**** BEGIN LOGGING AT Tue Aug 17 02:59:57 2010 Aug 17 03:00:14 ah, it seems target/imagebuilder/files/Makefile just loops through them all and enables all of them Aug 17 04:45:12 jow_laptop, xMff will you do me a favor? Aug 17 04:46:22 the oldwiki page for my device contains some bad information. if someone was to do what this page says it will brick the device. Aug 17 04:46:23 http://nuwiki.openwrt.org/oldwiki/openwrtdocs/hardware/netgear/wnr834b Aug 17 04:46:33 add an infobox plz :) Aug 17 09:02:57 jow * r22686 /trunk/package/iptables/Makefile: [package] iptables: partly revert r22582, fixes ip6tables extension packaging Aug 17 09:07:27 dirtyfreebooter: don't make it executable Aug 17 09:14:22 gmorning Aug 17 09:51:57 hi all Aug 17 09:52:20 still going on my 'battle' with supporting USB on ramips platform Aug 17 09:53:24 usb subsystem is now built, due to editing of the default platform config file Aug 17 09:53:53 I've managed to compile the driver and the two modules, LM and DWC_OTG are loaded with no errors Aug 17 09:54:14 but the driver does not detect any bus or devices Aug 17 09:54:17 any clues? Aug 17 10:29:03 <[florian]> Wipster: \o/ Aug 17 10:29:18 [florian], hey Aug 17 10:29:26 <[florian]> Wipster: I am preparing the patch for you to test Aug 17 10:30:32 [florian], oh excellent :) looking forward to seeing it Aug 17 10:39:32 nico * r22687 /packages/libs/gnutls/Makefile: [packages] gnutls: fix certtool dependency on libgnutls-extra (closes: #7776) Aug 17 11:44:01 <|KanjiMonster|> lfcorreia: do you create a platform device for it either in the system setup or a dynamic one in the driver's init method? Aug 17 11:44:36 <|KanjiMonster|> does the driver get loaded and attached? Aug 17 11:44:58 <|KanjiMonster|> driver's init => module's init Aug 17 11:54:48 [florian]: can the full git hash be used for acx-mac80211 in the future? are the firmware urls inactive? Aug 17 11:56:10 <[florian]> swalker: well, we can use the full hash if you prefer yes Aug 17 11:56:23 <[florian]> swalker: I do not think there are inactive firmware urls afaik Aug 17 12:06:22 [florian]: the full hash would mean one less package specific hack in my shell script; http://teknoraver.net/software/acx-mac80211/tiacx111c16 & http://ipkg.k1k2.de/files/tiacx111c19 are both 404s Aug 17 12:06:51 hi KanjiMonster ! Aug 17 12:07:30 I've created a separate package (for now) and this is how the driver is included with the firmware image and loaded on startup Aug 17 12:07:52 it depends on the usb subsystem and the LM module, also part of the same package. Aug 17 12:08:11 but it just loads, prints the driver version and sits there Aug 17 12:08:51 in one on my other experiments, I've created a platform driver for it, but didn't do the proper driver matching, and this is on planning base Aug 17 12:11:18 lfcorreia: does the driver's probe get called? Aug 17 12:12:53 unsure Aug 17 12:17:25 KanjiMonster: let me see where the driver name is printed Aug 17 12:21:45 hum... apparently the driver only inits Aug 17 12:26:28 since it waits for some (unknown to me) magic to happen, like the deviced being attached to buses... ?!?! wft Aug 17 12:26:49 deviced/devices Aug 17 12:47:16 lfcorreia: the driver needs a device to attach to, so you have to register one somewhere beforehand (usually in the platform setup code in case of platform devices), see http://www.mjmwired.net/kernel/Documentation/driver-model/platform.txt Aug 17 12:57:42 lfcorreia: so you probably need to add a platform device for usb, add a rt305x_register_usb(), and call that from the mach-* inits which have usb. Then you have to register a platform_driver in the module's init. It is important that the driver's struct contains '.driver = { .name = "" ...' , where is the name of the platform device you registered Aug 17 12:59:17 that was the work i0ve done as a test Aug 17 