| twb | lilstevie: do you know if TF has a "multi-channel sound card" ? | 04:07 |
|---|---|---|
| twb | I'm afraid the answer is no, which makes me sad | 04:08 |
| twb | I tried using two mpg123 processes at the same time, the second one gave up | 04:09 |
| infinity | twb: Try running mpg123 under padsp so it passes through a software mixer? | 04:50 |
| twb | infinity: pulseaudio? | 04:55 |
| twb | I don't allow pulse on my systems :P | 04:56 |
| twb | I was asking because I'm trying to get emacspeak up and running again | 04:56 |
| infinity | twb: Heh. | 05:00 |
| twb | I can't actually get emacspeak to work at all yet | 05:05 |
| === doko_ is now known as doko | ||
| === k1l_ is now known as k1l | ||
| jamespage | hello - anyone around who can help me with this armel/armhf build failure - bug 920871 | 12:20 |
| ubot2` | Launchpad bug 920871 in zookeeper "zookeeper 3.3.4+dfsg1-2ubuntu2 FTBFS on armel and armhf" [High,Confirmed] https://launchpad.net/bugs/920871 | 12:20 |
| jamespage | not quite sure where to start :-) | 12:21 |
| jamespage | (have reproduced locally) | 12:21 |
| ogra_ | jamespage, does this help https://wiki.ubuntu.com/ARM/Thumb2 ? | 12:28 |
| jamespage | ogra_, yes thanks | 12:35 |
| rbasak | I'm having bus powered USB disk issues on the pandaboard. Should I just give up? Does this work for anyone else? | 13:35 |
| taruti | rbasak: I have an usb disk but it wants external power with the pand | 13:36 |
| taruti | *panda | 13:36 |
| rbasak | so doesn't work without? | 13:37 |
| rbasak | even though it in theory should? | 13:37 |
| taruti | not sure why, that is just my experience | 13:37 |
| rbasak | I'm giving the panda a beefy enough PSU I think. But perhaps a varying load is causing issues. I get combinations of kernel panics and disk spindowns | 13:37 |
| ogra_ | rbasak, the PSU TI shipped with my first panda is 5V 4A, what do you use ? i wouldnt go below 3A for hub powered USb disks | 13:39 |
| rbasak | 5V 4A, and I also have been trying a 200W AT PSU | 13:39 |
| ogra_ | well, that should work, whats the issue you see ? | 13:40 |
| rbasak | kernel panics | 13:40 |
| ogra_ | actually USB related ones ? | 13:40 |
| rbasak | and sometimes the disk keeps spinning on and off in a fast (maybe 0.8s) cycle | 13:41 |
| rbasak | I don't think the panics are USB specific | 13:41 |
| ogra_ | well why blame the disk then ? :) | 13:41 |
| rbasak | I suspect maybe the disk is causing the board voltage to drop, though I've metered it I won't be able to see short interval drops | 13:41 |
| rbasak | only this disk does it. my sata/usb docking station which is externally powered does not have the same issue | 13:41 |
| ogra_ | hmm, well, might even be a driver issue | 13:42 |
| ogra_ | i think there are plenty people running host powered usb disks on their pandas ... not that i have empirical data though | 13:42 |
| rbasak | OK I'm pretty sure it's a power issue. taruti gave me an idea - I connected the second bus power connector to a usb hub powered by my AT PSU instead of into the panda. And powered the panda from the power brick I have. Now it seems to be going OK | 13:43 |
| rbasak | Perhaps I need to make a "USB power boost" harness which supplies 5V to those secondary connectors outside the panda | 13:44 |
| rbasak | perhaps this particular drive model causes more noise on the usb power affecting the panda? | 13:46 |
| rbasak | I just bought the cheapest one | 13:46 |
| rbasak | it's an intenso 320GB | 13:46 |
| === Guest71075 is now known as panda|x201 | ||
