/srv/irclogs.ubuntu.com/2010/08/25/#ubuntu-arm.txt

=== bjf is now known as bjf[afk]
DanaGWeird... trying to use a pl2303 usb-serial ON the beagleboard... and it's giving me garbage.00:53
DanaGpl2303_process_read_urb - tty_flag = 000:53
DanaGH�H�H00:54
DanaGsay, what's the practical difference between 2.6.35-17 (ubuntu) and 2.6.35-1002 (linaro)04:23
rsalvetinot much for now, I'd say04:30
rsalvetibut need to check the kernel tree04:30
DanaGhmm, stupid pl2303 gives garbage on my beagleboard.04:42
DanaGSay, if I'm using an aufs overlay, is there any advantage to having the underlying FS be ext4 instead of ext3?05:59
Baybalyes06:11
Baybalit would be faster06:11
persiaDanaG, Depends on what you're accessing.  If you're accessing stuff on the base FS, you'll get the normal advantages of ext3 vs. ext4.  If you're accessing stuff on the overlay, you won't see any difference between the two.  For most uses of union filesystems, actual usage is a mix of both.06:29
DanaGah, I don't even remember now what the advantages of the base were. =/06:31
persiahttp://en.wikipedia.org/wiki/Ext4 has fairly extensive discussion06:32
DanaGah, checksum is a big one.06:34
persiaNote that it *only* provides benefits to files unmodified from the base: anything in the overlay will naturally be handled based on the backing-store mechanism used for the overlay (often RAM, but potentially other things)06:38
DanaGI've also run into the blockage of hardlinks... http://patchwork.ozlabs.org/patch/52392/06:41
persiaWhat is this blocking?  I've played with all sorts of layed filesystems, and never ended up finding this painful.06:42
DanaGIt was blocking the "SingletonLock" Chromium links to wherever it links.06:46
persiaWith packaged chromium?06:48
DanaGYeah.  Though, there was a new update today... maybe that would've helped.06:48
persiaMy recommendation would be to try to catch fta: there may be a way to get it to work more easily.06:49
DanaGAnyway, it only happened when the lock was on the underlying fs.06:50
DanaGSay, is there any easy way to make ttyS2 start only if console=ttyS2 is present?06:50
persiaYou could write a custom upstart job...06:50
DanaGI need to use that port for other things in 'normal' (read-only) mode -- using the "user" button to boot read-write for servicing.06:50
DanaGyeah, though I have no idea of the syntax -- start-on?06:51
* persia doesn't know modern upstart syntax, not having worked on upstart jobs since jaunty06:51
persiaMy recommendation would be to review already installed jobs for clues, and ask in #upstart once you're sure you know the basics.06:52
DanaGhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/58638606:52
ubot2Launchpad bug 586386 in linux (Ubuntu Maverick) (and 1 other project) "omap3 kernel should hand over all comdline args to the init environment (affects: 1) (heat: 40)" [Medium,Fix released]06:52
DanaGPerfect, it's already been done.06:54
=== kmargar is now known as markos_
=== hrw|gone is now known as hrw
hrwhi09:17
=== amitk is now known as amitk-afk
cooloneylag: morning, man10:52
lagcooloney: They Bryan10:52
cooloneyi just pinged sebjan who is still working on the 2.6.35 patches.10:52
lagHey~*10:52
lagNo problem10:53
cooloneyit will be available soon.10:53
cooloneyhe helped a lot to fix coding style and duplicated patches. i think10:53
lagI'd rather them late and in good order than quickly and in a state10:53
lagI don't mind that at all10:53
lagKeep up the good work sebjan10:53
cooloneyafter sebjan boots the new kernel on panda, he will send it out.10:53
sebjanthanks lag :)10:56
lag:)10:57
=== amitk-afk is now known as amitk
ograndec, a ti-team ML for the package maintainer field sounds fine to me12:41
ndecogra: ok. so should I use tiomap-dev team on LP and create a ML with same name?12:42
ograyeah, i think LP offers that12:42
ndecogra: ok. I will do that.12:43
ndecogra: Mailing list requested and queued for approval....12:44
ograperfect !12:44
ograndec, was there ever a bug filed for your AAC issues ?13:42
ndecogra: no. in fact we are looking at using ffmpeg instead of faac by default. so i don't have any information to open a bug for now.13:45
ograok13:45
ndecogra: but do you confirm that packages cannot be compiled with neon optimizations by default?13:46
ndecogra: do you really support v7 h/w that does not have neon?13:46
ograif you do neon on a common package you should roll an additional binary with -neon in the name, yes13:46
ograndec, we support v6 HW that pretends to be v7 and has no NEON, yes :)13:47
ograndec, ask NCommander :)13:47
ograhe maintains the dove stuff13:48
ndecogra: so how do we automatically install these -neon packages in case the hw supports it?13:48
ograhmm, thats hard13:48
ograi think asac had a different idea of using runtime switches13:49
ograso the binary is the same but uses a different codepath if NEON is available13:49
