/srv/irclogs.linaro.org/2012/01/04/#linaro-android.txt

Dr_Who… "used render script in anger" … HAHA :-)01:41
vrbhello pundiramit06:25
vrbI am trying out the kernel 3.2 patches from linaro kernel..06:26
vrbafter applying the patches from linaro-android-3.2-agreen-rebase branch, I am getting error during `make clean`06:28
vrbthe problem seems to be commit b95db3b5e974b74a3a8c2bf1fb2a1b4e847b632906:29
pundiramitvrb hi06:29
vrbthis line: +obj-$(CONFIG_BCM4329)  += bcm80211/06:30
vrbthis directory does not exist06:30
pundiramitok06:30
vrbI changed this to +obj-$(CONFIG_BCM4329)  += brcm80211/06:30
vrbnow it works06:30
pundiramitcool06:31
pundiramitthanks for pointing it out06:31
vrbnp06:31
pundiramitbhoj, I received the shipment from china today06:33
vrbbtw.. the linaro patches are very helpful for us... we only got 6 conflicts out of 500+patches06:34
bhojpundiramit, like you :) ... did the cable work ?06:34
pundiramitbhoj, mini-A usb adapter worked fine06:34
pundiramitvrb, thats good to know06:34
bhojman ... I am jealous of you now : / . who delivered it ? I mean which courier guys ...06:34
bhojshould have ordered more of those and started a dealership in India ;P06:34
pundiramitvrb, if I'm not wrong most of the conflicts are in USB and driver/char subsystem06:34
pundiramitbhoj, thats a good business plan :)06:36
pundiramitbhoj, it has got a good market here06:36
pundiramitbhoj, what about idasystems now?06:36
pundiramitbhoj, got shipment through standard Indian post-office delivery06:37
bhojI have made the payment to them but they have told they will try to arrange one within this week or refund the money . . . Oh if its indian post-office then I think ill have to wait for another 2 days to reach my place.06:38
vrbone conflict in usb_rndis.. a couple in process, scheduler and in the fiq debugger... all easily resolved06:38
vrbthanks for all the help so far.. :)06:38
pundiramitvrb, no problem06:40
pundiramitbhoj, adapter is not stable06:41
pundiramitbhoj, attached mouse worked only for about 2-3 minutes06:41
pundiramitthen it went off06:41
pundiramitno dmesg/logcat log06:41
pundiramittried twice06:41
bhojpundiramit, thanks for the heads up :) . have you plugged only mouse ?06:42
pundiramitif I reconnect mini-A cable then it works again06:42
pundiramitplugged only mouse06:42
bhojok ...06:43
kenwsheh, https://github.com/tdm/android-x-server09:32
=== epatryd is now known as patrikryd_ste
pabhishekpundiramit, hi11:54
pundiramitpabhishek, hi12:23
pabhishekpundiramit, I was checking html5 video playback on Panda...12:23
pundiramitpabhishek, ok12:24
pabhishekpundiramit, http://html5video.org/ - it doesn't play - only some text appears on opening this link. - u mentioned sometime back that u were able to play12:24
bhojpabhishek, try the latest build #134 . it has audio I could play videos from youtube.12:24
pabhishekbhoj, ok12:24
bhojvideo playback was failing due to missing audio support ..I don't if html5 stuff would pplay though12:25
pabhishekbhoj, html5 is supported on ics browser12:25
bhojokay12:25
pabhishekbhoj, if you have your board flashed - can you try this link: http://samplemedia.linaro.org/VP8/big_buck_bunny_480p_VP8_VORBIS_25fps_1900K.WebM12:26
bhojhmmm ok .12:26
bhojits showing some garbage text thats it ...12:29
bhojpabhishek, its showing up garbage text on the browser12:30
pabhishekbhoj, yes- even I get same behavior12:30
pabhishekbhoj, so its not related to audio issue12:30
pundiramitpabhishek, bhoj can't we try it on a default media player?12:31
pundiramiti remember trying to play it once on media-player from gallery on gingerbread in my previous project12:31
pundiramitbrb12:32
bhojpundiramit, yes we can try it on gallery application. letme copy it into sdcard12:48
pundiramitbhoj, does it work?13:36