| tedg | Hey folks. I was curious if someone could build the HUD feature for ARM and check to see if it's responsive. | 15:48 |
| tedg | It's very much so on my computer, so I haven't spent much time optimizing, but I'd like to ensure that's across platforms. | 15:49 |
| tedg | I'm not super worried, but I want to make sure :-) | 15:49 |
| ogra_ | if it is as responsive as the dash it will take 10-15sec to come up | 15:49 |
| tedg | ogra_, On Unity or UnityQt? | 15:50 |
| ogra_ | can it run in unity-2d too ? | 15:54 |
| ogra_ | (GLES support for 3D unity isnt merged yet, so we can only run under -2d) | 15:54 |
| ogra_ | -2d == Qt | 15:54 |
| ogra_ | :) | 15:54 |
| tedg | ogra_, 2d is Qt only until Qt5 is out, then they're all 3d :-) | 15:55 |
| ogra_ | we only have the Qt version on arm atm until the GLES patches from linaro get merged into unity | 15:55 |
| tedg | ogra_, I thought Unity could run on the Pandaboards? no? | 15:55 |
| tedg | ogra_, Is that with the ES patches? | 15:55 |
| ogra_ | right, they are a) not merged (waiting since the dublin sprint) and b) still have a few bugs | 15:56 |
| ogra_ | so all we can currently run proper is unity-2d (Qt) | 15:56 |
| GrueMaster | rbasak: Most USB powered drives require two USB connectors to get enough power, but unfortunately even then the panda doesn't deliver enough power from the usb ports. All of my USB<>SATA drives have separate power source, which isn't easy to find on 2.5" enclosures. | 15:57 |
| tedg | ogra_, Hmm, okay. Is that plan updated? I thought there was a plan to merge those branches discussed in Budapest. Have you talked to the Unity guys about that since then? | 15:57 |
| rbasak | GrueMaster: aha, that'll be my problem then. I've tried connecting the second connector to an external source but that only seem to work some of the time. Thanks! | 15:57 |
| ogra_ | tedg, once a week, yep. its in the process and hopefully ready by feature freeze | 15:57 |
| tedg | ogra_, Cool, I'll leave you to it then! :-) | 15:58 |
| ogra_ | tedg, btw, the HUD is public stuff now ? | 15:58 |
| * rbasak wonders what to do with this drive that's going spare now | 15:58 | |
| * ogra_ didnt see any blogpost yet | 15:58 | |
| tedg | ogra_, Yup: http://www.markshuttleworth.com/archives/939 | 15:58 |
| ogra_ | ah, k, had no time to follow planet since budapest | 15:59 |
| tedg | ogra_, But you're a voracious reader of OMG! Ubuntu!, right? ;-) | 15:59 |
| ogra_ | heh, rather planet than OMG :) | 16:00 |
| LetoThe2nd | k1l: didn't know you also hang out here ;) | 16:15 |
| k1l | ;) | 16:15 |
| GrueMaster | ppisati: Any chance of getting the omap4 kernel fixed or at least removed from the pool today? | 16:23 |
| rbasak | GrueMaster, AIUI ppisati sent a fix to the list today | 16:30 |
| ppisati | GrueMaster: yep, sent, waiting for the upload | 16:32 |
| GrueMaster | Ok. | 16:32 |
| ppisati | btw, do anyone know why the arguments for the kernel cmdline changed? | 16:32 |
| ppisati | dpkg -S /boot/boot.script says it doesn't belong to any package | 16:33 |
| ppisati | so it must some hand-crafted stuff coming from the installer | 16:33 |
| ppisati | *it must be | 16:33 |
| GrueMaster | The boot script is generated during install, not by any packages. For preinstalled, it is created by jasper. | 16:33 |
| GrueMaster | For netinstall, it is generated by f-k-i (I think). | 16:34 |
| ppisati | ok | 16:42 |
| ppisati | and do we have a changelog of it? why an arg was added/deleted? | 16:42 |
| GrueMaster | What arg was added/deleted? | 16:43 |
| ppisati | GrueMaster: mem=456M@0x80000000 mem=512M@0xA0000000 | 16:44 |
