/srv/irclogs.linaro.org/2012/03/14/#linaro-android.txt

hgr1...and here comes my weekly bump for audio-recording on the Pandaboard! :)01:54
pfefferzhgr1, tgall_foo is listening bump away02:02
hgr1that was it! :)02:02
hgr1I want to do voip on Panda, and it is kind of hard without audio02:03
pfefferzhgr1, I know02:06
hgr1also, something that might be worth mentioning, using OpenSL-ES, it takes a good 1,5-2 seconds before I get any audio out on the speakers02:07
hgr1same test on a Samsung SII it is "instant"02:07
hgr1maybe something to look into02:07
pfefferzwould you mind fileing a few bugs02:07
pfefferzbugs.launchpad.net/linaro-android02:07
hgr1there is already one for recording02:08
pfefferzasac, won't be around for our sync up tomorrow, got anything for me?02:08
hgr1pfefferz: and for the playout-delay, I need to test a bit more to make sure it is not just me doing something wrong02:09
pfefferz:)02:09
pfefferzhey linarobot`02:14
pfefferzhey liuyq02:14
liuyqpfefferz, hi02:14
liuyqpfefferz, really around?02:15
pfefferzyup02:15
pfefferztianhong-wang, hey03:01
pfefferztianhong-wang, did you get the meeting invite03:01
dzinpfefferz: better good?03:02
dzinpfefferz: what's all that aboot?03:03
pfefferzoh just trying to collect up some people03:05
dzinpfefferz: last line of your email :-)03:06
mounirbpfefferz, https://blueprints.launchpad.net/linaro-android/+spec/test-biglittle03:07
pfefferzmounirb, https://blueprints.launchpad.net/linaro-android/+spec/auto-use-test03:20
pfefferzdzin, oh, I like my TSC layout03:20
pfefferzdzin, like https://linaro.papyrs.com/ANDROID2012-Q1-TOOLCHAIN03:21
dzinpfefferz: I do too. I wanted to get your opinion on "Blueprints" vs. "Work Packages"03:21
dzinI've been using "Work Package" to include the blueprints, plus individual acceptance criteria.03:22
pfefferzI mean we have BPs03:23
pfefferzwhich have acceptance03:23
pfefferzhey jackh03:23
dzinpfefferz: then the overall acceptance criteria section for the complete card functionality03:23
pfefferzsure03:23
jackhhi, pfefferz03:23
jackhi will test the patch samwook give03:23
pfefferzI think its whatever gives the TSC a complete picture03:23
pfefferzjackh, cool03:23
pfefferzLooking at a card the TSC should be able to tell how that cards going03:24
jackhmy mail on keymap, i broadcast it, no one replied me...03:24
pfefferzjackh, are you compiliong the entire Android platform?03:24
pfefferzthat file should be in the ANdroid platform03:25
pfefferzthen you caqn push a change to it03:25
jackhunder the device/linaro/origen03:25
pfefferzright03:25
pfefferzso when you compile the platform03:25
jackhi just modified the device.mk03:25
pfefferzit'll copy that file to the out directory03:25
jackhi know03:26
pfefferzand you'll be good to go03:26
pfefferzis that the answer you're looking for?03:26
jackhno...03:26
pfefferzk03:26
pfefferzhey pabhishek03:26
pabhishekpfefferz, hi03:26
jackhi already did these stuff, but since keymap stuff changed in ics, so it's differs03:26
jackhi am sure no one has ever tested the key.ics file03:27
dzinpfefferz: for big.LITTLE conflicting with the monthly cycles: is it sufficient to break up the blueprints to correspond to the monthly releases, or is there something else you were referring to?03:29
pfefferzbig.LITTLE is being planned with waterfall03:30
pfefferzour releases are agile with a roadmap03:30
pfefferzI've got 4 guys on the big.LITTLE schedule03:31
pfefferzand the rest of the team on the releases schedule03:31
dzinok03:31
pfefferzjackh, okay, so what is your exact question03:31
dzinI can see the conflict03:31
pfefferzdzin, and mounirb said, well we're going to have lots of these projects03:32
pfefferzso that's when my DANGER light went off03:32
dzinpfefferz: Understood. Let's chat when you get back.03:34
pfefferzdzin, sure03:34
pfefferzdzin, are you goin gto back fill Tony's PM duties for me?03:34
dzinpfefferz: where is Tony?03:35
pfefferzTonys gonna be leaving April 6th03:36
jackhpfefferz, my question is: how to write a correct .ics file for origen board03:41
jackhsorry, it's .kcm file03:41
