mwhudsonzyga: yes00:03
plarsI sometimes have to pause and think about it, but I'm pretty sure I've never typed lava-bzr-tool00:04
zygalt is not allocated00:04
plarsand you already know that I alias ldt=lava-dev-tool00:05
plarsthe name would expand to lava-dashboard-tool also of course, but I don't use that one as often00:05
zygalt dev 00:05
zygalt dash00:05
mwhudsoni think there is some work to be done in this area00:06
zygamwhudson,  I agree00:06
mwhudsonthe fact that lava-dev-tool has auth-add is pretty wrong :)00:06
zygamwhudson, if we have some time we can look at how commands get registered00:06
zygamwhudson, see plars' email 00:06
mwhudsonbut i'm not sure it's the most pressing thing00:06
mwhudsonzyga: which one? the v.l.o root page sucks one?00:06
zygamwhudson, yeah00:06
zygamwhudson, the one where management kills us for not producing viewable results00:07
mwhudsonzyga: tangentially, i don't know that having /lava-server/ in the url makes a lot of sense00:07
zygamwhudson, that's debian for you00:07
zygamwhudson, it's the debian webapp policy00:07
mwhudsonzyga: can we change that to /lava/ somehow?  ideally in a way that doesn't get overwritten in the next rollout00:07
zygamwhudson, it makes sense but we should customize it on v.l.o00:07
mwhudsonzyga: sure, it's a debian /default/ though right?00:08
zygamwhudson, perhaps, I believe it may get overwriten (I did not use ucs everywhere where it should be used)00:08
zygamwhudson, yes00:08
zygamwhudson, the parts to be hacked are in django-debian00:08
mwhudsonzyga: this is something i need to understand, we don't install /etc/apache2/lava-server.conf as a conffile do we?00:08
mwhudsonif that's even the right word00:08
zygamwhudson, we should and AFAIR we do automagically but I dont' know 00:08
zygamwhudson, AFAIR all /etc is treated like that00:09
zygamwhudson, but we don't install those00:09
zygamwhudson, we generate those postinst00:09
mwhudsonalthough, is that file created by the postinst of a package using django-debian?00:09
mwhudsonok 00:09
zygamwhudson, and as such one must use ucf to track them00:09
mwhudsonso django-debian's code will just blinding overwrite them currently00:09
mwhudsonzyga: ucf?00:09
zygamwhudson, I'm sure I missed that as using ucf is not trivial (I tried it but failed)00:09
zygamwhudson, another debian thing, update configuration file00:10
zygamwhudson, AFAIR the issue I found (afterwards) was me attempting ng to register an existing file00:10
zygamwhudson, where the correct use is to create a new file somewhere (perhaps with .dpkg-new suffix, perhaps not) and use ucf with both paths00:10
mwhudsonzyga: there only seems to be one invocation of ucf in django-debian currently00:12
zygamwhudson, yeah I'm sure it is broken00:12
mwhudson./postrm:44:            ucf -p "/etc/$django_debian_package/default_database.conf"00:12
zygamwhudson, I can look at it early tomorrow as it seems to be a pressing issue00:12
mwhudsonzyga: it's not pressing at all00:12
* mwhudson files a bug00:12
zygamwhudson, but I would love to retain /lava-server/RPC2 as a redirect or something that will still work00:12
mwhudsonzyga: ah yeah, xmlrpc and redirects00:13
plarscan't we just redirect if someone goes to /?00:13
zygamwhudson, POST does not redirect well, does it?00:13
zygaplars, oh00:13
zygaplars, that's easier00:13
zygaplars, yes we can00:13
plarsand it won't get overwritten :)00:13
mwhudsonzyga: well, browsers generally react to a redirect on POST with a GET00:13
mwhudsonzyga: but i don't know what xmlrpclib does00:13
mwhudson(probably it raises HTTPError with statuscode == 30x)00:14
mwhudsonzyga: i filed https://bugs.launchpad.net/django-debian/+bug/81436100:16
ubot2Ubuntu bug 814361 in django-debian "overwriting config files in postinst violates debian policy" [Undecided,New]00:16
mwhudsonbut we can forget about that again for now00:16
* zyga pushes final dashboard bits to 0.6-wip00:17
zygaI hate code review with launchpad and bzr :/00:18
zygait never seems to work well for my laziness00:18
zyga(and bzr crashing on bzr lp-propose-merge)00:18
zyga(bug reported)00:18
mwhudsonif you aren't really dedicated about 'one branch per feature' then it's a pain yes00:19
zygamwhudson, I'm lazy and I never found a good solution for the pipeline problem00:19
mwhudsoni would possibly argue that pushing you in this direction is a feature :)00:19
zygamwhudson, X depends on Y00:19
zygamwhudson, having 10s of branches 00:19
zygamwhudson, plars could we quickly look at https://code.launchpad.net/~linaro-validation/lava-dashboard/0.6-wip/+merge/68142 to see something obviously broken that I've missed?00:20
zygaI know it's 1.8K LOC00:20
zygasome of that is the new LEB view00:20
mwhudsonzyga: as we get more mature, i hope the level of relative churn will go down00:20
zyga(with 0 tests, sorry :-(00:20
zygamwhudson, I hope so too00:20
zygamwhudson, but it does not solve the workflow problem I have00:21
* zyga runs pyflakes00:21
mwhudsonzyga: well i don't see anything amazingly wrong, but of course i can't claim to understand it all00:26
zygamwhudson, thanks00:27
zygamwhudson, I found a few issues with pyflakes, fixing that now00:27
zygamissing imports00:27
zygaor unused imports00:27
mwhudsonzyga: you should get that integrated into your editor00:27
zygamwhudson, yeah, I should00:28
zygaI forgot about dispatcher.py00:29
zygaa whole copy of the dead dispatcher00:29
plarszyga: this doesn't break existing dataviews I assume right? just moves the code around a bit it seems?00:33
zygaplars, yes00:33
zygaplars, so it looks more like DataReport00:33
zygaplars, and uses the same dashboard_app.repository code00:34
plarszyga: my only comment before was that I didn't care for the name00:34
plarszyga: for the new report00:34
zygaplars, ?00:34
plarsbut that's not really critical I guess00:34
zygamwhudson, I get strangeness when I try to test on postgres00:34
plarsI don't think it's obvious to most people what there going to get there00:34
zygapsycopg2.ProgrammingError: autocommit cannot be used inside a transaction00:34
mwhudsonzyga: too new a psycopg200:34
