**** BEGIN LOGGING AT Mon Aug 15 02:59:57 2011
Aug 15 03:39:34 <CIA-89> swalker * r27985 /packages/net/httping/ (Makefile patches/002-version.patch): [packages] httping: update to 1.5.2, refresh patches
Aug 15 03:40:47 <CIA-89> swalker * r27986 /packages/utils/time/ (5 files in 2 dirs): [packages] time: add the GNU time utility
Aug 15 05:06:35 <CIA-89> swalker * r27987 /packages/ (ipv6/ahcpd/patches/ net/babel/patches/): [packages] remove empty patches directories
Aug 15 05:28:21 <builder_> build #61 of ppc44x is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/61
Aug 15 06:08:05 <builder_> build #65 of pxcab is complete: Failure [failed compile_10]  Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/65
Aug 15 08:58:46 <_trine_> nothing seems to select iptables-mod-conntrack-extra
Aug 15 08:59:06 <_trine_> iptables is supposed to do it but it doesnt
Aug 15 08:59:47 <jow_laptop> iptables is not supposed to select that
Aug 15 09:00:02 <_trine_> oh sorry,, what does then
Aug 15 09:00:44 <_trine_> http://dpaste.com/594595/
Aug 15 09:00:49 <_trine_> I was looking at that
Aug 15 09:01:31 <_trine_> have I mis-understood it?
Aug 15 09:01:34 <jow_laptop> yes
Aug 15 09:02:01 <_trine_> gimme a clue then :)
Aug 15 09:02:08 <jow_laptop> the clue is boolean logic
Aug 15 09:02:14 <_trine_> heh
Aug 15 09:03:18 <_trine_> hey jow_laptop if you are bored can you help me with setting up polipo and tor
Aug 15 09:04:05 <_trine_> and also kirkwood is still on 2.6.37 and I have been running 2.6.39.2 for ages now
Aug 15 09:05:11 <loswillios> _trine_: 3.0 works also fine
Aug 15 09:05:13 <jow_laptop> oh no, god will kill a kitten for every day we don't update the kernel ;)
Aug 15 09:05:33 <loswillios> there's too many of them on the internet anyway :p
Aug 15 09:06:47 <_trine_> jow_laptop, dont mess with Rottweiler's
Aug 15 09:07:03 <_trine_> they bite back
Aug 15 09:07:04 <_trine_> :)
Aug 15 09:53:26 <loswillios> _trine_: have you tried the ubifs-patch on your dockstar?
Aug 15 10:03:25 <_trine_> loswillios, no not yet, I dont think I have seen it
Aug 15 10:03:45 <loswillios> it's on -devel and thus in patchwork
Aug 15 10:03:48 <_trine_> I am compiling now but could stop if you would like me to try anything
Aug 15 10:06:03 <loswillios> would be nice to get it some testing
Aug 15 10:06:28 <_trine_> well if you could help me to apply it I will
Aug 15 10:06:38 <loswillios> http://thread.gmane.org/gmane.comp.embedded.openwrt.devel/10726
Aug 15 10:10:52 <_trine_> loswillios, would it mean updating uboot on the dockstar ?
Aug 15 10:11:44 <loswillios> If I read the last message correctly, the openwrt-uboot should boot from an ubifs volume
Aug 15 10:12:06 <_trine_> I dont use the openwrt uboot
Aug 15 10:12:38 <_trine_> I use the one from Jeff doodson
Aug 15 10:14:23 <loswillios> never heart of it
Aug 15 10:15:13 <_trine_> http://jeff.doozan.com/debian/uboot/
Aug 15 10:15:36 <_trine_> it does say :- Support for UBIFS partitions on NAND.
Aug 15 10:15:55 <_trine_> I typed his name wrongly at first
Aug 15 10:17:50 <_trine_> just going to make a coffee be back in a few mins
Aug 15 10:39:31 <_trine_> jow_laptop, what is it that I need to select in make menuconfig that would then select iptables-mod-conntrack-extra
Aug 15 11:07:53 <CIA-89> hauke * r27988 /trunk/package/kernel/modules/001-depends.mk:
Aug 15 11:07:53 <CIA-89> kernel: remove kmod-input-core just for some x86 profiles.
Aug 15 11:07:53 <CIA-89> Not all x86 profiles have CONFIG_INPUT=y in their configuration. Just remove kmod-input-core for them.
Aug 15 11:07:53 <CIA-89> Thank you Philip Prindeville for pointing out this problem.
Aug 15 14:13:59 <CIA-89> juhosg * r27989 /trunk/target/linux/generic/files/drivers/net/phy/ar8216.c: generic: ar8216: add support for get link status
Aug 15 14:14:04 <CIA-89> juhosg * r27990 /trunk/target/linux/ramips/files/ (13 files in 3 dirs):
Aug 15 14:14:04 <CIA-89> ramips: allow to specify port layout for the switch of the RT305x/RT3350
Aug 15 14:14:04 <CIA-89> Based on a patch by Roman Yeryomin <roman@advem.lv>
Aug 15 14:14:05 <CIA-89> juhosg * r27991 /trunk/target/linux/ramips/ (8 files in 6 dirs):
Aug 15 14:14:05 <CIA-89> ramips: add support for Argus ATP-52B
Aug 15 14:14:05 <CIA-89> [The ESW related changes has been removed - juhosg]
Aug 15 14:14:05 <CIA-89> Signed-off-by: Roman Yeryomin <roman@advem.lv>
Aug 15 14:14:06 <CIA-89> juhosg * r27992 /trunk/target/linux/ramips/base-files/lib/preinit/05_ramips_load-input_drivers: ramips: load the input drivers during preinit
Aug 15 14:14:07 <CIA-89> juhosg * r27993 /trunk/target/linux/ramips/ (rt288x/config-2.6.39 rt305x/config-2.6.39): ramips: sync kernel config
Aug 15 14:14:08 <CIA-89> juhosg * r27994 /trunk/package/button-hotplug/Makefile:
Aug 15 14:14:08 <CIA-89> button-hotplug: add autoload feature
Aug 15 14:14:08 <CIA-89> Reported-by: Philip Prindeville <philipp@redfish-solutions.com>
Aug 15 14:14:10 <CIA-89> juhosg * r27995 /trunk/target/linux/ramips/files/arch/mips/ralink/rt305x/ (devices.c devices.h):
Aug 15 14:14:10 <CIA-89> ramips: rt305x: add dwc_otg platform device
Aug 15 14:14:10 <CIA-89> Signed-off-by: Alexander Gordeev <lasaine@lvk.cs.msu.su>
Aug 15 14:14:11 <CIA-89> juhosg * r27996 /trunk/target/linux/ramips/files/arch/mips/ralink/rt305x/ (5 files): ramips: rt305x: register the dwc_otg device on the relevant boards
Aug 15 14:14:14 <CIA-89> juhosg * r27997 /trunk/target/linux/ramips/ (25 files in 6 dirs):
Aug 15 14:14:14 <CIA-89> ramips: rt305x: add dwc_otg driver
Aug 15 14:14:14 <CIA-89> Based on a patch by Layne Edwards <ledwards@astrumtech.net>
Aug 15 14:14:15 <CIA-89> juhosg * r27998 /trunk/package/kernel/modules/usb.mk:
Aug 15 14:14:16 <CIA-89> package/kernel: create package for the RT305x dwc_otg driver
Aug 15 14:35:30 <karlp> can anyone tell me if this is missing anything? http://patchwork.openwrt.org/patch/1304/
Aug 15 14:40:31 <loswillios> karlp: does it really have no other DEPENDS than libmosquitto?
