| DanaG | Feb 15 23:02:44 beagleboard NetworkManager: <WARN> device_creator(): /sys/devices/platform/musb_hdrc/gadget/net/usb0: couldn't determine device driver; ignoring... | 05:07 |
|---|---|---|
| DanaG | argh, so I can't use NM on the onboard usbnet. | 06:10 |
| DanaG | anyway, I've set the beagleboard up to be dhcp server now, instead of client. | 06:11 |
| DanaG | That's consistent with what WinMobile does, at least. | 06:11 |
| === DanaG1 is now known as DanaG | ||
| lool | apw: Hey | 08:42 |
| lool | apw: So I discussed linux-versatile on ubuntu-mobile@ | 08:43 |
| lool | apw: See <20100211101825.GA5444@bee.dooz.org>; had a reply from ogra | 08:43 |
| lool | apw: I think it's helpful to provide udebs and a meta | 08:43 |
| lool | apw: Do you want a bug report? | 08:43 |
| lool | or two rather | 08:43 |
| apw | bug report yes please | 08:46 |
| lool | apw: lp #522516 and lp #522515 | 08:49 |
| ubot4 | Launchpad bug 522516 in linux-meta (Ubuntu) "linux-versatile meta (affects: 1)" [Undecided,New] https://launchpad.net/bugs/522516 | 08:49 |
| ubot4 | Launchpad bug 522515 in linux (Ubuntu) "linux-versatile udebs (affects: 1)" [Undecided,New] https://launchpad.net/bugs/522515 | 08:49 |
| lool | apw: Also, just an unrelated heads up on lp #522308 which I filed and probably went under the radar when moving to 2.6.32 | 08:50 |
| ubot4 | Launchpad bug 522308 in linux (Ubuntu) "linux-source-2.6.32 is empty (affects: 1)" [High,New] https://launchpad.net/bugs/522308 | 08:50 |
| apw | no i suspect i broke that when i did the abstraction simplification | 08:50 |
| Sleep_Walker | hello ubuntu people | 09:44 |
| Sleep_Walker | I'd like to ask about Sharp PC-Z1 Netwalker a bit | 09:50 |
| Sleep_Walker | does anyone here knows about this device? | 09:51 |
| lool | Sleep_Walker: Some people here do, yes | 09:51 |
| Sleep_Walker | 1] is it based on some babbage board? | 09:52 |
| lool | Babbage is the name of the reference design boards from freescale used for development purposes | 09:52 |
| lool | It's based on the same SoC, but it's a different board | 09:52 |
| Sleep_Walker | I see | 09:53 |
| lool | for instance it has no NIC, one USB port, no MIC jack so it's different from the babbage boards | 09:53 |
| Sleep_Walker | 2] is someone working on merging drivers to upstream Russel's kernel? | 09:53 |
| ogra | someone upstream imx51 drivers where appropriate | 09:54 |
| ogra | *upstreams | 09:54 |
| lool | Sleep_Walker: ATM, I am not sure that any imx51 SoC based board boots with an upstream kernel, so that would need to happen first; the kernel trees are hard to merge for various reasons | 09:55 |
| * ogra doesnt know how big the portion beyond general basic mx51 stuff is though | 09:55 | |
| lool | IIRC, there were some people working on getting at least the basic boot stuff working for imx51 upstream, but that's a long way to go until we get to things like wifi or video drivers... | 09:56 |
| amitk | Sleep_Walker: it will boot with a 2.6.34, I've pushed the base mx51 code upstream | 09:56 |
| Sleep_Walker | and where are these people gathered? (mailing list, irc, forum...) | 09:57 |
| amitk | but there's a LOT to be done, in case you're looking to help | 09:57 |
| Sleep_Walker | amitk: yeah, I had something like that in my mind... | 09:57 |
| Sleep_Walker | (if there will be time) | 09:58 |
| Sleep_Walker | I noticed your patches so I was curious | 09:58 |
| Sleep_Walker | unfortunately I don't know much about Netwalker's HW | 09:58 |
