**** BEGIN LOGGING AT Fri Feb 13 02:59:57 2009 Feb 13 03:24:38 the removal of dnsmasq.conf and adding /etc/config/dhcp makes configuring dnsmasq a guessing work Feb 13 03:24:58 you can just continue to use /etc/dnsmasq.conf Feb 13 03:25:14 xMff: it's not there... Feb 13 03:25:21 you mean i can add one brandnew Feb 13 03:25:22 xxiao: create it Feb 13 03:25:50 I usually just keep dhcp stuff in /etc/config/dhcp and ad the advanced stuff to dnsmasq.conf Feb 13 03:25:55 *add Feb 13 03:27:03 xMff: i have one ethernet and one wifi card, so the ethernet is a WAN and wifi-AP assigns IP to clients at a wireless LAN Feb 13 03:27:32 ok Feb 13 03:28:20 the WAN takes 192.168.1.x and I'm assiging 192.168.2.x to the AP at /etc/config/network, but not sure how to let dnsmasq to pick up 192.168.2.100-150 Feb 13 03:28:59 http://mibbit.com/pb/y8cIXj Feb 13 03:29:26 basically the WAN ethernet get a IP from a wire LAN at 192.168.1.90 Feb 13 03:29:55 how did you called your wifi? Feb 13 03:30:22 nvm Feb 13 03:31:00 as I have no local LAN ethernet port, I added "config interface wifi" to /etc/config/network, not sure if it's the right thing to do Feb 13 03:31:04 "option interface" in the dhcpconfig refers to the logical name from the network config Feb 13 03:31:13 in your case it would be option interface wifi Feb 13 03:31:30 you see that's what i'm trying to do, i used wifi at dhcp Feb 13 03:31:36 oops Feb 13 03:31:44 i should not use wlan0 there Feb 13 03:32:01 yep, but wifi which descripes wlan0 Feb 13 03:32:46 and you can remove the gateway option from the wifi interface Feb 13 03:33:51 xMff: thanks. trying a reboot now. Feb 13 03:34:07 somehow my wifi setting is screwed up, the ssid is not shown in iwconfig Feb 13 03:34:40 is hostapd installed? Feb 13 03:34:44 or hostapd-mini Feb 13 03:35:24 hostapd is installed, not sure how to use them yet :9 Feb 13 03:35:25 :( Feb 13 03:37:57 do i need configure hostapd somehow? googled nothing out. Feb 13 03:40:59 could not find anything like hostapd.conf... Feb 13 03:41:10 nope Feb 13 03:41:22 the openwrt scripts will setup hostapd Feb 13 03:41:25 it's running though Feb 13 03:41:30 hm Feb 13 03:43:28 even though i put "ap" to /etc/config/wireless, i saw the wifi is brought up as "managed", and SSID is correct, then after cfg80211, it switches to "Master", but SSID disappears Feb 13 03:44:50 http://mibbit.com/pb/eDtta4 Feb 13 03:45:18 After "cfg80211: Calling CRDA for country: US" it switches to Master but SSID disappears Feb 13 03:52:56 xMff: it works now, i had to manually set up the option channel 11 Feb 13 03:53:17 if it's "option channel auto" then it did not work. thanks! Feb 13 03:53:30 np Feb 13 03:55:27 Bartman007: I'm ready :) Feb 13 03:55:29 Bartman007: yo! Feb 13 03:56:16 Bartman007: What should I do about the possibility of being locked out of the thing if it thinks the ethernet port is the wan? Should I flash it with an open ap in /etc/config/wireless ? Feb 13 03:56:51 What does everyone think? Feb 13 03:57:40 Bartman007: is there an OpenWrt presence at SCaLE? if not, a last-minute conundrum may make it a good idea Feb 13 04:02:04 noz: Around to help? Feb 13 04:07:01 I'm trying ath9k 802.11n, will /etc/config/wireless honor "option agmode n" now? Feb 13 04:07:46 xxiao: wasn't agmode changed to hwmode? Feb 13 04:08:04 sn9: i used agmode and it worked ok Feb 13 04:08:10 ok... Feb 13 04:08:24 i've been out of the loop a bit Feb 13 04:08:37 sn9: i do notice that "otion type mac80211" works, while "option type atheros" did not Feb 13 04:08:44 yes Feb 13 04:09:13 i helped make the mac80211 scripts usable Feb 13 04:09:46 i also think hostapd.sh has to be hacked for 802.11n, anyway i'm going to try it Feb 13 04:10:10 possibly Feb 13 04:12:21 alright Feb 13 04:12:45 I'm trying with the 330ge :) Feb 13 04:18:37 i want to use busybox/netcat