Aug 15 14:40:58 <karlp> the makefile makes three packages, mosquitto, mosquitto-client and libmosquitto
Aug 15 14:41:08 <karlp> mosquitto-client depends on libmosquitto and nothing else
Aug 15 14:42:48 <karlp> I had sent this earlier a couple of times, but back then it had been as a diff in the main trunk packages section, so I understand why it might have been ignored.
Aug 15 14:43:32 <loswillios> let me build it
Aug 15 14:48:57 <loswillios> karlp: do you also want to be maintainer?
Aug 15 14:49:02 <karlp> yep
Aug 15 14:49:11 <karlp> I'm already doing it within mosquitto,
Aug 15 14:56:34 <CIA-89> loswillios * r28001 /packages/net/mosquitto/ (5 files in 2 dirs): [packages] mosquitto: [PATCH] Add support for the mosquitto MQTT message broker, tools and library. Signed-off-by: Karl Palsson<karlp@tweak.net.au>
Aug 15 14:58:18 <loswillios> karlp: please ask thepeople to give you write-access to packages/net/mosquitto, then you can add a MAINTAINER line to the Makefile
Aug 15 15:06:36 <karlp> thepeople: can I please have write access to packages/net/mosquitto?
Aug 15 15:26:42 <karlp> are there any good examples in the packages of a packge that _only_ includes other packages?  there's the luci collections, but it seems a bit heavy
Aug 15 15:27:44 <loswillios> mtd-utils?
Aug 15 15:30:04 <fullstop> How does the linux kernel work with gpios?  On low-level micro controllers, I've always had ports A,B,C etc.. and they A1 would correspond to a particular pin.
Aug 15 15:30:34 <fullstop> How does, say, my system with a particular SoC know what gpio1, gpio2, etc actually are?
Aug 15 15:30:48 <Chocky> there's a mapping in the platform code
Aug 15 15:31:05 <fullstop> Chocky: clearly mine is incorrect.  :D
Aug 15 15:31:25 <Chocky> obviously it needs to match the actual hardware/datasheet :o
Aug 15 15:31:55 <fullstop> It's more than that.  I am able to export more gpios than I have defined.
Aug 15 15:32:58 <Chocky> probably you are picking up generic ones somewhere
Aug 15 15:33:45 <fullstop> So, regardless, even with a data sheet I won't know which pins are connected.
Aug 15 15:34:07 <fullstop> I would know which pins are available as gpio, but I imagine that many are multi purpose.
Aug 15 15:34:22 <Chocky> the datasheet for the device is necessary for that.
Aug 15 15:34:29 <Chocky> as in, a precise circuit layout
Aug 15 15:34:35 <Chocky> in other words, you may need to guess
Aug 15 15:34:40 <fullstop> I'm not going to have that either.  ;-)
Aug 15 15:35:03 <Chocky> you don't have the CPU datasheet?
Aug 15 15:35:28 <Chocky> (yes, I know ;-)
Aug 15 15:36:16 <fullstop> I may be able to find the cpu datasheet, but you never know.
Aug 15 15:36:22 <fullstop> I do, however, have dd-wrt source
Aug 15 15:36:41 <fullstop> and it appears to work for flipping the various gpios for the LEDs.
Aug 15 15:37:01 <Chocky> right.  the dd-wrt stuff was probably done in cooperation with the board manus, so that's best bet
Aug 15 15:37:34 <fullstop> It's ugly though.... ;-)
Aug 15 15:38:13 <fullstop> http://pastebin.com/mTZT6sSv
Aug 15 15:38:13 <Chocky> yes, no doubt.  good luck making it tidy
Aug 15 15:39:17 <fullstop> I honestly don't care about the various LEDs that much.. but I need to enable the USB port power.
Aug 15 15:40:38 <Chocky> right, that code is typical of GPIO handling
Aug 15 15:43:50 <fullstop> Yes, I've written similar looking code for dsPIC micros..
Aug 15 15:44:09 <fullstop> but I don't see the analogous code in openwrt.
