=== Jack87|Away is now known as Jack87 === rextsai is now known as chihchun === MrCurious_ is now known as MrCurious [05:01] lilstevie: hey, I think I must be doing it wrong; I'm trying to get into the SOS (single) kernel [05:01] With the system completely off, I hit voldn+power, and it pops up saying "press volup to confirm" with some android icons in the middle of the screen [05:02] At that point, no matter what I hit, it seems to boot into the LNX kernel -- at least I assume it's LNX not SOS, because X comes up [05:04] nm, I misread the message. [05:04] I have to hit power+voldn, then click voldn, then click volup [05:06] Except that STILL didn't work, WTF, why is X still there [05:20] So if you're curious, what I'm *now* trying is to copy the debian armhf rootfs onto a USB key, then put that on the TF101 as a chroot. And in *there* to do "make localyesconfig" and see what happens [05:22] localmodconfig turns off all the CONFIG_MEDIA_TUNER_*=m's [05:22] localyesconfig has no effect [05:22] I guess they aren't very useful unless you're running them on a kernel that's more like Debian/Ubuntu default, where almost everything is =m [05:23] heh [05:23] did you make sure your kernel command line on SOS is single [05:28] i think so [05:43] lilstevie: do you find you have an area of brightness near the dock connector, mostly noticable when the entire screen is set to #000 ? [05:43] twb`: that is due to the backlight layout [05:44] OK, so it's not just mine is defective [05:44] there are 4 spots around the screen which are noticeably brighter [05:44] I half-suspected it was normal; I'll live with it [05:44] it's cause the screen is sidelit rather than having a backplane that lights up [05:45] Can you selectively turn off some of those when your window is only e.g. in the top-left of the screen, to save power? :-) [05:46] Ref. Flinn and Satyanarayanan, (2004) [05:49] lol [05:49] no [05:49] it is all or nothing [05:49] Just checking [05:49] Since there's no rotating HDD, backlight and LCD will be some of the biggest power consumers [05:50] they are the biggest consumers :) [05:50] Especially if you find some way to power down the RF modules when they're not actively used [05:50] well bt isn't powered up if you don't load it [05:51] but both wifi and bt are rfkill compat [05:51] Does NM proactively power them down by default? [05:51] nm doesn't touch the wifi [05:51] on 2.6.36 [05:51] cause it is incompatible [05:51] on .36 "ip a" can't even see the wifi [05:52] but without an active scan working you cant scan :p [05:52] I get damn good battery life on mine though [05:52] I use about 10% per hour [05:52] all times [05:52] because of lack of suspend [05:52] So about 10 hours on ubuntu, with no real effort to tune/rice it? [05:53] 20 [05:53] 100% dock 100% tf [05:53] Wow, that's MORE than asus quoted [05:53] I was expecting more like 10hours since they quote 16 [05:53] I guess that's because android has all that phone-home spyware huh ? ;-) [05:56] So anyway, any idea why "ip a" and "rfkill" can't see wifi with the .36 kernel you provided? [05:56] *kernel and rootfs [06:00] because the driver is stuffed [06:00] like most bcm4329 drivers [06:00] androids method of rfkilling the wifi is to unload its module [06:00] OK [06:01] Which is decidedly effective. [06:01] Pfft [06:01] Only if rmmoding it also powers it down [06:02] well it unloads the firmware [06:02] :p [06:02] * twb` sighs [06:02] more specifically it triggers the reset gpio [06:02] Of COURSE there's a firmware blob, how silly of me [06:02] Yeah. It was actually the only good way I had (short of rebooting) to "reset" an ath9k on a netbook here. [06:02] 9k doesn't have blobs [06:03] trimslice has blobs too [06:03] for both its eht and wifi [06:03] I know because 9k works OOTB on debian [06:03] Which as the FSF "has finally taken a step towards being a free operating system, by removing binary blobs from its kernels" [06:04] that doesn't mean the blobs aren't slipstreamed into it :p [06:04] :-P [06:04] in /lib/firmware [06:04] Only if you enable non-free/restricted section [06:04] That's why iwl (intel) wifi doesn't work OOTB on debian, but does on Ubuntu [06:05] Not sure I claimed ath9k had a blob. [06:06] Just that rmmodding was an effective solution to