=== GrueMaster_ is now known as GrueMaster
alextisserantdoes anyone know if progress has been made on bug 532733?02:10
ubot4Launchpad bug 532733 in qemu-kvm (Ubuntu Lucid) (and 2 other projects) "apt/dpkg in qemu-system-arm hangs if a big task is installed (affects: 4) (dups: 1) (heat: 36)" [High,Incomplete] https://launchpad.net/bugs/53273302:10
ubot4Launchpad bug 532733 in qemu-kvm (Ubuntu Lucid) (and 2 other projects) "apt/dpkg in qemu-system-arm hangs if a big task is installed (affects: 4) (dups: 1) (heat: 36)" [High,Incomplete]02:10
=== XorA|gone is now known as XorA
loologra, NCommander: poke15:37
loologra, NCommander: Do you guys have any specs which are pending approval?15:37
loolNCommander: I'd like to discuss softboot loader (mukluk) with you15:38
loolNCommander: Does mukluk actually need *discussion*, or does it need implementation?15:38
ogralool, i submitted two for me and one that persia didnt want to run the BOF for15:38
loologra: Could you list them here?15:38
ograjust digging15:38
persiamukluk mostly just needs implementation.  Not a lot to discuss.15:38
* ogra looks for the panel one15:39
loolhttps://blueprints.edge.launchpad.net/ubuntu-arm/+spec/userspace-loop-mounting 404s15:39
loologra: The spec doesn't say why you'd want to loop mount an image?15:40
ograyeah, someone (i think asac) renamed it15:40
asacogra: can you fill out summary? ;M)(15:40
loolarm-m-userspace-loop-mounting > I'd like this to be worded towards the use case we're trying to solve15:41
ogralool, editing images you mean ? and running build tools as non-root ?15:41
loolarm-m-lightweight-panel-for-efl > sounds relatively good (except typos :-) but do we have a replacement software, do we intend to write it?15:41
* ogra can rephrase it indeed15:41
asaclool: both are already in the summit system fwiw15:41
ogralool, i suspect a bit of both15:41
loologra: It could be: "Building images as non-root" or "Remastering images"15:42
ograi'D suppose to pick something existing and modify it15:42
loolasac: You mean scheduled?15:42
loolpersia: Any spec / session from you I should be aware of?15:42
ograsince if we want to use the indicators we need applet support and most of the small panel imaplementations that exist dont have it15:42
ograbut thats for the discussion15:42
loolasac: Do you know of any other spec/session from the mobile team which needs attention?15:42
asaclool: NCommander's softbooloader spec15:43
loolasac: Ok, I pinged him earlier on this15:43
loolI dont think it needs to be scheduled for UDS15:43
asaclool: i wanted to approve it, but it just was gone from the url it previously had15:43
ograright, there schouldnt be any others15:43
asaclool: yeah. we unscheduled it ... but he came back15:43
loolasac: Lets get this straight with NCommander15:43
asacand explicitly asked to have that session still ... i am indifferent15:43
loolI'm fine if this is targeted for maverick, but I want to understand the need for a time slot at UDS15:43
asacmy argument was that there was enough discussoin and we should rather implement15:43
loolasac: +115:44
ogralool, i'D also happily run https://blueprints.edge.launchpad.net/ubuntu-arm/+spec/arm-m-soc-i.mx51 and https://blueprints.edge.launchpad.net/ubuntu-arm/+spec/arm-m-soc-omap15:44
asacbut he said he wanted to solicit more input now that we have more professional attention on arm15:44
asacwhich is a valid argument imo15:44
loologra: What I'd love for arm-m-soc-i.mx51 and arm-m-soc-omap is if you could gather input from Freescale and TI15:44
ograand as i sadi before, any other one i subscribed to15:44
asacogra: those are vendor sessions15:44
loologra: Ideally, they'd run the session with what they want to achieve in maverick, or in the future, or give a status update on where things stand -- whatever15:44