plarszyga: would prefer something a little more obvious like image status or something00:34
zygamwhudson, that's with psycopg2 from pip, some random fresh version00:34
zygaplars, ah, I can still change that00:35
mwhudsonzyga: yeah, don't do that00:35
zygaplars, just make up your mind quickly on how to call that00:35
zygamwhudson, let's add the right version to project.json shall, we?00:35
zyga(where the comma was late)00:35
mwhudsonzyga: pip install psycopg2==2.4.100:35
plarszyga: doesn't really matter what it's called internally, just call it image status for the link, I think that would be more clear than having a link called "LEB's"00:35
mwhudsonzyga: yes, good idea00:35
zygaplars, ok, shall I rename that to "Image Status"?00:36
zygaon the unrelated note -- all tests passed on my panda :-)00:37
mwhudsonzyga: lava-dev-tool terminology question00:55
mwhudsonzyga: what does 'focus' mean?00:56
zygamwhudson, I should draw a picture for that 00:56
zygamwhudson, it determines what gets used when component commands are invoked00:56
zygamwhudson, it technically defines the workind_dir00:56
mwhudsonzyga: is it related to track/hack ?00:56
zygamwhudson, you can have a component that is in a tarball or a branch and you can have multiple hacking branches 00:57
zygamwhudson, yes00:57
zygamwhudson, so the code may be from 1) tarball 2) trunk 3) arbitrary hacking branch (that's focus_target)00:57
zyga(focus_target may not be released yet, I don't remember)00:57
mwhudsonzyga: doesn't look like it's even in lp:lava-dev-tool00:57
zygamwhudson, nope, I wanted to write more tests and fix a design issue with actions00:58
mwhudsonoop, meeting time00:59
mwhudsonplars, zyga: is it just going to be us, though?00:59
zygaI think so01:00
rsalvetijames_w: how can I change a project group name at ubuntu-build?01:28
rsalveti11.05 and 11.11 is wrong and confusing, it should be Natty/Oneiric or Stable/Unstable01:29
mwhudsonrsalveti: i bet you can do it in the admin interface, if no other place01:36
rsalvetimwhudson: I'm not allowed to change nor rename the project group01:37
rsalvetiprobably lack of permission01:37
mwhudsonrsalveti: ok, i bet it is possible, for some people, in the admin interface :)01:37
mwhudsonnot necessarily you or i01:37
rsalvetimwhudson: yeah, never saw this interface but I believe it exists :-)01:38
rsalvetihopefully james_w should know better01:38
mwhudsonit's likely01:38
james_wrsalveti, seems I can only add and not edit02:08
james_wI go to edit one and it creates a new one with the new info and leaves the old one there02:09
rsalvetijames_w: can you add a project group called Natty and another Oneiric?02:09
rsalvetiI guess I can then move the projects around02:09
james_wrsalveti, do you want to maintain a images/hwpacks split?02:09
james_whi Dr_Who 02:09
rsalvetijames_w: yes, please02:09
james_wrsalveti, you should now be able to change the project groups of all of the projects02:10
rsalvetijames_w: great, thanks! :-)02:12
james_wrsalveti, no problem02:12
james_wrsalveti, I'll make you an admin02:12
rsalvetiDr_Who: where was our official live-build branch?02:13
rsalvetiDr_Who: it seems to be at your user still, am I right?02:13
rsalvetijames_w: great, thanks!02:13
Dr_Whorsalveti:  shouldn't be .. should be both my own and in ~linaro-maintainers02:14
james_wrsalveti, just note what it says in the description of https://launchpad.net/~linaro-offspring-admins02:14
rsalvetiDr_Who: I couldn't find it at ~linaro-maintainers02:14
Dr_Whomy fault then02:15
Dr_Whorsalveti: gotta bug ?02:16
rsalvetiDr_Who: no, was just looking for it to see what version we were using02:17
rsalvetiand also noticed that it seems that the ppa version is a little bit older than what is at your branch02:17
Dr_Whoyes that is correct ... I upgraded my branch and pushed that before I dove into the cross work02:17
Dr_Whoalso landed a patch or two from the ubuntu two IIRC02:17
rsalvetijames_w: sure!02:18
Dr_Whopfefferz would be proud... man vs food is in austin texas ... he was just at the Salt Lick ...  dang makes me hungry for BBQ02:19
mwhudsonplars: when do the jobs on http://validation.linaro.org/jenkins/ run?  seems like 6amish utc?02:25
plarsmwhudson: something like that, 8AM localtime iirc02:26
mwhudsonok cool02:26
mwhudsonplars: which machine is it, in fact?  the same as validation.linaro.org?02:27
plarsmwhudson: yes02:27
plarsmwhudson: we only have the one at the moment02:27
mwhudsongaaah i'm so used to jenkins builds on ec2, where the workspace gets annihilated for you02:36
rsalvetijames_w: can I remove projects by the admin interface or should we request an rt?02:36
mwhudsonoh and of course lava-dev-tool doesn't work with 2.702:38
james_wrsalveti, just disable the projects unless you really need them removed for some reason02:53
mwhudsonplars: i have to laugh: http://validation.linaro.org/jenkins/job/lava-selftest/ws/ldt/status.txt02:53
rsalvetijames_w: yup, better02:56
plarsmwhudson: hmm03:31
plarsmwhudson: I did test installing that, but now I doubt03:31
mwhudsonplars: otp sorry03:31
plarsmwhudson: if you are referring to the lava-dispatcher install failure, I think that's false, not sure why it fails in ldt03:32
mwhudsonplars: it was just all the terminal codes i was laughing at03:32
plarsodd, I'm not seeing them on that page03:33
plarsmwhudson: I just see ascii03:33
plarsmwhudson: I wonder if chrome stripped them for me03:33
mwhudsonplars: http://validation.linaro.org/jenkins/job/lava-selftest/lastSuccessfulBuild/artifact/ldt/status.txt03:33
mwhudsonplars: i fixed the problem, so maybe the copy in the workspace is updated03:35
Openfree` BUG: sleeping function called from invalid context at arch/arm/mm/fault.c:31003:46
Openfree`agreen-away, hi I'm using kernel3.0, master03:47
Openfree`commit: 5b14e370cca884fa86471e909512f65b06dcaec203:47
Openfree`or where's the better place to file a bug report?03:48
mwhudsonplars: lava-dispatcher fails because there's a missing dependency on pexpect03:55
rsalvetiand final 3 is really out03:57
=== nytowl is now known as nytowl_away
pfefferzDr_Who, I love the Salt Lick04:20