| Sleep_Walker | only that I was able to read from sources and from running system | 09:59 |
| amitk | Sleep_Walker: Just hang out on #ubuntu-kernel if you need some help and on linux-arm-kernel otherwise | 09:59 |
| persia | amitk: You pushed the Netwalker code upstream? | 09:59 |
| amitk | persia: Babbage base code (serial port, timers, clocks) | 09:59 |
| persia | Ah. Do you know if anyone is pushing the Netwalker patches? | 10:00 |
| persia | lool: There's actually two USB ports : one full-size, one mini. | 10:00 |
| Sleep_Walker | is it OTG capable? | 10:01 |
| Sleep_Walker | and can that be charged from USB? | 10:01 |
| amitk | persia: Netwalker is the pegatron stuff? | 10:01 |
| persia | amitk: Not at all (why does everyone think this?) | 10:01 |
| ogra | heh | 10:02 |
| amitk | persia: too many codenames... | 10:02 |
| persia | Sleep_Walker: The mini port looks like an OTG port, but I haven't tried to use it, actually. | 10:02 |
| amitk | with no context | 10:02 |
| persia | Yeah, well. Netwalker is it's own beast. There's kernel patches for the kernel that shipped, but I haven't heard about anyone porting them to newer kernels or pushing them upstream. | 10:03 |
| Sleep_Walker | persia: I tried to connect it with PC with no success | 10:03 |
| lool | persia: Well true, I meant one full size one compared to 4 on babbage boards | 10:03 |
| amitk | persia: is there a public git tree for the sources? Who did the kernel? What company made it? | 10:03 |
| lool | Both have mini-USB ports | 10:03 |
| persia | Sleep_Walker: I don't believe the USB Gadget driver is included in the shipping kernel: you might need to fiddle the kernel configuration. | 10:03 |
| Sleep_Walker | and that is the problem - I wasn't able to find any comunity about this device | 10:04 |
| persia | amitk: Not git, but there's a public apt-get source repo. | 10:04 |
| persia | (I heard the Netwalker kernel devs don't use git) | 10:04 |
| lool | amitk: netwalker is a sharp device | 10:04 |
| persia | Sleep_Walker: The community is almost entirely local. The device has seen very little adoption overseas. | 10:05 |
| Sleep_Walker | persia: I saw some SW support but I wasn't sure about HW support | 10:05 |
| persia | Sleep_Walker: Aside from accellerated audio/video codecs, everything seems to be open-source. | 10:05 |
| amitk | persia: and their kernel is based on freescale's SDP? | 10:05 |
| lool | http://www.ubergizmo.com/15/archives/2009/08/sharp_netwalker_unveiled.html | 10:05 |
| persia | amitk: I don't know precisely. When it comes to kernels, I'm just a user :) | 10:06 |
| persia | That's not quite right. It's 10 hours JEITA, which means 5-7 depending on load. | 10:06 |
| amitk | persia: if you can point me to the source, I could have a quick look when I have some time. | 10:07 |
| persia | Sharp just released a new "Dictionary edition" in the past couple weeks. The specs seem about the same (just different software load), but all the samples at the shop were password-locked, and I didn't want to try to hack them. | 10:07 |
| persia | amitk: Sure. Let me dig up what I have. | 10:08 |
| * persia fusses with apt sources | 10:08 | |
| persia | amitk: If you find it's not terribly hard to get up-to-date kernels, I'll give you one :) | 10:09 |
| persia | amitk: http://netbook-remix.archive.canonical.com/updates/pool/public/l/linux-fsl-imx51/linux-fsl-imx51_2.6.28-15.50fsl1araneo19.dsc | 10:10 |
| persia | That's from January. I'm unsure if there's a newer one on the new "Dictionary Edition" devices. | 10:11 |
