[06:19] asac: g++ -march=armv5te -pipe -pthread -fno-exceptions -fvisibility=hidden -Wall -D_FILE_OFFSET_BITS=64 -m32 -fno-ident -fdata-sections -ffunction-sections -fno-asynchronous-unwind-tables -fomit-frame-pointer -O3 -fno-rtti -fno-threadsafe-statics -fvisibility-inlines-hidden -DENABLE_LOGGING_AND_PROFILING -DENABLE_DEBUGGER_SUPPORT -D__STDC_FORMAT_MACROS -DDISABLE_NACL -DV8_TARGET_ARCH_ARM -DCHROMIUM_BUILD -DNDEBUG -DNVALGRIND -Iv8/src -MMD - [06:19] MF out/Release/obj.host/v8_nosnapshot/gen/libraries.o.d.tmp -c -o out/Release/obj.host/v8_nosnapshot/gen/libraries.o out/Release/obj/gen/libraries.cc [06:19] cc1plus: error: unrecognized command line option "-m32" [06:19] make: *** [out/Release/obj.host/v8_nosnapshot/gen/libraries.o] Error 1 [06:21] ah well, you hit the same issue :) === asac_ is now known as asac [09:51] * bizkut is away (i am away now) === bizkut is now known as bizkut-offline [09:59] * armin76 looks at lool [10:00] fta: seen the armel build failure of crhomium? [10:02] armin76: looks like v8 hasn't been set up to build natively for arm yet [10:02] armin76: do you know if fta has a bug open for that? [10:03] shenki: no clue [10:04] you guys are the only ones that are attempting to build chromium natively; upstream does cross builds, and so do I [10:04] (i've been babysitting the arm build for the past 6 months or so) [10:05] i know, buts its fun ricing! [10:05] :D [10:05] how much RAM do the build machines ahve? [10:09] 512mb, i guess [10:09] no clue [10:09] * armin76 is not ubuntu dev, nor related to ubuntu at all [10:09] but they should be either marvell dove or imx515, which means 800mhz && 512mb afaik [10:12] nice. so it takes about a week to link? :) [10:12] takes 10 hours to do all [10:17] shenki: have you tried the crossbuilds? [10:19] armin76: yeah, i wrote the cross build infrastructure for arm. it takes about half an hour on my phenom-II 3ghz, with 4G of ram [10:19] shenki: but does it work? :P [10:19] s/wrote/helped write/ [10:19] yeah, it does [10:19] fully? [10:19] have a binary i could test? [10:19] well, i don't build the tests [10:20] but yes, i have buildt and run binaries [10:20] i'll build you one now and upload it [10:20] thanks [10:20] what hardware do you have? [10:21] sheevaplug, efikamx and ssh access to marvell disco duo [10:21] i'm not familiar with the 2nd two. what architecture are they? [10:21] armv7a and armv5te [10:22] the latter has 3g of ram :) [10:22] ! [10:22] want [10:22] ubuntu has plenty of them, ask asac :D [10:23] ah, but it's only armv5 [10:25] * lool whistles [10:25] armin76: the chromium i build is for thumb2, is that okay? [10:25] shenki: guess so, will take a while to test it then, as the efika is a bit busy atm [10:26] hrm. i don't have a v5 rootfs, so it's a bit harder to build for that [10:26] don't worry [10:27] :D [10:29] armin76: what does the efika use for it's hdmi output? [10:29] and hdmi port :D [10:29] er, in terms of hardware. does it have a open driver? [10:30] nope [10:30] (i brought a openrd client and have been mega annoyed at not being able to use a modern kernel on it, due to no support for the video card) [10:30] does the beagleboard have open driver? [10:31] yeah [10:31] guess its the only one [10:31] afaik marvell dove has no open driver as of now [10:31] not for the 3d hardware, but it can display graphics no worries using the kernel driver. it's just been re-written too, presumably it's better [10:31] ok [10:32] that blows. silly embedded companies. [10:33] armin76, recent dss2 work has made settings modes quite nice, although it's not quite xrandr yet. [10:34] cwillu_at_work: do you know if the dss2 stuff can allocate framebuffer memory dynamyicly (as in, when it sees how big the display it's driving is). or do we still need to provide it with the kernel boot args? [10:34] err, excuse the typos [10:35] shenki, I don't know about going to bigger sizes than the initial framebuffer, no; I've mostly been concerned with 1280x1024x16 myself [10:36] cwillu_at_work: ok [10:36] it'd be easy enough to try though, although I don't have a spare board handy at the moment (my only board is tied up with some tests today) [10:36] yeah, i should just try it [10:37] shenki: i have sh board that has a siliconmotion gfx [10:37] the fun is that the x11 driver doesn't detect it :D [10:38] :/ [10:38] someone should make some embedded gpu IP, and write an open driver for it. i imagine they would do quite well out of it [10:52] armin76: The dove drivers are pretty open [10:52] https://code.launchpad.net/~lool/+junk/marvell-libgfx http://git.debian.org/?p=pkg-xorg/driver/xserver-xorg-video-dove.git [10:53] lool: junk :D [10:53] shenki: ^ [10:55] thanks [10:57] armin76: Yeah, I hate that name; it's the name of a special folder where you can push any branch [10:58] armin76: launchpad forces you to push to lp:$people/$project/$branch kind of trees [10:58] e.g. lp:~ubuntu-core-dev/apt/ubuntu [10:58] The special /+junk allows you to throw in any branch you like [10:58] Without having to register a project first [11:00] armin76: ah, you're on gentoo? i wonder if the ubuntu builds will run on your system [11:00] shenki: who knows [11:01] In theory they shold [11:01] should [11:11] not really, ubuntu/debian names some libraries differently which throws the dynamic linker out [11:13] also you need to be running similar versions to the ones i linked against in my rootfs [11:18] armin76: http://jms.id.au/~shenki/chromium-arm-r35329.tar.bz2 [11:19] shenki: got it, thanks [11:19] armin76: no worries. let me know how it goes [12:28] shenki: Most libs have identical SONAMEs in Debian and Ubuntu; which ones did you have in mind? [12:43] lool: not sure, im not a gentoo user. i was trying to convince a friend to use the google-built images and he had trouble [12:55] shenki: armin76: yeah. seems to be an issue with "toolsets" [12:56] like it sets -m32 if _toolset==host [12:56] not sure where in gyp that toolset thing is done [12:56] any idea? [12:56] grep :D [12:56] armin76: haha [12:56] already did that [12:56] didnt find the trick [12:57] i see where it sets -m32 if toolset==host [12:57] but i dont see how the hell toolset = host [12:57] _toolset=host [12:57] actually [12:57] maybe its host = toolset *G* [12:57] ho [12:57] ho [12:58] ho [12:58] merry christmas asac [12:59] * asac waits for shenki to tell me ;) [12:59] * asac kicks off a fresh build for a bug [12:59] file [13:00] armin76: http://code.google.com/p/chromium/issues/detail?id=31063 ... thats our snap with javascript bug ;) [13:03] :D [13:04] asac: sorry i forgot to run the backtrace, do you still want me to do it? [13:04] hm maybe someone wants to update https://wiki.ubuntu.com/ARM ? it still tells arm5t ? [13:04] don't like to ask around on IRC on multiple channels to find out my CPU is no longer supported [13:05] armin76: give it a try. though its not meanng much until we have a snapshot biuld [13:05] and thtahat has the same issue [13:06] christoph_debian: feel free to update that page? [13:06] ;) [13:08] done [13:08] hmm. not sure about karmic [13:08] karmic is armv6 you slacker [13:09] asac: tells mit it's a ineditable page? [13:09] christoph_debian: you need to log in ;) [13:09] i updatred it now [13:09] check if that is good enough for you [13:10] I'd had put it into FAQ as well (the difference point tells it's v5 opposed to debian's v4) but it's quite vissible this way [13:10] (login requires an account) [13:12] ok let me edit that too [13:12] done [13:13] check it please [13:13] (FAQ) [13:13] shenki: ping me when back ;) [13:13] * asac makes some coffee [13:31] shenki: armin76: http://code.google.com/p/chromium/issues/list?thanks=31274 [13:32] fta: ^ [13:33] asac, wrong link? [13:33] crbug.com/31274 ? [13:34] seems more like it [13:34] hmm [13:34] guess os [13:37] triagged [13:44] ok patched that stuff away ... lets see [13:50] 14:49 < jdstrand> asac: netbook-launcher-efl accepted [13:50] ole! [14:03] NCommander: did you check whether kexec sitll broken on arm? [14:08] fta: so triaging means its still left in unconfirmed state? [14:08] it was confirmed by armin76 and me ;) [14:12] asac, well, i usually semi-triage stuff, i won't assign a milestone as i'm not release manager