persiaCan we reliably detect the ability to run neon on a given machine?  If so, hwcap might work.13:50
ograyes, i think asac's idea involved hwcap13:50
ograbut he can probably elaborate13:50
persiaI'd hope it involved something similar to what we did for pango vfp for jaunty, rather than actually patching every source to have multiple codepaths.13:51
NCommanderndec: there was going to be some work i was going to suggest in Natty to work towards solving this issue13:52
ograNCommander, the missing NEON on dove ?13:53
NCommanderogra: no, having a mechanism in APT tohandle packages on a subarchitecture basis :-)13:53
ograthat wont happen i think13:53
ograand it would have to be on dpkg level i think13:54
ograiirc we discussed it before13:54
persiaAnd it doesn't help the neon bit anyway: we don't want that: we want opportunistic use of NEON, if available.13:58
ogra\o/14:08
* ogra dances 14:08
ogratelepathy-glib builds !14:08
ogra(when using -O0 :/ )14:08
persiaUm, how did you test that?  Everyone is reporting that it builds locally, and only fails on buildds.14:09
ograeveryone ?14:09
* persia has yet to actually replicate the test failures.14:09
persiaeveryone who told me about it.14:09
ograwho was that ?14:09
ograit doesnt build with -O2 here or on the porter box14:09
ograone test fails with "invalid utf-8" the other segfaults14:10
persiaHrm.  Right then.14:10
* persia suspects irregular hardware and incorrect environments14:11
pcacjrmorning14:19
* ogra files bug 623979 and uploads a telepathy-glib workaround14:30
ubot2Launchpad bug 623979 in telepathy-glib (Ubuntu) "telepathy-glib fails to build on armel due to two unsuccessfull selftests (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/62397914:30
persia\o/14:33
ogranow that were 2 very productive hours to fix that :)14:35
ografinally we can start to prepare for beta14:36
devilhornsmmm betas ... tasty :)14:37
devilhornsogra, any news/updates wrt doing the arm work ?14:48
=== zyga is now known as zyga-us-visa-stu
ndec1ogra: can we setup the PPA to send notification email to the team list for each upload and build success?15:28
persiaI think one only gets notification on build failures.15:28
persiaOr maybe I'm confused.15:29
ndec1persia: hi! I did receive an email of build failure for something that was uploaded by someone else. so you might be right15:30
ndec1persia: so there is no option to subscribe the team ML to get more notifications?15:30
* persia is asking15:30
persia(for reference, #launchpad is a good place to ask questions about LP)15:31
persiandec1, At least one other user reports they couldn't find any way to get richer notifications.  If you want it, please file a bug against LP (and subscribe me, please).15:35
ndec1persia: thx. will do.15:35
ndec1persia: Bug #624038 entered. you and ogra are subscribed15:43
ubot2Launchpad bug 624038 in launchpad "Team notifications of package upload and build success for PPA (affects: 1) (heat: 10)" [Undecided,New] https://launchpad.net/bugs/62403815:43
ndec1persia: cool.. didn't know about the ubot notification ;-)15:44
persiandec1, Thanks.  It may take a while to get implemented, but we now have a place to discuss it :)15:44
ograyeah15:45
ograthanks for that15:45
suihkulokkilool: there is a rebased qemu branch in gitorious (omap3-rebase) with your and matts changes included15:46
loolsuihkulokki: Thanks15:47
rsalvetiogra: ndec1: could any of you add me at tiomap-dev?15:47
rsalvetiI'm also interested on this, once we get the notifications15:48
loolsuihkulokki: I'm slowly moving this tree under a Linaro umbrella and we have asked one new Linaro starter to look at upstreaming these OMAP3 patches, amongst other things15:48
ndec1rsalveti: done15:49
rsalvetindec1: cool, thanks :-)15:50
=== bjf[afk] is now known as bjf
ograrsalveti, did you research the "Bad Magic Number" issues on omap4 more ?16:05
prpplagueogra: it is an issue with the old u-boot16:05
ograseems that really bites us badly now i cant even fatls on the panda atm16:05
prpplagueogra: updated u-boot from sakoman works with no issues16:06
prpplagueogra: i created a small script to reload the boot partition16:06
rsalvetiogra: nops, not yet16:06
ograprpplague, is there an updated 1.1.4 tree anywhere ?16:06
prpplagueogra: http://paste.ubuntu.com/483487/   for reloading16:07
ograwe're way paste feature freeze and i'm not really eager to change to a different u-boot version16:07
ograprpplague, thats not an option in release images16:07
prpplagueogra: no, there isn't a fixed 1.1.4 since the fat driver was re-written16:07
ograsigh16:07
rsalvetiyep, our version is *very old*16:08
rsalvetiat least the common u-boot part16:08
prpplagueogra: the issue only seems to occur when writing to the boot partition frequently16:08
rsalvetithat's why no one likes fixing that16:08
ograwell, it was the one version that i was told to use16:08
rsalvetiogra: probably because at the time sakoman's tree wasn't ready16:09
