[00:20] how much space is required to bootstrap ubuntu-minimal on ARM? === Dr_Who is now known as tgall_foo === Martyn is now known as Martyn^ === ericm-Zzz is now known as ericm|ubuntu === ericm|ubuntu is now known as ericm-afk [04:24] * mozzwald is away: snore... [07:17] how can i get a arm-ubuntu source list? [07:17] i got so many 404 when i use the kind of normal ones.. === ericm-afk is now known as ericm === hrw|gone is now known as hrw [08:21] morning [08:36] amitk: around? [08:36] cooloney: yes [08:36] amitk: i am looking at the OTG config issue you pointed out yesterday [08:37] from the git log, i think you revert the patch which enabling the OTG in the kernel [08:37] so why you revert that? [08:37] OTG driver doesn't work at all? [08:39] cooloney: that config was causing OOPs, IIRC. So we need to see if some more patches are necessary [08:40] amitk: yeah, i guess so. so firstly we enable the OTG as the omap3 beagle board defconfig [08:40] and test and file another bug? [08:40] i think OTG should be built-in [08:41] oh, maybe the dmesg.txt from ogra_cmpc in the LP is from a kernel which built-in OTG stuff [08:51] cooloney: if possible OTG should be a module so we can switch between different gadget drivers - audio, ethernet, gadgetfs, mass storage, etc. [08:52] cooloney: actually, only the USB Gadget drivers need to be modular, the HW-specific driver can be compiled in [08:53] cooloney: http://paste.ubuntu.com/449992/ is the diff for that should enable OTG [08:53] s/for// [08:56] amitk: i agree, gadget upper level drivers should be modulars, and TI OTG controller driver should be built-in [08:57] amitk: after a study of the oops from the dmesg, i think we missed to built-in the OTG transceiver driver as well as OMAP OTG driver [08:57] amitk: let me try to build in them and post a kernel for testing [09:01] Good morning every-peeps [09:03] morning lag [09:03] Hey amitk [09:08] lag and amitk morning. -:) [09:09] Morning cooloney, I'm just replying to your emails [09:13] cooloney: How far have you progressed with the Maverick - Panda kernel? [09:15] lag: I added some panda supported yesterday, and sebjan will review that. if that's ok, i will send out for review and pull [09:15] lag: for the bug #592295 [09:15] Launchpad bug 592295 in linux-ti-omap (Ubuntu) "omapdss DISPC error: SYNC_LOST_DIGIT (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/592295 [09:16] lag: is on .33 or .34? [09:16] and i think you may have some solution now [09:17] cooloney: 2.6.34-900-omap4 [09:18] If you send me a .33 kernel, I can see if the problem is a lasting one [09:18] cooloney: I have seen your solution, but I think I have a neater one [09:19] lag: ok, got it, but in the LP post, you said it is 2.6.33-900-omap4 [09:19] lag: ok, cool, i don't have the hardware. hehe [09:21] In which case, I have tested it on both [09:21] So it is .33 and .34 [09:22] cooloney: Have you read your emails? [09:23] lag: oh, just got it. [09:24] ;) [09:24] for the .34 repo, git://kernel.ubuntu.com/roc/ubuntu-maverick.git ti-ubuntu-2.6.34 [09:24] i think sebjan will do his work based on this branch [09:25] Does that boot? [09:25] hmmm, i think you told me it boots from you panda [09:26] hehe [09:26] If it's the same one you sent me yesterday it does [09:26] and after sebjan fixed his kernel cmdline, he boots that kernel on his panda too [09:26] However, I tried to build it and it didn't boot [09:26] lag: yeah, it is the same [09:26] lag: really? [09:26] Yeah, really [09:26] you built from the same tree, but does not boot? [09:27] oh, maybe you are trying the my old base [09:27] so which commit SHA is your base in you building tree [09:27] It's the one on kernel.ubuntu.com [09:27] Wait one [09:27] np [09:28] ea34fca089b7ab2986ff391f1fc036d4425e6995 [09:29] oh, sorry, that is not including the