getting it to unbork itself. [06:06] Sorry, that was me misreading [07:37] Does TF101 (Tegra 2) speak Thumb EE? If so, and I'm compiling a custom kernel specifically for that device, is there any reason *not* to enable CONFIG_ARM_THUMBEE? [07:43] twb`: I enabled it once and it wouldn't compile [07:44] OK [07:44] but yes tegra2 does speak thumb [07:44] just not ENON [07:44] er [07:44] NEON [07:44] thumb, thumb2, thumbee, thumb2-ee, or what? [07:45] at least thumb and thumb2 [07:45] :p [07:45] what ever is in the A9 mandate [07:53] NCommander: ping === vila is now known as babune === babune is now known as vila [08:15] lilstevie: what's tegra PWM do? [08:26] anyone using the panda board attached to an hdmi video? i'm missing some packages to make it work... === doko_ is now known as doko [08:52] ogra_, around? [08:54] ppisati, by hdmi video you mean a monitor via hdmi cable? [08:54] I don't think that needs special packages [08:59] janimo: yep [08:59] janimo: uhmmm [09:00] janimo: is your panda attached to an hdmi monitor? [09:00] janimo: and are you running latest kernel? [09:04] janimo, just finishing my mail, then i'll get to debian-äcd [09:04] lilstevie: it's annoying that when the screen blanks (like, DPMS), the only way to unblank it is hitting the power button -- you can't unblank it from the keyboard or touchscreen === Jack87 is now known as Jack87|Away [09:10] janimo, sigh, so even if i add the code, we cant build test images, cross your fingers the archive skew vanishes today [09:22] janimo, does the mx5 actually need a fat partition to boot ? [09:22] ogra_, that's how the linaro image does it [09:22] (code looks good, happy to merge it) [09:22] k [09:22] then let me merge [09:22] ogra_, if uboot knows about other FS types it may not be needed [09:23] ogra_, there's a raw non-fs area where bootloader is loaded from [09:23] and depending on what that can do the rest of the process is up to it [09:23] well, depends which device you bukd for i guess, i think the efikas can boot from /boot on ext2 [09:23] so I think uboot at fixed locaion then usbboot or whatever uboot is capable of should work [09:24] ogra_, ah this is quickstart only, I am not aware of any other mx53 board [09:24] right, thats a good start [09:24] I'd be happy though if this code worked for mx51 and older boards too [09:24] depends if the u-boot works on it [09:24] ppisati, I only connect the monitor via hdmi, and only when I need to test graphics (GLES) [09:25] ppisati, my connection is flakey sometimes it works, most times not [09:25] may be a timing issues [09:25] sometimes restarting the monitor helps [09:25] ppisati, what I usually found is it works better - almost always maybe - with stock images, not ones which I meddled with and upgraded [09:25] so stock latest linaro Ubuntu LEB for panda booted up fine in X [09:26] ogra_, ah indeed, those may be redboot only. [09:26] well, there is a u-boot that supports them [09:26] but no idea if thats the one we use :) [09:27] ogra_, then we couild look into it once mx53 works, likely in P [09:27] as long as the quickstart works its fine i'd say [09:27] if there is demand from users. I have no idea how popular those boards are [09:27] ogra_, which package is holding us up today? [09:27] is that why ac100 for today is not published? [09:32] What does "ld: no machine record defined" mean? [09:32] janimo, merged ... [09:33] janimo, i enmabled it in cdimage too and started a testbuild, but dont hold your breath, this morning we still had archive skew through unity [09:40] ogra_, thanks, let's see how far that goes [09:44] janimo: so you are using linaro images [09:45] janimo: that makes me think we miss something [09:45] janimo: for example, what's X saying about it? [09:45] ppisati, no, it worked with natty 11.04 too [09:45] I just used a recent linaro one to test my monitor and gles [09:46] It may be that a proper default kernel/X setup works. I mostly use the panda headless, then when I need a monitor I switch the one from my laptop manually [09:47] this may be one of the issues, I am not sure how good hdmi hotplug on the panda is [09:47] janimo: yeop, it worked in natty but lately we switched away from omafb [09:47] omapfb [09:47] also I am not sure if X is having a say in this, headless