ograprpplague, it happens on first reboot for me, after install (which writes the new initrd once to the fat)16:09
ograrsalveti, right16:09
prpplagueogra: ahh interesting16:09
ograbeta freeze is tomorrow and we have a ton of other issues to fix16:09
ogragdm isnt starting either :(16:09
rsalveti:-(16:10
ograGrueMaster, ^^^ do we have a bug for that ?16:10
GrueMasterLooking.16:10
ograi can run startx and get to a desktop session16:10
ograwell, i could before i tried to reboot16:10
ograand do we have a bug for the bad magic number ?16:11
* ogra digs16:11
GrueMasterI'm missing something here.  What am I looking for?16:11
robclarkogra: fwiw.. I think the issue updating the fat boot partition with new uImage is actually an issue with linux fat support...  if I write the uImage to an MMC card from macosx I can do it 100's of times and I haven't seen bad magic #16:12
ograGrueMaster, for a bug about gdm not starting on the panda16:12
robclarkso I think the best option is to dd an entire raw boot partition, rather than relying on fatfs16:12
GrueMasterI don't think so, but I might have.16:12
ograrobclark, why dont we have any issues on omap3 then ?16:12
robclarkI had same issue on omap316:13
robclarkat least on 3430.. I never tried 36xx16:13
GrueMasterNo, thats what I was working on last week.  Haven't finished narrowing it down yet.16:13
ograGrueMaster, we need bugs for the freeze execptions16:13
GrueMasterUnderstood.16:14
ograplease file something generic for each issue you find first, we can do detailed research later16:14
prpplagueogra: that is a good question, how are you mounting the boot partition? as vfat or just msdos?16:14
ograprpplague, vfat16:14
GrueMasterOk16:14
ograprpplague, well, we dont specify a fs actually we just mount it16:14
rsalvetiogra: you said we don't have this issue for blaze16:15
rsalvetithat's more interesting16:15
prpplagueogra: ahh, can you check to see what it is actually reporting after mounting?16:15
ogra                echo "Using u-boot partition: ${UBOOT_PART}" >&216:15
ogra                TMPMOUNT=$(mktemp -d)16:15
ogra                mount $UBOOT_PART $TMPMOUNT16:15
ograthats the code that mounts it16:15
ograwe rely on the kernel here to determine the fs16:16
ograi would, if i could boot :P16:16
ogra/dev/mmcblk0p1 on /mnt type vfat (rw)16:17
ograthats what i get on x8616:18
GrueMasterogra: Bug #624059.  I marked it as critical - incomplete to indicate that it is high priority but cause unknown.16:18
ograthanks16:19
=== zyga-us-visa-stu is now known as zyga
ubot2Launchpad bug 624059 in ubuntu "gdm fails to start on armel images updated from Alpha 3. (affects: 1) (heat: 8)" [Critical,Incomplete] https://launchpad.net/bugs/62405916:19
ograprpplague, same on panda ... "/dev/mmcblk0p1 on /mnt type vfat (rw)"16:21
ogramkdosfs -n "SERVICEV001" -F 32 /dev/mmcblk0p1 is the way we use to create the filesystem btw16:22
rsalvetiyep, creating it as a simple fat 32 partition is enough to get the problem16:23
prpplagueogra: can you email me a list of items that are of concern for the panda support?16:28
ograprpplague, you mean for u-boot ?16:28
prpplagueogra: any issues that you have for 10.10 release16:29
ograthe trashed filesystem is the only one atm16:29
ograthe other ones are rather homemade16:29
prpplagueahh ok16:29
ograi.e. busybox reboot doesnt work and things like that ... all stuff i need to fix in the packages16:29
GrueMasterI have discovered on this issue that I can reproduce it by just moving/renaming existing files and copying new ones over.16:30
prpplaguerobclark and i have seem to nailed down the issues with the hdmi/dvi stuff so we should be getting those merged soon16:30
ograGrueMaster, does the system dbus run if you try to start gdm ?16:30
prpplaguerobclark has his stuff posted on gitorious16:30
GrueMasterBut if I copy them to a different area, reformat the partition, and recopy them back with the new files first followed by the old ones it works.16:30
ograyeah, right thats another issue but we're waiting for the latest kernel16:30
prpplagueGrueMaster: yea that is what my script does16:30
GrueMasterIt appears to be an issue with seeking past a certain pont in the filesystem.16:31
ograwe cant reformat the whole thing on every initramfs update16:31
* ogra will not release with such hacks16:31
ograprpplague, how far is the new u-boot implementation feature wise ? shoudl it be on par with the 1.1.4 version ?16:32
ogra(we need hush shell and scripting support for our images)16:32
prpplagueogra: feature wise, it is 10000000 times ahead of 1.1.416:33
ograif it works as good as 1.1.4 we should probably update16:33
ograbut that would have to happen today16:33
prpplagueogra: you'd need to get with sakoman to double check the complete status16:34
ograhis wikipage didnt look like it would be comeplete16:34
ograhttp://omappedia.org/wiki/U-boot_Upstreaming_Project16:34
