/srv/irclogs.ubuntu.com/2013/01/17/#ubuntu-arm.txt

=== zz_chihchun is now known as chihchun
ipl31_anyone know if any of the -M options for qemu-system-arm support more than 1 nic?02:51
=== chihchun is now known as zz_chihchun
=== AmEv_ is now known as AmEv
AmEvHmmm... I wonder if it's the command line messing things up...04:00
=== zz_chihchun is now known as chihchun
dholbachgood morning07:26
sim590hey, I have deleted a tar file and I want to recover it. I'm using scalpel. I need to specefy the header of a ".tar" file, but I have no idea what it is..09:04
sim590anyone knows how to discover the header of a file like a tar file09:10
infinitysim590: "discover the header of a file"?09:16
infinitysim590: What are you actually trying to accomplish?09:16
sim590I want to find my deleted tar file09:18
sim590on my disk09:18
sim590I have to add something like this in the configuration file of foremost09:19
sim590 pdf     y       5000000 %PDF-  %EOF09:19
sim590where "pdf" is the extension09:19
sim590"y" is for case sensitive09:19
sim590and I also need to specefy the header09:20
sim590but I don't know what the *"!(/&*$ it is :D09:20
sim590I want to do this for "tar" file09:21
infinity# POSIX tar archives09:21
infinity257     string          ustar\0         POSIX tar archive09:21
infinity!:mime  application/x-tar # encoding: posix09:21
infinity257     string          ustar\040\040\0 GNU tar archive09:21
infinity!:mime  application/x-tar # encoding: gnu09:21
ubot2`infinity: I am only a bot, please don't think I'm intelligent :)09:21
ubot2`infinity: I am only a bot, please don't think I'm intelligent :)09:21
infinityThat's what file(1) knows about tar files.09:21
sim590man file09:22
sim590ok09:22
infinitysim590: apt-get source file && cd file-* and browse around magic/Magdir to see what it knows about these things.09:22
infinity(The tar stuff is in "archive")09:22
sim590ok.. I'll try that thanks.09:23
sim590infinity: I've not found my file yet, but I understand now what you wrote after looking the files you told me. But how the hell did you think about that? Did you just know that it was in these files? :p09:48
infinityI did, yes.09:50
sim590ok lol.09:52
=== Quintasan_ is now known as Quintasan
=== chihchun is now known as zz_chihchun
sim590#quit11:31
thervehi! I have some questions about 12.10 on pandaboard, can someone help?12:37
ogra_therve, how would we know if you dont ask your question ?12:47
therveone would wonder12:48
therveI've installed from the sd card on an external SD via USB12:48
thervecan I make the USB bootable to remove the install sd card?12:49
ogra_nope12:50
ogra_you could install to the SD itself (at least in raring, not sure if all the fixes were backported to 12.10)12:51
ogra_but the panda cant boot off anything else than SD or a special USB OTG setup that requuires a second PC12:51
thervewell I don't want to boot via USB12:52
ogra_thats a HW limitation12:52
therveI have a SD card plugged in a USB card reader12:52
ogra_<therve> can I make the USB bootable to remove the install sd card?12:52
ogra_oh, ok12:52
thervesorry I meant the SD used in USB right now12:52
ogra_well, if you have the exact same partitioning for the vfat partition, you chouls be able to just dd the contents from one card to another and swap them12:53
ogra_s/chouls/should/12:53
* ogra_ glares at his fingers12:54
thervehum12:54
ogra_the panda needs a vfat partition in the beginning of the card and that needs to have all the bootloader stuff in it12:54
therveok I see12:55
thervesounds like something doable, maybe by installing on a SD with the boot vfat ready12:55
ogra_right12:56
thervecool, thanks12:56
thervethe other question I had was about graphic support/video playback12:57
therveis there anything I can check to make sure I have the proper drivers enabled?12:57
ogra_duplicate the SD (write the image to both cards) then only delete the second partition but leave the first one intact ... after install dd the partition from one card to the other12:57
ogra_with 12.10 the drivers are oreinstalled and used from the beginning12:57
ogra_*preinstalled12:57
therveI'm asking because it feels pretty slow, so I was wondering if that was expected12:58