Aug 15 15:44:20 <fullstop> Which is why I feel that I am missing something.. .:D
Aug 15 15:45:12 <Chocky> this is ar71xx? There will be some generic platform code somewhere.
Aug 15 15:45:16 <Chocky> might even be in the headers
Aug 15 15:45:26 <fullstop> it is ar71xx, but the cpu is ar7242
Aug 15 15:45:53 <Chocky> you'll need to ask nbd etc.  I haven't looked at any of that code
Aug 15 15:46:04 <Chocky> all I can tell you is what's likely there
Aug 15 15:46:23 <fullstop> I've never gone this deep with a system having PCI..
Aug 15 15:46:28 <fullstop> all new to me.
Aug 15 15:47:51 <Chocky> back in the day, most systems in Linux were assumed to have PCI, and everything was connected via that.  That's not the case in most non-x86 hardware now of course
Aug 15 15:48:56 <Chocky> the problem with stuff at the level you're doing is really figuring out how all the platform specific structures tie together.  Once you get past that, it's easy.
Aug 15 15:49:58 <fullstop> Most of my embedded experience is bare-metal micros, but, you have the data sheet and you are in control of choosing gpios based on what peripherals you intend to use.  This is a bit different.  :)
Aug 15 15:50:26 <Chocky> well, put it on your resume once you are done
Aug 15 15:50:40 <Chocky> embedded devs are in demand right now
Aug 15 15:54:08 <fullstop> I'll add it, but I am not looking to move.  :)
Aug 15 15:54:55 <karlp> embedded is always in demand, just never very many at a time.
Aug 15 16:06:10 <fullstop> and, from going through the source, Gabor Juhos is a very busy.
Aug 15 16:06:28 <Chocky> he is
Aug 15 16:06:37 <Chocky> you don't have to move for embedded.  But where are you?
Aug 15 16:14:41 <fullstop> By move, I mean that I am happy with my current employment.   I'm in eastern Pennsylvania.
Aug 15 16:14:51 <fullstop> I don't mean physically moving.
Aug 15 16:15:29 <fullstop> I'll be back in a few minutes.
Aug 15 16:19:05 <wigyori> re
Aug 15 16:37:27 <_trine> using the latest trunk on my wndr3700 the maximum power setting seems to be only 17dBm
Aug 15 16:37:43 <_trine> is there a reason for this
Aug 15 16:38:05 <_trine> i have tried altering the country setting and it appears not to alter this
Aug 15 16:38:36 <_trine> I am referring to setting it with luci
Aug 15 16:39:01 <fullstop> My daughter, 3, was playing with a toy monkey the other morning, making him jump around and saying "I have the power" in falsetto.
Aug 15 16:39:19 <fullstop> and then she accidentally dropped him and responded "That's not the power"
Aug 15 16:39:42 <Chocky> kids these days
Aug 15 16:39:49 <fullstop> And that's what I woke up to.
Aug 15 16:41:58 <_trine> I am not what I say I am but I am what you would believe what I am not when I am really something else
Aug 15 16:42:05 <karlp> I've got a virtual package set up, that properly selects a bunch of depends, but at make time I get, "*** No targets specified and no makefile found.  Stop"
Aug 15 16:42:14 <karlp> I set up Package/blah/install to just have "true"
Aug 15 16:42:27 <karlp> but is there something else I need to fake out to not have any real compilation take place?
Aug 15 16:43:49 <Chocky> what are you typing, exactly?
Aug 15 16:44:40 <karlp> http://rme.tweak.pagekite.me/Makefile?view=/Makefile
Aug 15 16:45:18 <Chocky> ok, so full blown wget relies on libcrypto/libssl, which are fat.  If I make a wget which statically links to those, then I can save 700k.  assuming nothing else depends on them
Aug 15 16:46:23 <Chocky> but lighttpd does.  oh well
Aug 15 16:47:08 <Chocky> karpl: yes, but what are you typing
Aug 15 16:47:43 <karlp> "make"
Aug 15 16:47:53 <karlp> I selected the package in menuconfig, and then just make.