images sometimes come up with a green tty login on the monitor, sometimes not [09:47] I think it is mostly kernel [09:48] and i don't have any video output anymore since then [09:48] omapfb used to do console framebuffer too? [09:48] yep [09:49] ppisati, I know it worked for me with a stock 3.0 kernel 2 weeks ago in oneiric [09:49] haven't tested since [09:49] ok [09:49] right now X complains loudly: [09:49] Fatal server error: [09:49] [ 141.057] no screens found [09:49] [ 141.057] [09:49] and dies there [09:49] rsalveti, has been fixing DRM related things in that period, but those should be unrelated to the actual physical display support, which I think is the issue here [09:50] ppisati, ah, you may be misssing the 99-pvr.conf file for X [09:50] in /usr/share/X11/xorg.conf.d or such? [09:50] flag@omap:~$ ls -la /usr/share/X11/xorg.conf.d/ [09:50] total 28 [09:50] drwxr-xr-x 2 root root 4096 2011-09-12 17:59 . [09:50] drwxr-xr-x 5 root root 4096 2011-09-12 17:59 .. [09:50] -rw-r--r-- 1 root root 1099 2011-09-09 15:30 10-evdev.conf [09:50] -rw-r--r-- 1 root root 590 2011-06-29 18:42 11-evdev-quirks.conf [09:50] -rw-r--r-- 1 root root 364 2011-06-29 18:42 11-evdev-trackpoint.conf [09:50] or maybe unrelated, I know I had to edit such a file [09:50] -rw-r--r-- 1 root root 749 2011-07-05 15:56 50-synaptics.conf [09:50] -rw-r--r-- 1 root root 660 2011-05-25 09:33 50-wacom.conf [09:50] no pvr config here [09:51] it used to be there in older SGX packages [09:51] but went away recently as X is now supposed to figure out things without a static config [09:51] what does your X log says? [09:51] I even merged rsalveti's patch to Xorg last week, may not have been uploaded yet [09:51] I'll send you the pvr to place it there and see if it works [09:52] ok [09:52] ppisati, sent [09:55] janimo: got it and copied [09:58] * ppisati notes panda still takes ages to reboot... [09:58] * ogra_ watches the mx5 build go into the exciting phase === zyga is now known as zyga-afk [10:00] kernel installation seems to have worked ... next critical point is ubuntu-desktop^ [10:03] janimo: still no luck [10:04] janimo: is your panda up&running? [10:04] janimo: can you tell me what you X log looks like? [10:04] ppisati, it is now, but headless [10:04] janimo: X should be working anyway [10:04] ppisati, http://paste.ubuntu.com/688214/ [10:05] oh, awesome, no archive skew anymore \o/ [10:05] ogra_, quick, roll images. This may only last 20 minutes :) [10:05] lol, yeah [10:06] well, i notice it only because i already roll an image for you ;) [10:06] and watch the log [10:08] janimo: dpkg -S /usr/lib/xorg/modules/drivers/pvr_drv.so [10:08] janimo: what does it say on your panda? [10:13] ppisati, I have the linaro maintainers overlay PPA activated [10:13] I think only that had the updated sgx stuff [10:13] ah [10:13] but I am not 100% sure [10:13] that could be [10:13] ok [10:13] dpkg still running, I have a dist-upgrade going on [10:14] ppisati, pv4-omap4 [10:14] pvr-omap4 [10:14] uhm [10:14] but i'm running oneiric [10:14] crap [10:14] that should soon show up in the ti ppa too [10:14] I am running oneiric too [10:14] iirc rsalveti planned to copy these packages over [10:14] ok, let me try if it works [10:15] no idea, what their status is. Ideally they should be in the main archives, it is essential functionality [10:15] put in restricted if they have licensing issues [10:15] right, its just that nobody did the work for this yet [10:18] ogra_, are there any non-technical blockers for this? [10:19] lilstevie: hey, since you're compiling the kernel without PCI support, what is the backplane? How is e.g. the wifi chip connected to the CPU? [10:21] janimo, i dont think so, but better ask david about the licensing status [10:23] twb`: kernel is compile without PCI support because PCI is unused [10:24] wifi/bt is SDIO [10:24] crap [10:24] pvr compilation faield [10:24] ufff... [10:25] and the backplane I was talking about was the LCD one, most devices the whole back is covered in a film which supplies the backlight [10:25] this one uses 4 sets of 2 or 3 LED's that supply the backlight [10:27] ppisati, do you use the linaro ppa? [10:27] I know that had a fix for building with the 3.0 kernel [10:27] I wonder if new changes landed which broke it