Dr_Whopfefferz: me too :-D04:20
Dr_Whopfefferz: got a q or two for you but not at the mo .. I'm knocking off live-build3 cross build WIs04:21
Dr_Whomorning aviksil 04:22
aviksilDr_Who: morning!04:23
aviksilDr_Who: it must be evening your place? :)04:25
Dr_Whoaviksil: yeah ... 11:25pm ... just about to call it a day04:25
aviksilDr_Who: ah!04:26
rsalvetiaviksil: morning!04:27
rsalvetiI'm still on, as usual04:27
rsalvetilong night ahead04:27
Dr_Whorsalveti:  brazilian coffee keeping you strong!04:28
rsalvetiDr_Who: yeah :-)04:28
aviksilrsalveti: i know :)04:28
Dr_Whoheh .. just discovered a way to crash linaro-media-create04:30
Dr_Whobut .. it's not a l-m-c problem ... it's a bug in my stuff04:31
Dr_Whowell enough fun for one night ...  see you in the morning!  04:33
rsalvetic-ya, have a good night04:35
Dr_Whorsalveti:  one q ... does l-m-c create /etc/fstab or should that already be in the rootfs tarball handed off to the tool ?04:37
Dr_Whonm ... I have my answer04:38
Dr_Who(it exists but it's unconfigured)04:39
rsalvetiafaik l-m-c creates it04:41
jcrigbyrsalveti, linux-linaro-omap kernel is building here: https://launchpad.net/~linaro-maintainers/+archive/kernel/+build/263909007:08
rsalvetijcrigby: awesome, are you pushing other flavors than omap?07:09
rsalvetiwould be good to push the ones we support07:09
rsalvetilike mx51and vexpress07:10
rsalvetiplease make sure you're pushing the headers package and the meta :-)07:10
jcrigbyrsalveti, ok07:11
jcrigbyrsalveti, all three flavours and meta are currently building07:42
rsalvetijcrigby: awesome, you rock07:42
jcrigbyand aneesh fixed the spl problem07:42
jcrigbyhe sent an email if you did not yet notice07:42
jcrigbyso I'll get back to that tomorrow07:43
rsalvetijcrigby: yeah, just saw it07:43
rsalvetijcrigby: did you try it already?07:43
rsalvetilet me fire the build and see :-)07:43
jcrigbyok, his email was pretty convincing07:43
jcrigbyI'm done for today07:43
rsalvetijcrigby: yup, it works :D07:44
rsalvetijcrigby: yeah, me too07:44
jcrigbywow that was quick07:44
rsalvetiwas at the right console and the sd card was at my note already07:44
jcrigbyso do you know what tilt branch is supposed to be used for packaging?07:44
rsalvetijcrigby: not yet, let me fire an email to andy07:45
jcrigbyok, great. Thanks.  07:45
rsalvetialright, c-ya tomorrow, good night!07:45
jcrigbyit is so cool to see two version/date strings with the same info07:46
amitdkrsalveti,  Hi08:00
rsalvetiamitdk: hey, still alive but going to sleep08:00
rsalvetijcrigby: yup, lovely :-)08:00
amitdkamitdk, what is your linaro mail id08:00
rsalvetimine? ricardo.salveti@linaro.org08:01
rsalvetiare you planning to update your package today?08:01
rsalvetiif so just send me again and I'll review and push to our overlay and later to ubuntu08:01
rsalvetithis time we'll have enough time to do it properly08:01
amitdkrsalveti, i will send a mail metioning the details of my sources.08:02
rsalvetiamitdk: awesome, plan to go over the packages from the WG until my eod08:02
amitdkrsalveti, ok thanks08:03
rsalvetigreat, going by now08:03
rsalvetic-ya in a few hours08:03
* rsalveti &08:03
rsandiforamana: pong from yesterday08:20
ramanarsandifo: I was reading the auto-inc-dec patch and had a question though I seem to have forgotten it now. 08:21
rsandiforamana: ok08:21
amitkamitdk: nice work. Did you patch the .xml files too to build powertop?08:38
amitdkamitk, yes i added in xml files08:40
amitkamitdk: send the patch to the .xml files in your email too, to avoid any mistakes in branch names, etc. See reply to your email.08:42
sangwookdavepigott,  09:00
=== davidrusling_ is now known as davidrusling
=== agreen-away is now known as agreen
ramanarsandifo: remembered now - do you have any WIP address_costs patches. The ivopts patch appears to be working quite nicely in a number of places, the number of vfp moves drops dramatically in some of the SPEC2k6 benchmarks of the order of 20% or so in some cases. There is one case this doesn't handle - the POST_MODIFY_DISP and the PRE_MODIFY_DISP cases where it's auto-inc-dec we are fighting. 09:34
ramanaand not ivopts09:34
ramanafor e.g. this from scimark2. 09:36
ramana     fldd    d16, [r4, #-136]09:36
ramana        add     r4, r4, #3209:36
ramana        fmrrd   r2, r3, d1609:36
ramana        strd    r2, [r5], #3209:36
rsandiforamana: not address_cost, no, just rtx_cost09:36
rsandiforamana: what code would like you from the above sequence?09:37
ramanafor the above sequence I would prefer a ftd d16, [r5] followed by an add - it's no more expensive really. 09:38
ramana*fstd 09:38
ramanahere by choosing post_modify_disp we've ended up adding a move from the vfp reg bank to the integer reg bank and the increment of the address is still to happen. 09:39
ramanaso we've added an additional move by actually choosing this address form. 09:39
ramanaI suspect. 09:39
rsandiforamana: But the point is that we don't know before reload whether we'll have a VFP or a core register.09:40
rsandiforamana: If we increase the cost based on what happens for VFP registers, we might as well just disable automod for those modes.09:40
ramanarsandifo: That might be an option. 09:41
rsandiforamana: I suppose this comes back to the idea of running auto-inc-dec after reload.  I'd still like to run it before for NEON stuff, but a limited post-reload pass could be used to handle cases where the register really is core.  I.e. conditionally disable automod until after reload, then use the post-reload auto-inc-dec to clean up simple cases.09:43
Openfree`agreen, http://paste.pocoo.org/show/444032/ , any idea about this?09:43
rsandiforamana: In a similar vein, I wondered whether we should disable REG+CONST addresses for quad vector loads and stores.  I thought it would often be cheaper to have an ADD and VLDM/VSTM rather than two VLDR/VSTRs.09:45
ramanarsandifo: Yes I think a limited post-reload pass should end up cleaning up cases where the register is really core. 09:46
agreenhold on changing client09:48
ramanarsandifo: thinking about the neon case. 09:48
agreenFirst problem seems to be no wlan firmware09:49