Aug 15 16:48:38 <karlp> it enters the build directory, but (as expected) there's nothing there. (no PKG_SOURCE) or anything.
Aug 15 16:50:00 <Chocky> I think you may have to add override the default compile rule
Aug 15 16:50:17 <karlp> just Package/blah/compile  and make it true as well?
Aug 15 16:50:33 <Chocky> I don't know for certain, but that's what I would try.
Aug 15 16:50:33 <karlp> or is that guessing too easy?
Aug 15 16:50:38 <karlp> well, lets try then :)
Aug 15 16:51:33 <Chocky> I freely admint my wrt knowledge is incomplete, but I wrote a similar system that had to solve identical problems
Aug 15 16:51:37 <Chocky> -n
Aug 15 16:55:32 <karlp> beauty, that seemed to have worked
Aug 15 16:57:18 <karlp> now, the bigger problem, getting my netbeans dual compile (linux+owrt) makefile to work within the owrt package structure.
Aug 15 17:06:07 <philipp64|laptop> what's Gabor's handle?
Aug 15 17:11:04 <Chocky> jusog
Aug 15 17:11:08 <Chocky> er.
Aug 15 17:11:25 <Chocky> juhosg
Aug 15 17:12:31 <philipp64|laptop> ok, not around...
Aug 15 17:13:56 <Chocky> is it really necessary to build gettext-tools - at least for cross?
Aug 15 17:22:12 <Chocky> seems that gettext-tools build breaks when you turn of posix_spawn stuff in eglibc.  I'm not sure that any of the gettext tools end up in any package.  simply not building that would be an easy fix.  Otherwise, maybe some flag to deal with that eglibc cae
Aug 15 17:26:22 <Chocky> and how can I view package contents with host opkg?
Aug 15 17:33:35 <karlp> doesn't it work the same way?
Aug 15 17:33:44 <karlp> opkg files packagefile ?
Aug 15 17:34:49 <Chocky> don't think so
Aug 15 17:40:59 <builder_> build #57 of avr32 is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/57
Aug 15 17:41:29 <karlp> is it possible to set PKG_SOURCE_ to be a file link? for testing before I commit it all)
Aug 15 17:41:39 <karlp> I've tried just PKG_SOURCE_URL:=file:///blasdfsdf/
Aug 15 17:41:52 <karlp> but it tries to download it from openwrt.org,
Aug 15 17:42:13 <karlp> PKG_SOURCE_PROTO:=file doesn't seem to work...
Aug 15 17:43:09 <KanjiMonster> karlp: it doesn't if a matching file is in dl/, so you could just download it yourself and put it there
Aug 15 17:44:02 <karlp> hmm, I don't want to download anything at all, I have the source I want to compile on my local disk.  eventually, it will be PKG_SOURCE_PROTO:=svn, whcih I've got working, but I wanted to point it to local source for a bit while I tested out some changes
Aug 15 17:44:03 <KanjiMonster> karlp: "PKG_SOURCE_MIRROR:=0" also works
Aug 15 17:44:18 <karlp> what does MIRROR= 0 do?
Aug 15 17:44:49 <KanjiMonster> makes it not try to download a sources tarball
Aug 15 17:45:25 <karlp> are any of tehse documented anywhere?
Aug 15 17:46:28 <karlp> Download/default is missing the FILE field..  Stop
Aug 15 17:46:38 <karlp> I've got PKG_SOURCE_URL:=file:///blah
Aug 15 17:46:51 <karlp> PKG_SOURCE_MIRROR:=0 and PKG_SOURCE_PROTO:=file
Aug 15 17:48:52 <fullstop> KanjiMonster: Can you help me decode the gpio magic of openwrt?