rsalvetiogra: as I described at the bug report, it works fine :-)16:34
sakomanogra: for panda it should be fine16:35
rsalvetiogra: even using the current u-boot default tree16:35
prpplagueogra: i think that just means the upstream items16:35
ograrsalveti, would you have time to package it ?16:35
ograi think we need to switch for the es2.0 HW anyway16:35
rsalvetiogra: sure, if you think we're good to update it now16:35
sakomanfor sdp/blaze it is missing final ethernet/spi driver cleanup and emmc environment support16:35
ograrsalveti, well, if it saves us from bad hacks like reformatting the fat every update16:36
sakomanhush shell and scripting are there16:36
ograsakoman, do you have an ETA for landing teh blaze features ?16:36
rsalvetiogra: should I update it now or wait for es2?16:36
prpplagueogra: if you switch to the mainline u-boot to fixed the magic number issue, it will be a good reason to push for changing support on our end16:36
ograrsalveti, well, with the current state we cant roll beta images16:36
rsalvetiogra: ok, will test the latest one and pack it16:37
rsalveti*try16:37
sakomanogra: probably a month or two since I've had OMAP3 stuff moved up on the priority list by the LInaro folks16:37
ograprpplague, right, the thing is that we have time constraints and updating stuff gets harder every day, from tomorrow on each fix requires a ton of paperwork16:37
prpplagueogra: ahh16:38
ograsakoman, bah, thats bad, out omap3 u-boot works fine apparently16:38
sakomanogra: not for Beagle xM or C416:38
* ogra hast seen any such issues on beagles16:38
sakomanthat support is not upstream16:38
ograi only have a C4 to test (kernel issues on XM prevent me from testing there atm)16:38
sakomanmy task is to get all support upstream16:38
ograsakoman, i dont use upstream, i use your branches ;)16:39
sakomanogra: my omap4-exp branch has functional ethernet16:39
sakomanit is just not cleaned up for upstream16:39
ograso we should use that one in the distro ?16:39
ograrsalveti, would you update our u-boot-omap4 package from that one ?16:40
sakomanthat is my working branch and I do frequent rebases as I clean things up and submit upstrea16:40
rsalvetiogra: http://people.canonical.com/~rsalveti/maverick/kernel/linux-image-2.6.35-17-omap_2.6.35-17.23rsalveti1_armel.deb should work for your xM, if you want to test it16:40
rsalvetiogra: sure16:40
ograrsalveti, i will, but i'm currently focusing on panda16:40
ograand have a long list of broken stuff still here16:41
rsalvetii know, I'm pointing the link just in case you want to test it :-)16:41
ograyeah, i will16:41
sakomanthe omap4-exp branch should work with panda, blaze, beagle c4, beagle xm, and Overo (both 35xx and 37xx versions)16:41
ograsweet16:42
GrueMastercool.  One uboot to bind them.16:42
sakomanthe only caveat is that the omap4-exp branch is rebased fairly often as I prepare the patches, submit them upstream, and respong to feedback16:42
rsalvetias we're not seeing any issue with u-boot-omap3, I'd prefer updating just u-boot-omap4 with your current branch16:42
rsalvetiogra: what do you think?16:42
sakomanI try to keep it working at all times though16:43
rsalvetisure, np16:43
ograrsalveti, yeah, it will also be a lot easier to package16:43
rsalvetiok then16:43
ograif you build multiple binaries you need multiple configure and build runs16:43
ograi plan to switch to linaro u-boot in N anyway16:44
rsalvetithat's not a problem as linaro's package is probably doing that already16:44
rsalvetibut I'd say it's better to just update omap 4 now16:44
ograyeah, linaro uboot is not uploaded yet i think16:44
rsalvetiif we see any blocking issue for omap 3, then we merge both packages in one that's upstream based (or linaro)16:45
ograi think slangasek said something like that in yesterdays call16:45
rsalvetihm, ok16:45
sakomanogra: I know jcrigby is testing my branch for the linaro u-boot builds16:45
ograwell, if the linaro u-boot would work on omap4 we could even switch now16:45
ograbut i think thats plain upstream16:45
ograand i want all the additional sakoman love from the -exp branch ;)16:45
slangasekhmm, what did I say?16:46
jcrigbyogra: just uploaded new version to ppa16:46
ograslangasek, that u-boot isnt uploaded to the archive yet16:46
ograthe linaro one16:46
slangasekcorrect - I need to work with jcrigby today to get it uploaded in a feature-freeze-compliant manner16:46
ograslangasek, we have fat issues with the current omap4 version we use and i was wondering if it wouldnt make sense to switch to your version now instead of waiting for N16:47
slangasekif it helps you, we're happy to support your use of it16:47
ograsince i want to do that switch anyway at some point16:47
ograbut you are building from plain upstream16:48
ograwhich doesnt have everything i'd like to have :)16:48
slangasekah16:48
ogra(we need to support panda and blaze with the omap4 package)16:48