agreenThat's what's causing tne 60s business09:49
agreenWere you doing anything in particular when you got the BUG?09:51
agreenAlso it doesn't seem like you have CONFIG_HIGHMEM on09:56
ramanarsandifo: I think you're probably right on that one.09:59
davepigottspringz: How do I update to an old uboot, and which old uboot do I update to on the mx51?10:13
springzdavepigott, extract one from hwpack10:15
springzdavepigott, get it from deb, and burn it to sd card10:16
davepigottspringz: How do I burn the uboot onto the sd card?10:16
springzdavepigott, /bin/dd if=u-boot.bin of=/dev/sdb bs=1024 seek=1 skip=110:17
davepigottspringz: ok. thanks10:17
Openfree`agreen, hi, actually I'm doing pretty heavy compile10:18
Openfree`wait, previously I sent one mail to pandaboard group10:19
Openfree`which also use this kernel, same commit10:19
Openfree`but different error10:19
agreenI can see one of the processes affected was tar10:19
Openfree`I setup one swapfile, 10:20
agreenIs it possible the thing is getting too hot?10:20
Openfree`and this file located in usb-storage disk10:20
agreenWas it committing from swap?10:20
Openfree`too hot, em.. maybe, I've already have a heatsink10:22
agreenwell heatsink should be enough10:22
Openfree`also for the fw problem, should I abtain the firmware out of the kernel tree10:22
Openfree`I did make modules_install10:23
agreenyes there is a firmware tree on kernel.org10:23
agreenno it's a separate animal, linux-firmware iirc10:23
agreenyeah it's on git.kernel.org, pop it in /lib/firmware keeping any subdirs intact10:24
Openfree`agreen, also in my email, have you noticed that also one rtc/irq bug10:25
Openfree`because I enable RTC in kernel while pandaboard don't provide10:25
agreenhum dunno what that is about, twl6030 does expose rtc... maybe the irq is broken10:30
Openfree`checked the fw problem, actually there firmware "/lib/firmware/ti-connectivity/wl1271-nvs.bin"10:35
Openfree`drivers/net/wireless/wl12xx/main.c:950:ret = request_firmware(&fw, WL12XX_NVS_NAME, wl1271_wl_to_dev(wl));10:35
Openfree`ohad, for the firmware problem, I think that because I compile wl1271 into uImage, and it require the firmware before mount the rootfs10:38
agreenyeah makes sense... those guys work fine as modules; they should build as modules in our reference config omap4_defconfig10:49
Kalozmichaelh1|away: is the ppc build bug fixed with the new releases?10:51
fgiffplars: ping11:50
ams_csrsandifo: ping11:57
rsandifoams_cs: here12:03
ams_csrsandifo: I happened to notice that you have the lofty status of "RTL Maintainer" ;)12:04
rsandifoams_cs: acquired with much controversy...12:04
ams_csrsandifo: any chance you could take a look at this? http://www.mail-archive.com/gcc-patches@gcc.gnu.org/msg10065.html12:04
ams_csrsandifo: yeah, I noticed that also ...12:05
rsandifoams_cs: sorry, can't review that.  It's rtl optimisers only; expand and tree stuff don't count.12:05
ams_csrsandifo: Richard Guenther reviewed the rest of the patch series, but that one was too back-end for him12:05
ams_csrsandifo: oh, what sort of maintainer does cover that?12:06
* ams_cs has fallen through the cracks12:06
rsandifoyeah, I think you need a global maintainer who's interested in rtl12:06
rsandiforth, Bernd or Jeff maybe12:07
ams_csrsandifo: thanks12:09
james_wmorning all12:25
davidgilukramana: Hi12:31
ramanadavidgiluk: hi12:31
davidgilukramana: You remember that MCR/DMB define in gcc, the one where I added the !Thumb_T1 ?  It uses arm_archv6k - do you know why it has the k? pm215 pointed out that the copro registers look like their in all of 6 according to the ARM ARM12:33
davidgilukhmm - they're12:33
ramanadavidgiluk: no not really . I'll have to check that. 12:38
ramanathis is the TARGET_HAVE_DMB_MCR ? 12:39
davidgilukramana: Yes12:40
plarsfgiff: hi13:28
fgiffplars: I resubmitted my job to lava, but it disappeared again. Any ideas what might have happened to it?13:33
zygaplars, fgiff hi13:35
* zyga starts to wake up13:35
zygaI'll start deploying the bits 13:35
zygafgiff, this will give you good visibility to your jobs too13:35
fgiffzyga: Thanks!13:35
plarsfgiff: can you show me the json for it?13:36
plarsfgiff: Let me try to run it manually, that should provide some more insight13:45
plarsfgiff: ah, I think I might see the problem though13:45
plarsfgiff: there's an extra parameter for submit_results there... iirc it won't just ignore that, it will try to pass it in, and not recognize it as a valid param13:46
plarsfgiff: I'll be able to test that theory in a moment, but I suspect if we leave out "result_disk": "sdcard", it will work just fine13:46
plarsfgiff: wait, nm, that is valid... forgot that was added13:47
* plars is not quite awake yet13:47
fgiffplars: I got that parameter from here: https://wiki.linaro.org/Platform/Validation/LAVA/Documentation13:48
plarsyeah, it's fine13:48
plarsfgiff: it couldn't download https://android-build.linaro.org/jenkins/job/linaro-android_leb-panda/61artifact/build/out/target/product/pandaboard/boot.tar.bz213:51
fgiffplars: I see the problem, I cut out a '/'13:52
fgiffI'll fix it and resubmit13:52
=== nytowl_away is now known as nytowl
pfalconAnybody seen issue with cc: linaro-dev when posting to some other mailing list? I'm getting delivery failures in such cases since yesterday14:16
davidgilukwhat type of delivery failure ?14:19
pfalcondavidgiluk: usual SMTP bounce with "Delivery Status Notification (Failure)" in subject14:20
davidgilukthe contents should in principal have some detail as to what rejected it and if you're lucky why14:21
pfalcondavidgiluk: "550 550 Message headers fail syntax check (state 18)."14:22
davidgilukoh that's unusual14:23
pfalconI just forwarded such failure mail to linaro-dev@, it got thru %) (also cc: linaro-dev-owners). But cc:ing other parties I got such bounce 3rd time over last 2 days14:23
pfalconso I just wonder is anyone else saw this14:24
plarspfalcon: I got a bounce once, but it was because someone had typoed linaro-dev@linaro.org14:28