panda support [09:29] pls just git fetch [09:29] and rebuild, [09:30] That was the latest when you went offline last night [09:30] :( [09:30] I thought it was something I was doing wrong! [09:30] lag: oh, sorry, hold on, [09:30] could you post your git log --oneline? [09:31] I've just fetched [09:32] ok, [09:32] that should be the same as my kernel source, i think [09:33] you are using sbuild or just crosscompile [09:41] cooloney: It's still at ea34fca089b7ab2986ff391f1fc036d4425e6995 UBUNTU: SAUCE: fix changelog for our target release - maverick [09:41] Even after a fetch [09:48] lag: yeah, i think you got the same kernel source as me [09:48] And that should work? [09:48] yeah, i built the kernel package from that. [09:48] and uploaded to you and sebjan [09:49] so is there any error message when you was booting your own built kernel? [09:49] Nope, I just don't receive any output [09:52] lag: hmmm, too bad, no idea about that. you changed the kernel cmdline or something? [09:53] Nope, I'm using the same one [09:53] When I do a diff on or images, they 'differ' [09:53] Is ea34fca089b7ab2986ff391f1fc036d4425e6995 your final commit? [09:55] yeah, i didn't change anything since that [09:57] lag: you are using sbuild or just cross compiling? [09:59] I'm just using schroot [09:59] I'm using the build scripts [10:05] lag: the kernel i uploaded yesterday is cross compiled [10:07] Okay [10:07] Let me try it again [11:36] ogra_cmpc and amitk, i updated bug #566645 [11:36] Launchpad bug 566645 in linux-ti-omap (Ubuntu Maverick) (and 2 other projects) "OTG configuration is broken on omap kernel (affects: 2) (heat: 70)" [High,Confirmed] https://launchpad.net/bugs/566645 [11:38] http://people.canonical.com/~hrw/ubuntu-maverick-armel-cross-compilers/ [11:43] hrw: congrats! can you try powerpc or mips/el? [11:44] zumbi: thats dpkg-cross way [11:44] hrw: yes, but binary-*-cross.mk merged into binary-*.mk? [11:45] zumbi: no, this is still on a list [11:46] oh! ok, also biarch cross compilers were broken for multilib and non-multilib builds, on lenny used to work fine [11:47] anyway, well done :) [11:47] zumbi: gcc-4.5 cross work resulted in a set of fixes [11:47] in a couple weeks I hope to have some time available to help out [11:47] zumbi: https://edge.launchpad.net/ubuntu/+spec/arm-m-cross-compilers has them linked [11:48] i remember you showing me a patch [11:56] Hello, anyone here who can give some advice? [11:56] zumbi: We dont have mips/mipsel in Ubuntu and powerpc in not in the main priorities of Linaro, it's more a Debian thing, but that should just work -- problem is that the patches aren't in the Debian toolchain yet [11:58] JameswStubbs: don't ask to ask ;-) [11:58] lool: those patches are also not in ubuntu toolchain yet [11:59] hrw: Ok, did you apply any to build http://people.canonical.com/~hrw/ubuntu-maverick-armel-cross-compilers/ ? [12:00] Ok then, basically I've built a Ubuntu minimal rootfs using rootstock and have it mounted in qemu with networking, I'm trying to install matchbox. I've installed xorg, and all the matchbox applications but I need GTK, Is there a way to do this without installing the full gnome enviroment? [12:00] lool: 594534, 590696 were needed [12:01] I have a build with also 593281 applied to get libgcc1-dbg [12:02] libgcc1-dbg-armel-cross_4.5.0-5ubuntu1.lool1_all.deb [12:05] Bump (Sorry for bumping) [12:06] JameswStubbs, um, then don't bump? [12:06] this isn't ubuntuforums [12:06] lool1 that's me! [12:06] hrw: ~hrw1 please :-) [12:07] JameswStubbs: in theory you should just be able to apt-get install that [12:07] JameswStubbs: does it pull more than it should? [12:07] JameswStubbs: what specifically is pulled which shouldnt be? [12:07] apt-get install what? [12:07] JameswStubbs: matchbox [12:07] matchbox pulls everything it needs [12:07] But