Aug 15 17:49:12 <fullstop> I don't see where it is _doing_ anything.  There has to be a pin mapping somewhere, right?
Aug 15 17:50:00 <KanjiMonster> fullstop: usually gpio X = gpio X of the SoC
Aug 15 17:50:46 <fullstop> KanjiMonster: without the SoC spec sheets, do you know how they determine gpio?  Is it similar to other embedded micros where you have ports A, B, C, D?
Aug 15 17:50:58 <fullstop> or is gpio22 pin 22 of the part?
Aug 15 17:51:17 <KanjiMonster> I don't know
Aug 15 17:51:24 <karlp> most SoCs don't have that many gpios anyway
Aug 15 17:51:59 <fullstop> Don't have that many available or don't have that many at all?
Aug 15 17:52:05 <fullstop> not that it matters, really..
Aug 15 17:52:13 <karlp> at all, so they don't need to bother with A, B C,
Aug 15 17:52:17 <karlp> just gpio0..X
Aug 15 17:53:20 <fullstop> okay.. so gpio0 could be any pin on the SoC
Aug 15 17:53:38 <fullstop> and, in the end, without a datasheet, it doesn't really matter.  I just need to figure out which register to poke?
Aug 15 17:53:41 <karlp> well, on the SoC I've got, the SoC also names them GPIO0
Aug 15 17:53:50 <karlp> without a datasheet, good luck
Aug 15 17:54:07 <karlp> you're going to be looking at similar hardware's code, and then poking at somewhat random
Aug 15 17:54:15 <fullstop> Somehow I doubt Atheros will let me see one without plenty of $$ and a NDA.
Aug 15 17:54:21 <karlp> yeah :)
Aug 15 17:54:38 <karlp> is this a new atheros part?
Aug 15 17:54:57 <fullstop> OpenWRT has some support for it.  It's an AR7242
Aug 15 17:55:07 <karlp> some, but no gpio?
Aug 15 17:55:19 <fullstop> I even get GPIOs.. but flipping them around does nothing.
Aug 15 17:55:28 <fullstop> I can set input / output..
Aug 15 17:55:34 <fullstop> active low
Aug 15 17:55:45 <fullstop> and even change their values.. but the LED does not change.
Aug 15 17:56:04 <karlp> so you're recompiling and stabbing int he dark for the right addresses then I take it?
Aug 15 17:56:11 <karlp> doesn't sound like much fun :|
Aug 15 17:56:17 <fullstop> Anyway, I have working DD-WRT gpio code.. but I can't find where OpenWRT actually defines this.
Aug 15 17:56:33 <fullstop> The DD-WRT code is kind of ugly, but I think I can make it work.
Aug 15 17:57:46 <KanjiMonster> fullstop: you mean https://dev.openwrt.org/browser/trunk/target/linux/ar71xx/files/arch/mips/ar71xx/gpio.c ?
Aug 15 17:58:37 <fullstop> KanjiMonster: I need to get my eyes checked.
Aug 15 17:58:47 <fullstop> It has been 10 years since my last exam.
Aug 15 17:58:51 <fullstop> And..
Aug 15 17:58:53 <fullstop> clearly, I am blind.
Aug 15 17:59:20 <karlp> if I have a makefile that works properly on linux, what define flag can I use to only put "-I$(STAGING_DIR)/usr/include -L$(STAGING_DIR)/usr/lib" into the flags when building in openwrt?
Aug 15 17:59:58 <karlp> I take it "main.c:1: note: someone does not honour COPTS correctly, passed 0 times" is probably a thing I've done badly too?
Aug 15 18:00:37 <Chocky> likely
Aug 15 18:00:48 <Chocky> cross compiling is often a non-trivial thing
Aug 15 18:01:01 <karlp> you think :)
Aug 15 18:01:08 <Chocky> but there are various flags you can pick up to ifdef in a Makefile
Aug 15 18:01:24 <karlp> can you point me to a list of them?