davidgilukpfalcon: I'd not cc'd anyone to linaro-dev recently, and I've not seen it - but the only thing that looks a bit odd in there to me is that the To: line looks like it's split onto multiple lines oddly - but that might be harmless; I don't know enough to know if that's legal14:28
plarspfalcon: I just fixed the address in the reply and resent it, works fine for me otherwise14:28
pfalconplars: well, I've been sending to @lists, I checked14:28
pfalcondavidgiluk: it started just yesterday, I did like that (well, had few people in to:) before, no issues14:29
pfalconok, I hope lool will take a look into that14:29
zachtib pfefferz: And the internet is mine ;)14:46
zachtibanyone know: is there a setting in minicom that will make backspace work nicer when talking to the pandaboard?14:51
loolpfalcon: could you mime-forward me the original message and the bounce?15:00
loolor bounce15:00
pfalconlool: forwarded both15:02
salgadotgall_foo, about bug 814671, does it happen every time you run that command?  I can't seem to reproduce it here15:40
ubot2Launchpad bug 814671 in linaro-image-tools "failed to run command `linaro-hwpack-install'" [Undecided,New] https://launchpad.net/bugs/81467115:40
Dr_Whosalgado: yup15:40
Dr_Whosalgado: maybe there was a bzr merge hickup ...  I don't have any local changes so I'd like to think it should "just work"15:41
Dr_Whosalgado: there is one other possibility ... I am trying to use a rootfs that was generated with the new cross live-build3 tool that I'm working on (I'm trying to validate the image) ... yesterday I had run into a situation were I could crash the tool if /etc/fstab didn't exist  ... which even then l-m-c configures that file 15:43
Dr_Whoanyway so it it possible that perhaps there's some other file that isn't quite configured right that's tripping it up15:44
salgadobut that's weird because l-m-c copies linaro-hwpack-install into the unpacked rootfs and then runs it with chroot15:45
loolsalgado: are you running linaro-media-create with a full path?15:49
loolsalgado: "~tgall/bzr/linaro-image-tools/linaro-media-create" is what's being run, outside of the checkout dir15:49
loolthe other possibility would be usr/bin not being in PATH, which would be weird15:50
salgadoDr_Who, lmc copies linaro-hwpack-install to chroot/usr/bin, so maybe that's not in the PATH of your binary tarball?15:51
salgadolool, yep, I was running with the full path as well15:51
Dr_Whosalgado:   umm why would the binary tarball need to be in the PATH when it's directly reference with a full path ?15:52
loolDr_Who: no we mean "/usr/bin" wouldn't be in your PATH15:52
loolthe command which is being run is something like "chroot /tmp/foo linaro-hwpack-install /tmp/bar"15:53
loolafter copying linaro-hwpack-install to /tmp/foo/usr/bin15:53
salgadoDr_Who, in the PATH after you chroot into the unpacked rootfs15:53
Dr_Whook .. but the tool would be responsible for setting up the path in the chroot generally speaking15:54
loolit should be inherited from your env and a sane PATH would have /usr/bin somewhere15:54
Dr_Whoset | grep PATH15:54
loolyeah it's there15:55
loolDr_Who: env | grep PATH returns the same thing?15:55
loolDr_Who: well could you strace -f -etrace=process or -etrace=file the whole thing?  use -o foo.strace to save the result to a file15:56
loolwe should see what is failing to run15:56
salgadoDr_Who, you were running it as root, weren't you?15:56
salgadodoes root have the same PATH?15:56
salgadoor is the above from root?15:56
Dr_Whothat was as root15:57
Dr_Whoso yes15:57
=== bambee_ is now known as bambee
Dr_Whosalgado: lool : ok posted16:02
salgado32470 execve("/bin/cp", ["cp", "./linaro-hwpack-install", "/tmp/tmp8MccpT/binary/usr/bin"], [/* 18 vars */]) = 016:04
salgado32473 execve("/usr/bin/linaro-hwpack-install", ["linaro-hwpack-install", "/hwpack_linaro-panda-x11-base_20"...], [/* 18 vars */]) = -1 ENOENT (No such file or directory)16:05
looland also relevant:16:05
lool32473 execve("/usr/local/sbin/chroot", ["chroot", "/tmp/tmp8MccpT/binary", "linaro-hwpack-install", "/hwpack_linaro-panda-x11-base_20"...], [/* 18 vars */]) = -1 ENOENT (No such file or directory)16:05
loolwhat's this /usr/local/sbin/chroot ding?16:05
looloh sorry, it's another line16:05
lool32473 execve("/usr/sbin/chroot", ["chroot", "/tmp/tmp8MccpT/binary", "linaro-hwpack-install", "/hwpack_linaro-panda-x11-base_20"...], [/* 18 vars */]) = 016:06
loolThe only weird thing I can think of, but you'll hate me for this, is a linux bug16:06
loolhttps://lwn.net/Articles/452138/ https://lwn.net/Articles/452141/16:06
Dr_Whowell this is on a maverick box16:07
loolDr_Who: what's your kernel?16:07
Dr_Who 2.6.35-28-generic16:07
loolok, then I guess it's unlikely to be kernel related16:07
looloh unless we fail to copy the exec bit for some reason16:08
loolno, that'd be EPERM, not ENOENT16:08
Dr_Whomight be helpful to keep the temp dir around so we can have a look inside16:10
loolDr_Who: oh hold on, which rootfs are you using?16:10
Dr_Whobut at this point that shouldn't matter16:11
looldo you have /bin/bash in it?16:11
salgadoI tried with that as well16:11
Dr_Whowell ext for the sd card 16:11
loolsorry, I meant rootfs tarball16:11
Dr_Whoyes /bin/bash is in the rootfs16:11
Dr_Whonp ... 16:11
Dr_Whobut there is every reason to suspect that something isn't configured correctly in the rootfs16:12
loolyes; like some lib missing or something16:12
loolbut we would see it in strace16:12
Dr_Whoremember this rootfs was cross built with multistrap and the new live-build3 cross changes so I wouldn't be surprised if something was goofed16:13
loolDr_Who: where can I get the rootfs tarball?16:13
Dr_WhoId have to upload it somewhere ...  16:13
* Dr_Who wishes he had an account on people.l.o16:14
loolDr_Who: Yeah, it's likely; you could add some sleep or "read" before the chroot call and see what's going on in the chroot16:14
loolDr_Who: ask for one?16:14
loolDr_Who: do you have a RT for this already?16:14
Dr_WhoI don't believe so  or maybe I've forgotten I even have one!16:15
Dr_Whoby gosh I do!16:15
Dr_Whook .. be about 15 minutes or so...  16:20
davidgilukDr_Who: Can you chroot into that world and just run a shell?16:20