slangasekwell, we can discuss putting some sauce into the packaging16:48
slangasekoh, that sounds less like sauce and more like a side dish :)16:48
ograand as sakoman said above eMMC support for blaze is missing atm16:49
ograand eMMC is the main disk on the blaze :)16:49
ograndec1, whats your suggestion from a blaze POV ?16:49
rsalvetiogra: jcrigby can confirm, but it seems current linaro's tree is using most of current sakoman's patches from omap4-exp16:52
ograrsalveti, well, lets do it that way, i trust you on picking the right way and leave it in your hands ;)16:52
rsalvetiogra: haha, sure :-)16:53
rsalvetibut I'd say at the moment that linaro's tree is enough for us16:53
rsalvetibut still want to look for omap3 related patches16:53
* ogra needs to reasearch busybox, gdm and update the netbook-settings package 16:53
rsalvetias we could be missing patches16:54
ograok16:54
ograthe busybox breakage smells very much like toolchain :(16:54
GrueMasterogra: I'm continuing my Alpha 3 update-to-fail investigation right now.  Should have something by EOD (since I can focus ,more on it & have faster network).16:55
jcrigbyogra:I have all of sakomans latest patches16:55
jcrigbyon top of upstream16:55
rsalvetijcrigby: cool, will try your package later16:55
ograGrueMaster, i still see no systm dbus on ym images, gdm will not start without that running16:55
rsalvetiand check omap 3 support16:55
rsalvetiif everything is well, than we can just use your package16:56
rsalvetiand drop omap3 and omap4 ones16:56
ograyeah16:58
ndec1ogra: what do you mean by 'eMMC support is missing on blaze'?16:58
ograthat was my plan for N16:58
ograndec1, in u-boot16:58
ndec1ogra: eMMC on blaze is working with TI version of uboot16:58
ograndec1, which u-boot would you suggest to use, apparently 1.1.4 has massive issues16:58
ograsince it seems unlikely that we can get the whole fat driver replaced now, we're discussing which other tree to use16:59
jcrigbyrsalveti, I have tested it on the hw I have which  is C4, xM and mx51 babbage16:59
* jcrigby has no omap4 hw17:00
ograjcrigby, yeah, we need omap4 panda :)17:00
ograbut we have the HW17:00
rsalvetijcrigby: nice, but our old omap 3 tree had some extra patches, I'd say17:00
rsalvetiinitial zippy support and other stuff17:00
rsalvetijcrigby: I'll test it on my panda17:00
ograrsalveti, i think i only made config changes and used sakoman'S omap3 tree as is17:00
jcrigbyrsalveti, I think sakomans next task is expansion board support for mainline17:01
rsalvetijcrigby: cool, good to know17:01
rsalvetiogra: yup, but these patches are not at his current branch17:01
ograah17:02
ograok, so downgrading busybox to the last version makes reboot -f work again17:02
* ogra re-adds -marm to debian/rules17:03
rsalvetiwho could help testing it on blaze? ndec1, GrueMaster?17:03
ograi have a blaze but a) not set up and b) we need a special eMMC setup we dont have yet17:03
ogra(in flash-kernel)17:04
rsalvetihm, ok17:04
GrueMasterogra: and davidm have the blazes (I don't get cool toys, only guts).17:07
ndec1rsalveti: we can. I just not available atm to discuss since I am on another meeting. i want to understand more about this. perhaps we can discuss on Fri during our call.17:07
ograndec1, beta freeze is unfortunately tomorrow17:07
rsalvetiit's simply, just to test if the latest u-boot boots at blaze17:08
ograrsalveti, well, its more since the blaze should get a special setup later17:08
rsalvetiyeah, but for now, I mean17:08
ograit will boot from a raw partiton on the eMMC and the u-boot version we use needs to support that17:09
ograor at least be enhancable in a way that we can easily get a freeze exception17:09
davidmI can send the blaze to GrueMaster if that helps, just got it back from NCommander17:11
ograwell, to someone who will actually work with it would be good :)17:12
ograeither GrueMaster or rsalveti i'D say17:12
NCommanderdavidm: I think I forgot to send the power cord (i couldn't find it, but its a standard plug)17:13
GrueMasterProbably faster/easier to send it to me, although rsalveti would be the better choice for long term development.17:13
davidmGrueMaster, I can't send it to him the duties would be insane17:13
rsalvetiyep, expensive board17:14
rsalvetisend it to GrueMaster, I can work with him on that17:14
ograthen send it to GrueMaster17:14
rsalvetiat least it'd be good so he could test it when needed17:14
rsalvetiGrueMaster: I can get it directly from you later, if needed17:14
ograok, busybox uploaded, automatic rebooting should work again17:15
rsalvetiogra: just reverting last commit is enough?17:15
rsalvetire-adding -marm17:15
ograrsalveti, yeah, just building with -marm again17:15
prpplaguejust a note, the power management is working on the ES2.0 8-layer panda boards17:16
prpplaguelooks like the twl6030 silicon we had on the earlier es2.0 boards has an issue17:16
rsalvetihm17:16
ogratricky17:17
=== fta_ is now known as fta