pfefferzdo you mean how do you write a correct keymap file03:41
pfefferz?03:41
jackhpfefferz, my question is: how to write a correct .kcm file for origen board03:42
jackh.kcm file is the keymap file03:42
jackhthe exact name is gpio-keys.kcm03:43
jackhthe one now we are using is NOT correct03:43
pfefferzokay03:43
pfefferzI'm not sure03:44
pfefferzbut03:44
pfefferzfyou can always post to03:44
pfefferzandroid-platform@googlegroups.com03:44
pfefferzand CC linaro-android@lists.linaro.org03:44
jackhi posted03:44
pfefferzon android-platform?03:44
jackhi also posted03:44
jackhyes03:44
pfefferzk03:44
pfefferzno responses?03:44
jackhright03:44
pfefferzwell03:44
pfefferzthen you'll get to become the expert03:45
pfefferz;)03:45
pfefferzlook at the git log of that file03:45
pfefferzand where it may have come from originally03:45
pfefferzthen you can figure out who originally wrote it03:45
jackhahh, you are right,sire03:45
jackhsir03:45
jackhsounds like you will take some vacation?03:46
pfefferzyup03:46
pfefferzthe next 3 days03:46
jackhohh, another big walk?03:46
pfefferzbig walk?03:47
jackhforget it, i go lunch, man03:47
pfefferzcool man03:49
pfefferzhey bhoj03:52
=== Guodong-Xu is now known as guodong
bhojhi pfefferz03:53
=== guodong is now known as Guest33218
=== Guest33218 is now known as Guodong-Xu
pfefferzpabhishek, can we start 15 min late03:55
pabhishekpfefferz, ok np03:56
pfefferzpabhishek, ready to rock and roll?04:15
pabhishekpfefferz, yes04:16
pfefferzpundiramit, I've got some TODOs in the agenda for the meeting04:32
pfefferzfor big.LITTLE04:32
pfefferzwould you take a look?>04:32
pundiramitpfefferz, sure04:33
pfefferzthanks04:33
pfefferzbasically I'd like you to get some code into tests/linaro/biglittle/core04:33
pfefferzon android.git.linaro.org04:33
pundiramitpfefferz, do I have access to creat projects there?04:34
pundiramitpfefferz, if not then can you please create some place holder projects there? just empty directories?04:34
pfefferzactually I don't think so04:34
pfefferzI will04:34
pfefferzthanks for reminding me04:34
pfefferzwill create the starting branches as well04:34
pundiramitpfefferz, I need directories for stress, iozone, memtester. I will push these tests today.04:35
pfefferzpundiramit, test/linaro/biglittle/core created04:39
pundiramitpfefferz, does that mean all the tests are going to be a part of single "core" repository?04:43
pundiramitpfefferz, and not as individual projects/tests04:44
pfefferzall of your tests04:52
pfefferzshould be just individual files04:52
pfefferzthat call iozone04:52
pfefferzmemtester04:52
pfefferzwith one Android.mk04:52
pfefferzyoushould put the daemon in there too04:52
pfefferzthat'll be one part04:52
pfefferzjust about ready to send another mail about the other test locations04:53
pfefferzpundiramit, right04:55
pfefferzthose would go in external04:55
pfefferzcheck with bhoj about those04:55
pfefferzhe can create and push them in04:55
pundiramitpfefferz, ok04:55
pfefferzpundiramit, thanks05:01
=== lance_ is now known as Guest12843
kishoreHello everyone, Good morning... Can anyone help me in preparing master image for LAVA setup for linaro andrioid image.??05:48
kishoreHello pfefferz06:33
kishoreHello springz06:34
pundiramitkishore, pfefferz is not available on 14/15/1606:35
pundiramitkishore, anything we can assist you with?06:36
kishorepundiramit: Hi, Thank you for information.... Yes please...06:36
kishoreI want some information on making master boot image on 8gb sd card06:37
kishorefor linaro android06:37
pundiramitfor which board?06:37
kishorepundiramit: For origen board06:37
kishorepundiramit: I checked the link on linaro website.. but got information only on ubuntu06:38
springzkishore, refer to http://lava-dispatcher.readthedocs.org/en/latest/installation.html#creating-a-sd-mmc-card-with-master-image06:39
pundiramitkishore have you tried the instructions here https://android-build.linaro.org/builds/~linaro-android/origen-ics-gcc46-samsunglt-stable-blob-12.02-release/06:39
pundiramitand here https://wiki.linaro.org/Boards/Origen/AndroidSetup06:42