| persia | But that's definitely enough for basic HW enablement. | 10:11 |
| Sleep_Walker | psi died - sorry - did I miss something? | 10:12 |
| persia | Sleep_Walker: I posted the URL to kernel sources for the Netwalker. Dunno if those are useful to you. | 10:15 |
| persia | Nothing else meaningful. | 10:15 |
| Sleep_Walker | persia: does it differ from the ones I get with apt-get source | 10:16 |
| Sleep_Walker | or from the one provided by sharp? | 10:16 |
| persia | Don't think you. You're getting linux-fls-imx51 2.6.28-15.50fsl1araneo19 ? | 10:16 |
| persia | s/you/so/ | 10:16 |
| persia | But I haven't checked the uname on the newest edition, so I don't know if there's a new kernel available from Sharp. | 10:17 |
| Sleep_Walker | ....araneo18 | 10:17 |
| Sleep_Walker | I don't think so | 10:17 |
| Sleep_Walker | and I thought I bought finaly device for work and not to hack :b | 10:18 |
| amitk | persia: I don't see any 28-15.50 at http://netbook-remix.archive.canonical.com/updates/pool/public/l/linux/ | 10:19 |
| persia | amitk: `dget http://netbook-remix.archive.canonical.com/updates/pool/public/l/linux-fsl-imx51/linux-fsl-imx51_2.6.28-15.50fsl1araneo19.dsc` should get you want you want. | 10:19 |
| persia | It's in the linux-fsl-imx51 subdirectory. | 10:20 |
| Sleep_Walker | ok, I'll try to create wiki page about Netwalker (sorry, but not ubuntu one) to gather informations and possible community around Netwalker | 10:21 |
| Sleep_Walker | thanks for all your help | 10:21 |
| Sleep_Walker | bbl from work | 10:21 |
| persia | Sleep_Walker: Please share the URL when you get it together. | 10:21 |
| ogra | Sleep_Walker, if you create one, can you point us to it so we can at least link it from the ubuntu wiki ? | 10:21 |
| Sleep_Walker | of course | 10:22 |
| Sleep_Walker | I don't want to split efforts around t | 10:22 |
| amitk | persia: so it's called erdos... | 10:23 |
| Sleep_Walker | I'lll put it on hackndev.com - distro neutral area :) | 10:25 |
| persia | amitk: erdos? | 10:25 |
| amitk | persia: the board is called erdos in the kernel tree | 10:26 |
| persia | Ah. | 10:26 |
| amitk | and from a 5s look, the patches that I pushed upstream ought to be able to boot on it (given that IO mappings are identical to babbage) | 10:27 |
| persia | So I should be able to boot a babbage kernel? I can test with lucid if you like. | 10:27 |
| Sleep_Walker | erdos is name of Netwalker's board in kernel? | 10:28 |
| lool | persia: I don't think so, the bootloader board id needs to match | 10:28 |
| persia | lool: That's a bootloader thing or a kernel thing? | 10:28 |
| amitk | persia: naah, it'll require a tweak or two, but the kernel should probably be 99.99% identical | 10:28 |
| lool | Plus the board support file will try to load drivers at various I/O addresses where you might miss devices or have other ones, for instance the netwalker has builtin wifi and not babbage etc. | 10:28 |
| ogra | persia, upstream, not lucid | 10:28 |
| * persia is *not* overwriting the nice dual-boot support redboot | 10:28 | |
| amitk | lool: we're talking about the upstream (minimal) babbage kernel | 10:29 |
| Sleep_Walker | you can do it on kernel side | 10:29 |
| persia | lool: wifi is through separate modules. | 10:29 |
| lool | persia: The kernel needs to grow a new netwalker board file with the proper board id and this file needs to be tweaked to list the proper devices | 10:29 |
| lool | amitk: I'm not sure which babbage kernel persia meant | 10:29 |