Dr_Whodavidgiluk: well in order for the build to succeed it had to do that16:21
davidgilukDr_Who: You could edit the 1st line of linaro-hwpack-install to be #! /bin/bash -x   and then if it actually maanges to run the script (which from that trace I'm doubting) it should show you how far it got16:23
=== agreen is now known as agreen-away
* rsalveti lunch16:48
pfefferzkrtaylor, coming?17:03
pfefferzmounir, coming?17:04
zachtibpfefferz, you around?17:10
pfefferzzachtib, I will be, can I get back to you in just a little bit17:11
salgadodoanac, around?17:33
doanacsalgado: yep17:34
salgadodoanac, would you have a couple minutes to talk about prebuilt images?17:34
doanacsalgado: sure.17:35
salgado(that "pre-release build" email, to be specific)17:35
doanaci'm actually doing some hacking on that as we speak17:36
salgadodoanac, cool!  I missed the beginning of the thread, so maybe you could tell me a bit more of what you're doing?17:36
doanacsalgado: sure. I was asked to try and generate some images on the (i think new) server deneb.17:37
doanacbasically, I just created a small bash script that can generate these (nano and alip right now)17:38
doanacthen I tested the images to make sure they actually booted. (need more testing on this still)17:39
salgadooh, ok. does it do that for all hwpacks?17:39
doanacsalgado: it can (and its designed to extend). currently I'm just generating for overo, beagle, and panda17:41
salgadodoanac, also, is this to QA image building or something else?17:41
doanacsalgado: not sure I understand your question?17:41
salgadodoanac, I was wondering why you were asked to do this.  if it's just an automated QA of image building or something else (e.g. providing prebuilt images)17:42
davidgilukdoes anyone have an ARMv6 (preferably not t2 and not k) that they could try a small test on - I think the only thing I have is my phone and that's android and I don't have an android build env17:44
doanacsalgado: I think the goal was for me to generate pre-builts for 11.07 and make sure they worked. I wasn't sure where this would ultimately land, so I thought I'd try and make the script somewhat useful so it might be able to get used in the future17:44
salgadoI see17:46
salgadodoanac, is there anything I could help you with, or maybe something I could do to get us closer to having prebuilt images for everything?17:51
doanacsalgado: I think I have a handle on it now. Its pretty easy process to keep adding new stuff. I may ping you when I get them all in place to have you review the work though17:54
salgadodoanac, ok, cool. in case you haven't seen yet, https://wiki.linaro.org/Platform/Specs/11.11/PrebuiltImages describes how we were planning to do it17:56
doanacsalgado: can 2GB still hold the desktop image?17:58
doanacalso - i had to do 384MB for nano. Its grown beyond 256MB17:58
salgadodoanac, not anymore, apparently. plars reported a bug about it a couple days ago17:59
plarsdoanac: well, the bug was against lmc because of the default 2G image size, and another issue with leaking loop devices.  I don't know of any guarantees we have on image size, so I didn't file a bug about the images themselves, though I do wonder what caused them to grow18:01
plarsdoanac: perhaps rsalveti or someone from his team could answer that18:01
=== salgado is now known as salgado-afk
jamestunnicliffejames_w: salgado-afk: How realistic is it that we can get my merge reviewed this evening? Would hanging out in #infrastructure help so we can chat more quickly? I can paste the log into the merge request later so we have a record.18:33
suihkulokkiI guess we want these as part of LEBs18:37
davidgiluksuihkulokki: Might want to point those out to michaelh118:39
=== salgado-afk is now known as salgado
salgadojamestunnicliffe, I can do a final pass today, sure18:45
jamestunnicliffesalgado: I want to do a rename as well, but I am not sure if bzr will cope with both patches in flight18:46
jamestunnicliffesalgado: renaming 3 files and plumbing should be quick though18:46
rsalvetiplars: doanac: could be the live build 3 transition18:46
rsalvetiDr_Who: did you compare the image size with live build 2 against live build 3?18:46
rsalvetiguess it's the only real change we did that could affect the images18:47
salgadojamestunnicliffe, you can do these changes on separate branches.  I could review  the one with the code changes and once it's good to go you branch from it and do the renaming18:47
Dr_Whorsalveti: I checked them such that they were reasonably close, I wouldn't expect them to be exactly the same18:47
rsalvetisure, the issue it seems people are seeing is that something changed that they are bigger than the usual18:48
Dr_Whowhich image ?18:49
plarsDr_Who: ubuntu-desktop for sure, doanac said that nano is now 384MB though, so it could be all of them18:50
Dr_Whoplars: uh .. no : here's today's nano build : nano-n-tar-20110722-1.tar.gz22-Jul-2011 06:06 28M18:51
Dr_Whoplars: now not included in that is the hwpack ... if say you're using the panda-x11 hwpack,  that will add ALOT to the image size18:52
plarsDr_Who: yeah, not sure what he was using, just what he said earlier.  The panda (non-x11)+ubuntu-desktop is > 2GB now though18:52
Dr_Whoplars: well feel free to write up a bug ... tho I think this will ultimately land on the hwpacks ...  there was a note to the list about this jan/feb time frame .... 18:53
Dr_Whofeel free to assign to me and I'll do the homework on it18:54
james_wjamestunnicliffe, you are talking about the retry_broken_downloads branch?19:04
james_wjamestunnicliffe, and bzr will handle merging across the renames fine 19:04
jamestunnicliffejames_w: yep, that retyry_broken_downloads branch doesn't have the renames in. I am not sure how bzr will cope with merging another branch in that does the renames19:11
salgadojamestunnicliffe, I thought it was a final pass on the previous branch, but I just saw the retry-downloads one...  I'm wondering how important it is to have this feature in this release?  wouldn't it be better to delay it until the next one so that it gets more testing?19:11