Aug 15 18:02:09 <larsc> best is not to add any ifdefs at all. if you have a proper makefile it should just work fine
Aug 15 18:02:34 <Chocky> haha.  there are so many different styles of makefile out there
Aug 15 18:02:51 <Chocky> yes, I agree that's best, but neverless, typical simple ones often do require changes.
Aug 15 18:03:07 <Chocky> it's often the autoconf ones which do not, althought there are plenty of exceptions
Aug 15 18:03:22 <karlp> larsc: that's what I'm trying to do,
Aug 15 18:03:45 <Chocky> in any case, I woudln't expect to need explictl -I -L flags to those dirs
Aug 15 18:04:04 <karlp> I'm trying to use a makefile like this: https://github.com/karlp/mqtt_demos/blob/master/msg_signing/Makefile
Aug 15 18:04:07 <Chocky> the correct way is to pick up cflags which openwrt sets for you.  so, those two things are really the same
Aug 15 18:07:00 <karlp> ok, I took out my CFALGS; and I got all the openwrt ones,
Aug 15 18:07:01 <karlp> so that's cool,
Aug 15 18:07:22 <Chocky> you could do CFLAGS ?=
Aug 15 18:07:32 <karlp> now it's just "cannot find -lcrypto"
Aug 15 18:07:40 <Chocky> you need to add the dependency for that
Aug 15 18:07:52 <Chocky> openssl etc
Aug 15 18:08:12 <karlp> the package has DEPENDS:= +libmosquitto +iniparser +libopenssl
Aug 15 18:08:31 <karlp> libopenssl is the one providing crypto.... (i thought?)
Aug 15 18:08:34 <larsc> karlp: it's because you override LDFLAGS
Aug 15 18:08:46 <larsc> i guess
Aug 15 18:09:03 <karlp> ok, let me turn that off as well.
Aug 15 18:09:16 <karlp> does ?= define them only if they are not already defined?
Aug 15 18:09:35 <larsc> yes
Aug 15 18:11:39 <larsc> but if you want to link against those libs, you probably want to add them regardless of whether LDFLAGS is already set or not
Aug 15 18:12:25 <Chocky> you may need to use pkgconfig
Aug 15 18:14:55 <karlp> I guess.  with no LDFLAGS in my makefile, it compiles, but link fails with a mountain of undefined referenecs.  with my LDFLAGS, it's can't find lcrypto
Aug 15 18:15:42 <Chocky> indeed
Aug 15 18:15:49 <fullstop> Okay, I've found what I need.  It involves changing some of the base pointers in setup.c, which is used for all 71xx SoC
Aug 15 18:16:08 <larsc> karlp: try LDFLAGS+=
Aug 15 18:16:11 <Chocky> full: right, pretty standard
Aug 15 18:16:15 <fullstop> Is it okay to #ifdef in something like this, or would this be considered a new platform?
Aug 15 18:16:27 <Chocky> hm
Aug 15 18:16:35 <fullstop> It appears that everything is the same except for GPIO
Aug 15 18:17:25 <Chocky> if you can do some kind of platform detection, you can avoid ifdefs
Aug 15 18:17:36 <fullstop> Let me get it working first and then worry about that.
Aug 15 18:17:43 <Chocky> :o
Aug 15 18:23:12 <karlp> larsc: yeah, I did similar by defining KLDFLAGS to be the -lcrypto lines.  so now I can see that -L has the right staging/target/usr/lib and -l has the right library names, but it still complains about undefined references?
Aug 15 18:24:09 <Chocky> you might need to reorder your link
Aug 15 18:24:23 <Chocky> if you are sure it's being passed properly to compile
Aug 15 18:24:25 <Chocky> +r
Aug 15 18:25:03 <karlp> reorder the link? don't I just need to list all the objects?  the compile steps all seemd to compile ok.