| amitk | lool: the board id is mapped to babbage | 10:29 |
| lool | exactly | 10:29 |
| * persia was kinda hoping for the lucid babbage kernel, but will trust statements that this doesn't work (didn't work with karmic kernel) | 10:30 | |
| lool | amitk: Oh you mean netwalker uses babbage's? | 10:30 |
| amitk | MACHINE_START(MX51_BABBAGE, "SHARP PC-Z1") .phys_io = AIPS1_BASE_ADDR, .io_pg_offst = ((AIPS1_BASE_ADDR_VIRT) >> 18) & 0xfffc, | 10:30 |
| ogra | which should be fine for bringup | 10:30 |
| lool | Hmpf | 10:30 |
| ogra | just not for all devices | 10:30 |
| amitk | lool: right | 10:30 |
| lool | amitk: isn't this ugly? | 10:30 |
| ogra | lool, did you expect beauty ? | 10:30 |
| lool | persia: it seems that a minimal upstream kernel would work with the babbage id then; nervermind | 10:30 |
| amitk | lool: tell me about it, they were too lazy to even get their own board id | 10:30 |
| Sleep_Walker | :b | 10:30 |
| lool | persia: I wouldn't try booting a full blown kernel though, that might blow things up | 10:31 |
| persia | lool: As in physical damage? | 10:31 |
| lool | I can't exclude that | 10:31 |
| amitk | persia: I think there is no danger with physical damage with the minimal kernel going upstream in 2.6.34. Since it keeps most IO pins to their defaults | 10:32 |
| amitk | and we're no where close to a full-blown kernel yet | 10:32 |
| lool | amitk: I think persia intended to use the lucid babbage binary kernel against a netwalker | 10:32 |
| persia | Ah. I think I'll wait then, since I use this daily as a handheld, and it also houses my lucid pbuilder environment :) | 10:33 |
| lool | Which I fear has a small chance of being dangerous | 10:33 |
| amitk | true | 10:33 |
| === Sleep-Walker is now known as Sleep_Walker | ||
| lool | apw: Thanks for the quick fix | 10:58 |
| apw | we get into trouble if the source is missing ... | 10:58 |
| NCommander | lool: ping. for ARM softbootloader, I need to have kexec-tools available for kexec, but the package unfortunately then changes the installed system to use kexec for rebooting as well which is unfortunate. I want to split the package out so I can have kexec installed and on its own without having the restart script stuff, any ideas on how to best do that? | 11:26 |
| lool | apw: Ah? I saw only three rbdeps in universe | 11:26 |
| lool | NCommander: the restart stuff is disabled in Ubuntu by default | 11:27 |
| apw | people complain when that package is empty as they percieve it contains the source, and if its empty we arn't publishing it, even though its in the the 'source' package | 11:27 |
| lool | apw: Ah so a lot of people get it wrong, eh | 11:28 |
| apw | yep | 11:28 |
| persia | apw: Now you just need to find a way to make `apt-get source linux` work :) | 11:30 |
| apw | persia, define work | 11:30 |
| persia | apw: heh. DWIM : download the source for the source package named "linux". | 11:30 |
| persia | Current trick is to use `apt-get --only-source source linux` | 11:31 |
| NCommander | lool: hrm, that must be a recentish change. Ignore previous ping then :-) | 11:32 |
| lool | NCommander: It's not | 11:32 |
| lool | It might be that you used the Debian package during the sprint | 11:32 |
| NCommander | lool: It still did the kexec-load in karmic. | 11:32 |
| NCommander | Which was the last time I looked at this | 11:32 |
| lool | This was disabled in June | 11:33 |
| * NCommander shrugs | 11:34 | |
| NCommander | lool: sorry for the noise | 11:34 |