sim590Anyone has made the sound work on the TF101 under ubuntu with lilstevie's kernel? I've red a wiki speaking about this, but I don't know how alsamixer works.. I don't know where to find the "playback" and DMIC option13:52
sim590http://forum.xda-developers.com/showthread.php?p=25608362&highlight=sound#post25608362 -> A guy says he makes the sound work, but the playback thing.. Idk13:53
janimoorientation and backlight testing app for the nexus7 http://people.canonical.com/~jani/a14:28
janimoto be ran with DISPLAY=:0 set if via ssh and with sudo14:29
janimoit's a static executable , the golang source is https://code.launchpad.net/~jani/+junk/nexus14:29
ogra_janimo, wow, without kernel tweaks ?!?14:34
janimoogra_, well, apparently everything was there already14:34
ogra_yay14:34
janimojust hidden in sysfs14:34
ogra_i wish the backlight control would be there too14:34
ogra_the device is so bright it hurts my eyes after a few hours14:35
janimoogra_, well this app has backlight control14:35
ogra_oh, wow14:35
janimobt probably needs some tweaking14:35
ogra_ah, i see the code, yeah14:35
janimoI did not see it workin gonly starting with andorid 4.2.114:35
janimoand seeing it works there it made sense to look harder at where it is exposed14:36
janimofor a while I thought it simply is not working at all14:36
janimobut I need to ask around how to best integrate these in some daemons and control center14:36
ogra_i looked at the sysfs stuff for quite a while already, but only played with echoing stuff into the respective files14:36
ogra_which always failed with write errors14:36
janimoto have nice UI for it and non extra daemon running14:36
janimoright, some files are read only or einval14:37
ogra_yeah14:37
janimobirghtness is the only knob that should work14:37
janimobut it's enough for our purpose14:37
janimothe x input bugs are still annoying though, I think I'll have another go at those14:38
janimoafter a rotation or two or a few minutes taps are still not working14:39
ogra_woah14:39
ogra_the UI is really slow14:39
ogra_(in redrawing the content rotated)14:40
ogra_janimo, you rock !14:40
ogra_we should package that as a daemon with proper upstart file etc14:40
janimoogra yes it redraws slowly if the axis changes14:40
janimobut if it's a 180 degree rotation it is really quick :)14:41
ogra_i blame compiz14:41
ogra_heh14:41
janimoI guess copying is much easier if you do not have different line lengths and can do rectangle copies14:41
janimoas contiguous pixels in one case are still contiguous in 180 rotation, but not with 90 or 27014:42
ogra_funny that onboard fully wortks while rotated input doesnt in the rest of the UI14:42
janimoogra_, I'll mail ubuntu-devel for some input. While a daemon is fine I'd still rahter add this functionality to some existing plumbing daemons14:42
ogra_not even the restore button inside firefox responds in portrait14:43
janimootherwise it is yet another 1-2MB hit for our footprint14:43
ogra_ugh14:43
janimoI did not test apps too much TBH14:43
ogra_well, i doubt it is the app14:43
ogra_but onboard works flawless, i can type in a terminal no matter what the rotation is14:44
janimojust saw that cursor is where it should be, but may it be some apps get confused if screen size or orientation changes?14:44
ogra_launcher and FF dont take any taps14:44
ogra_they do if rotated back to landscape14:44
janimomaybe those that do their own drawing as oppposed to using a GUI toolkit get confused when screen rotates, no idea14:44
janimohmm, maybe we coudl try xrandr + firefox on x8614:45
ogra_oh, it was only the recovery button that didnt work14:45
ogra_i can scroll with grab n drag in FF and click links14:45
ogra_hmm14:46
ogra_i can klick links but they dont get executed14:46
ogra_(but undelined)14:46
ogra_i guess i'll need to test that with a mouse plugged in so the pointer gets visible14:47
ogra_janimo, why does the footprint raise so much ? GO ?14:48
janimowell GO - at least the official compiler - creates static executables and includes the Go runtime in them14:49
janimoso they are standalone14:49
janimothat makes the executable about 1.2 M, but runtime footprint I guess is also due to runtime bookkeeping14:49