kishorespringz: I refer that site... but in this it is explained with respect to the ubuntu images , because the partion which they have mentioned is rootfs, which i got in ubuntu images, but in android I am not getting any partition called rootfs06:42
springzkishore, master image is always ubuntu images06:43
springzkishore, we don't expect an android one to do the test image deployment06:43
springzkishore, you need to ensure three or more test image partitions for your android, for ubuntu test image, 2 partitions are enough06:44
kishorespringz: Sorry if I am wrong, I just want to test the android images on LAVA set up06:44
springzkishore, so android image is test image06:45
springzkishore, you need to setup one master image by using ubuntu image, then call lava-dispatcher to do the test image deployment/test06:46
kishorespringz: I don't know... I just downloaded the system.tar, userdata.tar, boot.tar and flashed it in 8GB sd card06:47
springzkishore, you need to deploy a master image first06:48
kishorespringz: You mean, I should first make master image using ubuntu image  and pass the userdata, systemdata, and bootdata of android from jobfile??06:48
springzyes06:48
kishorespringz:But isn't it strange, first we are using ubuntu images and then in jobfile passing the android images??06:50
springzkishore, yes, boot.tgz... is defined in test image field in jobfile06:51
springzkishore, master image will deploy them to the test image partitions, and switch to test image06:51
kishorespringz: And which date ubuntu images I should use to make master images.. is there any relation between ubuntu images and android images...06:52
springzkishore, no, use a known good one07:02
springzkishore, you can use one milestone release one07:02
springzkishore, is the mmc issue resolved on your board07:02
zuhHas someone booted the recent builds of https://android-build.linaro.org/builds/~linaro-android/snowball-ics-gcc46-igloo-stable-blob/ successfully? I'm about to file a bug about this but I'd like to know if others see what I do07:51
jackhkishore, i tried the build, it's fine08:48
jabjoepfefferz: Hi, well I've come in and the build has finished without errors. So now I need to stick in vendor tar and I'm ready to go. I think. I'm still not 100% clear what to do with it. Can you give me a step by step designed for an idiot? I'm clearly missing something some where, so safest to assume everything until the light goes on! ;-)09:02
jabjoepfefferz: Oh! I see, ignore me. Don't know Android build system, but I can see that .mk file does the business.09:11
kishorejackh: Hi, I also tried it today with my 9th sd card... Somehow its booted today.. I dont know the reason..09:19
kishore:-)09:19
kishorespringz: Yes mmc issue is resolved... I dont know the reson, Today I got a new sd card from my manger, It works...09:20
kishorespringz: Now I have issue with the LAVA test with linaro_android09:21
kishorespringz: Now Its failing with some error like09:21
kishorespringz: tar: boot/cmdline: Cannot change ownership to uid 0, gid 0: Operation not permitted09:21
kishoretar: boot/uInitrd: Cannot change ownership to uid 0, gid 0: Operation not permitted09:21
kishore[   56.435000] samsung-i2s samsung-i2s.0: i2s_set_sysclk:401 Other DAI busy09:21
kishoretar: boot/uImage: Cannot change ownership to uid 0, gid 0: Operation not permitted09:21
kishoretar: boot/u-boot.bin: Cannot change ownership to uid 0, gid 0: Operation not permitted09:21
kishoretar: boot/u-boot-mmc-spl.bin: Cannot change ownership to uid 0, gid 0: Operation not permitted09:21
kishoretar: boot: Cannot change ownership to uid 0, gid 0: Operation not permitted09:22
kishoretar: Exiting with failure status due to previous errors09:22
springzit doesn't matter09:22
springzownership can't work on vfat09:22
kishorespringz: You can see the complete log at http://paste.ubuntu.com/882986/09:22
kishorespringz: But deployment is getting failed ...09:23
springzkishore, it's not caused by ownership issue09:24
kishorespringz: This is the json link..... http://paste.ubuntu.com/882988/09:24
