusu is it works unless theres a lot of traffic, then we get a irq26: nobody cared and a horrible slowdown for minutes. Mar 23 01:29:23 theres a unhandled interrupt for some reason, then the ehci dog bites. Mar 23 01:29:38 What about hubs and devices? Mar 23 01:29:57 The current theory is that there is something fudementally busted in the PCI interrupt handler, so gurus are looking at it. Mar 23 01:30:05 Hubs work Mar 23 01:30:19 How many devices have you had connected at one time? Mar 23 01:30:34 it sometimes get confused which drive is supposed to be sda or sdb and swaps them. Mar 23 01:30:43 Hmm, I've had at most 3. Mar 23 01:30:53 rwhitby had numerous. Mar 23 01:31:10 I could try throw more stuff on. Mar 23 01:31:38 So if we can't iron out the 2.4 USB issues, 2.6 (provided the gurus fix the IRQ problem) can come to the rescue for some of us. Mar 23 01:32:08 I'm not really famliar with the "not enough connections" problem. Mar 23 01:32:17 I'm tempted to sit down an look at the 2.4 problems, but then I need to get another slug. Mar 23 01:32:21 I think rwhitby was using unslung wiht a hub. Mar 23 01:32:31 but he had like serial adapters and toppy's plugged to it Mar 23 01:32:39 (ie: non storage devices) Mar 23 01:32:46 I can too, but not at once Mar 23 01:33:21 I've connected and run with printer, cam, DVD, tape, mem stick and BT dongle Mar 23 01:33:39 and it all works until what condition? Mar 23 01:33:55 like copying from the DVD to the tape or similar? Mar 23 01:34:11 Until more than two devices are connected at once Mar 23 01:34:32 Kernel traps, lockups etc. Mar 23 01:35:06 Just doing cat /proc/bus/usb/devices is enough to induce a crash Mar 23 01:35:11 Don't need the drivers loaded Mar 23 01:35:33 That is with hub connecte to port #2 Mar 23 01:36:14 Hmm. I wonder if theres a patch for that. Mar 23 01:36:59 I've looked around and the problem is found elsewhere, but the USB kernel source we use is very different from the patches I've seen Mar 23 01:37:43 The "usb.c: not enough configurations" is not unique to us Mar 23 01:41:22 whats the minimum number of devices you see the problem with? 3? Mar 23 01:41:28 Yes Mar 23 01:41:32 + HUB Mar 23 01:41:55 is that a 4 port hub? Mar 23 01:42:09 Yes Mar 23 01:42:12 See here: Mar 23 01:42:14 Aha. Mar 23 01:42:17 Jan 1 01:00:26 Naseem hub.c: new USB device 00:01.2-2, assigned address 3 Mar 23 01:42:18 Jan 1 01:00:26 Naseem hub.c: USB hub found Mar 23 01:42:20 Jan 1 01:00:26 Naseem hub.c: 4 ports detected Mar 23 01:42:22 Jan 1 01:00:26 Naseem hub.c: new USB device 00:01.2-2.3, assigned address 4 Mar 23 01:42:23 Jan 1 01:00:26 Naseem usb.c: not enough configurations Mar 23 01:42:25 Jan 1 01:00:26 Naseem hub.c: new USB device 00:01.2-2.3, assigned address 5 Mar 23 01:42:26 Jan 1 01:00:26 Naseem usb.c: not enough configurations Mar 23 01:42:31 Aha? Mar 23 01:42:56 Try patching include/usb.h to change USB_MAXCONFIG to something bigger than 8. Mar 23 01:43:06 like 9 Mar 23 01:43:12 and see if 4 devices works but 5 doesnt. Mar 23 01:43:27 include/linux/usb.h Mar 23 01:43:32 I will try. Mar 23 01:43:44 I might be barking up the wrong tree. Mar 23 01:44:13 but the default is 8, and thats your 3 devices, plus the hub, plus the internal 5 port hub. Mar 23 01:44:29 I'll set it to 12. Mar 23 01:45:01 Currently running a CVS compile, but I'll make one straight afterwards. With announcement of downtime I can test it today Mar 23 01:45:19 Okay Mar 23 01:45:45 I really hope it works! Mar 23 01:45:50 Well, good luck. I hope I'm not wasting your time barking up that tree. Lemme look at a 2.6 tree and see what it says. Mar 23 01:47:11 Compiling now Mar 23 01:49:32 2.6 is also set to 8 Mar 23 01:49:39 dyoung-zzzz: Sure it is sufficient to delete the stamp for kernel-do_compile? Mar 23 01:49:40 it says its a arbitrary limit Mar 23 01:49:46 Yes - I saw that Mar 23 01:49:54 it should be sufficient to blow away the do_compile stamp. Mar 23 01:50:07 I have a gut feeling that it will have no effect. Mar 23 01:50:10 but we can hope. Mar 23 01:50:28 The kernel is made - I'll just check the date/time for the relevant object modules. I share your gut feeling. Looked at the source and when the error was issued. Mar 23 01:50:49 I wonder if its some endian issue, because it says if blah < 1 Mar 23 01:50:57 then return the error. Mar 23 01:53:47 Okay so for whenever licensing folk come back online. Mar 23 01:54:06 I was trying to make my wireless card work in the new IBM T42 lappy. Mar 23 01:54:18 So I went to the ipw2200.sourceforge.net page Mar 23 01:54:21 A T42? Lucky you Mar 23 01:54:26 I think what we want it something similar to here: Mar 23 01:54:28 http://ipw2200.sourceforge.net/firmware.php Mar 23 01:54:28 Running? Mar 23 01:54:54 when you click the firmware, it takes you to a bigass legal sounding page you have to agree to before getting the card firmware. Mar 23 01:55:08 Yeah, I like my T42 Mar 23 01:55:12 Its happy. Mar 23 01:55:23 wirless is busted for linux though Mar 23 01:55:29 I'm so happy I live Norway when it comes to this. Mar 23 01:55:33 I gave up around 15 driver versions ago. Mar 23 01:55:39 so I decided to try again. Mar 23 01:55:50 I dont have a result yet. Mar 23 01:56:01 Shrink wrap licences and unreadable "Agree" licence pages have to legal value. Mar 23 01:56:03 but they tagged it for version 1.0.0 so it must kinda work. Mar 23 01:56:26 I solved my TP problem by running XP. Some would say that is not really solving, though. Mar 23 01:56:37 Have CoLinux on top Mar 23 01:56:55 Yeah, the other annoyance with the t42 is it cant display on a projector under linux. Mar 23 01:57:05 it sounds dumb., but it just plain doesnt work. Mar 23 01:57:24 Using VGA output? Mar 23 01:57:29 so I'm starting to go down the same road and putting colinux on Mar 23 01:57:44 It works amazingly well Mar 23 01:57:48 Yeah, when you press the button to make it display both internal and external, it croaks. Mar 23 01:57:53 hard crash. Mar 23 01:58:03 Ah - ATI chipset? Mar 23 01:58:06 It does it on the t41 too, so I'm not alone. Mar 23 01:58:16 I got X31 Mar 23 01:58:17 Uh... Actually I have no idea what chipset is in there. Mar 23 01:58:34 But you can run VGA only? Mar 23 01:59:28 Not sure. it hard crashes when I press the button to make it display out the external port. Mar 23 01:59:30 At least I could - set up external only in BIOS Mar 23 01:59:44 Ah, I never tried that. Mar 23 01:59:55 Don't know T42, but check it out. Mar 23 02:02:38 rwhitby: I have additions for defconfig. The three compiled in drivers for traffic shaping and a number of modules. Can you check it in for me? Mar 23 02:02:57 rwhitby: I'm confident wrt the compiled in drivers Mar 23 02:04:32 dyoung-zzzz: let's steal that ipw2200 page Mar 23 02:05:52 Okay, I havent been privy to the communications with Intel, so I'm not sure if that covers everything that needs covering. Mar 23 02:07:11 bob_tm: send'em to me, or send me an rsa1 key Mar 23 02:07:31 rwhitby: You got the key - I'm registered, just incompetent Mar 23 02:07:57 rwhitby, while youre in there, I think its safe to remove unslung-4.x-symlink-package and oe Mar 23 02:08:22 rwhitby: You want pure additions or the whole defconfig? Mar 23 02:09:26 .... or diffs Mar 23 02:11:54 either Mar 23 02:13:44 looking after elise at mo, so may not get to it till later Mar 23 02:14:43 dyoung may be able to gert it in quicker Mar 23 02:22:07 rwhitby: Mailed it just now. dyoung-zzzz: You want to do it? Mar 23 02:22:26 It= check in a new defconfig Mar 23 02:23:17 Well, gotta get a shower. Then time to write some deep thoughts about language and mind. Good luck with the T42, dyoung-zzzz . Mar 23 02:24:07 dyoung-zzzz: interesting thought about the configurations - I get that error too Mar 23 02:24:51 which file? Mar 23 02:30:47 oop, sorry was away. Mar 23 02:30:50 was that question for me? Mar 23 02:31:17 tmp/work/unslung-kernel-foo/linux-2.4.22-xfs/include/linux/usb.h Mar 23 02:33:04 will try it later ... Mar 23 02:33:28 after looking for the error message in the source and finding causality Mar 23 02:33:43 its in the patch. Mar 23 02:39:09 is it a linksys mod? Mar 