| ppisati | GrueMaster: these two are not of boot.script anymore and one of the feature of our kernel (CMA) craps out without it | 16:45 |
| ppisati | s/it/them/ | 16:45 |
| GrueMaster | That arg is only necessary when using the closed source video decoder. Shouldn't have any affect when running server/headless. | 16:45 |
| ppisati | it has with CMA | 16:45 |
| ppisati | without these args it doens't initialize the pool at boot, and when the usb ask for some DMA buffers, kaboom | 16:46 |
| ppisati | that's what you get yesterday | 16:46 |
| ppisati | i sent a revert of this CMA | 16:46 |
| ppisati | since it's not mandatory and it's not even in mainline | 16:46 |
| ppisati | but the camera driver needs it | 16:46 |
| ppisati | so | 16:46 |
| ppisati | we either drop the camera driver, or we revert these args | 16:46 |
| GrueMaster | Then that is a new change, as I have been running w/o those settings on all headless images since Maverick. | 16:47 |
| ppisati | yep | 16:47 |
| ppisati | it was bropught in with the last TI bits | 16:47 |
| ppisati | btw i had these args since i started working here | 16:47 |
| ppisati | that's why i didn't notice CMA panicng at boot | 16:47 |
| GrueMaster | As to the netinstall cmdline parameters, they were never added when netinstall was created. I'd have to look, but I'm not sure they are even in the server preinstalled images. | 16:48 |
| ppisati | uhm ok | 16:48 |
| ppisati | anyway the revert is on its way | 16:48 |
| ppisati | so... | 16:48 |
| GrueMaster | Those parameters create a hw memory hole so that a specific hw device can use that address range. Not sure why it can't be a bit more intelligent and request it on module load. | 16:49 |
| ppisati | perhaps because if you don't explicitely say so at boot time, the entire memory would be mapped (and couldn't be reclaimed later) i guess | 16:50 |
| GrueMaster | Ok, those parameters are on the server preinstalled (which makes sense as they are hardcoded in jasper). | 16:51 |
| GrueMaster | I'll bug NCommander to add them to the netinstaller when he comes online. | 16:52 |
| ppisati | no no | 16:52 |
| GrueMaster | But as I said, we were told they were only needed for the closed source decoder bits. | 16:52 |
| ppisati | i already asked for the revert | 16:52 |
| ppisati | these won't be necessary anymore after nex upload | 16:52 |
| GrueMaster | But is that what TI wants going forward? | 16:53 |
| ppisati | well, CMA is not even mainline | 16:53 |
| ppisati | so when it will hit mainline, we'll see | 16:53 |
| GrueMaster | If their patches are now enabling it, then we should make sure all images are consistent. | 16:53 |
| === apachelogger is now known as sadpanada | ||
| === sadpanada is now known as sadpanda | ||
| GrueMaster | (we should make sure netinstall is consistent anyways). | 16:54 |
| ppisati | i was wondering why linaro has no such problems | 16:54 |
| === sadpanda is now known as apachepanda | ||
| GrueMaster | They may have that parameter as default on their images. | 16:56 |
| ppisati | could be | 16:56 |
| ppisati | i tried to ask it to agreen | 16:57 |
| ppisati | but he didn't answer yet | 16:57 |
| GrueMaster | For now, if that is all that is required for netinstall to work, that is easy to fix (w/o reverting the patch). Netinstall can be preseeded to include those parameters during install and the netinstall image can be fixed to have them on boot. | 16:57 |
| ppisati | yes, that would fix it | 16:57 |
| GrueMaster | My concern now is that by reverting it, are we causing issues to TI's development work down the road? | 16:58 |