kishorespringz: Then wat is problem??09:24
springztar extraction09:26
springzyou can check the md5sum09:27
kishorespringz: But why tar extration operation is not permitted??09:27
springzkishore, yes, it's a warning09:27
springzon boot partition, it's vfat format, no owner ship09:27
kishorespringz: The steps which i am folowing is correct.. right? have seen my jobfile??09:29
springzkishore, yes09:30
kishorespringz: http://paste.ubuntu.com/882988/   ... this is the link09:30
kishorespringz: Is is correct??09:30
springzright09:30
springzkishore,  you can try manually, on the board, get boot.tar.bz209:31
kishoreSo what you will suggext me to do next.... I am not getting any idea??09:31
springzmount /dev/disk/by-label/testboot /mnt/lava/boot09:31
springztar --numeric-owner -xjf boot.tar.bz2 -C /mnt/lava09:31
springzthere is also issue09:31
springzI mean ownership issue09:32
springzyou can try run the job again.09:32
kishorespringz: Ok I will check this once...09:32
kishorespringz: I tried ... I ran it two times.. but both the times it got failed on deployment itself09:32
springztry to extract the boot.tar.bz2 on your machine09:33
springzsee if there is an error09:33
kishorespringz:No on my machine I is extracting fine09:35
jackhkishore, so the reason is the sd card...09:36
kishorejackh: Yes, This was my 9th SD card......09:37
kishorejackh:  My bad luck....09:37
springzkishore, you can check the md5sum09:39
springzkishore, download the boot.tar.bz2 like lava-dispatcher09:39
springzwget http://192.168.0.10/images/???/boot.tar.bz209:39
kishorespringz: md5sum ??09:40
springzand check md5, see if it is same with your machine09:40
springzmd5sum boot.tar.bz209:40
kishorespringz: I gave the above command on the target through monmux...09:40
kishoreSome binary values stated displying on the screen , so i gave ctrl +C09:41
springzI don't understand, you can input the cmd manually09:41
kishorespringz: From conmux, I use the command on target directly..09:42
springzright09:42
springzkishore, use md5sum to check if the file is the same as original09:43
kishorespringz: Ok I downloaded.... last time some error was there...09:43
kishorespringz: How can i use md5sum to check the file is orignal or not??09:44
kishoreI never used md5sum command so have no idea on this....09:45
springzuse 'md5sum filename' on both side09:46
kishorespringz: Ok... from x86 machine output came...waiting the result from target09:47
kishorespringz: Yes. both the output is same....09:50
springz<springz> mount /dev/disk/by-label/testboot /mnt/lava/boot09:51
springz<springz> tar --numeric-owner -xjf boot.tar.bz2 -C /mnt/lava09:51
springzI think you can see same error of ownership09:52
kishorespringz: Getting the same error , operation not permitted...09:53
kishorespringz: This is the output09:53
kishoreroot@master:~#  [rc=32]: tar --numeric-owner -xjf boot.tar.bz2 -C /mnt/lava09:53
kishoretar: boot/cmdline: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: boot/uInitrd: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: boot/uImage: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: boot/u-boot.bin: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: boot/u-boot-mmc-spl.bin: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: boot: Cannot change ownership to uid 0, gid 0: Operation not permitted09:53
kishoretar: Exiting with failure status due to previous errors09:53
springzoh!09:53
springzwhat's the return value of tar?09:54
kishorespringz: 209:54
springzit's strange, I saw it was a warning not error for the operation09:57
kishorespringz: But we are getting the same message in the output log also, right.. and after that it was terminating....09:59
springzkishore, I don't have such line: tar: Exiting with failure status due to previous errors10:01
kishorespringz: Yes it is there....in the log file....10:02
kishorespringz: please see this link: http://paste.ubuntu.com/883034/   ... Line :91310:04
springzkishore, right, I see, it's true, but I'm successful to do it before10:06