it depends on GTK which is gnome's toolkit I think [12:07] JameswStubbs: You need Gtk+ for what, a custom application? [12:08] Matchbox-session needs gtk [12:08] lool: then maybe sparc? -- I just wonder if biarch cross compilers are fixed. I saw you having some mails with doko about it, but he was wanting to fix it in the right way, while you wanted a quick hack for linaro [12:08] not seeing the problem [12:08] JameswStubbs: Gtk+ is the toolkit used in GNOME, but it's not the full GNOME environment [12:08] gtk isn't all of gnome [12:08] zumbi: This is incorrect [12:08] So what do I need to enter into apt-get to get gtk? [12:08] apt-get install gtk?? [12:09] zumbi: Linaro is investing time in fixing it the right way, hrw is working on that, but it will take a while so I've asked him to produce intermediate builds the "old" way (regular Emdebian way) in the mean time [12:09] zumbi: there is no additional hack than the existing Emdebian binary-cross stuff [12:09] JameswStubbs: the matchbox-session package should depend on the libgtk package [12:10] JameswStubbs: So this should just all work [12:10] JameswStubbs: I fail to understand your actual problem :-( [12:10] lool: patch was yours [12:10] lool: and I do not provide them outside of my hdd [12:10] hrw: You mean the build fixes? [12:10] lool: sure, but Emdebian way only works for armel (also sh4) but the rest of arches were broken [12:10] I'd copy and paste it all but it's within 2 vm's :) , When I enter matchbox-session from the prompt, it says something like gtk: couldnt open [12:10] hrw: These are not additional hacks though, they just fix bugs in the current approach, do you agree? [12:11] lool: i agree [12:11] lool: they are fixes, yes [12:11] zumbi: We didn't look into that, we don't have people blocking on cross-compilers for the arches you mention [12:11] zumbi: I would hope that everything will work once hrw is done, but we don't need to produce daily cross-compilers for sparc or powerpc in Linaro right now [12:12] zumbi: We do need to hand out some armel cross-compilers while hrw continue on the quest to clean cross-compilers builds [12:12] JameswStubbs: Ok, you need to be more specific though [12:12] Ok I'll try [12:13] lool: while I understand you focus on armel (which is primary) I'll try to work on the rest of arches (at least test builds) [12:13] JameswStubbs: (IIUC, you installed only Ubuntu packages in your rootfs, but you see some error messages with gtk, can only comment if we know what the error is) [12:13] zumbi: As I said, I expect the *final* solution to work on other arches [12:13] `buildcross` could easily be used to test all cross compiler arches [12:13] zumbi: But I asked hrw to provide hand-built packages *right now* because we need them, we don't need them for other arches than armel [12:13] lool: i got it. thanks :) [12:14] Ok :-) [12:14] I've made a rootfs using rootstock. I've then opened it in qemu. I then installed xorg. I then installed matchbox. Now when I enter matchbox-session is returns: (matchbox-desktop:434) : Gtk-Warning **: cannot open display: [12:15] JameswStubbs: Do you run matchbox-session from within Xorg? [12:15] Should I enter startx [12:15] Then when I get the x shell then enter matchbox-session? [12:15] JameswStubbs: matchbox-session should be started from within a Xorg context, otherwise it doens't know which Xorg DISPLAY to connect to or which credential (XAUTHORITY) to use [12:16] JameswStubbs: I'm not sure what the proper way is to start matchbox-session on boot or to start Xorg + matchbox-session, perhaps startx will do the trick, perhaps you need to setup more things [12:16] JameswStubbs: One way to quickly test matchbox-session is to start Xorg with a xterm and then run matchbox-session from there, but that's certainly not what you would want in the end