janimoit's not as much as it would be in python but C (or even dash) may be leaner14:50
ogra_heh, lux sensor seems to work fine too ...14:50
ogra_i can use a torch and point at the camera area and the display reacts14:50
janimothe formula is ad-hoc, a better fit is likely possible14:50
ogra_but its slow14:50
janimoyes, I used my desklamp for testing, went up to 40000 lux14:50
janimopassing -v shows lux and accel reading values14:51
ogra_ah, nice14:51
janimoslow in reacting to it?14:51
ogra_yes14:51
janimowell both accel and light are poll loops with 1s delays14:51
ogra_my torch is a tiny LED one pretty focused14:51
ogra_ah, thats it then14:51
ogra_1s seems to match what i see14:52
ogra_i wonder if we could make it trigger uevents so you dont need to poll14:52
janimothere may be interrupt based interfaces for these sensors, but maybe not via sysfs14:52
ogra_but that would indeed be kernel side14:52
ogra_and probably still cause latency14:53
janimoright, eventually having triggers is better than waking up for no reason periodically14:53
janimobut, on android actually the lag is a few seconds14:53
ogra_but i think thats for 13.10 ;)14:53
janimoso  I guess it adapts once it decides this is really ambient light change and not a transitory flashlight14:53
ogra_i'd set it to 0.5s though, will likely feel a lot smoother14:53
janimomaybe eve 5 seconds lag on android14:54
janimoa bit confusing for test purposes but for real use cases it may be better14:54
ogra_wow, 5s is long14:54
janimoespecially since light adjusts very seldom and incrementally14:54
janimowell, if you read and the sun slowly sets you don't really care about a few seconds14:54
ogra_indeed14:55
janimoI think are tests are artificial14:55
janimoour tests that is14:55
Tassadarwell, on Android 4.1, when the backlight is turned to maximum because of ambient light and then you put it to dark place, the brightness never goes down again14:55
=== rsalveti_ is now known as rsalveti
Tassadarso you can't really do worse than that)14:55
janimoTassadar, I know my nexus did not seem to change backlight at all until this week14:55
janimonot sure if due to 4.2.1 (which I had before), a clean reinstall, or just me being more determined to wait for the change14:56
=== chihchuno is now known as chihchun
Tassadarjust in case you posted it because you want someone to test it - it works for me too, but that touchscreen input in portrait orientation is really weird)15:07
ogra_well, onboard works just fine, thats the puzzling bit15:07
Tassadaryou can also drag icons in dash15:08
ogra_right15:08
ogra_and i can scroll firefox pages with drag'n grab enabled15:08
ogra_links get marked if i tap them but never get executed15:09
TassadarI can open terminal menu (the one on top bar) if I keep the finger on screen15:09
ogra_heh, xchat copes fine though15:11
Tassadarkernel crashed when rotating the screen :x15:11
ogra_kernel ?15:12
ogra_how do you know ?15:12
ogra_did you get an oops ?15:12
Tassadarhttp://pastebin.com/AY6Cg9R615:12
Tassadarit was after i managed to bug the touch input by clicking too fast15:12
ogra_that looks like cpufreq issue ...15:13
ogra_nothing in my dmesg, looks all fine15:15
Tassadarmaybe just some rare crash15:16
Tassadardunno which kernel does Ubuntu use as base, but 4.2 kernel is not that stable even on Android15:16
=== doko_ is now known as doko
achiangogra_: can every Ubuntu package in 12.04 and beyond assume that NEON is supported in the hardware?16:42
ogra_achiang, nope, NEON is not allowed at build time at all, if a binary wants to use NEON it has to have runtime support16:43
achiangjanimo: ^^16:43
achiangogra_: interesting....16:44
ogra_by default all packages in the archive need to be non-NEON16:44
janimoachiang, what ogra said. Even if we do not have official images, people run Ubuntu on Tegra2 which has no neon16:44
qenghoachiang: yo.16:44
janimoAnd ac100 actually had 12.04 images16:45
achiangqengho: hi, let me paste a bit of scrollback for you re: neon16:45
ogra_and as long as we potentially (or actively) support something like tegra2 or armadaxp it will have to stay that way16:45
qenghoNo neon.  Saw that.16:45
janimoogra_, oh armadaxp too, I forgot16:45