| dmart | dyfet: ping | 11:41 |
| dyfet | *yawn* | 11:42 |
| dyfet | morning | 11:43 |
| dmart | Hi... wasn't sure if you'd be up | 11:43 |
| dmart | I had a question on gmp | 11:43 |
| dyfet | oh I remember that... | 11:44 |
| dmart | Did you try to build the asm code for Thumb-2 in the end? | 11:44 |
| dyfet | I had some trouble with coming up with a patch for configure. They reject using try_compile, and try to do everything by the gnu target architecture tags alone | 11:45 |
| dmart | Actually, I had an idea for that... you can maybe get the predefined macros out of GCC and munge that. I wrote some notes on the Thumb-2 howto wiki page. | 11:46 |
| dyfet | Oh, okay, cool! I did not notice that | 11:46 |
| dmart | It was late yesterday :) | 11:47 |
| dyfet | But that is kind of what I need to do for that one :) | 11:47 |
| dmart | However, if you do try to build this code for Thumb-2, we do need to check that the function symbols in the asm are properly tagged as function symbols, otherwise they would get called as ARM accidentally. | 11:48 |
| dmart | I think that the PROLOGUE() m4 macro used in the asm does this, but I didn't fully track down where it's defined. | 11:48 |
| dyfet | Ah.... | 11:49 |
| asac | great. we have 1 builder again ;) | 11:49 |
| asac | https://edge.launchpad.net/builders | 11:49 |
| dmart | Did you have >1 or 0 builders before? | 11:50 |
| asac | perfect timing in a3 week ;) | 11:50 |
| persia | dmart: 0 | 11:50 |
| dmart | (I'm guessing >1) | 11:50 |
| asac | dmart: this morning we had 0 ;) ... usually we have 7 | 11:50 |
| dmart | Oh, OK | 11:50 |
| dmart | What's the problem? | 11:50 |
| dyfet | For me, lack of coffee :) | 11:51 |
| asac | dmart: not sure. our is knows about it and are investigating. most likely the aweful pegatrons died again | 11:51 |
| dmart | Hum | 11:52 |
| * asac hopes for new build machines ;) | 11:53 | |
| dmart | dyfet: To check whether a symbol is a proper Thumb code symbol, you need to use readelf -s <object> | 11:54 |
| dmart | 6: 00000001 0 FUNC GLOBAL DEFAULT 1 f | 11:54 |
| dmart | Crucially, the symbol type if FUNC, and the value is an odd number (bottom bit set) | 11:54 |
| dmart | objdump helpfully masks of the bottom bit so as not to confuse you, so it's no good for this check :P | 11:55 |
| dyfet | This should be described on the wiki page too... | 11:55 |
| dmart | Yeah, I'll post it. (I was just figuring out how to check... | 11:55 |
| saeed | asac | 11:56 |
| asac | saeed: hi | 11:57 |
| saeed | hey | 11:57 |
| saeed | I want to install lucid img on dove | 11:58 |
| asac | right | 11:58 |
| asac | whats the prob? | 11:58 |
| saeed | http://cdimage.ubuntu.com/ports/releases/lucid/alpha-2/ has only imx images | 11:58 |
| asac | saeed: just pick latest daily | 11:59 |
| saeed | link? | 11:59 |
| asac | and yes. we didnt publish alpha2, because at that time we had severe issues with dove ;) | 11:59 |
| asac | saeed: http://cdimage.ubuntu.com/ports/daily-live/current/ | 11:59 |
| asac | http://cdimage.ubuntu.com/ports/daily-live/current/lucid-desktop-armel+dove.img | 11:59 |
| saeed | ok | 12:00 |
| persia | saeed: If /current/ doesn't work for you, there's often an archive of the past couple days which ought work if current doesn't. | 12:00 |
| asac | yeah. just navigate one up in the tree | 12:00 |
| asac | and you will find it | 12:00 |
| persia | (end the URL at .../daily-live/ to see the (short) archive list. | 12:00 |
| asac | but current should work afaict | 12:00 |