to get an image, but nc can not listen? what does "nc iP port" means? i normally do "nc -lp port > filename" Feb 13 04:19:05 it has to have been compiled in Feb 13 04:19:37 sn9: but nc is in, you mean i have to configure busybox for a full-future nc or something? Feb 13 04:20:16 nc listen support is a separate option from nc Feb 13 04:20:51 sigh...thanks, will do just that Feb 13 04:21:09 want to get cpu-bench on the board Feb 13 04:23:40 i hate to say that scp is already in the flash :) Feb 13 04:28:12 Test 1 worked, it has a hold-reset failsafe mode just like the wl-5x0gx family Feb 13 04:28:28 I flashed with the latest asus firmware, it accepted it on 192.168.1.220 oddly enough Feb 13 04:28:58 I'm just building a custom trx to test openwrt with that leaves the wifi open (since I'm not sure what's gonna happen with the ethernet port) Feb 13 04:30:53 brb Feb 13 04:35:12 xxiao: I don't think agmode has n, but haven't really looked. ht_capab or wahtever in hostapd.sh for sure works Feb 13 04:39:15 RoundSparrow: by works, you mean hostapd.sh will ignore whatever mode is in /etc/config/wireless? Feb 13 04:39:49 can you show me /etc/config/wireless, as no cocument was found for 802.11n on openwrt Feb 13 04:40:52 so far g mode works... Feb 13 04:41:36 http://mibbit.com/pb/pla3f7 my /etc/config/wireless Feb 13 04:53:20 works great Feb 13 05:00:59 root@OpenWrt:/tmp# cat /proc/diag/model Feb 13 05:00:59 ASUS WL-500g Deluxe Feb 13 05:01:26 The power LED is working perfectly in /proc/diag/power -- and the wireless and ethernet leds are used internally for traffic Feb 13 05:01:52 Bartman007: ping ping ping ping ping :P Feb 13 05:02:37 yes, ping ping Feb 13 05:03:18 sn9: What do you know about probing GPIO ? Feb 13 05:03:38 on which board? Feb 13 05:03:53 broadcom? Feb 13 05:05:14 Yeah Feb 13 05:05:19 Standard broadcom board Feb 13 05:05:58 not much to know there Feb 13 05:06:01 same/similar as the wl-520gu, I think Feb 13 05:06:14 sn9: I just want to get to the reset button properly detected Feb 13 05:06:21 CFE controls it some Feb 13 05:09:55 Yeah, it's the same board as the wl-520gu -- the board got USB support, Feb 13 05:11:16 can you trace the lines on the board? that's how i found the reset on the airlink101 ar525w Feb 13 05:11:57 I haven't opened it up Feb 13 05:15:55 gpio was the same as WL-520gU Feb 13 05:15:56 development without opening things up is usually a losing proposition Feb 13 05:16:04 sn9: I won! Feb 13 05:16:39 gpio was same as WL-520 minus WLAN LED, which I think is given to the wireless chip from CPE Feb 13 05:16:41 good Feb 13 05:17:24 CFE Feb 13 05:17:30 er, I always move up that acronym Feb 13 05:21:29 very cool, in the Asus toolchain, the only differences between the configuration for CFE between the WL-520gU and the WL-330gE are comments on how the WLAN module controls the LED Feb 13 05:21:41 And nvram default settings for board version, etc Feb 13 05:23:35 gpl tarballs are good Feb 13 05:23:58 They're a freaking mess with knowing what to grep for Feb 13 05:27:07 s/with/without/ Feb 13 05:56:16 xxiao: you figure it out? Feb 13 05:57:25 xxiao: I think you can look at iw & iwconfig to figure out chip capab. http://www.google.com/search?hl=en&q=site%3Aopenwrt.org+ath9k+ht_capab&btnG=Google+Search&aq=f&oq= Feb 13 05:57:52 ah, left room an hour ago Feb 13 06:06:59 batman-advanced it build-fail'ing (x86-net48xx) Feb 13 06:07:04 is* Feb 13 06:07:29 install -m0755 /aux/src/openwrt/build_dir/target-i386_uClibc-0.9.29/batman-adv-r1220/ipkg-install/usr/sbin/batmand-adv /aux/src/openwrt/build_dir/target-i386_uClibc-0.9.29/batman-adv-r1220/ipkg/batmand-adv/usr/sbin/ Feb 13 06:07:29 install: cannot stat `/aux/src/openwrt/build_dir/target-i386_uClibc-0.9.29/batman-adv-r1220/ipkg-install/usr/sbin/batmand-adv': No such file or directory Feb 13 06:07:29 