kishorespringz: What s the permission we should for boot.tar.gz2 in our systems10:08
springzkishore, it's no matter with its permission, but the internal files permission10:08
kishoreShould I try with some other images...??10:09
springzkishore, after it extracted, will apply 0:0 but no ownership with vfat10:09
springzno10:09
springzhttps://bugs.launchpad.net/lava-dispatcher/+bug/95487010:09
ubot5Ubuntu bug 954870 in LAVA Dispatcher "android deployment failed due to a untar ownership issue" [Undecided,New]10:09
springzfiled a bug, will fix it10:09
kishoreSpringz: The link you gave , I am getting the same wrror...10:11
springzkishore, ?10:11
springzkishore, I just filed the bug10:11
springzkishore, I will go off work now, you can send mail to me if something more wrong10:12
kishorespringz:Oh.... :-) ..   "after it extracted, will apply 0:0 but no ownership with vfat"  .. I didn't get this point...10:12
kishorespringz: Can I have your email id please... ??/10:12
kishorespringz: Ok i got it from link10:13
springzkishore, --numeric-owner will keep the file ownership in tar, but vfat don't support it10:13
kishorespringz: Thank you...10:13
bhojdmart, Hi !\10:17
jabjoepfefferz: I can't get the SnowBall to show any sign of life with my build or the downloaded build. I've got /dev/ttyUSB0 but nothing seams to be happening over it. I'm clearly missing something some where again.11:04
zuhjabjoe: Have you flashed the board or using an SD card? Nothing on the console sounds like the board's eMMC is either empty or contains wrong binaries11:08
jabjoeI'm trying to use a SD card.11:08
jabjoeI've not read this, but I expected it to be a bit like the SheevaPlug. That Uboot outputs to the TTY. But nothing, not even Uboot. Sounds like the board isn't flashed to what I need then. Is this correct?11:09
jabjoeIf so, can you point me to something about flashing the board?11:10
zuhYes, Snowball _always_ boots from the eMMC. You need boot binaries and u-boot to be flashed on the board.11:10
jabjoeHow do this work? Via the usb/tty I take it?11:11
zuhNo, there's a dedicated tool for that. The best way to do that is getting the latest Ubuntu release from http://www.igloocommunity.org/download/ubuntu/images/2012.02/ and then flashing it according to http://www.igloocommunity.org/support/Flashing_howto11:12
zuhThe Ubuntu image regardless of which you target; the uboot in there will be good for both Android and Ubuntu11:14
jabjoeI briefly saw that. Just wasn't clear on how it physically is connecting to the board. It must be over the tty/usb serial connection. I'm not connected to the board any other way. Anyway, I'll give it another go, unless you come back and tell me I need to connect differently to the board, or I read it of course. Cheers for the pointers. :-)11:15
zuhThe flashing works over the OTG port and requires power cycling, so if you are not in the same room with the board it will be a bit tricky ;)11:16
zuhAnd actually I was lying, you could get the -2 version of http://www.igloocommunity.org/download/android/images/20120201/ too.11:17
jabjoeIt's right here on my desk, so that shouldn't be a problem.11:17
jabjoeCheers, I'll try that image instead.11:18
jabjoev7 and v5?11:20
zuhjabjoe: oh wait, don't. That's the GB image X)11:20
zuhhttp://www.igloocommunity.org/download/android/ics/images/2012.02/ <- this one11:20
jabjoecheers, downloading now.11:20
zuhsorry for confusing :)11:20
jabjoeNot at all. Cheers for the help! I was getting lost.11:21
jabjoenice! http://www.youtube.com/watch?v=gRa8NjI5umc11:25
jabjoeWell that was easy. It lives!11:34
EsperssonHi, I have a question about the daily Android builds, how can you find out what is the diff between one build and the previous on?11:42
Esperssone.g. https://android-build.linaro.org/builds/~linaro-android/snowball-ics-gcc46-igloo-stable-blob/#build=200 and 199?11:43
dmartbhoj: hi there11:51
bhojdmart, Hi .As per your howto  Can we change to "ASYNC ?= FALSE" in big-little/Makefile in your tree  since the android builds with ASYNC=true won't boot on a15x4-a7x4 ?11:57
bhojdmart, or are we getting new 2.2 version soon  that has the fix ?11:58