| persia | Usually does. | 12:00 |
| saeed | can you update me which issues still unresolved with dove | 12:01 |
| asac | saeed: i planned to do some thorough testing on dove this week ... afaik all bad issues are fixed with the X0 | 12:04 |
| saeed | great | 12:04 |
| asac | NCommander: any unresolved dove issues for saeed ? | 12:04 |
| NCommander | asac: saeed, X0 isn't here yet (I wasn't home yesterday to get the delivery) | 12:04 |
| NCommander | saeed: I did see the patch to fix kexec()'s decompression speed, thanks for the fast turnaround on that. | 12:05 |
| persia | saeed: I can't find a good list of *all* the issues with dove, but https://bugs.launchpad.net/ubuntu/+source/linux-mvl-dove probably includes a good chunk of them. | 12:05 |
| saeed | you're welcome | 12:05 |
| asac | NCommander: can you throw mrpt on one of your many babbage boards and check if that builds? | 12:06 |
| asac | NCommander: seems that package killed all our biulders | 12:06 |
| NCommander | asac: ugh. I'm still not setup on any of the boards, and may need to do a purchase order to get them all up | 12:06 |
| * NCommander is kinda still coming off unpacking from the long weekend | 12:07 | |
| asac | k | 12:07 |
| saeed | NCommander: when I tried kexec, the bootargs were not passed to the kexeced image | 12:08 |
| saeed | I've had to append all the command line in order to make it work properly | 12:09 |
| NCommander | saeed: I thought that's the way kexec is supposed to work but I'm not 100% sure | 12:18 |
| * persia is 100% sure : one may well want to have completely different bootargs from the bootloader and to the target kernel | 12:21 | |
| NCommander | dmart: is there a handy list of arm opcodes? I think I need to tear some code apart by hand without a disassembler | 12:32 |
| * NCommander feels like crying | 12:32 | |
| asac | NCommander: for ooo=? | 12:34 |
| NCommander | asac: uh huh :-/ | 12:35 |
| suihkulokki | NCommander: not handy but IIRC the only place to find opcode->instruction mapping is arm archictecture reference manual (aka ARM ARM) | 12:35 |
| * NCommander is trying to determine where this code is blowing up | 12:36 | |
| suihkulokki | alternatively, if you can run the binary under qemu linux-user, qemu-arm -d in_asm ./binary can give good insight | 12:37 |
| dmart | NCommander: really best to use a disassembler ;) (You could create an assembler file with the data words in it and disassemble that.) | 12:38 |
| dmart | If you really want to decode instructions by hand, you need to refer to the ARM ARM | 12:38 |
| NCommander | dmart, suihkulokki its just a buch of hexcodes in a C file. No specific binary to take apart ;.; | 12:38 |
| dmart | Ah, is this in the kexec implementation? | 12:39 |
| NCommander | dmart: OpenOffice | 12:40 |
| * NCommander is trying to track down where it explodes | 12:40 | |
| dmart | oh! Which file? I think I have that unpacked somewhere... | 12:40 |
| asac | its in the uno bridge | 12:41 |
| asac | we currently need to ship a jaunty .so fo rthat | 12:41 |
| asac | because otherwise it fails | 12:41 |
| asac | we have a binutils bug open for that | 12:41 |
| asac | iirc | 12:41 |
| NCommander | asac: its not clear that binutils is the issue | 12:42 |
| NCommander | the debugger breaks and cries though when you try and solve this, so I'm just scattering debug printfs | 12:42 |
| dmart | Can you point me to the affected file in OOo? | 12:42 |
| NCommander | dmart: we don't know specifically where its going bust | 12:43 |