| ppisati | revert the patches? | 16:58 |
| ppisati | this is just for 3.2 | 16:58 |
| ppisati | and since TI won't support it/as, we stay as it is | 16:59 |
| ppisati | for 3.3 we will get it, and bunch of new drivers exploiting it (i guess) | 16:59 |
| GrueMaster | Then we are better off leaving the kernel as is and fixing netinstall. Less pain in the long run. | 17:00 |
| ppisati | GrueMaster: when you say the binary driver, do you mean the pvr-omap4 dkms? | 17:01 |
| GrueMaster | Not sure if it is that one or the gstreamer code (think it was gstreamer though). | 17:02 |
| GrueMaster | ogra_ would know. | 17:02 |
| GrueMaster | Or ndec. | 17:02 |
| ppisati | k | 17:03 |
| * ogra_ looks up | 17:03 | |
| ogra_ | ah | 17:04 |
| * ndec too | 17:04 | |
| ogra_ | the cmdline is created at buildtime, only the root= parameter is added by jasper | 17:04 |
| ogra_ | the memory hole is needed for ducati (the multimedia framework) | 17:04 |
| ndec | ppisati: GrueMaster: i saw the email on kernel list, i am not sure i understand what the proposed fix is | 17:05 |
| ppisati | ndec: the problem is in the new memocry allocator (CMA) | 17:05 |
| ndec | does it require the memory hole in bootargs? | 17:06 |
| ppisati | this CMA has never been in mailine, but we got it via tilt | 17:06 |
| ogra_ | GrueMaster, the netinst images should have that cmdline params too, if not, NCommander should add them at build time | 17:06 |
| ppisati | because the camera driver needs it | 17:06 |
| ndec | yes, it's in tilt because we use it for v4l2 camera driver. | 17:06 |
| GrueMaster | ogra_: Ok. I'll file a bug and assign to him. | 17:06 |
| ndec | and later for MM with syslink3.0 | 17:06 |
| ppisati | since the usb crap was due to this memory allocator, i reverted it and now everything works | 17:06 |
| ogra_ | GrueMaster, great, if he doesnt have the time, infinity or i can do it as well | 17:07 |
| ppisati | dunno why it doesn't init the pool at boot time if we don't pass the args stuff | 17:07 |
| ndec | the camera driver should not need it. or there is something else missing. | 17:07 |
| ppisati | that's what agreen told me | 17:07 |
| ppisati | but i handed a couple of kernels without this cma to rbasak today, he testd it | 17:07 |
| ppisati | and all the panicing on boot&c are gone | 17:07 |
| ppisati | so i pushed the revert since we are close to kernel freeze | 17:08 |
| ndec | i think the memory hole is just helping to hide another issue ;-) | 17:08 |
| GrueMaster | ogra_: iirc, we (NCommander and I ) had this discussion last cycle. Iirc, there was a problem getting others to accept the change. | 17:08 |
| ndec | I will look into this, i would prefer we make the real fix instead. | 17:08 |
| ppisati | ndec: that's why i don't want it in (i mean the CMA) | 17:08 |
| ndec | and our MM stuff with syslink3 will no longer require the memory hole anyways. | 17:08 |
| ppisati | but that's 3.3+ and we talking about a P kernel | 17:09 |
| ndec | ppisati: no, i think we should keep it, but you might be missing another config somewhere. | 17:09 |
| ndec | if you revert this one, you might want to revert the v4l2 camera driver as well, to ensure that nobody will attempt to use it. | 17:09 |
| ppisati | let me check one thing | 17:10 |
| ndec | on the other hand if you keep it, and if we fix it properly, then all the boards that have a camera (blaze have, and some panda do) will get support for camera out of the box with 12.04... | 17:10 |
| ppisati | ndec: does our kernel boot on that boards? | 17:10 |
| ndec | yep | 17:10 |
| ndec | sebjan maintains blaze support in tilt tree | 17:11 |