bhojdmart, I don't want to keep a mirror of your code base just for this particular change .11:59
manssonpatrikryd, can you check if you have read access to the  TSC cards on todays agenda? I can't see most of them.12:59
patrikrydmansson, I can read the one about browser... Will check the others..13:00
patrikrydmansson, I can load all the pages.13:01
manssonpatrikryd, and you can see them listed at https://linaro.papyrs.com/pages/alphabetical/ ?13:03
patrikrydmansson, No13:03
patrikrydmansson, "This page is not available to guests. Log in to view this page."13:03
manssonWhich page gave you that?13:04
patrikrydmansson, https://linaro.papyrs.com/pages/alphabetical/13:09
manssonI hate Papyrs.13:12
manssonBut yo can log in can you?13:13
jackhwhat's papyrs?13:13
manssonThats our requirements gathering web site. It controls what eventually becomes Blueprints for us to work on.13:14
jackhwho manages the input for that?13:15
manssonjackh: Long story. Here: https://wiki.linaro.org/Process/Roadmap What's called "cards" is what we are supposed to look at today in papyrs. But as so many times before the permissions to those pages are totally impossible to get right.13:17
jackhohh...agree...make me crazy13:18
patrikrydmansson, Should have a separate user and password of is it somehow connected to linaro or launchpad?13:20
patrikrydmansson, just checked if I could get a password -> no user registered for my email address.13:21
bhojmansson, I too have problem with login same as patrikryd  :) . I don't understand how I get logged in automatically at times .13:22
jackhbhoj, how to get an ES board?13:23
manssonpatrikryd, bhoj, unfortunately I think they are not connected to launchpad.13:24
bhojmansson, I  don't understand what username password needs to be given .  I have previously edited this page https://linaro.papyrs.com/ANDROID2012-q1-FAST-MODEL-PLATFORM . but I don't know how I got logged into it .13:25
bhojjackh,  you can order it from http://www.digikey.com/product-highlights/us/en/texas-instruments-pandaboard/686#tabs-213:25
jackhorder??money??killing me...13:26
jackhAnnamalai123, have you had a chance to check my mail on key.kcm?13:27
bhojjackh, I think if you could talk to pfefferz he could find some  spare panda ES and get it shipped to you .13:27
jackhbhoj, will check with zach, thanks13:27
jackhbhoj, i just confused by how to reproduce a lava issue?13:56
bhojjackh,  are you talking about the bug in origen ?13:57
jackhy13:57
bhojjackh,  the patch that tushar is referring is the actual fix and  builds above #214 are working fine . I think you can just close the bug .13:57
bhojjackh,  in general for a lava issue you could just do a normal boot ... that is what is lava does in an automated form.13:58
jackhok, but how to do a reproduce of that?13:58
jackhbut for normal boot, kernel will not crash...13:59
bhojjackh,  if you pick the build number mentioned in the bug   it should crash.14:00
jackhbhoj, since tushar want me to test that patch, you mean he did it himself already?14:00
bhojjackh, if you see the builds  214 and above the tests have been run succesfully on LAVA . I think tushar just wants a confirmation that the bug is fixed. I think you could close it with a note that images are booting in LAVA and tests are being executed and results are visible in the android-build page.14:03
=== pundiramit is now known as pundiramit_away
=== zyga is now known as zyga-afk
springzkishore, hi15:27
mpoirierbero: good day15:54
berompoirier: hi15:54
mpoirierbero: just the guy I need to talk to...15:55
mpoirierbero: help.15:55
mpoirierbero: for a little while graphics on snowball is broken15:55
mpoirierbero: let me get you a pastebin...15:55
zuhbero: http://pastie.org/3584115 <- this way15:55
* zuh is one step ahead of mpoirier, again ;)15:56
* mpoirier can't compete with the best.15:56
mpoirierbero: with special attention to 1497 cannot locate '__aeabi_llsr'15:56
beroThat looks like libUMP.so (probably a proprietary binary?) doesn't like the latest libgcc... We can fix that by providing __aeabi_llsr elsewhere15:57