12:59:52 but just a question, that .name = "something" is just some string, right? Aug 17 12:59:57 yes Aug 17 13:00:18 if my driver is called dwc_otg then it must correcpond to whatever is in the platform code Aug 17 13:00:33 this is something like the "bus" name for it Aug 17 13:00:41 i could call it "rt305x_usb" Aug 17 13:01:19 so, basicly you're telling me my driver is mostly ok, but it lacks the plarform code for it all to be 'glued' Aug 17 13:03:26 you made sure the platform_device get's registered (through e.g. printk debugging)? And it's a platform_driver you are are registering? Aug 17 13:04:15 yes, it was visible under /sys/drivers/platform ( or something) let's just say it was in the same folder as the rt*_wmac Aug 17 13:04:29 ah ok Aug 17 13:04:34 this sounds good Aug 17 13:05:03 one other thing, i've got a SMC router, also based on rt3050, I might create a config for that Aug 17 13:05:27 it should be as easy as the last patch that was made for the ramips platform one or two days ago Aug 17 13:05:45 and I'll just muck about in that config Aug 17 13:25:09 nbd * r22688 /trunk/package/mac80211/patches/560-ath9k_baw_fix.patch: ath9k: remove an unnecessary BUG_ON in the aggregation code and clean up block ack window tracking to use less memory Aug 17 13:25:56 nbd * r22689 /branches/backfire/package/mac80211/patches/560-ath9k_baw_fix.patch: ath9k: merge aggregation fix from r22688 Aug 17 13:34:50 lfcorreia: it should looks something like this => http://pastie.org/1097644 Aug 17 13:40:56 great, I'll keep that example code Aug 17 13:42:44 KanjiMonster: my only doubt is what to put in the probe and remove code, but I don't have the rest of the code here now... i'll see it later on Aug 17 14:16:50 lfcorreia: depends on how the driver/device works. If it uses interrupts, you would probably register the interrupt handler (and setup any private data structures), probably register an usb port (somehow, I don't know how the usb subsystem oder drivers work), while in the remove you would unregister them Aug 17 15:14:45 [florian], ping, how did that patch come along? Aug 17 15:15:57 <[florian]> Wipster: got trapped by work, hold on Aug 17 15:17:09 [florian], ah cool I cant test till I get out of the lab and go home anyway, thought I'd check :) Aug 17 15:32:05 florian * r22690 /trunk/target/linux/rdc/ (3 files in 2 dirs): [rdc] allow 1MB extra space for JFFS and other enhancements (#7706) Aug 17 15:33:02 <[florian]> Wipster: http://openwrt.pastebin.com/SjVaRypU Aug 17 15:33:21 <[florian]> Wipster: I do not guarantee it works, I could not yet compile test it Aug 17 15:35:21 mb * r22691 /packages/lang/python-sip/ (7 files in 3 dirs): Add SIP C/C++ Bindings Generator for Python Aug 17 15:42:05 [florian], ah yes ok thanks I had tried once with diabling the EPHY bit but I didn't include that magic number ioremap as I had no idea what that did. Aug 17 15:43:04 <[florian]> I am trying to find the corresponding GPL tarball to have it being documented Aug 17 15:45:47 [florian], always mystified how people go 'I know' and that comes out, thing is when I have disabled the fixed phy in platform I dont get the falling back to switch mode message Aug 17 15:46:31 <[florian]> Wipster: you won't get it I think Aug 17 15:49:05 [florian], sorry not following Aug 17 15:52:15 <[florian]> Wipster: even if you remove the fixed phy registration you won't get the fallback to switch mode (I think) Aug 17 15:52:21 <[florian]> Wipster: have you just tried the patch? Aug 17 15:53:59 [florian], no I shall try when I get home... in about an hour, but from what I have tried I dont get the message when phy_id == PHY_MAX_ADDR Aug 17 15:54:22 <[florian]> Wipster: ah, and you did not either set external_switch,right ? Aug 17 15:54:45 <[florian]> Wipster: if not, that means that your switch might be succesfully detected Aug 