jamestunnicliffesalgado: I have done a lot of testing on it today, but I don't mind putting it off.19:12
salgadojamestunnicliffe, what concerns me is the size of it, but given that it only changes fetch-image stuff it may not be a big deal19:13
jamestunnicliffesalgado: A lot of it was moving some functions out of one class into another. The quantity of new code isn't massive.19:13
salgadothe size of the diff and the fact that we only have a few days left19:14
james_wit's something new, so we don't really have to worry about regressions19:20
james_wso the question is, would fetch_image be more useful to people with or without this change?19:20
salgadowell, if it's more useful without this feature then we shouldn't consider it even for future releases19:25
Amaranthrsalveti: So the previous unity release doesn't build either :/19:28
AmaranthMay be able to fix this but don't know how I'm going to put it in bzr19:28
davidgilukright, that's my panda busy for the weekend - time to go19:29
jamestunnicliffesalgado: james_w: It is a rather useful feature to have, but not critical. It will avoid creating images with bad downloads.19:29
jamestunnicliffesalgado: james_w: Without it, if you quit half way through a download, it isn't marked as part done, and the program will assume it has been cached the next time you run it.19:33
Amaranthhrm, crap, old unity requires old nux19:34
Amaranthrsalveti: I don't even think it's worth trying to do a release if we can't update unity, if we have old nux and old unity the only change is to fix flickering transparency in compiz19:35
james_wjamestunnicliffe, you've tested the various scenarios?19:36
james_wjamestunnicliffe, what about the other stuff in here, like taking a date one day after today?19:36
jamestunnicliffeThe only se19:37
james_wsalgado, yeah, I meant more if it was likely to be buggy it would be less useful, not that the feature isn't useful19:37
jamestunnicliffejames_w: the onluy scenario I haven't is repeated downloading of bad signatures or files. It is simple enoug to download them to the cache and break a file, but I didn't have another server set up to try broken files on server.19:38
jamestunnicliffejames_w: the date stuff is just so I could recycle a well tested function that searches forward/back in the database from a given day, not including that day.19:39
james_wit doesn't look to intrusive, so I think we can merge this for this release19:43
james_wjamestunnicliffe, please submit smaller chunks in future to aid review19:44
james_wjamestunnicliffe, this should have been four or so submissions19:44
james_wone just rearranging the code and making no functional changes19:44
james_wanother doing the redownload stuff19:44
james_wand one or more for the unrelated things like the date stuff19:45
jamestunnicliffejames_w: Noted. I was aiming for a smaller change! Hopefully my aim will improve.19:46
salgadojamestunnicliffe, I rarely manage to do that, but whenever I see a change is getting too big I stop and split into multiple branches;  I think that's a lot more feasible than trying to guess the size and splitting the work before you start or doing it all at the end19:50
salgados/it all/the split/19:50
jamestunnicliffesalgado: good point. So far I have seen branches as indivisible. Clearly I need to take more care to split them up when they get large.19:52
salgadojamestunnicliffe, yeah, it takes getting used to it. lately I've been using stgit (as I'm working on patchwork), but the bzr-pipeline plugin (http://wiki.bazaar.canonical.com/BzrPipeline) is quite nice as well19:57
jamestunnicliffesalgado: that looks rather cool and definitely would have helped today. I will have to give it a try.20:12
doanac+1 for stgit20:17
rsalvetiAmaranth: what is needed to fix the flickering transparency issue?20:52
rsalvetiargh, why old unity requires the old nux?20:53
Amaranthrsalveti: in vertexbuffer.cpp (in the opengl plugin) the lines to set opacity, saturation, and brightness need to be changed to attrib.opacity / 65535.0f and etc20:53
Amaranthnux had a big API/ABI change for 1.020:53
rsalvetiargh :-(20:54
Amaranthrsalveti: http://git.compiz.org/~amaranth/mobilebling/commit/?id=c7a0a222b7b1f49d1f62532e7fc56072f6e885c520:54
rsalvetiAmaranth: so, let's do a stable release with this fix and updating the compiz bits, what do you say?20:54
rsalvetiAmaranth: we should keep both stable/development branches, as they are natty/oneiric20:54
rsalvetias it seems to be quite complicated to port unity from oneiric to natty20:55
rsalvetiAmaranth: and we can do a stable release, basically with this additional patch and also a dev release, with the branches you already have20:55
rsalvetithat are based on oneiric20:55
rsalvetiI believe that compiz can always be updated20:55
rsalvetiat least, what do you say?20:55
Amaranthrsalveti: Compiz can be always be updated, porting unity to any changes isn't hard21:04
Amaranthrsalveti: I don't have an oneiric arm install to test with right now21:05
rsalvetiAmaranth: so, based on the 11.06 release, do we need to any kind of update/bufixing for unity and nux?21:06
rsalvetianything important it'd be good to backport?21:06
Amaranthrsalveti: No, that was all syncing with upstream iirc21:06
rsalvetiAmaranth: maybe fixes from unity stable branch, but problem is that we would need to pick the fixes by hand21:07
Amaranthunity will probably just be a constant syncing with upstream from this point on, nux has at least one more bug to fix but will hopefully be getting merged upstream soon21:08
rsalvetiAmaranth: ok, cool21:08
rsalvetiAmaranth: so for 11.07 and 11.08 I believe we should still be using the same unity/nux packages we created for 11.06 when using Natty21:09
rsalvetiwe can keep updating at least the compiz packages21:09
AmaranthWell unity needs some tweaks for the new compiz21:09
rsalvetiand at the same time we can integrate your new branches against Oneiric21:09
rsalvetioh hell21:09
Amaranthbut just a VERSION file being created and removing the dist and distcheck commands from the CMakeLists.txt21:10
Amaranthat least I think, let me double check21:10