beroAny chance we can recompile libUMP.so?15:57
mpoirierbero: not a chance.15:57
mpoirierbero: but it can be recompile by STE.15:57
mpoirierbero: something I'd rather avoid.15:57
beroWell, the best thing to do would be to get a recompiled version that uses our compiler and -O3, to make sure it doesn't require __aeabi_llsr to be exported...15:58
mpoirierbero: first, is it related to your latest checkin in libbcc ?15:58
beroThe other thing we could do is providing __aeabi_llsr in a separate library and make sure that gets linked in along with libUMP15:58
mpoirierbero: libbcc: Don't implement __ashldi3 and __lshrdi3 with current gcc versions15:59
beroRelated yet, but not caused...15:59
beroRelated yes even15:59
beroThe thing is...15:59
beroSome stuff relies on gcc implementation details that can change a lot15:59
mpoirierbero: is that checkin mandatory ?16:00
mpoirierbero: what happens if we don't have it ?16:00
beroI'd think so -- it fixes compilation at -O3 and it fixes the crash when enabling live wallpapers16:01
jabjoeIs this why I'm getting no graphics?16:01
beroAnd of course reverting it wouldn't affect __aeabi_llsr in any way16:01
beroLike I said it's related (both issues have the same cause) but not causally related16:01
mpoirierbero: sounds like we need it.16:01
beroLet me dig up a proper implementation of __aeabi_llsr that we can link in...16:01
mpoirierbero: that would be great. It will be hard justifying the recompilation of those libs.16:02
beroBtw, AFAIK __aeabi_llsr calls can get generated only if the code requests not to make use of an FPU16:04
beroSo chances are if they recompiled it the right way (with -mfpu=neon), that could improve the speed quite a bit (by doing FPU work on the FPU instead of emulating it on the CPU)16:05
mpoirierbero: do we need hard float turn on in the rest of the system for that to work ?16:06
beroHardware float yes, but not hardfloat ABI (we can't support the latter anyway because of binary modules)16:07
berohttp://www.codebrowse.net/source/xref/linux-3.0/arch/arm/lib/lshrdi3.S <--- One implementation of __aeabi_llsr16:07
zuhbero: One thing is unclear to me, this used to work up to a point and the binaries haven't been changed. The only changes in the time frame between the broken builds is your checkin and mpoirier's framebuffer change in the kernel. Yet, with an older tc this was still reproducible.16:11
zuhWhat could have triggered it if not your change (although it indeed doesn't seem like it would directly cause it)16:11
beroAre you sure the libbcc change is the only change in the time frame? That sounds extremely unlikely because it doesn't touch __aeabi_llsr at all (it does touch __ashldi3, which is rather similar, so if something aliases __aeabi_llsr to __ashldi3, that could make it possible)16:12
beroI think the more likely change to trigger this was the change to enable -O3 globally16:13
beroThat happened around the same time as the __ashldi3 change16:13
jayabharathFolks: I had a question on Linaro-andriod builds on PandaBoard ES. Does the DVI port (not HDMI port) work in the linaro android LEBs ?16:13
bhojjayabharath,  there is a patch to get DVI working on linaro-android builds : have a look at this bug https://bugs.launchpad.net/linaro-android/+bug/89300016:14
ubot5Ubuntu bug 893000 in Linaro Android "DVI out port is not working with staging-panda or tracking-panda" [High,New]16:14
mpoirierbero: point me to the -03 commit pls.16:14
* jayabharath looks16:15
jabjoeI've yet to get graphics for Android on the SnowBall either from images I've downloaded or ones I've built. Anyone got any idea what I'm doing wrong? I've also tried Ubuntu and got that up and running without problem, so it's not the hardware setup. It's something about the Android setup.16:16
zuhbero: now that you cast a doubt, I realize my search methodology has a flaw16:16
jayabharathbhoj: thanks.. It helps.16:17
bhojjayabharath, np .16:17
zuhbero: and I guess build/ commit 258794bf688309357254f3326225dcfd614fae16 will be a likely culprit based on your comments16:17