ograi dont think we'll get the 8 layer yet17:17
rsalvetiogra: i'd be good if people could test or ask to test arm specific changes on packages17:17
rsalvetibut don't know how hard was to find this bug17:18
ograit was just a look at the changelog and some experience :)17:18
ograthe gdm issue seems to be rather tricky17:18
ograwhat would be good would be to have enough HW to give it to ara so she could set up automated testing17:19
ograi.e. not require the maintainer to test but just have automated tests that send regression alterts17:19
ogra*alerts17:20
sakomanjcrigby: correct -- adding the expansion board stuff is on the agenda next17:20
sakomanshould be in by end of this week17:20
GrueMasterHopefully I can work with her on this post release.  My systems are idle at night anyways.17:20
rsalvetisakoman: nice17:21
ograGrueMaster, not your systems :) for N we should have huuuuge amounts of HW17:22
ograGrueMaster, but its surely worth a spec for N17:22
ograhey zul !17:22
ograyou in -arm ?17:23
=== hrw is now known as hrw|gone
zulhey ogra, yep just lurking17:23
ographew ... i was fearing server breakage :)17:23
ogragood to see you here :)17:23
GrueMasterogra: I've added a task to my list to write up a blueprint for test automation, however I'll need someone to proxy for me as I won't make UDS.17:25
ograhuh ? you wont ?17:25
ograthats bad, but i can proxy for you17:25
zulogra: hehe17:26
ograGrueMaster, or worst case i'll delegate to NCommander he has nothing to do anyway ... just that little dove thing :)17:26
GrueMasterI had already booked a cruise that week.  Booked it once we had the release date finalized at the last UDS.17:26
GrueMasterheh17:26
ograah, year i remember now, you told me at the sprint17:26
GrueMasterWhen I told the wife I should go to UDS, she gave me a look that would have given an exorcist pause.17:27
* sakoman has no idea what this N thing is17:27
* NCommander embeds a golf club in ogra's head17:27
GrueMasterNext Release.17:27
ograsakoman, next release after mmmmmaverick :)17:27
NCommandersakoman: its a letter17:27
NCommander:-)17:27
sakoman:-)17:27
* sakoman imagines Sesame Street music in the background17:28
sakoman"this release brought to you by the letter N"17:28
ogralol17:28
ograit actually *is* like sesame street :) you learn a lot about adjectives and animals following ubuntu releases17:29
ogramaverick meerkat :) natty narwahl17:30
* ogra takes a break17:31
GrueMasterI thought it was narcissistic narwhal.17:32
GrueMasterGuess I should read the web a little more.17:33
mpoirierhas anyone been able to use the ftrace's "set_ftrace_filter" on OMAP3 ?18:54
prpplaguesebjan: ping19:24
slangasekogra: are you guys using the linux-fsl-imx51 kernel at all in maverick?  It was inherited from lucid when the archive opened, hasn't been updated since, and if it's not needed / worked on, it's better to cull it from the archive than to let it drift19:29
slangasek(it's already missing two security revisions from lucid)19:30
loolslangasek: it's removed already19:31
loolin maverick19:31
slangaseklool: no, it isn't19:31
ogra_cmpcslangasek, we dont support imx51 at all19:31
ogra_cmpcin maverick19:32
loolslangasek: Oh I don't see the source in rmadison output19:32
slangaseklool: I'm looking straight at the archive; not sure why rmadison doesn't show it for you19:32
slangasekogra_cmpc: so you're ok with me nuking it?19:32
loolslangasek: Uh yes I do, I'm just assuming it's sort alphabetically </slap>19:32
loolslangasek: I'm just tired, sorry19:32
ogra_cmpcslangasek, totally19:32
GrueMasterslangasek: I do get security updates for lucid to test.  And we do still support it on lucid.19:33
slangasekGrueMaster: yes, we don't remove packages from released versions after the release19:33
GrueMasterok, just making sure.19:33
ogra_cmpcGrueMaster, only maverick19:34
GrueMasterMight also check with oem to see if they have any strange projects going on that require it.19:34
ogra_cmpcoh, that might be, but for imx51 they will surely have their own kernel/bootloader19:35
ogra_cmpcsince ours definitely only supports babbage boards19:35
slangasekand hasn't even been getting security updates19:36
ogra_cmpcslangasek, lamont or IS might have some interest in having an imx51 kernel since our buildds are currently all babbage 3.0 boards19:36
slangasekif OEM were using it, I expect they would've interfaced with the kernel team on this before now19:36
slangasekogra_cmpc: IS doesn't run non-LTS on the buildds19:36
slangasek(if they can help it)19:36
ogra_cmpcwell, armel is spethial :)19:36
slangasekit's nuked already, so anyone who needs it should've spoken up sooner19:37
slangasekarmel isn't *that* special, no19:37
ogra_cmpcwell19:37
ogra_cmpcwe had buildds where IS had no control about the kernel at all for the last few releases19:37
slangasekwhich is quite a separate thing from wanting to run a kernel from a *newer* non-LTS release19:38