17 15:55:50 [florian], I think I have tried with and without, I know without it being set I dont get a fallback message so the PHY search is running and getting something? Aug 17 15:56:16 <[florian]> Wipster: if the phy search works you should see which PHY driver has been picked up Aug 17 15:58:01 [florian], I dont see anything popping up in dmesg about that, I get cpmac_mii: probed then no phy detected at 0:00 and 0:01, wish I saved a log Aug 17 15:58:18 <[florian]> Wipster: ok, strange Aug 17 16:02:59 [florian], I see that the mdio bus cpmac-mii was moved to id 1 to stop conflicting but the probe still reverted to 0... possibly the fixed mdio bus I was seeing so just has some fake registers in it... removed the fixed mdio *I think* and put the cpmac-mii back to 0 and will test that later, or is that barkin up the wrong tree Aug 17 16:03:46 <[florian]> Wipster: ok, makes sense to do that change Aug 17 16:08:13 [florian], so what is that ioremap doing setting a bit somewhere on a bus? Aug 17 16:11:16 mapping device space into kernel addresseable ranges iirc Aug 17 16:11:22 *device memory Aug 17 16:13:26 yeh okay well time to packup, bbl Aug 17 16:20:59 hey nbd how many percent do you have for your mac nat thing? :) Aug 17 16:22:52 Memphiz: its done Aug 17 16:23:03 nbd.name/relayd.git Aug 17 16:23:09 git://... Aug 17 16:23:24 still needs integration though Aug 17 16:24:02 ohh ... cool Aug 17 16:24:18 i'll give that a try at the end of the month :) Aug 17 16:33:41 jow_laptop: is someone working on the integration? or should i put a package together for it... Aug 17 16:34:08 Memphiz: afaik none yet Aug 17 16:34:34 is there an url of a source tarball? Aug 17 16:34:48 or is it possible to make the package check it out from git ... Aug 17 16:36:52 git checkou is possible Aug 17 16:37:03 have no example around though Aug 17 16:37:08 k Aug 17 16:37:39 should work with PKG_SOURCE_PROTO=git, PKG_SOURCE_REV=md5sum, PKG_SOURCE_URL=git:// Aug 17 16:37:56 ahh ok ... i'll try that Aug 17 16:40:34 it's not completely done yet Aug 17 16:40:41 still working on fixing some stuff Aug 17 16:40:49 ah ok ... Aug 17 16:49:20 jow_laptop: PKG_SOURCE_REV <- should be the md5sum of what? Aug 17 16:57:30 sorry, sha1 Aug 17 16:57:40 the git revision identifiers Aug 17 17:39:32 jow_laptop, will u copy/move the page http://wiki.openwrt.org/oldwiki/openwrtdocs/hardware/netgear/wnr834b to the new toh so i can make edits. Aug 17 17:43:52 plz :) Aug 17 18:08:07 A document called wnr834b already exists in toh:netgear Aug 17 18:14:11 RealOpty: wiki is going offline briefly in 10 minutes Aug 17 18:14:22 server maintainence Aug 17 18:14:34 xMff, kk ty for notice. im editing that page now Aug 17 18:14:37 almost done. Aug 17 18:15:24 k all done Aug 17 18:15:27 just in time ;p Aug 17 18:20:17 xMff, what kind of maintenance btw? Aug 17 18:20:29 kernel update, reboot Aug 17 18:23:37 relayd should work properly now Aug 17 18:23:50 fixed up route handling for the default gw based on dhcp response packets Aug 17 18:24:04 and added options for setting the defaultgw and/or static routes manually on the command line Aug 17 18:26:56 xMff, hopefully that will fix some problems with 'lag' :) Aug 17 18:56:17 [florian], ping Aug 17 19:05:24 <[florian]> Wipster: pong Aug 17 19:05:28 <[florian]> Wipster: how did it go? Aug 17 19:07:10 [florian], unfortunately the added code doesn't get executed :( get this on dmesg http://pastebin.com/hsZJi8mJ Aug 17 19:10:05 <[florian]> Wipster: have you enabled all switch drivers? Aug 17 19:10:44 [florian], well only the one thats for my chip Aug 17 19:11:06 [florian], the one which worked with 8.09.2 Aug 17 19:12:14 <[florian]> ok Aug 17 19:13:24 [florian], rejiggling the code atm so it turns off the internal phy first checks then re-enables. Do I need to call reset on that phy bit or will enable