ogra_well, NEON, but only at runtime :)16:45
qenghoGot it.  Thanks!16:45
achiangqengho: is the chromium build FTBFS due to NEON?16:46
ogra_not sure which packages have actual code for this, Qt could be and possiblly pixman16:46
janimoQt does indeed16:46
ogra_(if you want to look at code for runtime support)16:46
achiangogra_: qengho is the chromium maintainer for ubuntu engineering, and he's trying to get something going for 12.0416:46
qengho13.04 mostly, but anything I can.16:47
achiangqengho: are you seeing a NEON-related build failure?16:47
janimoachiang, chromium/v8 probably works without neon too but their build system is complicated so some flags may get set inconsitently16:47
janimosome with neon support some without16:47
ogra_qengho, did you talk to micahg already ? he used to do the chromium builds in the past16:47
ogra_probably something wheer exxperience can help ....16:48
qenghoachiang: Yes.  janimo has it right.  It's mostly build flags.16:48
janimoI also had (scarring) chromium FTBFS experiences16:48
qenghoogra_: I did.16:48
janimoqengho, the only thing I remember is flags either not being passed correctly from debian/rules or build files needed patched to turn neon off16:49
janimoqengho, Launchpad may still preserve chromium changelog history and you could find neon FTBFS bugs there16:50
janimoit may be a recurring one due to dropped patches or changed upstream build system16:50
qenghoIt's changing upstream, AFAIK, janimo.16:51
qenghoOkay, thanks all.16:51
qenghoEr, it looks like I might have to write a function that does what I want. What's the best way to test for NEON support programatically at runtime?16:55
=== nemik_ is now known as nemik
* qengho hopes it's not "parse /proc/cpuinfo".16:56
qenghoOh good. Assembly.16:58
ogra_iirc it was somewhere in /proc/self17:00
ogra_ /proc/self/auxv ? not sure17:01
ogra_actually i'm pretty sure libjpeg-turbo has the code for runtime switching, just grep for auxv in the source17:03
morphisogra_: thanks for your response about gstreamer/ti/omap some days ago17:03
ogra_welcome17:04
morphis:)17:05
morphisogra_: seems like I have to investigate a little bit more17:11
ogra_well, the guys that used to maintain the TI PPA packages are still around in #pandaboard (some of them at least) they can probably tell you what happens to the PPA stuff and if anyone goes on maintaining it17:12
morphisogra_: good hint, will join there and ask17:15
=== yofel_ is now known as yofel
qenghodoes someone have a device that lacks NEON support?  I want to test something.  http://pastebin.ubuntu.com/1542106/17:40
qenghoogra_, achiang, ^17:40
TassadarI have raspberry sitting in here, if that's enough17:41
ogra_not near me today, sorry17:41
qengho$ make foo; ./foo17:41
qenghoTassadar: Ooo, that just might!17:41
ogra_i will have my ac100 (tegra2) around tomorrow again though17:41
Tassadarokay17:42
ogra_what OS is on there ?17:42
Tassadarthat debian wheezy they have on their download page, raspbian or whatever17:42
Tassadaryou only need to test if it correctly detects the CPU, right?17:42
Tassadarit should be enough17:43
* ogra_ isnt sure if a test on an armv6 debian rebuild is actually reliable17:43
qenghoTassadar: yes, just the CPU.17:43
ogra_your kernel definitely needs hwcap support though17:43
ogra_(or was that libc ... its so long ago i had to tinker with it)17:45
Tassadarcompiled okay, says "17:45
Tassadarpi@saffron ~ $ ./test17:45
Tassadarthis CPU has no NEON support.17:45
Tassadar"17:45
qenghoTassadar: Thanks.17:45
achiangqengho: i don't have a device like that near me, unfortunately17:47
TassadarMy LG P500 (~same cpu as raspberry) with debian in chroot also says "no neon support"17:51
ogra_for a positive match,, i get proper output on the chromebook17:57
ogra_ogra@chromebook:~$ ./foo17:57
ogra_this CPU has NEON support.17:57
qenghoogra_: ah, thanks.  Good data all.  Thanks.17:57
vhadilhy friend18:23
vhadilbesides cmus what applications you install18:24
=== XenGi_ is now known as XenGi
=== XenGi is now known as XenGi_
=== XenGi_ is now known as XenGi
=== XenGi is now known as XenGi_
=== XenGi_ is now known as XenGi
=== XenGi is now known as XenGi_

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!