ogra_cmpcif we run into HW related probs like we did with the former setup it might require an update19:38
ogra_cmpcbut for that we can indeed have a separate kernel if neede19:39
ogra_cmpcd19:39
ogra_cmpc(the pegatron buildds with the binary only kernel had issues building thumb2 for example)19:40
prpplagueogra_cmpc: have you got a L24.9 repo you guys are using for testing your ubuntu builds from?19:40
ogra_cmpcprpplague, for kernel ?19:40
prpplagueyea19:40
prpplaguersalveti: said he was waiting on sebjan to do some testing with L24.919:41
ogra_cmpcprpplague, we get a specially merged tree from sebjan where he rebases the TI releases on top of the ubuntu tree19:41
ogra_cmpcwe're waiting for the 2.6.35 rebase, zes19:41
prpplagueogra_cmpc: ahh ok19:41
ogra_cmpcs/z/y/19:41
prpplagueogra_cmpc: ok, i was kinda hope'n to test your builds19:42
ogra_cmpcwell, we should get the rebased tree this week19:42
ogra_cmpcthe only working image we currently have is a few weeks old if you are intrested19:43
ogra_cmpchttp://cdimage.ubuntu.com/ubuntu-netbook/ports/releases/maverick/alpha-3/19:43
GrueMasterslangasek: I'm not sure why the kernel updates haven't been getting in there.  I know they have released some to proposed fairly recently (July I think).19:43
ogra_cmpcjust gunzip, dd to SD card and boot away19:44
GrueMasteractually, I am currently running a kernel built on 8/19.19:44
ogra_cmpcGrueMaster, because the mx51 kernel was in a separate tree19:44
ogra_cmpcin maverick nobody maintains that tree19:44
slangasekGrueMaster: because publishing to the current dev release isn't part of the security team's process for kernels; so if nobody's maintaining the kernel in maverick, it would be guaranteed to continue falling farther behind between now and release19:44
GrueMasterok.  Starting to understand.19:45
ogra_cmpcsmb updates the lucid tree19:45
ogra_cmpcbut cooloney doesnt update the dev tree anymore19:45
slangasekif somebody wants imx51 in maverick, they can work on getting support upstreamed and into linux-linaro :-)19:46
ogra_cmpci thought linux-linaro has imx51 support already19:46
slangasekyes, as much support as is present upstream19:46
ogra_cmpcah19:46
ogra_cmpcwhich is likely sparse :)19:47
slangasekquite19:47
ograsigh, now my panda stopped booting completely20:30
GrueMasterIs there an easy way to backrev a package with apt?20:31
GrueMasterOr does it always load the latest.20:32
ograyou can define a version iirc, but if its only one package i'd just wget and dpkg -i it20:33
GrueMasterok20:38
rsalvetisakoman: does 'reset' at u-boot work with you at beagle xM?20:48
rsalvetiat mine I just get stuck20:49
sakomanrsalveti: I don't have a "final" xM yet (due tomorrow)20:49
sakomanrsalveti: let me try on an early prototype I have20:49
sakomanrsalveti: yes, just hangs20:50
rsalvetisakoman: mine is a p8, not final too20:50
sakomanrsalveti: let me try on an Overo 37XX20:50
rsalvetihm, so, so it seems we got a bug20:50
rsalvetiok20:50
sakomanwe can at least see then if it is a 37XX related issue, or is xM specific20:51
loolsakoman: I'd love knowing if the XM comes with a MAC address in eeprom in the final version20:51
sakomanlool: perhaps we'll find out soon :-)20:52
sakomanrsalveti: same situation with the Overo 37XX20:52
sakomanrsalveti: I'll investigate20:52
rsalvetisakoman: ok, thanks :-)20:53
sakomanrsalveti: looks like the reset command uses a generic arm library used by all arm processors21:07
rsalvetihm21:08
sakomanI'll have to ping some TI folks to see if they know why this code doesn't work on the 37XX21:08
=== robclark_ is now known as robclark
sakomanrsalveti: found a potential fix.  will test.21:12
rsalveticool21:12
sakomanit is indeed -- in involves doing a "cool" reset21:13
sakomanso your choice of words was amazingly accurate :-)21:13
rsalvetihaha :-)21:16
rsalvetilool: so no mac address in eeprom for us21:16
rsalvetithe eeprom pins are  n.c.21:16
sakomanbummer :-(21:16
rsalvetimaybe for panda, but not for xM21:17
rsalvetiprpplague: do you know if we get an smsc eeprom for panda?21:18
rsalvetior is it just n.c. as xM?21:18
sakomanrsalveti: the reset fix worked on my 37XX Overo, let me check for regressions on the 35XX versions21:19
rsalvetinice21:20
prpplaguersalveti: one sec21:20
sakomanrsalveti: worked fine on 35XX also, I'll push to my omap4-exp branch shortly21:22
sakomanI'll ping you when it is done21:22
rsalvetisakoman: great!21:22
loolrsalveti: sadness21:23
loolrsalveti: So I definitely think it's worth talking to the usbnet maintainer21:23
sakomanlool: the Overo's have eeproms on their network chips21:24
loolGood!21:27
GrueMasterHA.  I found a dependency issue in the pool that slipped by the awesomeness of debian packaging.  gnome-terminal now depends on libvte but libvte failed to upgrade at the same time.21:27
prpplaguersalveti: the eeprom is no connect on the panda21:34