again [10:31] janimo: yep, just enabled it [10:31] hm [10:31] twb, did you get that? [10:37] ok, my bad [10:37] now it's properly installed [10:37] let's see... [10:44] ok, now is a bit better [10:44] but X still complains [10:45] [ 141.016] (EE) Couldn't get PVR Services status [10:45] [ 141.016] (EE) No devices detected. [10:45] hmm, watching the buuld log for mx5 it strikes me that we should review the set of default languages we ship [10:46] seems we currenntly only have zh, en, es and pt [10:46] zh? [10:47] chinese [10:47] you should add Sanskrit, i head it's the way to go! :) [10:47] i heard [10:47] nah, swaheli [10:47] or what was that language that was supposed to suprsed everything else... [10:47] or that click tune the african bush people speak ... i think we have a translation for it [10:48] * ogra_ tries to remember the name [10:48] ah, xhosa was it [10:48] mark invested a lot into that translation years ago [10:48] really? :) [10:48] yep [10:48] hired 4 people around hoary [10:48] took them two cycles to get everything together [10:49] but i think its rotting since then [10:49] ah yes [10:49] the Esperanto! :) [10:50] we have esperanto too [10:50] and klingon [10:51] then if have Klingon we are good :) [10:51] language-pack-tlh i think [10:51] might be funny to roll an image that defaults to it [10:51] asd :) [10:55] btw, one day you'll have to teach me how to roll a release [10:55] you mean building images ? [10:56] we had a spec for writing a howto, but that fell off the shelf for O [10:56] we'll re-visit it in P [10:56] well, actually i would like to know how do the entire process [10:57] i mean, setup my own builders, build a list of known packages and then create an img out of them [10:57] one part of that is trivial... sadly the other is so hard that not even i would run my own builders [10:58] for building the livefs you just need to use livecd-rootfs and call the right command ... [10:58] ok [10:59] to make an image out of that you need to have a setup of cdimage and debian-cd ... both of thzem expect a certain environment (servers to find packages on, etc) [10:59] setting up the latter part isnt easy [10:59] uhm, i see [11:00] i usually do test builds in the datacenter [11:00] i know NCommander once set up a local build env ... but that took him days [11:02] isn;t the raspberry pi an armv6? [11:03] so it's limited to... karmic? [11:03] i guess people would like something more up to date [11:03] and would love to "recompile" their own version of ubuntu [11:04] for me it was something i would like to know [11:04] rebuilding the archive ? [11:04] that will take you ages unless you have a good amount of arm builders [11:04] ah right [11:05] no cross compilation... [11:05] and you need to know a lot about packages if you run into issues, i wouldnt suggest to anyone to even think about attempting it currently [11:05] i know that one of linaros masterplans is to make that easy [11:05] that would be awesome [11:05] but i suspect that will still take multiple cycles [11:06] for cross building we will at least get a bunch of big packages that will support it in the future [11:06] that would help a lot [11:06] (see the ubuntu-devel ML discussion about that) [11:07] i think i missed the thread [11:08] subject: "What would you like to cross-compile?" === zyga-afk is now known as zyga [11:21] we have audioooooooooooooooooooooooooooooo [11:21] :) [11:30] ppisati, from pulse ? [11:31] ogra_: aplay [11:31] yeahm, that was announced in fridays call already [11:31] doh [11:31] wasn't there [11:32] pulse is more critical (and the reason why i was so pushy to get sound devices back) [11:32] well, if aplya works it means the kernel is good [11:32] i mean [11:32] driver&c is ok [11:32] pulse is not running on my setup right now [11:32] and i wonder why... [11:33] ok [11:33] pulseaudio running now [11:33] let's see [11:41] uhm [11:54] rsalveti: when you are up&running, can you take a look at this http://paste.ubuntu.com/688274/ [11:55] rsalveti: can't get any video output on my oneiric panda [11:56] ppisati: the pvr xorg driver is not expected to work with latest kernel [11:57] rsalveti: uhm, so no X for oneiric? [11:57] ppisati: I still didn't fix it because ti was saying that they would release a new