make[3]: *** [/aux/src/openwrt/bin/packages/target-i386_uClibc-0.9.29/batmand-adv_r1220-1_i386.ipk] Error 1 Feb 13 06:08:35 looks like some kind of makefile typo Feb 13 06:09:40 ipkg-install doesn't exist Feb 13 06:09:55 oh, wait Feb 13 06:10:08 strike that Feb 13 06:11:10 How should I handle /etc/init.d/netconfig when I (a) have only one ethernet port that's a 5-vlan taggable switch under the hood, and (b) want no wan Feb 13 06:34:27 hmm. it looks like the to-be-installed files aren't getting compiled first Feb 13 06:36:34 jumps straight to install Feb 13 06:50:35 is there any way to automate the make menuconfig picking with consistent behavior? I guess the buildsystem is for that, but almost like a uci for the makefile "makeset package ath9k=n" Feb 13 06:52:55 edit the makefiles Feb 13 06:56:17 yha. the probably more direct issue is ar71xx platform needs some profiles. perhaps: generic-ath5k, generic-ath9k, AP81, AP81USB Feb 13 06:56:36 right now it defaults to madwifi for all Feb 13 06:57:17 right Feb 13 07:16:08 kupesoft: Sorry - I'm in GMT, so wasn't around at 4am! Looks like you've made good progress though. Feb 13 07:18:27 11:18 pm in Bartman007's timezone Feb 13 07:23:57 noz: Yup, was pretty straightforward, Feb 13 07:24:13 noz: I'll be in GMT in about 24 hours ;P Feb 13 07:24:33 noz: Just a few questions about treating a one-port device in netconfig Feb 13 07:24:40 Really just one, how to do it? :P Feb 13 07:25:46 but that can I wait, I have to get some sleep and pack, Feb 13 07:25:50 happy hacking Feb 13 07:27:02 noz: well I have pinged kaloz twice.... once yesterday... and once today...... Feb 13 07:27:08 no answer.. Feb 13 07:27:28 so I am thinking the arm mailing list Feb 13 07:28:15 for what is so terribly different about an ixp42x A0 stepping and a B0 stepping... with the NPE's. Feb 13 07:29:09 Google is pissing me off....... i can't seem to find what I need using it. Feb 13 07:36:25 johnrw: http://loveyuna.egloos.com/1898401 Feb 13 07:40:26 RoundSparrow: thanks for that! Feb 13 07:40:40 I guess I should be googling for A0 silicon! Feb 13 07:47:14 So what it is saying is that on A0 silicon... both coprocessors are automatically available... Feb 13 07:48:13 funny because noz had suggested and threw in a print_npe right around the feature check... Feb 13 07:48:33 <------><------>struct npe *npe = &npe_tab[i]; Feb 13 07:48:33 <------><------>if (!(ixp4xx_read_feature_bits() & Feb 13 07:48:33 <------><------> (IXP4XX_FEATURE_RESET_NPEA << i))) { Feb 13 07:48:33 <------><------><------>print_npe(KERN_ERR, npe, "%d not present\n", i); Feb 13 07:51:12 but that didn't produce any msg... either because it went ok... or wasn't exec'ed Feb 13 08:07:34 ok... so where is ixp4xx_read_feature_bits() defined? Feb 13 08:11:19 okay... got it... arch/arm/mach-ixp4xx/include/mach/cpu.h:static inline u32 ixp4xx_read_feature_bits(void) Feb 13 08:20:26 RoundSparrow: Would this be the same for both steppings? Feb 13 08:20:32 /* Processor id value in CP15 Register 0 */ Feb 13 08:20:32 #define IXP425_PROCESSOR_ID_VALUE 0x690541c0 Feb 13 08:21:41 the kernel identifies it correctly as a A0 stepping on bootup... (rev 0) Feb 13 08:33:00 sn9: hey, long time no see. I was hoping I could make it down to SCaLE, but I cannot. nslu2-linux will be there. Feb 13 08:33:53 Bartman007: got a last-minute booth-staffing cancellation Feb 13 08:35:04 Bartman007: http://scale7x.socallinuxexpo.org/dotorg/silicon-valley-linux-users-group Feb 13 08:35:19 freenas and m0n0wall Feb 13 08:35:37 perfect opportunity for openwrt Feb 13 08:38:49 sn9: I'm not sure who would be able to make it. probably best to speak to ka6sox of the nslu2 project to see what their presence is going to be Feb 13 08:40:08 I'm a bit stretched time-wise, and any usable liquid funds were likely blown via ar71xx fever. Feb 