=== JaMa is now known as JaMa|Zzzz === bjf is now known as bjf[afk] [05:57] ericm|ubuntu: oh, you're reading lkml [06:00] ericm|ubuntu: and I would welcome some common practise, Linus, too, I guess [06:27] ukleinek, replied [06:43] ericm|ubuntu: oh, missed the __initdata [06:44] * ukleinek is away, the children are up, so time for breakfast [07:20] ericm|ubuntu: what do you need .dev_name for? [07:20] ukleinek, nothing - just to distinguish between dev_name and drv_name :-) [07:23] ukleinek, I hope in the end I can use dev_name only - when all platform driver accepts a list of supported platform_device_id[] === JaMa|Zzzz is now known as JaMa [07:55] ericm|ubuntu: why is platform_device_id[] better than a dev_name? [07:56] ericm|ubuntu: just to let you know, this is how I register devices on imx (currently) [07:56] http://article.gmane.org/gmane.linux.ports.arm.kernel/83340 [07:57] http://article.gmane.org/gmane.linux.ports.arm.kernel/83342 [07:59] 83285 is more interesting than 83342 though [08:10] ericm|ubuntu: regarding "Introduce 'struct machine_class' for SoC level abstraction", do you really want to have boot_params in struct machine_class? I didn't recheck, but isn't that legacy cruft? [08:11] (i.e. for platforms not passing r2) [08:24] ukleinek, well - I don't like that either - but for backward compatiblity consideration, it can be removed later as well === hrw|gone is now known as hrw [08:35] morning [08:36] Morning hrq [08:36] Grrrr [08:36] Morning hrw! [09:58] ogra: hi [10:29] cooloney, does the new omap4 package have the videoram fixes from TI so we can actually boot with a screen ? [10:31] ogra_cmpc: oh, no, [10:31] ogra_cmpc: is there any bug about that videoram issue? [10:32] ogra_cmpc: i can talk with sebjan about that [10:33] cooloney, there isnt a bug, i think sebjan's tree has the fix though, vram needs to be 32M instead of 8 [10:33] my first gfx card had 2MB ram... [10:34] well, your first gfx card dint have three different outputs i guess :) each needs 8M [10:34] yep - vga only [10:34] and I used 1152x864-16 with it [10:34] wow [10:35] my first gfx card didnt do more than 640x480 [10:35] and that was an expensive one back then [10:35] ogra: it was 2000 [10:35] pci ati mach64 card which was quite obsolete at that time [10:35] yeah, mach64 cards didnt exist when i started [10:36] * ogra_cmpc isnt even sure ati existed back then [10:36] ogra_cmpc: http://kernel.ubuntu.com/git?p=roc/ubuntu-maverick.git;a=commitdiff;h=5c5862e6dfc3a46e72d82c46ea8534b38c4ac3a8 [10:36] ogra_cmpc: is this commit which will fix this issue? === JaMa is now known as JaMa|Wrk [10:36] i am not sure about that, since there is no bug information in the commit log [10:37] ogra: at same time I had Hyundai 386sx/25MHz with ati onboard graphics. but we used only 720x480 mode for text console with it [10:38] cooloney, hmm, afaik it was just a config option [10:39] ogra: http://kernel.ubuntu.com/git?p=roc/ubuntu-maverick.git;a=commitdiff;h=984ab103c27b1811e05a60e0a1a7129fed3b5250 [10:39] this is the only config change in the commit [10:39] weird [10:39] but i failed to see any videoram fixes [10:39] there must be a default for VRAM [10:39] ogra: so the testing kernel fixed that? [10:39] setting that to 32M should fix the display issues [10:39] i havent tested it yet [10:40] ok, [10:40] setting vram=32M on the cmdline gives me HDMI with all kernels though [10:40] that's a very useful info [10:40] looking at apt-cross code makes my head explode [10:42] cooloney: I planned to set the VRAM to 32MB by default in next kernel sometimes this week (not included in the patches you pulled today) [10:43] sebjan: ok, got it. from ogra's info, HDMI issue can be fixed by that [10:43] yes, the vram size can be overriden through the command line [10:53] sebjan: sorry, i am not debugging on it, it seems it doesn't