tixypatrikryd: do you know why bootfiles.mk copies u-boot to $(PRODUCT_OUT)/boot rather than just having u-boot built directly into that directory?13:38
asac_i guess we are not yet back to -O3 for ICS builds, right?13:41
asac_bero|2: ?13:41
bero|2asac_: No, not yet, but I do have a local build with -O313:59
=== bero|2 is now known as bero
patrikrydtixy, There is probably no good reason. I guess it was the easiest thing to do when I did it a long time ago.14:01
tixypatrikryd: Thanks, I'm adding building of device tree blobs and also thinking about UEFI, as well as trying to work out how to have three different u-boot versions built into the same boottarball14:04
asac_bero: do you have a blueprint?14:05
asac_bero: one or many blueprints that would contribute to the goal "get to -O3" ?14:06
asac_bero: we have this roadmap card http://status.linaro.org/card/TCWG2011-GCC-O3.html ... which doesnt have any of the android blueprints that directly or indirectly contribute to it in it14:07
beroThere is one, let me check...14:07
beroasac_: https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-11.12-optimize14:08
beroThat includes -O314:08
asac_bero: we shouldnt name blueprints like that14:10
asac_blueprints should be about a tangible deliverable14:10
asac_not about random monthly chunks of work on some higher level topic14:10
asac_but ok14:10
tixyshould we be doing all new work in the 4.0.3 branch ?14:11
asac_anything else in blueprints that might have contributed to O3?14:11
asac_bero: how about blueprints about strict-aliasing patchset etc.? isnt that a subtopic of it?14:12
asac_that would be nice to highlight individually?14:12
beroasac_: I think strict-aliasing is a subtopic of optimize14:13
asac_right14:13
asac_are there no blueprints?14:13
asac_guess not14:13
asac_leave it to me/zach14:13
asac_we will see how we can beautify this effort14:13
asac_(beautify on communication side)14:14
mpoirier_danilos: good day14:18
beroAll builds are switched over to 4.0.3 now15:08
asac_o/15:08
berobhoj: What's the status of your software GL patch? That's the last bit needed for https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-remove-target-specific-branches aside from testing15:11
bhojbero, the patch is included in framework/base in 4.0.1 . Its all there. its just that the board specific branches were created before cleaning up the software GL patch.15:15
berobhoj: Ok, I'll check if it's pulled into 4.0.3 and then mark it done15:15
bhojbero, looks like its pulled http://android.git.linaro.org/gitweb?p=platform/frameworks/base.git;a=shortlog;h=refs/heads/linaro_android_4.0.315:15
bhojVishal Bhojframeworks/base: Disable H/w GL rendering based on...15:16
beroYes, looks good15:16
tgall_foowhat's the correct way to make use of the hwcaps.h include on android?    on linux it's in the compiler path at <asm/hwcap.h>  ,   I see the include in ./kernel/arch/arm/include/asm/hwcap.h, but it kinda feels wrong to try and use it from the kernel dir16:33
asac_tgall_foo: might be that you need to ship a copy or something. wonder if and how bionic uses kernel headers16:48
tgall_foolink.h is another include I'm wondering about as well ?.16:49
tgall_foothat's another glibc header but bionic oddly doesn't provide it16:50
asac_some stuff is in libc/arch-arm/include/ but i guess a very minimal selection16:50
asac_tgall_foo: which link.h is that?16:51
asac_also from kernel?16:51
tgall_foodpkg-query for me says from libc6-dev16:51
asac_tgall_foo: what do you need link.h for?16:52
asac_the link.h i see here on ubuntu seems to be something about elf16:52
asac_so maybe some runtime linking funcs?16:52
tgall_fooit contain's the macros for ElfW which you need for working with auxv16:52
asac_tgall_foo: right. but why do you need to work with auxv? whats the topic?16:53