| asac | NCommander: can you please give dmart the bug id ;) | 12:43 |
| NCommander | dmart: https://bugs.edge.launchpad.net/ubuntu/+source/openoffice.org/+bug/417009 | 12:43 |
| ubot4 | Launchpad bug 417009 in openoffice.org (Ubuntu Karmic) (and 3 other projects) "all openoffice apps die in 'com::sun::star::ucb::InteractiveAugmentedIOException' on armel in karmic (affects: 1)" [Low,Won't fix] | 12:43 |
| asac | bug 436617 | 12:44 |
| ubot4 | Launchpad bug 436617 in binutils (Ubuntu Karmic) (and 2 other projects) "ARM unwind table linker processing broke OO's uno2cpp (affects: 1)" [High,Won't fix] https://launchpad.net/bugs/436617 | 12:44 |
| asac | i think thats the bug | 12:44 |
| asac | dmart: NCommander: ^^ | 12:44 |
| dmart | thanks | 12:44 |
| asac | hmm ... the sata disk i have is really really slow here for imx51 | 12:45 |
| * asac break before meeting | 12:45 | |
| dmart | NCommander: Is this any help? http://pastebin.ubuntu.com/377573/ | 12:52 |
| NCommander | dmart: I'll play with it in a moment | 12:52 |
| dmart | It should allow you to decode invidual opcodes | 12:52 |
| * NCommander is making some headway | 12:52 | |
| dmart | ok | 12:53 |
| asac | dmart: the links to the quick references in the asm intro you posted dont exist | 15:11 |
| asac | like http://www.arm.com/pdfs/QRC0001H_rvct_v2.1_thumb.pdf | 15:11 |
| asac | dmart: hmm for libv4l compiler wth thumb2 complains about cbnz r5, .L2 ... with "jidctflt.s:74: Error: branch out of range" | 15:16 |
| asac | i found a quick refernce and that refers to CBNZ as T2 :/ | 15:16 |
| asac | with -marm it doesnt fail | 15:17 |
| asac | cat jidctflt.s | pastebinit | 15:19 |
| asac | http://pastebin.com/f350081ab | 15:19 |
| asac | thats the full asm generated | 15:19 |
| asac | cc -Wp,-MMD,"jidctflt.d",-MQ,"jidctflt.o",-MP -c -I../include -I../../../include -fvisibility=hidden -fPIC -DLIBDIR=\"/usr/local/lib\" -DLIBSUBDIR=\"libv4l\" -g -O1 -Wall -Wno-unused -Wpointer-arith -Wstrict-prototypes -Wmissing-prototypes -o jidctflt.o jidctflt.c | 15:19 |
| asac | /tmp/cctyjob8.s: Assembler messages: | 15:19 |
| asac | /tmp/cctyjob8.s:74: Error: branch out of range | 15:20 |
| asac | thats the error | 15:20 |
| dmart | asac: Thumb-2 has different range limits for some instructions. If cbnz can't branch far enough, you may be able to move the branch destination closer, or recode using cmp <Rn>, #0 // bne (I think) | 16:09 |
| dmart | Hmmm, actually CB(N)Z is Thumb only | 16:12 |
| NCommander | saeed: ping? | 16:25 |
| NCommander | saeed: I just got my X0, it won't boot; kernel hangs at Uncompressing; the X0 I used in Portland worked just fine with our existing images | 16:25 |
| asac | dmart: hmm. i think that .s is generated by gcc | 16:27 |
| asac | i only get that with -save-temps | 16:27 |
| asac | (already found that range thing in the quick reference) | 16:27 |
| asac | dmart: yeah. i think for -marm its probably not generated by gcc at all | 16:28 |
| asac | will produce a .s with -marm and compare | 16:28 |
| dmart | Oh, right. That's a compiler bug then. Can you raise a launchpad bug on gcc and stash the preprocessed source there? | 16:28 |
| dmart | The compiler should not generate out-of-range branches in its own code... | 16:29 |
| asac | dmart: yep | 16:29 |
| asac | will give you a bugid when filed (once off the call) | 16:29 |
| dmart | I posted info on the porting wiki page showing where to find the up-do-date instruction set quick references btw (in case you didn't already find them) | 16:29 |