version with the fixes and some other improvements [11:57] ppisati: you can have X, but without the pvr driver [11:57] rsalveti: usign which driver? [11:57] fbdev [11:58] ppisati: I can probably fix the driver, will see, but wasn't going to do it because of the new driver [11:58] rsalveti: no prob [11:58] but you know how it works, it can be the case that TI will only release it after the release [11:58] rsalveti: if i can get any ouput with another driver, i'll use that [11:58] yeah [11:58] ppisati: just remove the package that provides /usr/lib/xorg/modules/drivers/pvr_drv.so [11:59] ppisati: will give it a try later today [11:59] let you know if it works [11:59] rsalveti: wiuthouyt that pkt, was not working [11:59] ppisati: what was the errot then? [11:59] it should always work with fbdev [12:00] unless the drm driver is broken in some sort [12:01] removing pvr, wait... [12:02] ppisati, fyi https://bugs.launchpad.net/ubuntu/+bug/848154 [12:02] Ubuntu bug 848154 in ubuntu "ARM version not supporting V6 RaspPi" [Undecided,Invalid] [12:03] so we already have a bug :) [12:04] yeah, i just closed it :) [12:06] flag@omap:~$ sudo X -configure [12:06] X.Org X Server 1.10.4 [12:06] Release Date: 2011-08-19 [12:06] X Protocol Version 11, Revision 0 [12:06] Build Operating System: Linux 2.6.31-608-imx51 armv7l Ubuntu [12:06] Current Operating System: Linux omap 3.0.0-1204-omap4 #9-Ubuntu SMP PREEMPT Mon Sep 5 19:29:18 UTC 2011 armv7l [12:06] Kernel command line: console=ttyO2,115200n8 root=/dev/sda1 rootwait ro vram=32 elevator=noop earlyprintk [12:06] Build Date: 09 September 2011 11:41:55AM [12:06] xorg-server 2:1.10.4-1ubuntu1 (For technical support please see http://www.ubuntu.com/support) [12:06] Current version of pixman: 0.22.2 [12:06] Before reporting problems, check http://wiki.x.org [12:06] to make sure that you have the latest version. [12:06] Markers: (--) probed, (**) from config file, (==) default setting, [12:06] (++) from command line, (!!) notice, (II) informational, [12:06] (WW) warning, (EE) error, (NI) not implemented, (??) unknown. [12:07] (==) Log file: "/var/log/Xorg.0.log", Time: Tue Sep 13 14:06:04 2011 [12:07] List of video drivers: [12:07] sisusb [12:07] cirrus [12:07] r128 [12:07] mach64 [12:07] radeon [12:07] ati [12:07] fbdev [12:07] vesa [12:07] No devices to configure. Configuration failed. [12:07] ddxSigGiveUp: Closing log [12:07] it seems it doesn't care about fbdev [12:07] pastebin !!!!!!!!!! [12:07] for more than three lines please [12:08] and i'm not sure X -configure works if you dont have a PCI bus [12:12] well at one point the X log says [12:12] "falling back to old method to probe for " [12:13] so i guess it's trying hard to find the gfx chip [12:13] well, then it should generate a config for "" :P [12:13] but afaqik it only scans the PCI bus [12:13] it should, but it doesn't... :) [12:13] even USB isnt supported [12:14] asac once added a hack to make it walk /dev/fbX but that might have been dropped [12:14] i'll try with a dummy xorg.conf then [12:14] why do you use an xorg.conf at all ? [12:14] i don't [12:14] that's the point [12:14] it will autodetect fbdev just fine without [12:14] nope [12:14] it doesn't [12:15] how doesnt it ? [12:15] wait [12:15] let me paste the entire x.log here [12:15] ok [12:15] kidding :) [12:15] :9 [12:15] wait [12:18] http://paste.ubuntu.com/688285/ [12:24] thats what you get if you start lightdm ? [12:25] janimo, hmm, livefs build seems to be fine but somehow it tries to use a tar.gz instead of the .ext3 image [12:31] ogra_: sorry, wrong buffer [12:31] there's no /dev/fbX [12:31] oh [12:31] http://paste.ubuntu.com/688296/ [12:32] well, then its no surprise that it fails [12:32] but isn't /dev/fbX provided my omapfb? [12:32] because that was turned off in latest kernl upload [12:32] because IIRC clashed with omap drm [12:33] (but i'll have to check) === zyga is now known as zyga-afk [12:37] ogra_, you mean cdimage tries to use tar.gz? [12:37] yes, just discussing the bug with colin in #ubuntu-release [12:37] I am not sure how those interact, I tried making it look like the omap script [12:37] seems we did soemthing wrong in our implementation [12:37] its not your fault [12:38] NCommander wrote the code in question and i nodded it off [12:38]