Aug 15 18:26:13 <Chocky> the order the libraries are passed sometimes matters
Aug 15 18:28:59 <karlp> hah
Aug 15 18:29:13 <karlp> libiniparser was only in the usr/lib dir as libiniparser.so.0
Aug 15 18:29:22 <karlp> with no link as libiniparser.so
Aug 15 18:29:28 <Chocky> that's a bug then
Aug 15 18:29:41 <karlp> well, yeah, but a  bug in my code, libiniparser is something I've added :)
Aug 15 18:30:05 * karlp checks his Build/InstallDev
Aug 15 18:33:55 <karlp> I have $(INSTALL_DATA) $(PKG_BUILD_DIR)/libiniparser.{a,so*} $(1)/usr/lib
Aug 15 18:34:07 <karlp> is there any better way to make the symlink to .so magically?
Aug 15 18:34:18 <karlp> or is that somethign that should be handled by the makefile of iniparser itself?
Aug 15 18:39:41 <fullstop> Huh.. this is interesting..
Aug 15 18:40:37 <fullstop> the dd-wrt code accesses the gpio registers through PCI memory mapped addresses.
Aug 15 18:40:54 <fullstop> not as the APB
Aug 15 18:52:19 <Chocky> right
Aug 15 18:55:01 <fullstop> Anyway, I used this base address for the gpios, but I can't export them now.. so I probably did something bad.  ;-)
Aug 15 18:58:09 <CIA-89> juhosg * r28002 /trunk/target/linux/ramips/files/arch/mips/ralink/rt305x/mach-argus-atp52b.c: ramips: rt305x: make config and factory partitions read-only on the ATP-52B
Aug 15 18:58:09 <CIA-89> juhosg * r28003 /trunk/target/linux/ramips/Makefile: ramips: add input drivers to the default package list
Aug 15 18:58:10 <CIA-89> juhosg * r28004 /trunk/package/maccalc/ (. Makefile src/ src/Makefile src/main.c): package/maccalc: a simple mac address manipulation utility
Aug 15 18:58:11 <CIA-89> juhosg * r28005 /trunk/target/linux/ramips/Makefile: ramips: add maccalc to the default package list
Aug 15 18:58:13 <CIA-89> juhosg * r28006 /trunk/target/linux/ramips/base-files/ (etc/uci-defaults/nw718 lib/preinit/06_set_iface_mac): ramips: rt305x: fix mac addresses on the Netcore NW718 board
Aug 15 19:02:43 <CIA-89> juhosg * r28007 /trunk/target/linux/ramips/base-files/etc/uci-defaults/nw718: ramips: use maccalc instead of hexdump + sed
Aug 15 19:04:56 <karlp> well, it now works for building for linux in netbeans and from teh command line, and for openwrt within the owrt toolchain.  but no longer cross compiled from netbeans.
Aug 15 19:05:00 <karlp> that project can wait for another day
Aug 15 19:21:25 <Chocky> geez. turning off stuff in eglibc is Russian Roulette
Aug 15 19:39:12 <fullstop> hey, it worked.  I have GPIO, USB and LEDs
Aug 15 19:40:48 <Chocky> \o\ \o/ /o/
Aug 15 19:42:43 <fullstop> I should be clear.  I have LED.  Just one.  :D
Aug 15 19:42:54 <fullstop> but I can power and unpower the usb port.
Aug 15 19:43:17 <fullstop> which, you know, would be a handy feature to have at runtime.
Aug 15 19:48:09 <Chocky> so, now what?
Aug 15 19:49:21 <fullstop> well, figure out all of the gpios
Aug 15 19:49:35 <Chocky> you're just a gigolo. er, gpio
Aug 15 19:49:36 <fullstop> and my wan port is not correct for some reason.
Aug 15 19:49:48 <fullstop> f-ing magnets
Aug 15 23:13:15 <thepeople> karlp: send me an email thepeople@openwrt.org
**** ENDING LOGGING AT Tue Aug 16 02:59:57 2011