asacboth are in the process to appoint a lead for that esssion ... but definitly subscribe you15:44
ogralool, though imx51 should probably be 5315:44
asacogra: imx5x15:44
ograyeah !15:45
asacogra: lets not change it now, because fsl has that url to think about15:45
loolasac: IMHO, softbootloader is a pre-production developers solution; not going to be deployed on production systems due to the boot time impact15:45
asacand then it would be gone (or are there automatic rediects)15:45
asaclool: right. it clearly falls in the developer (tools?) part15:45
loolasac: As such, I dont think it needs attention from another round of people; it should be one of these nice tools we use while we want to stay open in terms of boot options15:45
loolasac: Ack15:46
asacyeah. i think we are on the same page. lets make ncommander understand that its approved for maverick implementation, but that we probably dont want a session15:46
ogralool, description changed (for the loop stuff)15:47
=== jmcgee|gone is now known as jmcgee
* persia doesn't see why mukluk wouldn't be a production solution in the future: coreboot seems popular with many. Mind you, mukluk is insufficiently mature to be used now.15:48
ograpersia, its slooow15:48
ograimagine 10secs from u-boot plus 3 seconds for booting an initial kernel plus 5 seconds for second kernel and initramfs15:49
persiaogra: coreboot isn't.  The demo I heard about from a couple weeks ago wasn't.15:49
ograpersia, i was talking mukluk15:50
persiamukluk just isn't yet mature enough.15:50
ograand its slow by design :)15:50
loologra: I updated the description again15:52
loolpersia: One way we could have mukluk is as a fallback bootloader15:53
loolBut let's walk before we run15:53
ogracrawl rather :)15:54
loologra: Happy with the new description?15:54
ograyup, fine15:54
loologra: I think I will rename the blueprint as well15:54
ografine too15:54
loologra: what about lightweight panel?15:54
ograit was registered before we had ubuntu-arm or any naming scheme15:54
lool16:41 < lool> arm-m-lightweight-panel-for-efl > sounds relatively good (except  typos :-) but do we have a replacement software, do we intend to  write it?15:54
ogra<ogra> lool, i suspect a bit of both15:55
ogra<ogra> i'D suppose to pick something existing and modify it15:55
ogra<ogra> since if we want to use the indicators we need applet support and most of the small panel imaplementations that exist dont have it15:55
ogra<ogra> but thats for the discussion15:55
loologra: Is it really ARM specific?15:56
ogra(from above)15:56
loolI'm fine discussing it in the ARM track, but I wonder if it's going to impact desktop or whether they have differing plans15:56
ogralool, well, specific to efl using images with low RAM i'd say15:56
loolI'd hate ending up with one panel for ARM and one panel for rest or such15:56
ogranot so much for ARM as a hardware, but very much for the possible ARM hardware that might be supported in the future15:57
ograand i doubt desktop has *any* plans in that direction15:57
loologra: It doens't sound like a too good idea to have multiple implementations of the panel; moving parts, multiple code bases to maintain, ...15:57
persialool: certainly.  I'm not advocating mukluk is ready now, just that the concept shouldn't be rejected for production in a general way.15:57
loologra: So I think you should involve desktop and find a way which works for everybody15:57
ograwell, but you dont need the majority of things the current panel pulls into your memory15:58
loolNew name https://blueprints.edge.launchpad.net/ubuntu-arm/+spec/arm-m-image-builds-without-root15:58
ograat least on efl15:58
loologra: So if you dont need them with ARM, you dont need them for other hardware15:58
ograbut other hardware might have less speed/size constraints15:58