berompoirier: http://review.android.git.linaro.org/#change,156116:18
berozuh: Yes, that looks right16:18
mpoirierzuh: you have code syn'ed on your machine right ?16:19
zuhmpoirier: yeah16:21
mpoirierzuh: do you have time to start a c ompilation ?  otherwise I'll do it but I'll have to sync first.16:21
zuhSure, but what do you want me to test? Reverting the -O3?16:22
mpoirierzuh: ya, to make sure that's the culprit.16:22
zuhok, I'll kick a build going16:23
mpoirierzuh: cool16:23
zuhhmm16:23
* bero will check out the source to see where we can inject an __aeabi_llsr implementation16:23
zuhDo we know what we could rebuild or do I need to do a clean one?16:23
beroSince everything is built with -O3, it's best to do a clean rebuild16:24
zuhYeah, ok.16:24
=== zyga-afk is now known as zyga
jabjoepfefferz: Any idea why I'm not getting graphics on any image I try?16:40
jabjoepfefferz: Least not for Android. I got that missing something feeling again.16:40
zuhjabjoe: DVI-D or HDMI? https://bugs.launchpad.net/igloocommunity/+bug/89865016:43
ubot5Ubuntu bug 898650 in Linaro Android "Snowball needs HDMIservice" [Undecided,Confirmed]16:43
jabjoezuh: ooo that looks like it might be it.16:44
ramanabero, can you reduce that testcase with the problem with __aeabi_llsr . what's going wrong there ? Why is it statically linking with libgcc.a shouldn't it be dynamically linked with libgcc_s.so ?16:52
ramanabero, It sounds like a toolchain error from what you are doing - but how are you linking it - why isn't this definition coming from libgcc_s.so ?16:53
beroramana: Android thing -- for some reason Google is paranoid about not shipping a libgcc_s.so and forcing everything to link statically with libgcc.a17:14
ramanabero, right - and how are you configuring your toolchain ?17:15
ramanabero, and what is the value of --target ?17:15
beroramana: And since the problem occurs in a binary-only blob we're getting from STE, we don't really know what they do to generate it, but it looks like they force software fp emulation17:15
ramanabero, __aeabi_llsr is *not* for FP emulation17:15
ramanait is for right shift of long long - 64 bit quantities.17:16
ramana__aeabi_llsr is something that should be provided by libgcc.a and it should get statically linked in somewhere.17:16
ramanaIt looks like something else is broken here.17:16
ramanaSee pm for other issues -17:16
ramanaso if the toolchain isn't providing it in libgcc.a then something is probably broken in the toolchain but I'm surprised that it's showing up this month with 4.6.17:17
ramanaWhat is the value of --target when you configure your toolchain ?17:17
beroToolchain is configured with --target=arm-linux-androideabi --host=i686-pc-linux-gnu --build=i686-pc-linux-gnu --enable-languages=c,c++ --disable-libssp --disable-nls --disable-libmudflap --disable-libstdc__-v3 --disable-sjlj-exceptions --disable-shared --disable-libquadmath --with-fpu=vfp --with-arch=armv5te17:18
beroSame thing AOSP does17:18
ramanabero, ah that explains it.17:18
bero(If it weren't for compatibility with them, we should probably go for --with-arch=armv6 at least)17:18
ramanabero, that's what it probably is.17:18
ramanabero, can you mail me the libgcc.a that you get now ?17:19
ramanabero, did you see my other note ?17:19
beroyes to both17:20
tixybhoj: Are you working on Android for RTSM in general, or is work work specifically geared up for bigLITTLE ? Or put another way, if someone wanted Android on an A15 model, would your work be useful?17:36
joeyOh look, just found a video of zach. :-)  http://www.bgr.com/2012/03/14/99-diy-kit-overclocks-your-brain-video/18:33
zuhmpoirier: bero: FWIW, reverting the commit in question indeed produces a working image19:24
mpoirierzuh: thanks for the update.19:24
mpoirierzuh: please update the bug while bero decides what to do.19:25
zuhmpoirier: already did19:26
mpoirierzuh: cool.19:26
mpoirierzuh: thanks for your time on this.19:26
zuhThanks to bero for identifying the problem!19:27

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