23 02:41:13 I'm not sure, I'm guessing not Mar 23 02:41:36 but the linksys diffs are so ... bulky, it doesnt really surprise me its there Mar 23 02:42:11 is it the same in openslug? Mar 23 02:42:36 Yes Mar 23 02:42:47 that one is in in drivers/usb/core/core.c Mar 23 02:43:06 and labelled "arbitrary value" Mar 23 02:43:45 bah, config.c Mar 23 02:50:05 have you found where the error is printe? Mar 23 02:52:07 yes Mar 23 02:52:35 for unslung, linux-2.4.22-xfs/drivers/usb/usb.c Mar 23 02:53:16 :2119 Mar 23 02:53:38 got it Mar 23 02:55:06 I cant imagine why that value would return < 1 Mar 23 02:55:42 If you have a openslug running, you can peek into it somewhere in /sys Mar 23 02:56:22 where does it say blah < 1 ? Mar 23 02:56:41 on line 2119 Mar 23 02:56:44 oh right Mar 23 02:58:42 I get the error on unslung Mar 23 02:58:48 Hmm Mar 23 02:59:00 recreating error now Mar 23 02:59:25 According to the 2.6 file, that get_configuration stuff is for hubs, both real and virtual root hubs Mar 23 02:59:37 in which case 8 isnt enough. Mar 23 02:59:48 because the nec chip has 3+5 already. Mar 23 03:00:14 but I'm still trying to understand what its all for. Mar 23 03:00:17 that's 8 per device though, isn't it? Mar 23 03:00:41 not related to number of ports Mar 23 03:01:15 ask me again in 10 minutes. Mar 23 03:10:50 ok, on a running unslung system, with a storage device and a toppy on a 2.0 hub, I plug in a irtans.com device and a usbnet adapter (pegasus), and get the following in dmesg: Mar 23 03:10:59 hub.c: new USB device 00:01.2-2.7, assigned address 6 Mar 23 03:11:00 usb.c: not enough configurations Mar 23 03:11:00 usb.c: unable to get device 6 configuration (error=-22) Mar 23 03:11:00 hub.c: new USB device 00:01.2-2.7, assigned address 7 Mar 23 03:11:00 usb.c: not enough configurations Mar 23 03:11:01 usb.c: unable to get device 7 configuration (error=-22) Mar 23 03:11:02 hub.c: new USB device 00:01.2-2.2, assigned address 8 Mar 23 03:11:04 usb.c: not enough configurations Mar 23 03:11:06 usb.c: unable to get device 8 configuration (error=-22) Mar 23 03:11:07 hub.c: new USB device 00:01.2-2.2, assigned address 9 Mar 23 03:11:09 usb.c: not enough configurations Mar 23 03:11:11 usb.c: unable to get device 9 configuration (error=-22) Mar 23 03:11:13 # Mar 23 03:11:58 hmm, lemme try find some stuff to plug into openslug. Mar 23 03:15:48 then a segfault on cat /proc/bus/usb/devices Mar 23 03:16:56 Okay I just plugged in everything I own into it. Mar 23 03:17:24 thats a hard drive, a MMC adapter, a camera, a wireles dongle, and a Pegasus Mar 23 03:17:35 and catted /proc/bus/usb/devices Mar 23 03:17:55 its all there Mar 23 03:18:06 so openslug is fine? Mar 23 03:18:11 Seems so. Mar 23 03:18:26 although your error message reminded me of soehting Mar 23 03:18:43 when ehci was out; thats what it would do. Increment the device number Mar 23 03:18:53 address rather Mar 23 03:18:57 over and over and over...... Mar 23 03:20:02 The code in usb/core/config.c is pretty much the same for that section. Mar 23 03:24:14 03ingeba * 10unslung/sources/asterisk/editline.makelist.patch: Removed explicit path to awk to facilitate cross compile Mar 23 03:25:50 03rwhitby 07 * r1.3148.1.1 10openembedded/packages/linux/ (unslung-kernel_2.3r25.bb unslung-kernel-2.3r25/defconfig): Added lots of modules for traffic shaping and other various things to OpenSlug kernel Mar 23 03:26:11 s/openslug/unslung ?? Mar 23 03:26:17 yeah Mar 23 03:26:48 rebuilding unslung oe feed now .... Mar 23 03:28:07 if ever you need to do it, it's just "make" in /home/unslung/firmware on nudi Mar 23 03:32:49 rwhitby: I have problems getting the loader to find the staged libs if they are not explicitly linked. Example: I have -lssl, but libssl needs libcrypto, but cannot find it unless -lcrypto is also specified. Mar 23 03:33:23 Is there a cure except for patching the makefile? Mar 23 03:33:24 that sounds right Mar 23 03:33:40 not that I know of Mar 23 03:33:45 Shouldn't the linker look for libcrypto ? Mar 23 03:34:01 It knows the correct file name Mar 23 03:34:24 dunno - you're asking the wrong guy here ... Mar 23 03:34:35 Ok - who would be the right guy? Mar 23 03:35:04 lennert, tiersten, jp30 ? Mar 23 03:35:16 peteru might know too Mar 23 03:35:19 Ah - will check with them Mar 23 03:35:36 ...once they get online :-) Mar 23 03:35:50 probably lots of other people know more about that stuff than I do Mar 23 03:37:25 rwhitby: there is Mar 23 03:37:31 np Mar 23 03:37:45 Just an annoying difference between cross and natice Mar 23 03:37:57 T: Bus=01 Lev=02 Prnt=114 Port=06 Cnt=03 Dev#=-1070610016 Spd=12 MxCh= 0 Mar 23 03:37:57 D: Ver=40.00 Cls=00(>ifc ) Sub=00 Prot=24 MxPS= 0 #Cfgs= 0 Mar 23 03:37:57 P: Vendor=0000 ProdID=0000 Rev= 0.00 Mar 23 03:38:17 that's what I get as the last entry in /proc/bus/usb/devices when the not enough configurations happens Mar 23 03:39:04 rwhitby: And the same device gives #Cfgs when connected in other ways? Mar 23 03:39:19 Usually #Cfgs= 1 I guess Mar 23 03:39:37 dunno - can't do that kind of rearrangement at the moment - it may wake the baby Mar 23 03:40:05 but I expect it would normally be =1 Mar 23 03:40:08 Ok -I'll try. Mar 23 03:40:27 Risky - it is running, but I feel like gambling Mar 23 03:43:04 Confirmed Mar 23 03:43:49 so something weird happening when its enumerating the ports Mar 23 03:44:36 Yes - look at the diff before and after connecting the third device: Mar 23 03:44:59 - //Naseem/opt/tmp/> diff devices.* Mar 23 03:45:00 45,48c45 Mar 23 03:45:02 < T: Bus=01 Lev=02 Prnt=03 Port=02 Cnt=01 Dev#= -1 Spd=12 MxCh= 0 Mar 23 03:45:03 < D: Ver= 0.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 0 #Cfgs= 0 Mar 23 03:45:05 < P: Vendor=0000 ProdID=0000 Rev= 0.00 Mar 23 03:45:06 < T: Bus=01 Lev=02 Prnt=03 Port=03 Cnt=02 Dev#= 4 Spd=480 MxCh= 0 Mar 23 03:45:08 --- Mar 23 03:45:10 > T: Bus=01 Lev=02 Prnt=03 Port=03 Cnt=01 Dev#= 4 Spd=480 MxCh= 0 Mar 23 03:45:59 bob_tm: unslung oe feed is updated with all the new modules Mar 23 03:46:11 thnks Mar 23 03:47:55 I unplugged the errant device, and got a kernel oops Mar 23 03:48:08 in khubd Mar 23 03:48:44 in the 2.4 tree theres all sorts of "ifnecdev type conditionals. Mar 23 03:48:51 Like "Naseem Process khubd (pid: 8, stack limit = 0xc03a4368)" ? Mar 23 03:49:16 Followed by reg dumpt and stack trace Mar 23 03:49:41 yep Mar 23 03:50:02 I'm especiall fond of the later note in two out of three such oops'es : "Backtrace: no frame pointer" Mar 23 03:50:26 It spells stack corruption to me. Mar 23 03:54:18 It might be connected to the fact that the device gets address "-1" after the "not enough..." problem Mar 23 03:54:42 A connect, disconnect says: Mar 23 03:54:47 Mar 23 12:48:02 Naseem hub.c: new USB device 00:01.2-2.4, assigned address 12 Mar 23 03:54:48 Mar 23 12:48:02 Naseem usb.c: not enough configurations Mar 23 03:54:50 Mar 23 12:48:03 Naseem usb.c: USB disconnect on device 00:01.2-2.4 address -1 Mar 23 03:58:53 anyway, I'm getting nowhere fast on this problem. Lucky for me I don't need either of the two devices plugged in that cause the problem Mar 23 03:59:17 back to unslung diversion scripts for me - need to make the recognise sda1, sdb1 as rootfs candidates. Mar 23 03:59:42 rwhitby: Good luck Mar 23 04:13:28 we can remove /linuxrc from unslung rootfs now, right? Mar 23 04:15:19 Probably. Mar 23 04:59:04 so in the case where unslung is running from an external disk, should we or should we not create ramdisks for /dev and /var ? I'm leaning towards no .... Mar 23 04:59:35 <[g2]> I not aware of a reason to create one Mar 23 05:00:06 <[g2]> also we need should consider mounting noatime Mar 23 05:00:33 <[g2]> I'll initially do that by default for Openslug Mar 23 05:00:47 <[g2]> then ppl running on flash sticks won't burn them out Mar 23 05:00:54 The /.ext3flash flag file does that for unslung already Mar 23 05:01:16 since early 3.x Mar 23 05:01:25 <[g2]> and that triggers no mounting of swap to ? Mar 23 05:01:29 <[g2]> cool Mar 23 05:01:38 yep Mar 23 05:02:20 our assertion is that unslung with /.ext3flash only writes to flash on boot or ipkg install or user action Mar 23 05:03:04 <[g2]> has that be verified via /proc/diskstats over time ? Mar 23 05:03:24 * [g2] tested overnight with openslug and the noatime Mar 23 05:04:22 yep, I tested it when I made that change a long time ago ... Mar 23 05:04:35 <[g2]> btw I don't know if I mentioned it, but the Gigabit is running pretty well Mar 23 05:05:42 I'm gonna leave the ramdisks there for the moment, cause I probably want them for flash disks, and it's too late to work out how to handle all those conditions ... Mar 23 05:05:56 <[g2]> nod. that's fine Mar 23 05:07:03 <[g2]> does unslung always umount /initrd ? Mar 23 05:07:14 nope Mar 23 05:07:31 <[g2]> that's left for the user to do Mar 23 05:08:27 03rwhitby 07 * r1.3150.1.1 10openembedded/packages/nslu2-binary-only/ (20 files in 2 dirs): Unslung 4.7-alpha, with support for external disks on sda1, sda2, sdb1, sdb2. Mar 23 05:17:26 Time to obsolete the unslung-feeds.ipk by putting all three feeds in the base firmware. Mar 23 05:28:13 03rwhitby 07 * r1.3154 10openembedded/ (2 files in 2 dirs): Fixed the unslung patch for rc.1 Mar 23 05:42:51 rwhitby: What about OE feed? Where are the kernel modules intended to go? Mar 23 05:43:51 what do you mean? Mar 23 05:44:29 The module-ipk's made in OE - aren't they going into a IPK feed? Mar 23 05:44:58 yes - that's the oe feed I've been talking about Mar 23 05:45:07 Oh, so that is one of the 3? Mar 23 05:45:44 Unstable, cross, native? Mar 23 05:46:00 cross, native, oe Mar 23 05:46:10 Ah - got it. Mar 23 05:46:18 Yes, smart move anyhow. Mar 23 05:47:02 * bob_tm goes to write the damned paper. Mar 23 05:50:36 here's a tip - don't unsling to /share/hdd/data until after the quota check has finished :-) Mar 23 05:51:04 ByronT-Away: that one will need to go in the README Mar 23 05:59:04 I guess we probably don't even want a ramdisk for /tmp if we're running from an external disk .... Mar 23 06:02:06 03rwhitby 07 * r1.3154.1.1 10openembedded/packages/nslu2-binary-only/unslung-rootfs-2.3r25/ (unsling create-ramdisks.patch): Made Unslung *not* create ramdisk for /var and /dev when you've unslung to an external disk Mar 23 06:02:25 will think about the /tmp question Mar 23 06:02:58 it's easy enough to just umount /tmp in a diversion script if you want ... Mar 23 06:03:31 # mount Mar 23 06:03:31 "/dev/sda1 on / type ext3 (rw) Mar 23 06:03:31 "/proc on /proc type proc (rw) Mar 23 06:03:32 "usbdevfs on /proc/bus/usb type usbdevfs (rw) Mar 23 06:03:32 "/dev/sda1 on /share/hdd/data type ext3 (rw) Mar 23 06:03:32 "/dev/sda2 on /share/hdd/conf type ext3 (rw,sync) Mar 23 06:03:33 "/dev/sdb1 on /share/flash/data type ext3 (rw) Mar 23 06:03:35 "/dev/sdb2 on /share/flash/conf type ext3 (rw,sync) Mar 23 06:06:31 night all Mar 23 06:06:40 <[g2]> sweet dreams Mar 23 06:06:45 oh - Unslung 4.7-alpha ready for testing ... Mar 23 07:21:10 * perlguru Leaving Mar 23 08:22:08 dyoung-zzzz, PING! Mar 23 08:22:31 man, I get some rest and come back, and you guys leave work all over everywhere for me... ;) Mar 23 08:29:14 re: the license issue, that page (meaning format) for the ipw2200 looks usable... but the license presented is completely wrong - for one thing, we only need the end user portion -since the binary is only usable for a nslu2 Mar 23 09:03:09 I've got a question about which version of base-passwd package should be used. Mar 23 09:03:34 I'm doing my first 'bb openslug-image' and it failed looking for base-passwd-3.5.7. Mar 23 09:04:16 I see that the current ver on this pkg is 3.5.9 but openslug-image.bb DEPENDS on 3.5.7, causing bitbake to fail. Mar 23 09:04:49 So I changed openslug-image.bb to only DEPEND on base-passwd. Any reason why this won't work? Mar 23 10:32:21 03jp30 * 10unslung/Makefile: promote bitchx ctags snownews muxsshssl Mar 23 11:52:24 * ruDelph -> sleep(); Mar 23 11:52:26 ÂÂ Mar 23 11:52:59 03cdoban * 10unslung/Makefile: asterisk ready for testing Mar 23 13:29:21 <[g2]> well my dev system may have gotten toasted to day :( Mar 23 13:30:44 bummers Mar 23 13:30:47 what happen? Mar 23 13:31:17 <[g2]> power went out at on the house level Mar 23 13:31:45 <[g2]> I've got a nice surge suppressor, but something seems tweaked Mar 23 13:31:53 in the middle of a flash operation or soething? Mar 23 13:32:03 <[g2]> no in the middle of a build Mar 23 13:32:22 <[g2]> it's the PC dev system that's not 100% Mar 23 13:32:46 <[g2]> when I run bb or sometimes other things it just locks-up :( Mar 23 13:33:57 ohh, I thought you meant the ixp422 board. Mar 23 13:34:04 bummers. Mar 23 13:34:04 <[g2]> I copied the data over to another drive and the same thing happened so it's not the drive Mar 23 13:34:40 <[g2]> that leaves Power-supply, Motherboard, memory Mar 23 13:34:41 I blew a 512M stick of ram in a similar situation. Mar 23 13:35:00 <[g2]> I'm wondering if its memory Mar 23 13:35:12 <[g2]> I guess I'll reboot with 512M@0 Mar 23 13:35:14 <[g2]> :) Mar 23 13:35:23 did yu change the swap area too? Mar 23 13:35:36 <[g2]> I didn't change the swap Mar 23 13:35:49 <[g2]> I can zero and recreate the swap Mar 23 13:36:00 <[g2]> it's this machine I'm typing on right now Mar 23 13:36:04 Good Plan. Mar 23 13:36:33 <[g2]> I nfs'd the data over to another box running koppix from CD Mar 23 13:36:34 dyoung-zzzz, did you get that email? Mar 23 13:36:45 <[g2]> it's acutally bitbaking now :) Mar 23 13:36:56 <[g2]> from the knoppix cd Mar 23 13:37:32 ByronT, Thanks. Receieved. Mar 23 13:37:47 Boggled about it for 10 minutes Mar 23 13:37:59 decided to leave it to you guys. Mar 23 13:38:01 <[g2]> I'll be back later Mar 23 13:38:11 [g2], good luck Mar 23 13:38:14 <[g2]-away> thx Mar 23 13:39:04 that ipw2200 license is much more restrictive than the one for the ipx/npe Mar 23 13:54:49 [g2]: did you push the base-passwd fix? Mar 23 13:55:56 rwhitby, what part of australia are you located? Mar 23 13:57:10 adelaide, why? Mar 23 13:57:32 setting up some clock widgets on my desktop... Mar 23 13:57:43 easier for me to tell what time you're in.... Mar 23 13:59:41 EST: 1700, PST: 1400, RTZ, 0830. Mar 23 13:59:48 RwhitbyTimeZone Mar 23 14:00:23 don't forget ZZZT (dyoung time) Mar 23 14:00:40 in honor of the -zzzz Mar 23 14:01:25 ACST/ACDT is my timezone Mar 23 14:15:23 bbl Mar 23 15:26:51 03ingeba * 10unslung/make/proftpd.mk: Now creates folder for storing socket and scoreboard. Note: Must find new place to store the 'empty'-folder as one shouldn't write to the flash - maybe /opt/home/empty Mar 23 15:27:27 siddy-away: Thanks for traffic shaping - runs great on latest Unslung 4 alpha! Mar 23 15:53:56 03ingeba * 10unslung/sources/openvpn/ (S20openvpn openvpn.conf openvpn.up openvpn.xinetd): Initial check-in Mar 23 15:59:53 03ingeba * 10unslung/make/openvpn.mk: Inital checkin. Did some mods - latest RC and control file generation Mar 23 16:08:09 03ingeba * 10unslung/make/lzo.mk: Cleanup and control file generation Mar 23 16:10:23 03ingeba * 10unslung/sources/lzo/control: Control file now generated Mar 23 16:10:52 bob_tm-away: ping Mar 23 16:11:01 ping Mar 23 16:11:25 rwhitby-away: ping Mar 23 16:12:01 would you be able to add a /opt/etc/xinetd.d/proftpd.conf file to the proftpd package? Mar 23 16:12:11 Yes Mar 23 16:12:18 Just move the one I got Mar 23 16:12:32 Doesn't matter if there is no xinetd there? Mar 23 16:12:38 and then make it depend on xinetd Mar 23 16:12:51 I want us to start moving packages over to depend on xinetd Mar 23 16:13:01 and have the packages drop in an xinetd conf file Mar 23 16:13:02 Sound good - Mar 23 16:13:11 What about standalone? Mar 23 16:13:17 Wasted memory? Mar 23 16:13:29 perhaps you can write it up on the packaging best practises page? Mar 23 16:14:04 Yes - it is what politicians would call a "courageous decision". Mar 23 16:14:17 Did you get it to work then? Mar 23 16:14:41 yeah, let's assume that xinetd will pretty much always be running instead of inetd for anyone who is using unslung packages. Mar 23 16:14:49 yes, it is magically working today :-) Mar 23 16:14:52 :-) Mar 23 16:14:54 if the package can work as a daemon as well as through xinetd, a Recommends would probably be better. still installs xinetd if you install the app, but the user is then free to uninstall it if they want to run it as standalone Mar 23 16:15:10 but then I have reinstalled unslung 4.7 and unsling'ed again since then Mar 23 16:15:32 kergoth: good idea Mar 23 16:16:05 we should do that for apache's dependency on openldap too Mar 23 16:16:11 Hmmm.... very strange - I looked at the code and can't see how it could fail without error unless it failed to see the includedir command Mar 23 16:16:32 if apache can run without openldap, it shoudlnt Depends on it. thats the rule of thumb Mar 23 16:16:38 What about dropbear - I think there are good reasons not to run dropbear from inetd's of any kind. Mar 23 16:16:41 bob_tm-away: it was late at night :-) Strange things always happen then Mar 23 16:16:47 is it functional (yes, thats ambiguous, but it depends on the userbase), without it Mar 23 16:16:52 bob_tm-away: agreed about dropbear Mar 23 16:17:20 So let's say we let developers use good judgement. If there is good reason not to run it via xinetd, then don't Mar 23 16:17:56 yeah, but it has to be a good reason due to the ram limitation Mar 23 16:18:06 (Y) Mar 23 16:18:34 Should we depreciate inetutils inetd? Mar 23 16:18:43 yes indeed. Mar 23 16:18:49 I feel a calling Mar 23 16:19:32 xinetd should be the one true unslung packages inetd, and all packages that intend to be called from an inetd-like thing should drop in an xinetd conf file Mar 23 16:19:53 How should we go about that? If I release a new inetutils without inetd, then those running it would be quite suprised after upgrading. Mar 23 16:20:03 As I assume it will be gone Mar 23 16:20:14 then the user sets up the rc.xinetd diversion script *once*, and all packages work correctly after that. Mar 23 16:20:30 bob_tm-away: we remove it from the package, and announce the policy on the mailing list. Mar 23 16:20:30 Yes! That is nice indeed Mar 23 16:20:44 those who don't read the mailing list deserve to be surprised Mar 23 16:20:48 Will you do the popal honour? Mar 23 16:21:06 we can also write in in the inetutils package wiki page Mar 23 16:21:30 Ok - I will do (a) best practices, (b) wiki on inetutils and (c) proftpd change. Mar 23 16:21:35 yes, I will make the announcement once we have one package working with it to show as an example Mar 23 16:21:43 Goed so. Mar 23 16:22:13 I will go to sleep real soon now, so don't expect much to happen until, let's say in about 8 hours time Mar 23 16:22:16 we will do the inetutils change after the announcement Mar 23 16:22:34 no hurry, but we will get this done before the 4.x beta Mar 23 16:22:53 Ok - xinetd is good software. It is rare that such small things can excite me. Mar 23 16:23:05 Yes we do Mar 23 16:23:08 I'm cool like rwhitby now. Mar 23 16:23:33 the ability for other packages to drop in the configuration files is the key thing. we should also make sure that people name the configuration file according to the package name. no exceptions. Mar 23 16:23:45 KsiTreo: BlueSlug? Mar 23 16:24:15 One question: I have brushed up openvpn, but have no time to deal with it now. I want the kids to have an easter toy - so it is in CVS and compiles cross here. Shall I mark it ready for testing even though I don't know if it works? Mar 23 16:24:22 CDMA Mar 23 16:24:28 I'm in a car. Mar 23 16:26:19 poof go I. Mar 23 16:26:20 KsiTreo: Don't tell me you're driving? Mar 23 16:27:30 rwhitby-web: I made the decision - the answer is yes. I will inform on the list that it is pre-alpha. Mar 23 16:28:04 good plan Mar 23 16:28:18 lots of decisions today :-) Mar 23 16:28:32 Decisions are good, at least when they are the right ones :-) Mar 23 16:28:40 Sometimes they are good anyway Mar 23 16:29:19 has anyone built 4.7-alpha yet? Mar 23 16:29:49 I'm especially looking for testing of external rootfs on sda1 or sdb1 Mar 23 16:30:01 and how that interacts with stock apps Mar 23 16:37:54 03ingeba * 10unslung/Makefile: OpenVPN and lzo ready for cross testing Mar 23 16:43:06 bob_tm-away: Papal decree sent. Mar 23 16:43:26 rwhitby-web: Proud of you Mar 23 16:43:36 aww gee Mar 23 16:43:39 :-) Mar 23 16:45:28 I think xinetd should enable telnet by default. Mar 23 16:46:02 Read it - good call. Yes - I thought the same. They can always to "disable=yes" once they have a stable dropbear. Mar 23 16:46:19 security conscious people can easily disable it. and now that we have conffiles working, their modification will stick Mar 23 16:46:44 What about conflicting files? We let people fix it manually? Mar 23 16:46:57 what conflicting files? Mar 23 16:47:00 eg vsftp conf vs. proftpd conf Mar 23 16:47:08 or.. (good heavens) ftpd conf Mar 23 16:47:16 put Conflicts in the ipk Mar 23 16:47:38 Yes - I will add that. Maybe ftpd should either die or become a package of its own then? Mar 23 16:47:42 we should deprecate ftpd out of inetutils too Mar 23 16:47:50 Yes - agreed Mar 23 16:48:06 The two others are both good sw Mar 23 16:58:01 03ingeba * 10unslung/make/findutils.mk: Very minor change - just want it in sync with CVS and don't plan to make further changes for a while Mar 23 17:07:21 rwhitby-web: Please remind me that we should do a brushup of /dev to have device nodes for the device driver modules we expect to be used. Also we should consider Josh's pty change if it is not already in. Mar 23 17:09:02 yes, we can add as many /dev nodes as we want. I've already added 4 ttyUSB nodes Mar 23 17:09:49 I'm looking for josh to commit to testing the pty change before it goes in. as far as I can see, it's not absolutely essential for the 4.x-beta, but I'm happy to be convinced otherwise. Mar 23 17:16:47 bob_tm-away: you need to get bk working, so you can add /dev nodes to openembedded/files/device_table-unslung.txt yourself ... Mar 23 17:17:02 Yes (*blush*) Mar 23 17:33:42 Just had the thought thought that ftpd might be worth saving for those who want to run on flash only - do we intend to support that? Found in the process that inetutils does not strip execs. Mar 23 17:43:18 if you're running flash only, what do you have available to provide via ftp? only the internal flash contents - not very useful Mar 23 17:43:27 back later Mar 23 17:46:22 has anybody tried connecting a usb sound card to the slug? Mar 23 18:18:57 novacane_22: Not that I know of. Don't forsee any trouble with it however assuming the applications you run are okay with it being a big endian system Mar 23 18:19:24 wasn't there some discussion in the mailing list? Mar 23 18:19:34 Maybe. Can't remember Mar 23 18:19:47 :) Mar 23 18:20:01 If so, feel free to search the archives at yahoogroups.com :) Mar 23 18:20:55 or use the HTTP <==> Email Gateway to search them Mar 23 18:29:03 thanks, I did search the archives the other day and there was some posts on usb sound back in Dec, but it does not look like anything come of it Mar 23 19:08:04 hey dyoung-web Mar 23 19:08:19 tried 4.7 yet? Mar 23 19:14:35 Nope not yet. Mar 23 19:14:46 Thats for the weekend I think Mar 23 19:14:57 maybe tonight I can look Mar 23 19:39:16 we need to test maintenance mode (the fact that it shouldn't appear as an option if running from external disk) Mar 23 23:22:54 03bzhou * 10unslung/ (Makefile make/xterm.mk): xterm ready for testing (cross) Mar 23 23:26:11 03bzhou * 10unslung/ (Makefile make/bc.mk): bc ready for testing (cross) Mar 23 23:47:15 bob_tm-away: no problem! ;) Mar 23 23:50:58 Hi siddy-away Mar 23 23:52:03 hey dyoung! Mar 23 23:52:16 CSR 1.5 got the better of me. Mar 23 23:52:40 Were you on Holiday BTW? Mar 23 23:53:17 No, no holiday, I've been quite busy last days. :\ Mar 23 23:53:31 exam, cebit, work, etc pp... ;) Mar 23 23:53:38 sorry to hear that, I thought might be on holiday! Mar 23 23:54:00 What's the improvement of CSR 1.5? Mar 23 23:54:34 Far as I can tell no salient feature improvement for our applications. the big thing is its LE Capable. Mar 23 23:55:10 ah, so we could switch the whole openslug to le? Mar 23 23:59:01 In theory; but that would require changes to the bootloader. Mar 23 23:59:08 And I'm not sure how worth it it is. Mar 23 23:59:18 Gut feeling tells me, "waste of time". **** ENDING LOGGING AT Wed Mar 23 23:59:56 2005