loologra: Could you make sure you discuss this with didrocks and the plans are agreed upon on the two teams (desktop and15:59
ograasac was initially proposing to actually enhance the efl launcher15:59
ograto be able to embed the bits and pieces we want15:59
ograpossibly thats a better idea, not sure15:59
loolI added didrocks as participation essential15:59
loologra: You might want mterry?16:00
looladded him16:01
sveinseA little OT: Do you know if there's there an ongoing effort to make a new (upgraded) version of the beagle board?16:01
persiaI can't imagine there isn't, but I'd recommend asking in #beagle16:01
ogralool, ++16:01
loolsveinse: as in hardware?16:01
loolsveinse: Google Beagleboard XM16:02
sveinselool, yes. Specifically one with SATA16:02
sveinseThanks, I'll check it out16:02
ograi dont think there are plans for SATA support16:02
ograbut there are several addon boards that might enable you to attach a self built SATA solution16:03
ograprpplague might be someone who can point you to any16:03
suihkulokkiomap has no sata so the only practical way for sata support would usb-sata bridge..16:03
* prpplague looks in16:03
loolNo SATA on OMAP316:04
XorAprpplague: see I told you guys was asking for SATA :-)16:04
sveinsejoin #beagle16:04
loolsveinse: The SoC doesn't provide SATA; the only ARMv7 one I know of is the Dove one16:04
prpplaguea standard PATA->SATA bridge can be added to the omap3 gpmc , problem is that the beagle doesn't bring it out16:04
sveinseInteresting though... Maybe TI has something like that in its pipe. The generic base board of my AM3517 eval kit (a Zoom2 kit) has a SATA connector. But for OMAP3 its NOOP :(16:06
prpplaguesveinse: interesting, i'd like to see the schematic on tat16:16
sveinseprpplague: It's just routed to some GPIO on the SoC.16:21
sveinseprpplague: This is the OEM mfg which makes the evalkit for TI: http://www.logicpd.com/products/development-kits/zoom-am3517-evm-development-kit16:22
prpplaguesveinse: hmm, the AM3517 has Sata ?16:22
sveinseprpplague: No. It does not. It only routed on the common baseboard which they obviously share between a number of SoCs16:23
sveinseprpplague: I was speculating if newer cores (OMAP4 perhaps?) had SATA16:23
kblinhey folks16:23
sveinseprpplague: ..since they have added it to the board I mean16:23
kblinrootstock is failing for me on a current lucid system16:24
prpplaguesveinse: oh so the base board has a sata connector but no sata support?16:24
kblinthe error message is "/usr/bin/rootstock: 792: Bad substitution"16:24
kblinwhich seems a bit weird16:24
sveinseprpplague: yes. The SoC is on a daughter board (which they name SoM) and then you plug it into a base board. The pins connected to the SATA is simply routed to some GPIO pins on the SOM. No PHY or anything16:25
kblinis there any way of telling rootstock to not delete downloaded files?16:26
prpplaguesveinse: ahh ok16:27
=== XorA is now known as XorA|gone
aboganiSorry for off-topic: Are there someone TI employee here?18:16
ojnabogani: you can find a bunch of TI folks at linux-omap18:17
ojn#linux-omap that is18:17
vstehleabogani: yep. May I help?18:17
aboganiojn: Thanks and sorry again :-)18:17
aboganivstehle:  May I disturb you for a simple question in private? (it is really off-topic here)18:18
vstehleabogani: Sure.18:19
rcn-eekblin, that 792: is a bashism i introduced, fix is waiting for review: http://bazaar.launchpad.net/~beagleboard-kernel/+junk/rootstock-bashism/revision/9618:33
loolNCommander: Hey you got our pings earlier?18:37
loolrcn-ee: ${KERNEL_IMG_REL%%_*} might be nicer (no fork)18:38
looland isn't a bashism18:38
loolecho ${KERNEL_IMG_REL%%_*}18:39
lool=> foo18:39
rcn-eecool, that's a lot nicer lool..18:39
rcn-eeyeah, that works sweet..  keeps learning new things..  lool, have you had any insights to the qemu lockup?18:43
loolrcn-ee: qemu lockup > I'm afraid not18:48