13 08:40:29 _more_ ar71xx? Feb 13 08:41:36 well, all the ones I got are actually ar91xx, but they use the ar71xx target :) Feb 13 08:41:56 even wnr2000? Feb 13 08:42:04 Bartman007: you went to ebay! shame, shame, triple shame... :) Feb 13 08:42:15 johnrw: amazon Feb 13 08:42:50 you said you were gonna deadroute ebay... so amazon worked... Feb 13 08:42:52 sn9: it is booting via tftpboot, I haven't gone back to it in a long time though Feb 13 08:43:24 hmm, well i should probably sleep now Feb 13 08:43:28 that's funny... Feb 13 08:43:51 on the wnr2000 netgear/dni hacked u-boot to checksum the rootfs on boot, and the checksum behavior is hardcoded into u-boot Feb 13 08:44:10 sn9: sorry for interrupting... I am outa here... Feb 13 08:44:46 ugh, bootloader replacement necessities suck... Feb 13 08:45:29 fixing it is simple enough, just remove a #define in the source, but then it's a one-shot flashing attempt due to lack of jtag Feb 13 08:46:35 well, catcha l8r... Feb 13 08:46:38 so I have other ar91xx devices of similar design with jtag to test it on :) Feb 13 08:46:43 adios. Feb 13 08:46:56 Bartman007: no jtag on the board ? Feb 13 08:46:59 or not supported ? Feb 13 08:47:34 blogic: both Feb 13 08:47:35 no jtag on the netgear wnr2000 unless it is hiding in various testpads/resistors on the board. Feb 13 08:48:21 they made a nice device that runs openwrt but soiled it by omitting jtag and using rootfs checksums in the bootloader. Feb 13 08:48:45 though if the latter wasn't there I'd have no need for the former :) Feb 13 08:51:11 so you'd have to do one-go bootloader tries :P Feb 13 08:56:19 juhosg * r14493 /trunk/target/linux/ar71xx/ (config-2.6.26 patches-2.6.26/): [ar71xx] remove 2.6.26 specific stuff Feb 13 08:59:13 juhosg * r14494 /trunk/target/linux/ar71xx/ (3 files in 3 dirs): [ar71xx] remove netif schedule from the ag71xx driver Feb 13 09:27:28 hi, any dev bored and wants to take a look at ticket #4571 and #4562? both come with patches and i would like some feedback on them. if they suck i will consider improving them. Feb 13 10:43:43 juhosg * r14495 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ (5 files): [ar71xx] ag71xx driver: update phy connecting code Feb 13 11:17:28 juhosg * r14496 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ (ag71xx.h ag71xx_main.c): [ar71xx] ag71xx driver: fix return code of the napi poll routine, was broken since [13545] (patch by Ubiquiti Networks) **** BEGIN LOGGING AT Fri Feb 13 12:02:41 2009 Feb 13 14:37:15 nbd * r14497 /trunk/target/linux/x86/ (config-2.6.23 config-2.6.24 config-2.6.25 config-2.6.26): nuke old x86 kernel configs Feb 13 14:37:30 nbd * r14498 /trunk/target/linux/x86/ (config-2.6.27 config-2.6.28): disable SMP by default on x86 Feb 13 14:38:56 nbd * r14499 /branches/8.09/target/linux/x86/ (config-2.6.23 config-2.6.24 config-2.6.25): merge r14497, r14498 to 8.09 Feb 13 15:13:18 markus * r14500 /trunk/target/linux/ (2 files in 2 dirs): Feb 13 15:13:18 - new upstream kernel 2.6.28.5 Feb 13 15:13:18 - the fix-serial_h patch does not apply anymore and is integrated Feb 13 15:13:18 upstream now Feb 13 17:38:07 markus * r14501 /trunk/target/linux/brcm47xx/ (Makefile config-2.6.28): Feb 13 17:38:07 changed default kernel to 2.6.28.5, it at least runs on Feb 13 17:38:07 Asus WL-500g Premium. Feb 13 17:38:07 The earlier version 2.6.28.4 is also running on WRT54G3G Feb 13 17:38:07 so I don't expect any trouble here. Feb 13 17:43:01 \o/ Feb 13 17:44:01 :) Feb 13 17:44:10 _ Feb 13 17:46:24 is markus here? Feb 13 17:46:34 Thanks. :) I was running 2.6.28 anyway :) Feb 13 17:55:34 mb * r14502 /trunk/include/kernel-version.mk: Add 2.6.28.5 md5sum Feb 13 18:13:46 who is marek? Feb 13 18:16:06 i have a patch for the batman build damage i found last night Feb 13 18:17:21 russell_: could you pastebin it? Feb 13 18:17:36 