work on lag's side [10:53] cooloney, TI said there might be issues with some monitors [10:53] http://www.amazon.co.uk/LG-W2261VP-inch-LCD-Monitor/dp/B0028KGKJA [10:53] This is the one I'm using [10:53] funnily ndec said specifically with samsung ... my samsung works fine though [10:53] Fairly standard [10:54] ogra: What do you see on your monitor? [10:55] ogra: mine is viewsonice HDMI 1080P, but don't have hardware to test, -:< [10:55] teh kernel messages and some plymouth console errors (as i said my rootfs doesnt work) [10:56] i'll try with a proper rootfs later today [10:56] At least you see something [10:56] What kernel are you using? [10:56] currently i'm busy getting the omap3 images working at least [10:56] Linux version 2.6.34-900-omap4 (buildd@hubbard) (gcc version 4.4.4 (Ubuntu 4.4.4-4ubuntu1) ) #1-Ubuntu SMP PREEMPT Fri Jun 18 23:51:15 UTC 2010 [10:57] Same as me [10:57] the latest archive kernel [10:57] That sucks :( [10:57] did you try booting without setting a serial console on cmdline [10:58] (your paste chopped off the cmdline string) [10:58] setenv bootargs root=/dev/mmcblk0p2 rootwait ro mem=463M console=ttyO2,115200n8 vram=32M; mmcinit 0; fatload mmc 0 0x80200000 uImage; bootm 0x80200000 [10:58] Fail [10:58] err [10:58] indeed [10:59] console=ttyO2,115200n8 and no other console= option forces serial only [10:59] try dropping all console= options for a start [10:59] that should default to tty0 [10:59] ogra: yeah, i agree [11:12] setenv bootargs root=/dev/mmcblk0p2 rootwait ro mem=463M vram=32M; mmcinit 0; fatload mmc 0 0x80200000 uImage; bootm 0x80200000 [11:12] Fail [11:12] try setenv bootargs root=/dev/mmcblk0p2 rootwait ro mem=463M console=ttyO2,115200n8 console=tty0 vram=32M; mmcinit 0; fatload mmc 0 0x80200000 uImage; bootm 0x80200000 [11:12] thats what i used with my last boot [11:13] if that doesnt work, tell ndec that LG monitors have issues too [11:13] Tried that already - Fail [11:13] I think he would have heard you [11:13] ndec: ACK -^ [11:17] * ogra ponders buying something like http://www.mp3car.com/vbulletin/mp3car-blog-talk/141162-hardware-review-lilliput-669gl-70np-c-t-7-hdmi-monitor.html for travelling [11:21] ogra: to car or for desk? [11:21] for using my omap boards while travelling [11:21] its small enough to fit in a laptop bag along with beagle or panda [11:22] and needs 12V [11:22] comes with power supply [11:22] so you need power socket to use it [11:22] indeed [11:22] dpkg-shlibdeps: error: couldn't find library libc.so.6 needed by debian/libgcc1-armel-cross/usr/arm-linux-gnueabi/lib/libgcc_s.so.1 (ELF format: 'elf32-littlearm'; RPATH: ''). [11:22] shit [11:25] It's more expensive, but: http://www.cinema5d.com/news/?p=3063 [11:26] heh, $600 [11:27] As I said ... [11:28] hrw: Yeah, same thing as last week [11:28] hrw: Did you make progress on this issue? Did you chat with slangasek about it [11:28] Probably better to buy a 12v battery pack - not sure how long it would last though [11:28] It's quite a subtle issue, and resolution is not easy [11:29] lool: I just got hit by that [11:29] lag, well, i'm more intrested in having a display when sitting in some hotel room at a conference or sprint [11:29] beyond that the lilliput has a touchscreen too :) [11:29] hrw: Isn't that the same thing as last week, where we discussed shlibs and building a fake shlibs package? [11:29] lool: no [11:29] It looks good, and I'm sure you'd be able to power it if you really wanted to [11:30] lool: thats I have in chroot which has all cross libs installed [11:30] hrw: I dont see the difference [11:30] (by batteries I mean) [11:30] yeah [11:30] hrw: Do you have binutils-multiarch installed? [11:31] yes [11:40] * ogra fires off an omap3 livefs build ... [11:42] lool: I found a bug [11:42] side effect of merging cross rules === rsalveti_ is now known as rsalveti [12:24] