**** BEGIN LOGGING AT Tue May 28 03:00:00 2013
May 28 08:50:00 <champtar> Hi, i've sent this patch http://patchwork.openwrt.org/patch/3692/, is OPENSSL_WITH_EC ok (name convention), and is it ok to use +@OPENSSL_WITH_EC
May 28 08:50:00 <champtar> which imply +52Ko on openssl package
May 28 08:52:49 <DonkeyHotei> turnaround time is not quick
May 28 08:53:10 <nbd> champtar: doesn't authsae only use some low level elliptic curve crypto parts of openssl?
May 28 08:53:30 <nbd> if so, it might be a good idea to patch those parts into the authsae package and leave the normal openssl package alone
May 28 08:53:39 <nbd> makes the whole thing carry a lot less bloat
May 28 08:57:36 <champtar> nbd: don't know, but it mean convincing authsae authors to add these stuff, and maintain them
May 28 08:58:06 <champtar> i'll look at what part is used
May 28 08:58:48 <nbd> or making a small openssl-static package that builds only the EC parts
May 28 08:58:53 <nbd> and only changing includes and linker flags for authsae
May 28 08:59:50 <champtar> don't know if authsae only use EC
May 28 09:01:37 <DonkeyHotei> nbd: would you agree or disagree with blogic's responses to me on the ML for my patchset?
May 28 09:02:39 <champtar> nbd: or compile openssl with only an EC subset (patching openssl)
May 28 09:04:53 <nbd> DonkeyHotei: i agree with blogic's responses
May 28 09:05:03 <nbd> DonkeyHotei: btw. you can make things a lot easier for you wrt. patch submission by using git
May 28 09:05:33 <DonkeyHotei> nevertheless, the canonical repo is still svn
May 28 09:05:44 <nbd> no
May 28 09:05:50 <DonkeyHotei> no?
May 28 09:05:55 <nbd> see https://dev.openwrt.org/wiki/GetSource
May 28 09:06:07 <DonkeyHotei> i must have missed the switchover
May 28 09:06:13 <nbd> for users, git is the primary recommended way of getting the sources
May 28 09:06:30 <nbd> the svn is still the master, but it's meant to be used by devs
May 28 09:06:31 <nbd> typically via git-svn
May 28 09:06:50 <DonkeyHotei> hmm
May 28 09:07:16 <nbd> btw. if you use git, you will no longer have to use your email client to send patches
May 28 09:07:21 <nbd> you can just make some local commits
May 28 09:07:24 <DonkeyHotei> i already had it checked out, merely did an update, and saw some of the feeds went to git all of a sudden
May 28 09:07:30 <nbd> and send those in one batch using git send-email
May 28 09:08:09 <DonkeyHotei> i'm so used to svn and still haven't got the hang of git
May 28 09:08:11 <nbd> you should also be aware that the subject of an email makes it into the commit message as well
May 28 09:08:38 <nbd> and there are also reasons why you should choose an email subject that actually makes sense when sending patches
May 28 09:08:52 <DonkeyHotei> github promised an svn frontend to git repos, but it's incomplete
May 28 09:09:07 <nbd> svn frontend for git repos? sounds like a bad idea
May 28 09:09:18 <DonkeyHotei> nevertheless, they have one
May 28 09:10:27 <DonkeyHotei> anyway, be things as they may, the svn patch submission howto still specifies that the base directory should be exactly what i made it
May 28 09:10:53 <nbd> i think the base directory thing doesn't matter too much
May 28 09:11:07 <nbd> it's just that svn for some stupid reason always generates patches that apply with -p0, not -p1
May 28 09:11:10 <DonkeyHotei> he made a big point of it on the ML it seemed
May 28 09:11:16 <nbd> using git for generating patches fixes this
May 28 09:11:40 <nbd> so ignore the base directory thing, but the other points were more important anyway
May 28 09:12:14 <DonkeyHotei> agreed, but there weren't that many other points
May 28 09:12:42 <DonkeyHotei> the patches were still separated into things that go logically together
May 28 09:13:08 <DonkeyHotei> and "the big one" was quite atomic except for three little lines
May 28 09:13:48 <nbd> the patch that changes some target kernel patches turns them into the wrong patch style
May 28 09:14:05 <DonkeyHotei> it was done with quilt refresh
May 28 09:14:22 <nbd> you should either use make target/linux/refresh to refresh patches, or refresh individual patches using this quilt command:
May 28 09:14:25 <nbd> quilt refresh -p ab --no-index --no-timestamps
May 28 09:14:35 <DonkeyHotei> hmmm, did not know that
May 28 09:14:54 <DonkeyHotei> thanx
May 28 09:15:07 <DonkeyHotei> you're much more informative than the ML
May 28 09:15:36 <nbd> make sure you look at your patches closely to see if there are changes in there that seem out of place or useless
May 28 09:15:42 <DonkeyHotei> the first patch was the result of make kernel_menuconfig
May 28 09:15:54 <nbd> right
May 28 09:16:05 <nbd> maybe a refresh patch before the explicit change would be a good idead
May 28 09:16:12 <nbd> so first make kernel_oldconfig
May 28 09:16:13 <nbd> make a patch
May 28 09:16:19 <nbd> then make kernel_menuconfig with your changes
May 28 09:16:32 <nbd> makes it easier to keep the patch in shape for review
May 28 09:16:36 <DonkeyHotei> the existing kernel config for 3.8 was just cp'ed from 3.7 and things changed that were not addressed. the patch fixes that
May 28 09:16:58 <DonkeyHotei> ok
May 28 09:17:35 <nbd> and when you use git, it'll be easier for you to maintain a series of patches
May 28 09:17:54 <nbd> since you won't have to re-diff whenever you change something
May 28 09:17:56 <nbd> just change your local commits
May 28 09:18:19 <DonkeyHotei> i can resubmit in the style you just enumerated. Noltari mentioned he might have an opportunity to test the patchset as is on different hardware, but idk when he'll be back
May 28 09:19:21 <DonkeyHotei> once i do switch over to git, i'll just put it on github and submit pull requests
May 28 09:19:28 <DonkeyHotei> is that ok?
May 28 09:20:20 <nbd> no, you should send patches, not pull requests
May 28 09:20:37 <nbd> but send patches using git (e.g. with git send-email)
May 28 09:20:56 <DonkeyHotei> well, since i'd have it locally anyway, that shouldn't be an issue
May 28 09:21:26 <DonkeyHotei> and if pull requests are not the way to go, then maybe github isn't, either
May 28 09:22:54 <nbd> right, you don't need github for openwrt development
May 28 09:23:24 <champtar> nbd: authsae use a lot of part of openssl
May 28 09:23:30 <champtar> grep -R "openssl" . give me:
May 28 09:24:58 <champtar> bn.h, sha.h, ec.h, evp.h, hmac.h, rand.h, crypto.h
May 28 09:25:25 <nbd> hm, maybe i should look into it and see if i can make it use something small instead
May 28 09:38:05 <champtar> nbd: i will look at OPENSSL_NO_EC2M
May 28 09:50:43 <champtar> nbb: with OPENSSL_NO_EC2M libopenssl ipk is  35Ko larger (was 52Ko)
May 28 09:51:02 <champtar> nbd*
May 28 10:03:58 <nbd> ok
May 28 10:04:19 <OpenWrt-Commits> mirko r36740 packages/admin/ (6 files in 3 dirs) * This patch add 3 extra packages for zabbix-agentd
May 28 10:07:07 <OpenWrt-Commits> mirko r36741 packages/admin/zabbix/Makefile * Update zabbix to version 2.0.6.