loolrcn-ee: I actually wonder whether anybody tried reproducing and then installing a bunch of dbgsym from apt and dpkg and getting a backtrace?18:48
loolthere's also the possibility of offering a gdb server in qemu (-s flag I think)18:48
rcn-eeokay.. what's weird, randomly 1/50 I'll get past that lockup and as soon as apt-get is almost done configuring it'll segfault...18:49
loolrcn-ee: I wonder whether it could be related to the memory in use18:49
loolrcn-ee: One thing to try would be adding swap18:50
rcn-eeyeap, tried both those.. added more swap, having qemu access a spare fast hardrive instead of image.. i was playing around with the apt-dbsym's jsut not sure how to change the apt-get call to work under gdb..18:51
loolrcn-ee: You dont need to change it18:51
loolrcn-ee: just attach to the hung process18:51
loolgdb /usr/bin/apt-get and then "attach 1234" where 1234 is the pid18:51
loolor hack rootstock to apt-get install the -dbgsym first, and run gdb apt-get -c 'set args install xyz; run'18:52
rcn-eeokay, that's what it is... "-c <stuff>; run"  I'll fire up another run.. (i have quick hack to get the dbsym's installed first already)18:54
loolrcn-ee: Sorry, it's -x with an input file, no -c (command) in gdb18:57
loolrcn-ee: Or --args18:57
* lool disappears18:58
rcn-eeI'm going to have to hold up anyways.. the ddebs's i used previously have disappeared off ddebs.ubuntu.com...18:58
rcn-ee(older versions exist)18:58
NCommanderlool: pings recieved w.r.t. to ARM softbootloader18:58
loolNCommander: Ok, so do you agree it doens't need discussion at UDS?19:10
loolNCommander: But just implementation in maverick?19:10
loolrcn-ee: Yes, ddebs are large, we only keep the last one19:10
NCommanderlool: I suspose. I think there are still points that could be discussed such as making it the default boot mechanism, etc.19:11
rcn-eeyeah: http://ddebs.ubuntu.com/pool/main/a/apt/ except it only has karmic's... lucid's is (which was there a week ago.) .;)19:11
=== hrw|gone is now known as hrw
NCommanderlool: BTW, be expecting another spec from me on improved ARM subarch detection so its something a bit less specific than the hardware field; I'll still drafted it however19:12
rcn-eeNCommander, btw any chance can we get an override, (for external kernels that share the same hardware id)?19:16
NCommanderrcn-ee: that's what we need to fix19:17
rcn-eeahh. cool..19:17
NCommanderrcn-ee: which board is this for?19:17
NCommander(and for lucid or maverick?)19:17
rcn-eepretty much any ti board i work on.. (specificly the beagle and beagle xm)19:17
NCommanderrcn-ee: I thought beagle XM won't boot a standard OMAP3 kernel (yet)19:18
NCommanderrcn-ee: can I PM you?19:18
rcn-eesure no problem... yeap it doesn't boot my or ubuntu's kernel but it does boot on angstrom psp.. although the ram settings are still iffy last i talked to jason and gerald...19:19
ograrcn-ee, did you forget to push your changed to the branch ?19:32
* ogra doesnt see a change19:32
rcn-eeogra, launchpad is error-ring out all morning today for me.. ;) (it's not refreshing any of my branches on the http interface)19:33
ograah, k19:34
ograthere was an update today and it was r/o for quite a while19:34
ogra(lp update)19:34
rcn-eeyeap... but i'll try pushing that last diff (lool's version) i emailed you to a branch, or you could apply it as is.. (for an sru do you need a bug report?)19:35
ograbug would be helpful but i can also file it myself19:36
ograoh, i missed that there was an attachmendt in the mail, sorry19:37
rcn-eelaughs... that's what happens when i quickly top post'd... ;)  heads back to work.19:39
loolNCommander: Please register the spec ASAP then20:02
=== hrw is now known as hrw|gone
=== sveinse is now known as sveinse_EE

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