| dmart | thanks | 16:29 |
| asac | dmart: oh on top of the asm intro? | 16:31 |
| asac | good | 16:31 |
| * asac checks | 16:31 | |
| dmart | yes | 16:31 |
| asac | found | 16:31 |
| NCommander | plars: GrueMaster, I'm reminded of what happens when the bootloader machine id and the kernel machine id fail to match :-/ | 16:39 |
| asac | dmart: bug 522717 | 16:40 |
| ubot4 | Launchpad bug 522717 in gcc-4.4 (Ubuntu) "libv4l code compiles to invalid asm: jidctflt.s:74: Error: branch out of range (affects: 1)" [Undecided,New] https://launchpad.net/bugs/522717 | 16:41 |
| dmart | asac: Can you attach the preprocessed source? This makes it easier for the compiler guys to reproduce the problem. | 16:42 |
| asac | right ;) | 16:42 |
| asac | done dmart | 16:44 |
| dmart | asac: cool, thanks | 16:47 |
| saeed | Ncommander | 16:48 |
| NCommander | saeed: ah, your around! | 16:51 |
| saeed | what it's the board rev? | 16:52 |
| plars | saeed: mine is 1.4, same problem | 16:53 |
| saeed | NCommander, please try the patches I just sent you be email | 16:54 |
| plars | GrueMaster: did you see the posting that just came across ubuntu-qa? how does that relate to the libtest stuff you've been doing? any help at all? | 16:55 |
| NCommander | saeed: will try as soon as I can | 16:58 |
| saeed | ok | 16:58 |
| GrueMaster | just a sec. | 17:02 |
| GrueMaster | plars: which channel? | 17:10 |
| plars | GrueMaster: ubuntu-qa mailing list | 17:10 |
| GrueMaster | Forward to me. I don't appear to be on that list. | 17:12 |
| plars | sure | 17:15 |
| GrueMaster | It might be useful. It does more api level testing, whereas the test suite I am working with does more low level testing (like at the fpu level). | 17:26 |
| GrueMaster | It is also very new. | 17:27 |
| GrueMaster | Wiki is dated this month. | 17:27 |
| GrueMaster | plars: A good read on the different test suites would be http://ispras.linux-foundation.org/index.php/LSB_Tests. | 17:31 |
| GrueMaster | It lists the different types of tests and compares them. | 17:32 |
| plars | GrueMaster: cool, will take a look. I was mostly just wondering if that one would also be useful | 17:32 |
| plars | or if it added anything really | 17:32 |
| GrueMaster | My understanding is that the tests mentioned in the email are essentially smoke tests. They will quickly tell you if there is a problem with a library function. What they don't give you is an underlying understanding of the problem (i.e. is it a toolchain issue, hardware issue, etc). | 17:35 |
| asac | anyone can install netbook-launcher and see if it fails to start? | 18:00 |
| asac | the 3d one | 18:00 |
| asac | idea is to understand if we need another probing on top ... or can just rely o nthat failing to determine if we want to go for 2d | 18:00 |
| ogra | asac, why dont we have the compiz wrapper in netbook-launcher ? that works pretty relaibly | 19:11 |
| asac | ogra: thats too slow | 19:17 |
| asac | for une | 19:18 |
| asac | takes 2 seconds or something on boot time | 19:18 |
| ogra | oh, i wasnt aware it takes 2sec | 19:35 |
| JamieBennett | GrueMaster: https://wiki.ubuntu.com/MobileTeam/Meeting/2010/20100223 | 19:53 |
| GrueMaster | Cool. Send me that link on 20100222 and I'll be good. | 19:54 |
| JamieBennett | :) | 19:54 |
| ojn | Wow, i.MX51 cell phones announced? Not having POP in that form factor must hurt. | 20:01 |
| ojn | (yeah, off topic, I know :) | 20:01 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!