yes, standby Feb 13 18:19:07 xMff: http://pastebin.ca/1336276 Feb 13 18:19:27 problem was looking for config=y, whereas i had config=m Feb 13 18:19:39 ic Feb 13 18:19:39 (afaict) Feb 13 18:19:57 looks good, I'll commit it Feb 13 18:20:01 thansk Feb 13 18:22:13 jow * r14503 /packages/net/batman-advanced/Makefile: [packages] fix batman-advanced compilation as package, thanks russell Feb 13 18:41:53 In file included from dsl_qos_queue.c:15: Feb 13 18:42:29 /aux/src/openwrt/staging_dir/target-i386_uClibc-0.9.29/usr/include/linux/netfilter.h:44: error: field 'in' has incomplete type Feb 13 18:42:29 /aux/src/openwrt/staging_dir/target-i386_uClibc-0.9.29/usr/include/linux/netfilter.h:45: error: field 'in6' has incomplete type Feb 13 19:09:39 hmm. also running into a uvc_v4l2.c error that looks awfully familiar. Feb 13 19:09:58 juhosg: do you think prom.c should change LS-SR71 to LSSR71... Feb 13 19:10:39 juhosg: it's more consistent as other places we use UBNT_LSSR71 Feb 13 19:48:35 Can someone familiar with the build system tell me where/how the different directories in target/linux//[-] are chosen? Feb 13 19:50:03 In some of the target directories, I just see files/ and in others I see files-2.6.24/ files-2.6.25/ etc Feb 13 19:51:12 Do the make scripts try to find a files-/ that matches what is in the version specified in the target/Makefile for LINUX_KARCH? Feb 13 19:53:19 target/linux//files is applied, the target/linux//files- is applied Feb 13 19:53:39 s/the/then/ Feb 13 19:54:28 Bartman: ok, so it applies whatever is in files/ first, then looks for a files-/ and if found, then applies that Feb 13 19:54:45 does it work the same for the config files? Feb 13 19:55:15 or does it prefer a config- over a config-default? Feb 13 19:55:28 config- is used over config-default Feb 13 19:55:57 great, thanks for the info Feb 13 20:20:23 What is a 'profile' in the context of the OpenWRT build system? Feb 13 20:21:53 What kind of board variations would dictate the need for a 'profile'? Feb 13 20:24:59 are vmware/vbox trunk images still broken even with https://dev.openwrt.org/changeset/14460? Feb 13 20:26:39 profiles are just different sets of packages afaict Feb 13 20:32:21 swalker: thanks. i see a lot of targets don't even have a profiles folder, then only a few that do actually have anything in them Feb 13 20:34:21 Is the Kernel Modules ---> section in OpenWRT configuration menu just a shortcut to adding the support through the kernel's config? Feb 13 20:35:03 pitbullpc: yes Feb 13 20:35:16 but not all options are covered there Feb 13 20:36:12 yeah, just looked like a very special subset for things related to adding functionality to a WAP --- go figure :) Feb 13 20:36:41 juhosg * r14504 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ (ag71xx.h ag71xx_main.c): [ar71xx] ag71xx driver: fix a kernel crash caused by r14496 Feb 13 20:36:45 was the intent to not really have to do a make kernel_menuconfig? Feb 13 20:39:14 Also, is there an OpenWRT developer familiar with powerpc that might be interested in reviewing/guiding me on the proper way to add a new extensible target platform to the build system? Feb 13 20:39:37 pitbullpc: Kaloz and [florian] Feb 13 20:40:30 I am adding support for the Freescale MPC8313E-RDB, which really means that I should make it flexible enough to handle MPC831X based boards Feb 13 20:40:53 what/when is the best way to interact with them? Feb 13 20:41:54 @openwrt.org or via /msg Feb 13 20:42:11 /msg while take n hours to get a response, but the savest way imho Feb 13 20:42:24 /msg will take $n hours to get a response, but the savest way imho Feb 13 20:42:35 thanks blogic Feb 13 20:43:45 pitbullpc: :) Feb 13 20:50:40 has anyone been playing the ubiquity's routerstation? Feb 13 20:51:57 i've got one here with it's stock fw which is r13xxx something kamikaze, no link lights on LAN2 Feb 13 20:54:26 blogic: do you know how to protect the dl/ ? I thought I read somewhere that builds could be sped up by keeping a local dl/ instead of it being deleted everytime? Feb 13 20:55:04 yes Feb 13 20:55:06 erm Feb 13 20:56:10 [*] Advanced configuration options (for developers) ---> Download Folder Feb 13 21:05:44 so, if specify the path there, it wont try to kill it? Feb 13 21:15:05 ok, i specified a path for dl/ and packages/. i currently have nothing in them. however, i did a make kernel_menuconfig to see how it would try to use them. it first looked in the dl folder, for .tar.bz2, but didn't find it --obviously, its empty. But what I noticed was the preference of tar.bz2 over tar.gz. Is this common among packages or just the kernel? Feb 13 21:15:45 no idea Feb 13 21:15:53 i personally would use bz2 Feb 13 21:16:32 ok, educate me on _blogic_ :) Feb 13 21:17:03 bz2 is better at compressing text Feb 13 21:17:17 quite a bit better Feb 13 21:17:23 maligor: indeed Feb 13 21:17:40 fantastic, knowledge is appreciated Feb 13 21:17:50 so uuencode your binaries first! Feb 13 21:17:59 (joke) Feb 13 21:18:11 lol, yeah, i got it Feb 13 21:19:02 you know, I have actually seen that done though Feb 13 21:19:56 i had the displeasure of working on a project where someone thought it a good idea to turn structs into ASCII and pass that between processes Feb 13 21:20:17 heh Feb 13 21:20:38 but xmlrpc might qualify for that Feb 13 21:20:51 but it's usually used remote Feb 13 21:21:12 exactly, this was code on an embedded target Feb 13 21:21:43 passing structs around to various tasks in the _same_ address space Feb 13 21:22:09 so it was a experiment on how to make the program use as much memory as possible without resorting to unused memory allocations :P Feb 13 21:22:47 haha, yeah, you could say that :) Feb 13 21:23:08 xmlrpc on embedded? overhead:payload = 10:1 ;-) Feb 13 21:24:21 actually, the geniuses did try to build xerces into the app Feb 13 21:25:07 tried to handle 6M xml files of data extracted from database Feb 13 21:34:50 pitbullpc, did they also solder a few... er...... 20 more memory modules to it? Feb 13 21:36:47 128M RAM, but 16G of flash Feb 13 21:37:06 nope, still no luck with .28 on a WGT634U. Hangs after diag: Detected 'Netgear WGT634U' Feb 13 21:38:09 pitbullpc, the sloppiness caused by modern pc's eh Feb 13 21:41:18 yeah, something along the lines of calling an electrician to build you a microprocessor Feb 13 21:50:01 nice, it does work without kmod_diag Feb 13 21:53:37 phy0: device does not respond! p54pci: probe of 0000:00:01.0 failed with error -16 <- that doesn't look too good Feb 13 22:10:16 pitbullpc: the idea is that for any file that will be compressed once and transferred over a network many times you should compress it as small as possible. obviously a little bit of extra CPU time when generating the official kernel tarball is well worth the 14MB (for 2.6.28.5) size difference multiplied by the thousands of times it is downloaded. Feb 13 22:12:17 Bartman007: so I picked up a dir-615c1 before it is replaced by the d1 (already reported by OpenWRT user). I see that it uses an ar8216 switch chip? I thought we didn't have drivers for that Feb 13 22:13:16 RoundSparrow: we do have drivers for it, the trendnet boards use it also Feb 13 22:13:47 I asked juhosg why it doesn't work on some boards and it is due to two different configuration methods Feb 13 22:15:40 huh, ok, I just assumed he was saying the Planex had a unique chip; didn't realize it was an issue of same chip but init issues Feb 13 22:18:08 most devices (like the trendnet boards) use an mdio interface to configure the switch, but some (like the planex) use ethernet packets, and we don't know what the payload of those packets is Feb 13 22:18:19 