suffice? Aug 17 19:17:14 wiki is back online Aug 17 19:23:04 hi all Aug 17 19:23:22 a question about choices on menuconf Aug 17 19:23:49 when a package is chosen as "M", how do I make it be part of the image? Aug 17 19:24:02 going to retest brb Aug 17 19:24:34 lfcorreia: M == package, * == package and include in image Aug 17 19:25:18 RealOpty: btw, it's probably 264 Mhz (48 Mhz * 5.5) ;) Aug 17 19:25:49 Bartman007, thanks, but how do I use the M package? loaded with ipkg? Aug 17 19:26:15 lfcorreia: correct. they can be installed with opkg (the replacement for ipkg) Aug 17 19:26:47 you can either host the package in a repository on a webserver and add the repo to /etc/opkg.conf, or you can copy the package to /tmp/ on the router and install it directly. Aug 17 19:26:48 Bartman007, thanks, now I know Why TF the packages weren't making it to the image... Aug 17 19:28:29 <[florian]> Wipster: I do not quite understand what is happening now Aug 17 19:37:15 anybody interested in testing a patch to get the rtc to work on the dockstar? Aug 17 19:38:31 re Aug 17 19:38:35 ping xMff Aug 17 19:39:16 qry Aug 17 19:42:35 <_trine> KanjiMonster, I didnt know it had an RTC Aug 17 19:43:00 _trine: the soc has one integrated, but it definitely lacks a battery Aug 17 19:43:28 so it propably works for keeping time through reboots, but not for blackouts Aug 17 19:43:40 <_trine> would you know how to connect a battery though Aug 17 19:44:07 unfortunately no Aug 17 19:44:30 <_trine> I just have mine updating the time off the net Aug 17 20:36:07 [florian], think might be on to a winnar Aug 17 20:41:01 [florian], took the external_switch bit out and just doing the PHY checking with internal phy enabled and disabled, http://pastebin.com/m8qHpnFw the right registers in MII but the switch driver didn't load hmmm Aug 17 20:43:57 looks like the internal PHY doesn't need to be disabled... or what ever that patch was doing Aug 17 22:02:48 more testage bbl Aug 17 22:12:27 ah my bad, of course the cpmac driver gets registered for high and low, so when one disabled internal phy the second test was already in that state opps heh, it does need to be disabled to see the switch. Interesting it could see it in 8.09.2 without the disable though Aug 17 22:34:00 oo try to fix cpmac! Aug 17 22:37:01 heh Aug 17 22:38:45 speak russian Aug 17 22:39:35 AndyI, sorry only language I know is English.... badly Aug 17 23:19:53 is there a make command that just patches everything up and sets up build_dir before compiling? Aug 17 23:20:22 the kernel tree? Aug 17 23:21:06 yeh Aug 17 23:21:58 make target/linux/{clean,prepare} V=99 [QUILT=1] Aug 17 23:22:29 actually just target/linux/preare Aug 17 23:22:33 prepare Aug 17 23:23:41 brilliant so simple, that saves me some time Aug 17 23:46:57 can I run this by someone just so I understand it, *(unsigned long*) ioremap(0x08611A08, 4) |= 0x00000001; remaps that magic number into some addressable space then sets the lsb... or msb, bit 1 Aug 17 23:57:55 it maps 4 bytes of memory at addr 0x08611A08, returns some void pointer which is casted to unsigned long and bit-or-ed with 0x1 Aug 17 23:58:01 at least I read it that way Aug 17 23:58:35 so its basically flipping a bit at address 0x08611A08 Aug 17 23:59:33 excellent flipping it back being, &= FFFFFFFE; Aug 18 00:00:27 thanks Aug 18 00:05:33 jow * r22692 /trunk/package/uhttpd/ (8 files in 2 dirs): Aug 18 00:05:33 [package] uhttpd: Aug 18 00:05:34 - fix parsing of interpreter entries in the config file, fixes serving of static files as .cgi with X-Wrt Aug 18 00:05:34 - better cope with connection aborts, especially during header transfer Aug 18 00:05:34 - fix return value checking of TLS reads and writes, solves some blocking issues Aug 18 02:55:13 xMff: FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF **** ENDING LOGGING AT Wed Aug 18 02:59:57 2010