May 28 10:35:00 <champtar> mirko: thanks for applying zabbix patch, can you just delete https://dev.openwrt.org/browser/packages/admin/zabbix/files/zabbix_agentd.conf
May 28 10:35:15 <champtar> (empty file)
May 28 11:01:27 <champtar> nbd: OPENSSL_NO_EC_NISTP_64_GCC_128 is already selected by default, so 35Ko seems to be the minimum
May 28 11:04:02 <OpenWrt-Commits> mirko r36742 packages/admin/zabbix/files/zabbix_agentd.conf * [packages/zabbix]: delete empty file zabbix_agentd.conf
May 28 11:04:55 <champtar> nbd: is it ok? or should i use @OPENSSL_WITH_EC instead of +@OPENSSL_WITH_EC (forcing authsae users to compile there own image)
May 28 11:05:09 <nbd> i'd say enable the option by default
May 28 11:05:14 <nbd> to avoid snapshot build issues
May 28 11:05:47 <jow_laptop> meh
May 28 11:05:48 <nbd> make sure you add the option to PKG_CONFIG_DEPENDS
May 28 11:06:00 <jow_laptop> can't we make a build variant or something?
May 28 11:06:26 <nbd> openssl is bloated anyways, 35k is not much compared to the total lib size
May 28 11:06:29 <jow_laptop> ok
May 28 11:06:46 <nbd> when i find the time for it, i want to make authsae work without openssl
May 28 11:06:51 <nbd> then we can disable it again
May 28 11:07:04 <nbd> but that shouldn't hold up including the package, imho
May 28 11:08:18 <champtar> nbd: now that most device have 8Mo, having a big openssl shared between hostapd, authsae, openvpn is a good thing
May 28 11:09:50 <nbd> not really
May 28 11:10:14 <nbd> hostapd with internal crypto, openvpn with polarssl and authsae with builtin stuff will be smaller than sharing openssl between the three
May 28 11:10:22 <nbd> significantly smaller
May 28 11:10:34 <champtar> nbd: really
May 28 11:10:56 <nbd> yes, openssl is *that* bloated
May 28 11:11:35 <champtar> nbd: can't we disable more openssl feature?
May 28 11:11:57 <OpenWrt-Commits> nbd r36743 trunk/package/ (8 files in 2 dirs)
May 28 11:11:57 <OpenWrt-Commits> mac80211: add support for "active" monitor interfaces which allow userspace tools to connect to APs via injection
May 28 11:12:12 <nbd> champtar: why bother trying to make openssl smaller, when replacing it is easier
May 28 11:12:15 <nbd> openssl code is crap anyway
May 28 11:13:17 <champtar> nbd: is it more performant at least (on routers)?
May 28 11:13:49 <champtar> (asm code ...)
May 28 11:14:18 <nbd> where performance matters, we can port the asm code to polarssl or something
May 28 11:14:27 <nbd> unlike openssl, polarssl actually has some decent code
May 28 11:14:29 <nbd> in addition to being small
May 28 11:14:33 <DonkeyHotei> nbd: is the new resubmit better?
May 28 11:14:48 <nbd> it hasn't reached my inbox yet
May 28 11:20:20 <DonkeyHotei> nbd: it's all in patchwork now
May 28 11:21:33 <nbd> looks better, let's see what john says
May 28 11:24:17 <champtar> nbd: sure about OPENSSL_WITH_EC default y, dependencies should be sufficient
May 28 11:25:03 <champtar> (does dependencies bug that much)
May 28 11:57:09 <champtar> nbd: http://patchwork.openwrt.org/patch/3707/ authsae v2
May 28 13:22:57 <mirko> champtar: welcome and thanks
May 28 13:43:15 <matteo> hi all
May 28 13:43:17 <matteo> anyone has ever used a WRT54G3G-EM?
May 28 13:44:20 <SwedeMike> it's better if you ask the real question instaed of the "anyone" question.
**** BEGIN LOGGING AT Tue May 28 14:16:22 2013
May 28 14:20:27 <matteo> A friend has a WRT54G3G-EM, I haven't one. I think that he bricked his device, anyone knows the rescue procedure?
May 28 14:20:50 <matteo> usually is power+reset button and then tftp or wathever
May 28 14:27:34 <matteo> I haven't one, otherwise I would solder the serial on it
May 28 16:08:32 <brnt> the wiki says that "swconfig allows you to configure /configurable/ switches". the rt3052 isn't in the list of examples of configurable switches. can anyone confirm that it is/isn't configurable?
May 28 16:08:45 <OpenWrt-Commits> nbd r36744 trunk/package/mac80211/patches/300-pending_work.patch * ath9k: fix some issues in PA predistortion (still disabled by default)
May 28 16:18:32 <champtar> nbd: can you commit authsae http://patchwork.openwrt.org/patch/3707/ , so we (someone on the mailing-list seems to have a patch) can move on with authsae integration
May 28 16:21:36 <stintel> that patch should probably better be splitted into two patches. 1 for enabling the openssl EC stuff, and a 2nd one that adds the new package
May 28 16:22:00 <brnt> second question: in ramips_esw.c (around line 1075--see pastebin below), rt3050_esw_data gets defined using two big hex values. are those flags documented somewhere?   http://pastebin.com/zjBjFnF6
May 28 16:23:19 <champtar> stintel nbd: do you see anything else?
May 28 16:24:07 <champtar> (to avoid sending 10 versions)
May 28 16:29:30 <OpenWrt-Commits> cyrus r36745 trunk/package/network/ipv6/odhcp6c/Makefile * odhcp6c: fix or debug disappearing IPv6 addresses
May 28 16:30:55 <OpenWrt-Commits> cyrus r36746 branches/attitude_adjustment/package/odhcp6c/Makefile * AA: odhcp6c: backport r36745
May 28 16:58:56 <builder> build #243 of sibyte is complete: Failure [failed shell_12]  Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/243
May 28 17:09:26 <builder> build #283 of orion is complete: Failure [failed compile_8]  Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/283
May 28 17:12:01 <builder> build #248 of xburst is complete: Failure [failed shell_12]  Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/248
May 28 17:19:18 <OpenWrt-Commits> tripolar r36747 packages/net/openssh/Makefile * [packages] openssh: update to 6.2p2
May 28 17:33:05 <champtar1> nbd: http://patchwork.openwrt.org/patch/3711/    http://patchwork.openwrt.org/patch/3712/
May 28 18:33:16 <OpenWrt-Commits> cyrus r36748 trunk/package/network/config/netifd/Makefile * netifd: fix IPv6-addresses disappearing due to lifetime-overflows
May 28 18:34:16 <OpenWrt-Commits> cyrus r36749 branches/attitude_adjustment/package/netifd/Makefile * AA: netifd: backport r36748
May 28 18:52:22 <_trine> hey people trunk is bricking my nanostations
May 28 18:52:48 <_trine> after flashing trunk I cant connect to it
May 28 18:53:30 <_trine> unfortunately the first one I bricked is at the top of my outdoor mast
May 28 18:53:33 <_trine> :(
May 28 18:54:06 <_trine> I have serial coonected to the second one
May 28 18:54:22 <strasidlo> _trine: please, submit a bug report to dev.openwrt.org or you can also ping one of the devs here
May 28 18:54:39 <strasidlo> the first solution is also better, you can also do both
May 28 18:55:33 <_trine> it will get picked up by someone in good time I guess
May 28 18:55:53 <_trine> I dont know why I cant connect to it
May 28 18:56:06 <_trine> I can ping it
May 28 18:56:37 <_trine> I am connected to the second one vis serial
May 28 18:56:59 <strasidlo> on the first one - no telnet? no serial?
May 28 18:57:10 <_trine> correct
May 28 18:57:17 <strasidlo> did you preserve configuration or is it clean?
May 28 18:57:36 <_trine> I cant connect serial to it as it is at the top of my outdoor mast
May 28 18:57:50 <_trine> I preserved
May 28 18:58:22 <_trine> I can cat the files on the second one via serial connection
May 28 18:58:57 <_trine> I cant connect with luci either
May 28 18:59:09 <strasidlo> it is trunk, isnt it?
May 28 18:59:17 <_trine> yes
May 28 18:59:25 <strasidlo> trunk does not have luci by default
May 28 18:59:33 <strasidlo> you need to install it first
May 28 18:59:51 <_trine> mine has luci
May 28 19:00:04 <_trine> BARRIER BREAKER (Bleeding Edge, r36745)
May 28 19:00:13 <_trine> is bust
May 28 19:00:18 <strasidlo> gimme a sec
May 28 19:00:35 <strasidlo> (headache, going to get some pills)
May 28 19:01:37 <_trine> dropbear and network all look ok
May 28 19:03:27 <strasidlo> do you see uhttpd in netstat -tunap?
May 28 19:03:37 <strasidlo> or something else running on port 80?
May 28 19:03:50 <_trine> why
May 28 19:03:58 <strasidlo> because of luci
May 28 19:04:13 <strasidlo> if it's running, it should be there
May 28 19:05:40 <OpenWrt-Commits> jogo r36750 branches/packages_12.09/net/asterisk-chan-sccp-b/Makefile * AA: asterisk-chan-sccp-b: fix PKG_SOURCE_URL, URL, maintainer and dependency
May 28 19:07:05 <_trine> I am just getting network refused from everything
May 28 19:07:39 <groz> ok, I have a dumb question
May 28 19:07:51 <groz> why would you flash the one on the mast first ?
May 28 19:08:48 <_trine> complacency
May 28 19:09:27 <_trine> I only experiment with them it does nothing else
May 28 19:09:33 <groz> hehe, for which we all pay the price once in a while
May 28 19:10:23 <groz> I am in the process of gathering in all my equipment, moving van shows up Friday morning at 0830
May 28 19:10:38 <_trine> I was like that with all the girls :)))
May 28 19:10:40 <groz> and we head up island later in the day, with all of our worldly posessions in tow
May 28 19:33:24 <_trine> there is no output from ifconfig either and trying to invoke it seems to stop any further output on serial until I reboot the nano
May 28 19:33:50 <_trine> oh well it must be whisky time :)
May 28 19:34:19 <_trine> whisky and tele what a nice end to a day
May 28 19:34:23 <strasidlo> It is not. It was a beer time a moment ago.
May 28 19:34:55 <_trine> actually I may start with a pyms
May 28 19:35:04 <_trine> and lemonade
May 28 21:21:38 <OpenWrt-Commits> nbd r36751 trunk/package/mac80211/patches/300-pending_work.patch * mac80211: fix a crash issue introduced in the active monitor code
May 28 21:56:28 <DonkeyHotei> nbd, blogic: updated patchset submitted, omitting the ones that had no comments last time. comment added for the esi patch before that can be resubmitted
May 28 22:02:08 <mazilo> Does anyone here know why gcc-4.7.x Linaro was removed from OpenWRT SVN trunk?
May 28 22:04:29 <nbd> to have one less version to maintain
May 28 22:04:39 <nbd> gcc versions were going out of sync wrt. patches
May 28 22:05:15 <mazilo> nbd: I thought gcc-.4.7.x is pretty stable while gcc-4.8.x may have some optimization issue.
May 28 22:05:33 <nbd> 4.7.x has received only limited testing, because it was never enabled by default
May 28 22:06:08 <mazilo> nbd: IC. Now, I have to fall back to gcc-4.6.x :(
May 28 22:06:59 <nbd> what target was failing for you with 4.8?
May 28 22:07:18 <mazilo> nbd: Kirkwood.
May 28 22:07:31 <nbd> ok, i'll see if i find the time for building it tonight
May 28 22:08:41 <mazilo> nbd: Oh, it builds OK, but I believe the optimization causes some problem with freeswitch package.
May 28 22:09:24 <mazilo> nbd: Right now, I am rebuilding with gcc-4.6.x to see if the problem I encountered on freeswitch compiled with gcc-4.8.x is there.
May 28 22:09:30 <nbd> ok
May 28 22:31:05 <builder> build #253 of avr32 is complete: Failure [failed compile_5]  Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/253
May 29 01:28:43 <OpenWrt-Commits> luka r36752 trunk/target/linux/ (7 files) * kernel: generic: remove upstream patches for 3.10-rc3 and refresh patches
May 29 01:28:44 <OpenWrt-Commits> luka r36753 trunk/target/linux/ (14 files in 8 dirs) * imx6: add new target
May 29 01:28:46 <OpenWrt-Commits> luka r36754 trunk/package/ (5 files in 3 dirs) * [package] uboot-imx6: add new package
**** ENDING LOGGING AT Wed May 29 02:59:58 2013