Bartman007: yeah, i got it thanks Feb 13 22:18:27 weird, my kvm freaked out and I wasn't able to reconnect to this machine. Feb 13 22:27:38 loswillios: I have a little experience with p54 wireless... does it download the firmware? Feb 13 22:28:34 looks like it never makes it to that part of the sequence Feb 13 22:31:27 johnrw: yes, fw loads: http://pastebin.com/m7a3cb22f Feb 13 22:31:32 Bartman007, they use out of kernel or bootloader?... or just not linux atall Feb 13 22:32:14 johnrw: I fear it's something on a lower lever, maybe the kernel doesn't like the bus or something Feb 13 22:32:31 where did you get the firmware?? Feb 13 22:32:54 from wireless.kernel.org Feb 13 22:33:07 maligor: I assume both the bootloader and firmware (linux) do it, but we don't have any gpl tarball for the planex ar91xx boards Feb 13 22:33:52 the ar8216 datasheet would be nice also :) Feb 13 22:33:57 Bartman007, violators huh? Feb 13 22:34:14 okay... i guess that's the same... as http://linuxwireless.org/en/users/Drivers/p54#firmware Feb 13 22:34:34 yes Feb 13 22:34:58 actually, scratch that part with the kernel. looking closer at the logs it seems to detect it Feb 13 22:36:20 need to figure out what error -16 is. but probably it just means something runs into a timeout (when the card doesn't respond) Feb 13 22:37:41 Planex has not responded to two emails sent in past 48 hours to GPL request (sent to tech support) Feb 13 22:38:14 RoundSparrow: I think you can give up counting in hours Feb 13 22:38:34 yha, I know, but they could at least acknowledge the support request ;) Feb 13 22:38:42 we could use the theme of the day and count in epoch(s) :P Feb 13 22:38:50 ;-) Feb 13 22:38:51 right... that error 16 is googleable... remove the - Feb 13 22:39:41 - means in googlespeak... to NOT include the item following... the - Feb 13 22:40:43 yeah Feb 13 22:42:17 Hi, I'm creating a package but I've got a doubt. The library I want to port compile (NOT links) with some .c files that came from a "compatibility package". Feb 13 22:43:19 Should I "embedd" those .c files inside my package , or should I copy all these compatibily files inside the staging dir ? (or what ? :) ) Feb 13 22:46:36 loswillios: In the pastebin... it says... device does not respond! Feb 13 22:49:19 loswillios: what kernel are you running? Feb 13 22:51:02 2.6.28.5 johnrw Feb 13 22:51:14 according to the page... Unless you are running 2.6.29+ you should be using this firmware... http://daemonizer.de/prism54/prism54-fw/fw-softmac/2.13.1.0.arm Feb 13 22:51:27 yeah, tried that as well Feb 13 22:51:53 but with compat-wireless from five days ago, it's probably the .29+ one Feb 13 22:52:44 ok... just looking for obvious stuff... Feb 13 22:52:54 maybe the maintainer knows where to start.. I wrote him an email Feb 13 22:53:48 well if the card doesn't respond... Feb 13 22:53:59 ah, he responded already Feb 13 22:56:12 the card boots the firmware but doesn't respond afterwards, so -16 means busy/timeout Feb 13 22:56:45 compat-wireless is well into 2.6.30 Feb 13 22:57:17 he suggested to try other firmwares and/or increase the timeout Feb 13 22:58:26 have to go now, laters Feb 13 23:16:49 I have JTAG cables now, so I may attempt my first jtag work... I have Trendnet 652 that I overwrote u-boot and all by putting 6 kilos of crap into a 4 kilo bag. Feb 13 23:17:21 shoulda used cramfs Feb 13 23:17:41 I enabled debug kernel symbols ;) Feb 14 00:09:29 nbd * r14505 /trunk/toolchain/gcc/patches/ (3 files in 3 dirs): fix gcc 4.3.x segfault triggered by using -frename-registers (reproduced on arm, might be relevant for other platforms as well) Feb 14 00:10:10 good evening everyone Feb 14 00:12:35 after more than 8 time that i get a syntax error while compiling on my native system (debian) and then on a VM (clean debian) based on vbox Feb 14 00:12:49 and