there [14:13] just moved it to untriaged, meaning confirmed but not reviewed for priority & assignment [14:15] where do i see "untriaged"? [14:15] i always say "unconfirmed" ;) [14:15] see [14:15] status [14:15] what does available mean? [14:16] you can't set it yourself, you have to be bugcontrol [14:16] available is fully triagged, but not assigned [14:16] ah [14:16] felt like "fixed" ;) [14:16] which would have been odd [14:17] fta: do you see where _toolset is defined? [14:18] asac, http://www.sofaraway.org/ubuntu/tmp/chromium-status.png [14:18] its odd as i see toolset without _ being tested elsewhere [14:18] and i see _toolset nowhere being set [14:18] neither gyp code, nor chromium [14:19] are there .py extensions to .gyp in the chromium tree? [14:19] i only grepped in .gyp there [14:19] grep gyp and gypi [14:19] hmm. i think i also tried that [14:19] let me check [14:19] i need to run, be back later [14:20] there are only tests [14:20] no sets [14:22] hah [14:22] found it [14:22] LoadAutomaticVariablesFromDict [14:22] http://paste.ubuntu.com/348585/ [14:31] quick [14:34] still ;) [14:34] no clue how this whole thing works [14:34] e.g. is that injected from outside ... detected during build time? etc. [14:57] hmm ... now the build failed with the same "pure virtual function called" [16:17] asac: what did you do? [16:18] armin76: i dropped the -m32 flags hackishly [16:20] hah [16:26] armin76: http://paste.ubuntu.com/348631/ [16:48] asac: which gcc are you using? [16:51] armin76: tried both: gcc-4.4 and gcc-4.3 [16:51] gcc-4.4.real (Ubuntu 4.4.1-4ubuntu8) 4.4.1 [16:51] Copyright (C) 2009 Free Software Foundation, Inc. [16:51] This is free software; see the source for copying conditions. There is NO [16:51] warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. [16:51] hmm [16:51] though i didnt build all with g++-4.3 yet [16:51] maybe i should try [16:51] i now disabled arm_thumb=0 alltogether [16:51] seems the v8 code generation has special casing [16:52] so could be that the combination armv7=0 arm_thumb=1 isnt good [16:53] ok kicking a new build with AVOID_GCC44 [16:57] armin76: http://pastebin.com/f73b8edfb [16:57] thats the hackish patch to continue [16:57] you didnt get the virutal func issue before, so maybe it just works for you with that === bizkut-offline is now known as bizkut [19:22] asac: export LD_LIBRARY_PATH=/var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/lib.host:/var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/lib.target:$LD_LIBRARY_PATH; cd v8/tools/gyp; mkdir -p /var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/obj.target/geni; "/var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/mksnapshot" "/var/tmp/portage/www-client [19:22] /chromium-9999/work/chromium-9999/out/Release/obj.target/geni/snapshot.cc" [19:22] /bin/sh: line 1: 14154 Segmentation fault "/var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/mksnapshot" "/var/tmp/portage/www-client/chromium-9999/work/chromium-9999/out/Release/obj.target/geni/snapshot.cc" [19:22] make: *** [out/Release/obj.target/geni/snapshot.cc] Error 139 [19:39] armin76, looks familiar, could you get a backtrace? [21:07] * bizkut is away (i am away now) === bizkut is now known as bizkut-offline [21:15] armin76: so its still building with gcc 4.3 [21:15] nice [21:15] armin76: do you get a reasonable backtrace for that seg? [21:16] the sigill is just rubbish. maybe thats a better hint to see whats going on [21:16] asac, ogra_: http://paste.ubuntu.com/348756/ [21:16] Will send to qemu-devel@ shortly [21:19] nice [21:20] I had also noted that I should fix qemu to sigill on instructions not in the emulated CPU, but actually it does [21:20] qemu-arm takes a -cpu arg, just like the system one [21:20] It's just the man page which sucks [21:52] cool ;) [21:52] armin76: mine is still spinning ;) [21:52] something must be wrong :) [22:21] lool: funny away msg! :P [22:21] fta: asac: will check tomm, i'm tired atm [22:27] * lool is away (not reading armin76's temptations) === jmc93739653 is now known as jmc93739653_ === bizkut-miau is now known as bizkut-redhat