| ndec | ppisati: let me reply on the mail and let's try to sort it out. you can use the mem hole as a temp workaround. since it's really no longer needed moving forward, i would prefer to get rid of it... | 17:12 |
| * ogra_ wouldnt be opposed | 17:13 | |
| ogra_ | its one hack less we have in the image builds | 17:13 |
| ppisati | ok, but then remember to get the memory hole back in place | 17:14 |
| ogra_ | ppisati, i thought ndec just proposed to get rid of it | 17:14 |
| ppisati | ogra_: actually the opposite | 17:15 |
| ndec | ;-) | 17:15 |
| ppisati | ogra_: he wants to reintroduce it to workaround this problem | 17:15 |
| ogra_ | ppisati, anyways, we didnt change anything wrt kernel cmdline in precise yet | 17:15 |
| ppisati | ogra_: yep | 17:16 |
| ogra_ | its still there, hasnt gone anywhjere | 17:16 |
| ppisati | ogra_: it's just that this cma came via last kernel update | 17:16 |
| ogra_ | unless you used a netinstall | 17:16 |
| ppisati | ogra_: and my default boot.scr has those mem args in it | 17:16 |
| ogra_ | right | 17:16 |
| ppisati | that's whty i didn't notice the panic on boot | 17:16 |
| ogra_ | you will only see it on netinst since apprently NCommander didnt take them into account when building the image *or* in f-k-i | 17:17 |
| ogra_ | not sure where they are missing there exactly | 17:17 |
| GrueMaster | So, what is going to be the goal here? a) fix netinstall to include the hole, or b) fix the kernel cma to not need the parameter on the cmdline? | 17:18 |
| ogra_ | sounds like b) as the final goal and a) as the interim | 17:19 |
| ppisati | right | 17:19 |
| ogra_ | not sure how long b) will take | 17:20 |
| ogra_ | if its only a few days, changing netinst might be wasted time | 17:20 |
| ogra_ | GrueMaster, so the answer is "depends ... " | 17:21 |
| GrueMaster | Well, we can work around the netinstall issue for now, but I wouldn't recommend it long term. | 17:21 |
| GrueMaster | If the kernel can be fixed before freeze, then netinstall workaround is the way to go. | 17:22 |
| ogra_ | right | 17:22 |
| GrueMaster | If not, a more permanent netinstall fix is required. | 17:22 |
| GrueMaster | Ugh. Now I remember why it wasn't in the netinstall (or at least on my testing). u-boot cmdline is limited to ~256 characters. I'm getting ready to verify it now, but if that is still the case, it screws up automation in a huge way. | 17:26 |
| GrueMaster | Grrr. And netinstall doesn't seem to care about debian-installer/add-kernel-opts in the preseed. grumble. | 17:33 |
| ndec | GrueMaster: does latest 12.04 image boot on panda and es? | 17:40 |
| ndec | i was told (didn't try) it wouldn't boot | 17:40 |
| ndec | on es | 17:40 |
| GrueMaster | I just booted today's daily-preinstalled desktop image ok. | 17:41 |
| GrueMaster | And I have a couple of systems running netinstall now. | 17:41 |
| ndec | on panda ES? | 17:41 |
| GrueMaster | The desktop is running on my 4460 ES. Haven't seen any issues. | 17:41 |
| ndec | ok. thx | 17:44 |
| GrueMaster | I haven't done a "full" test, but desktop is installed and running. I'm currently sitting in unity2D with an open terminal. Couple of kernel warnings in dmesg (early during boot), but nothing serious. | 17:44 |
| === Quintasan_ is now known as Quintasan | ||
| ogra_ | bug 920618 | 18:22 |
| ubot2` | Launchpad bug 920618 in debian-installer "Setting "d-i base-installer/kernel/image none" fails to ignore installation w/o kernel" [Undecided,New] https://launchpad.net/bugs/920618 | 18:22 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!