Amaranthrerunning the build now with 2011.06 nux21:11
rsalvetiAmaranth: cool21:11
AmaranthI think it'd be worthwhile to make whatever unity we use work with the latest nux too, for the next release21:14
rsalvetiAmaranth: yeah, true21:16
jamestunnicliffesalgado: I hope that merge request is nice and self contained :-)21:34
michaedwWho can educate me on Pango/Cairo/Pixman rendering using OpenGL ES2?21:43
michaedw(And on the Linaro work/goals in this area)21:44
rsalvetimichaedw: alf_ is your man21:54
rsalvetimichaedw: check the related blueprints at https://blueprints.launchpad.net/linaro-graphics-wg21:56
michaedwrsalveti: thanks!21:56
=== apachelogger is now known as birthdaylogger
michaedwalf_: I am specifically interested in using Pango/Cairo to do font rendering into Qt surfaces using GLES222:14
michaedwwith Qt back-ended on the GLES2 rendering engine and the eglnullws gfxdriver22:15
michaedwalf_: I may be in a position to help with GL_OES_texture_npot on OMAP422:18
rsalvetiAmaranth: any news with your build?22:31
Amaranthrsalveti: got some oddities going on with installing it22:31
Amaranthand my panda is running slower than normal22:31
michaedwalf_: note this: https://mail.gnome.org/archives/commits-list/2011-April/msg06160.html22:32
rsalvetiAmaranth: hm, will try here22:33
Amaranthhrm, it failed to load the plugins for some reason22:34
Amaranthoh, i think it just froze22:34
JesseBarkermichaedw, those extensions don't do exactly the same thing22:39
michaedwJesseBarker: I'm aware of that22:40
JesseBarkerAmaranth, rsalveti, I'm going to be afk for a while, send me email to let me know what's going on with the release...22:41
rsalvetiJesseBarker: sure22:41
michaedwJesseBarker: the reference indicates what the IMG_texture_npot extension can be used for22:41
michaedwJesseBarker: i. e., try the cogl paths that use npot textures, and see what breaks ;-)22:42
michaedwJesseBarker: I have the SGX library sources, and am investigating from the inside22:43
JesseBarkermichaedw, i see.  my recollection is that it's the min/mag stuff that's in there, but not the extra wrap modes.22:44
michaedwcomment claims:  "Can't have GL_NEAREST_MIPMAP_LINEAR or GL_LINEAR_MIPMAP_LINEAR minification filter on non-power-of-two textures"22:45
JesseBarkermichaedw, alf_ really is your guy for the questions you have, but it's quite late for him, so perhaps email...22:45
rsalvetijcrigby: around?22:49
rsalvetijcrigby: about the meta package, it's building for u8500 and s5pv310, even if we don't have the proper kernel packages22:50
rsalvetijcrigby: I remember you had problems with those packages in the past22:51
rsalveticaused by lack of proper support at upstream22:51
rsalvetido you know if they are in a better shape already? if not, we can just remove them from the meta22:51
michaedwthe comment is erroneous; trilinear minification does appear to be allowed on NPOT textures unless you disable it with a hint in powervr.ini23:03
michaedwbut there is code elsewhere that implements this:  "A non power of 2 texture is inconsistent if using a non-clamp-to-edge wrapmode"23:05
michaedwso the actual meaning of the IMG_texture_npot extension appears to be "can do trilinear mipmapping on NPOT, but cannot wrap NPOT with REPEAT or MIRRORED_REPEAT"23:06
Amaranthrsalveti: seems to work, so I need to make compiz, plugins-main, and unity tarballs23:12
michaedwrsalveti: can you explain why the Cairo GLES2 back end would want NPOT textures with REPEAT/MIRRORED_REPEAT wrapping?23:14
rsalvetimichaedw: not so sure, maybe Amaranth knows it, guess alf_ is the only one working at Cairo23:16
rsalvetiAmaranth: ok, great23:16
rsalvetiAmaranth: let me know when you're done, I'll package them and test23:16
Amaranthrunning compiz through distcheck now, already have a unity tarball23:17
Amaranthnice, distcheck runs all the tests sam added23:18
jcrigbyrsalveti, yes I added them but they those hwpack have never used the linaro kernel so if that is going to be the case then we should drop them23:18
jcrigbyrsalveti, also the name of the samsung soc should probably change to be the same as the arch name in the kernel23:19
rsalvetijcrigby: yeah23:19
jcrigbythat changed near the end of the .38 cycle23:19
rsalvetifor now let's skip them, and we ask for the LT about the upstream support later23:19
rsalvetiwhen it's useful at least, we can add them back23:19
jcrigbyrsalveti, ok next build of the meta will have them removed23:20
rsalvetifor now I don't have a hardware to test, so nothing we can do23:20
rsalvetiunless you want to spend some time on it23:20
rsalvetijcrigby: cool, thanks23:20
=== jconnolly is now known as jconnolly|away
jcrigbyrsalveti, I have a snowball but not sure how to use it, thats another thing I should chase down one-on-one at linaro event23:21
rsalvetijcrigby: yeah, better23:21
jcrigbyI don't think the samsung hw is available to many people yet23:21
michaedwcould somebody ls /var/db/fontconfig so I can see whether the fontconfig "arch" ought to be le32d4 or le32d8?23:28
michaedwer, probably /var/cache/fontconfig23:32
rsalvetijcrigby: yeah23:39
rsalvetijcrigby: next time you push new packages to the ppa, can you push the ~ppa ones to oneiric first and then also push with ~ppa~natty1 for natty?23:40
rsalvetijcrigby: this way we can already start seeing hwpacks for oneiric23:40
rsalvetijcrigby: just change the version, series and push, no other change :-)23:40
jcrigbyrsalveti, sure23:43
rsalvetiDr_Who: can you put the blocking issue at the whiteboard? https://blueprints.launchpad.net/linaro-ubuntu/+spec/linaro-platforms-o-cross-build-support-known-arch-live-build23:43
rsalvetijcrigby: great, thanks!23:44
michaedwthe correct answer seems to be le32d823:48
michaedwat least with my toolchain23:49
Dr_Whorsalveti: I'm unblocked now  ...  23:52
Dr_Whorsalveti:   granted I have a different mystery Im' trying to get to the bottom of but when you get the rest of the bug comments you'll see why23:52
Amaranthrsalveti: https://launchpad.net/unity-gles/2011.11/2011.0723:54

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