prpplaguersalveti: you can do one of two things21:34
prpplaguersalveti: either have linux driver get a random mac21:34
prpplaguersalveti: or you can use rob's patch to assign one manually as a boot arg21:34
sakomanrsalveti: I added the fix to the set of patches I am preparing for submission -- staged in the omap4-exp branch21:34
sakomanrsalveti: in particular: http://www.sakoman.com/cgi-bin/gitweb.cgi?p=u-boot.git;a=commitdiff;h=ce927c809c6b3a619a2cd712c47563676d79f8f021:35
rsalvetiprpplague: yep, also saw that patch21:36
rsalvetijust wanted to confirm21:36
prpplaguersalveti: ahh ok21:36
rsalvetiI'm fine with the random one for the moment21:36
rsalvetiouch, can't boot my kernel on a b5 (external abort on non-linefetch)21:37
rsalvetihm... works at my c421:37
rsalvetisakoman: nice, thanks21:37
rsalvetiprpplague: but thanks for checking it21:40
prpplague:)21:40
sakomanrsalveti: thanks for testing and the bug report21:42
rsalvetiprpplague: ndec: don't know yet, but do we have access to any hardware documentation for panda? (public or canonical)21:42
sakomanthere's a lot of common code, so I need to test on 6 different OMAP3/OMAP4 boards21:42
ndecrsalveti: jayabharath might know.21:42
rsalvetisakoman: compiling new u-boot right now :-)21:42
jayabharathrsalveti: what documentation would you like.. we can load it up on the wiki21:43
sakomanrsalveti: I'm sure you'll let me know if it doesn't work ;-)21:43
rsalvetijayabharath: things like the schematics21:44
rsalvetithis sometimes can help debugging kernel issues21:44
rsalvetimostly gpio, pins and etc21:45
rsalvetinot anything specific at the moment, can ping you if I need anything else21:45
jayabharathSure.. will load them up on 'the' wiki21:45
rsalvetisakoman: sure :-)21:45
rsalvetijayabharath: cool, thanks :-)21:45
rsalvetijayabharath: would you mind sharing the doc for es1 and es2?21:46
rsalvetiif not possible, just es2 related is enough21:46
jayabharathOk. will load up all the schematics for all revs :)21:46
rsalvetijayabharath: awesome, thanks a lot21:46
prpplaguersalveti: i have the lilluput/iMO display working on the panda22:43
rsalvetiprpplague: haha, cool!22:47
rsalvetiargh! another fried sd22:48
rsalvetisakoman: do you know if reset worked for panda before?22:49
* robclark wants to see it22:49
* rsalveti can't remember22:49
* rsalveti can't see it :-(22:49
robclarkrsalveti: fwiw, the reset button works much better on es222:49
sakomanrsalveti: don't recall, will have to check22:50
rsalvetiwell, doesn't seems to work at all at my es1 hah22:50
rsalvetisakoman: at least with your latest patch it doesn't work22:50
rsalvetibut don't remember if it worked or not before22:50
rsalvetitook a while to test because one of my most used sd card got fried :-(22:51
rsalvetilots of i/o errors22:51
sakomanrsalveti: OMAP4 uses different reset code -- the arm reset library function calls an architecture dependent assembly language reset function22:52
rsalvetioh, ok, makes sense22:52
sakomanjust tried sdp -- that hangs too22:52
sakomanI may need to do a similar patch for OMAP4, will investigate22:53
rsalvetisakoman: haha, ok22:53
rsalvetibut at least I can confirm that it's working fine for all beagles I have22:54
rsalvetir5, c4 and xm22:54
rsalvetiwith your latest patch22:54
sakomanrsalveti: oops - I was mistaken OMAP3 and 4 share the reset code.  I tested with a pre-patch u-boot on SDP, so I need to rebuild and retest22:55
rsalvetihm, ok22:55
sakomanrsalveti: it hangs with my change, and it hangs with TI's internal u-boot23:05
sakomanso I suspect that either the HW is bad, or the required reset code has changed for OMAP4 (or perhaps both)23:06
sakomanI'll ping some TI folks about that23:06
rsalveticould be both23:07
rsalvetisakoman: do you have an es2?23:07
rsalvetihm, needs other stuff23:07
sakomanrsalveti: no, all my machines are ES123:07
sakomanprpplague: does the u-boot 'reset' command work with the latest OMAP4 silicon/boards?23:08
* rsalveti things he is the only one who actually uses this command23:10
rsalveti*thinks23:10
* rsalveti needs more coffee23:10
sakomanrsalveti: I used to use it a lot on OMAP3 after changing the u-boot environment in nand23:11
sakomanrsalveti: I am roasting some coffee, cause I need some too :-)23:11
sakomanrsalveti: I don't use it some much on newer 37xx/OMAP4 boards because they don't have nand!23:12
rsalvetihahah, true23:12
sakomans/some/so/23:12
sakomansee I do need coffee!23:12
rsalvetihaha :-)23:12
sakomanin any event, I'll try to track down why reset command is broken on OMAP423:13
prpplaguesakoman: not sure i'll have to test23:18
Baybalmmm23:33
Baybalare there any omap4 COMs?23:33
IowIm booting ubuntu-10.04.1-minimal-armel.tar.7z, it loads up to the language selection but it wont take input from my keyboard23:49
rsalvetitime for dinner, hungry23:52
IowWhich line is the init for the usb keyboard?23:55

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