* tgall_foo takes another way of looking ait ? let's see if those definitions are cleverly hidden somewhere else16:53
tgall_fooasac_, runtime neon detection16:53
asac_oh i see16:54
tgall_fooI'm starting to think that bionic doesn't support auxv16:55
asac_i would think the same16:55
tgall_fooyeah standards ? who needs them....16:55
asac_what i would propose would be to talk to android-... group and ask them what to do for a lib where you want to do runtime neon detection16:55
asac_you might either get a big push back/rant or maybe a great new topic for linaro :)16:56
asac_i would hope for the latter16:56
asac_for now i would think that android is embedded in the sense that you build for a specific board/soc16:56
asac_hence you use ifdef :/16:56
asac_there should be some NEON ifdef be available to your code already16:57
asac_grep -r ARCH_ARM_HAVE_NEON  external/skia/16:57
asac_grep: external/skia/.git/svn: No such file or directory16:57
asac_external/skia/Android.mk:ifeq ($(ARCH_ARM_HAVE_NEON),true)16:57
asac_external/skia/Android.mk:ifeq ($(ARCH_ARM_HAVE_NEON),true)16:57
asac_tgall_foo: ^^16:58
tgall_foodoing a little more wider grep, both binutils and the elfutils appear to be making use of it so that's a good sign16:58
asac_thtas what skia probably uses to set build time optin16:58
asac_right16:58
tgall_fooyeah that's another way to do it16:58
asac_ifeq ($(ARCH_ARM_HAVE_NEON),true) LOCAL_CFLAGS += -D__ARM_HAVE_NEON17:00
asac_endif17:00
asac_thats what they do17:00
asac_i think thats best practice then?17:01
tgall_foowell in the land of android, do as they do17:01
tgall_fooit'll work17:01
asac_yep17:02
=== asac_ is now known as asac
tgall_fooany skia-bench knowledgeable types around ?19:06
=== ramana is now known as ramana_away
tgall_foois there a way to get skia-bench to output to someplace other than the log?   the log seems to fill up rather quickly thus discarding results20:58
doanactgall_foo: no good way. You'll have to patch the source21:10
tgall_fooew21:11
tgall_foois there a way to run sshd on ICS on panda ?21:11
doanactgall_foo: I considered patching it because it made doing the LAVA test a pain to parse21:12
tgall_fooyeah the way it stands currently skia-bench seems pretty much unusable21:12
tgall_fooleast with tjbench i can get numbers21:13
tgall_fooah ha ? skia-bench is segving21:13
doanactgall_foo: FYI: here's the code I use to run/parse skia in LAVA: http://bazaar.launchpad.net/~doanac/lava-android-test/v8-skia-tests/view/head:/lava_android_test/test_definitions/skia.py21:14
tgall_fooyeah that's not going to help much without having some way to either slogin, scp or something on the pandaboard21:20
tgall_fooscp seems to be there but the network isn't up and running21:21
tgall_fooifconfig doesn't seem to have the ability to bring up the network21:21
tgall_fooI'm not sure if that's just the state of ICS on panda or if I'm doing something wrong21:21
doanactgall_foo: android uses "netcfg" instead of ifconfig21:46
doanacyou can run "netcfg eth0 dhcp" for example21:46
tgall_foohhh21:46
tgall_foogreat thanks21:47
doanacand the adb tool has push/pull commands to transfer files21:47
tgall_foodoanac, is it possible to setup sshd to be able to ssh in ?21:49
doanactgall_foo: I've never tried that. I normally use a combination of adb and serial console.21:49
tgall_foowell reason why I'm asking is there is so much gorp getting dumped to the serial cons it's just not useful .. specially if I need to fire up gdb and do some serious debug21:50
doanactgall_foo: I just noticed android's busybox includes telnetd - that might work for you21:50
tgall_fooah ? now that'd work!21:50
doanacor you can run "adb shell"21:51
tgall_fooyeah looks like i'm going to have to run via gdbserver anyway21:51

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