then an other vm (ubuntu) and the last test is on a native ubuntu every time i get a syntax error while compiling the trunk Feb 14 00:13:17 please show me the error Feb 14 00:13:21 i would like to ask if anyone of you guys running his dev env in a vm img which he can share Feb 14 00:14:50 i don't do any dev work in a vm Feb 14 00:15:14 so it might be helpful if you could just copy&paste the error to a pastebin site Feb 14 00:15:15 I'll probably put together a dev vm at some point, I just haven't gotten around to it. Feb 14 00:15:57 Bartman007 you were there as i pasted the full error log in #openwrt right? Feb 14 00:16:17 nbd i reinstalled my system from debian to ubuntu and i have no env now Feb 14 00:16:34 I think so, it's been a while, it was an ld related error IIRC. Feb 14 00:16:46 syntax error in ld.script Feb 14 00:17:04 similar to this: http://forum.openwrt.org/viewtopic.php?id=18076 Feb 14 00:18:55 oh yeah. I've tried to reproduce it and cannot. Feb 14 00:31:13 maybe someone here is having an image which have mroute and multicasting enabled ? Feb 14 00:31:48 i just can't get my iptv stuff working with those two thing enabled :s Feb 14 00:32:02 s/with/without/ Feb 14 01:07:29 hey Bartman007 i just set up a new Debian / KVM based Image and wanted to re try compiling from the scrach, anyone would mind to guide my steps ?? Feb 14 01:07:50 Debian Etch Feb 14 01:11:59 nbd, do you happen to have a link for me (compiling openwrt and dependencies?) Feb 14 01:12:58 yassine: I use Knoppix's latest dvd iso... and grub4dos to boot it up on a windows xp machine. This is about half of it, the other half being - one must create a knoppix "persistent home" on the xp drive... which is done in the knoppix menu on the start bar. Feb 14 01:13:47 It creates a single file... knoppix.img that holds any changes you make... from reboot to reboot. Feb 14 01:14:09 my knoppix.img is 30GB Feb 14 01:14:24 and i build openwrt on it all the time. Feb 14 01:15:23 I can have you booting an environment in a half hour... that you can delete by deleting 2 files... Feb 14 01:16:14 johnrw, i was thinking since now many distros supporting vm that such an image is just easy to managed ? Feb 14 01:16:34 i did have to install quilt from apt... Feb 14 01:16:52 the problem with vm... is speed Feb 14 01:17:24 a pseudo environment has a false processor Feb 14 01:18:04 would you like a vm that takes hours more to build...? Feb 14 01:18:07 johnrw, i have installed nothing sofar Feb 14 01:18:22 i wanted to install only the minimum necessary Feb 14 01:18:30 svn, buildutils etc Feb 14 01:18:45 well are you running xp on your host? Feb 14 01:18:55 no ubuntu Feb 14 01:19:22 and the vm is debian based Feb 14 01:20:05 well I know MateIn4 here... uses Ubuntu... but I just use knoppix. Feb 14 01:21:18 I know that Gilles Van Ruymbeke make's vm's that run knoppix... in the vm Feb 14 01:22:39 johnrw: decent vm solutions still execute usermode code natively, they hook the kerenl stuff. Feb 14 01:23:11 but that often means that ./configure is slow which much of the compilation runs at a decent speed. Feb 14 01:24:03 Bartman007, "apt-get install build-essential subversion" i guess this is everything needed right? Feb 14 01:24:12 okay... Gille's is really high up there on my list of people I can learn from... his post announcing a knoppix 5.3.1 vm (with persistence) is somewhere in this thread... http://www.knoppix.net/forum/viewtopic.php?p=119802#119802 Feb 14 01:24:31 yassine: the buildsystem should yell at you for everything else it needs. Feb 14 01:24:58 yep the libz etc Feb 14 01:25:13 adios. Feb 14 01:25:22 good night Bartman007 Feb 14 01:26:24 ok... here is the vm announcement... http://www.knoppix.net/forum/viewtopic.php?p=118954#118954 Feb 14 01:27:14 yassine: did you try make V=99 prereq Feb 14 01:27:53