I guess [12:16] I think I need to add match-box session to xinitrc [12:17] Yes, it is indeed possible you have to do that [12:28] Hm, All I'm getting after adding matchbox-session to xinitrc is a blackscreen and a mouse, which I can't move. [12:31] Any ideas lool? [12:57] JameswStubbs: Check the xinit man page, and/or read through /etc/X11/Xsession.d/* scripts and /etc/X11/Xsession [12:57] JameswStubbs: I don't have a recipe for you, as I didn't do this myself, but it should be easy to look into what the shell scripts are doing [12:57] JameswStubbs: I think one important thing is to have .xinitrc +x or something like that [12:57] and you need a shebang too if you do this obvisouly [12:58] shebang? !#? [12:58] #!/bin/sh [13:00] JameswStubbs: I checked and if you look at startx, it will pass xinitrd to xinit which expects to be passed a client program [13:00] So if you pass it a text file, it wont work [13:47] ogra_cmpc: where is rcn? isnt he usually hanging out here? [13:48] asac: 16:10 < cwillu_at_work> okay, that's a week since rcn was last on, anybody want to come with me on a rescue mission? :p [13:49] asac: 17:09 * cwillu_at_work found rcn [13:49] asac, he should be back today [13:49] took an unannounced vacation [13:50] ah ... one of the weak ones ;) [13:51] he's running from you :D [13:51] *away [13:51] I was wondering about that actually :p [13:53] omg [13:53] that's embarrasing [13:53] I've got a webapp I run in an undecorated fullscreen firefox window [13:53] been having horrible performance [13:54] firefox using 60-80% cpu, with the backend server using the other 20% [13:54] (backend has to poll a bunch of serial devices and push data to firefox) [13:54] figured it was just the overhead of updating things / the push system [13:54] finally got around to getting venkman on it [13:56] a completely unrelated piece of code was in a hard mutually recursive loop: initializing a drop-down menu was triggering a request to get data, and the request handler was triggering an init of the drop-down [13:57] the embarrassing part is that I've been working under the false assumption for long enough to have actually shipped this to customers :p [15:22] http://people.canonical.com/~hrw/ubuntu-maverick-armel-cross-compilers/ contains now amd64 and i386 cross toolchains for armel target === bjf[afk] is now known as bjf [16:22] hrw: is there a branch for the source used to build your cross toolchain ? [16:22] rsavoye: I used ubuntu sources [16:22] will add README [16:22] so no Code Sourcery patch yet ? [16:22] not yet [16:23] you hacked libstdc++ and libiberty to cross configure ? [16:23] hrw: there is a nice patch in GCC trunk for Android you might want to add when you get to patching 4.4 [16:23] all it does is cleanup linking mostly [16:24] rsavoye: url? [16:24] the thread is at http://www.pubbs.net/201005/gcc/37284-patch-06-add-support-for-android.html [16:25] it's based on my older Android patch, and adds bionic support in a nice generic way [16:25] hey rsavoye [16:25] howdy [16:26] rsavoye: Might make sense for you folks to discuss this on #linaro, but here is fine too [16:26] ah, didn't think of that, thanks [16:26] I was just hacking on ARM toolchains this weekend for Gnash, so pipped up when I saw the link to debs === hrw is now known as hrw|gone === ericm|ubuntu is now known as ericm-Zzz === rbelem is now known as rbelem_Brazil === jcrigby is now known as jcrigby-afk === rbelem_Brazil is now known as rbelem [21:16] Hi. Just as a follow up of last time I connected (About atlas build on arm): I'll upload atlas as someone has been able to build it in qemu [21:16] in case it loops, who should I poke to kill it? === jcrigby-afk is now known as jcrigby [22:06] ogra: I'm still getting a fair number of qemu segfaults with lucid === tumbleweed_ is now known as tumbleweed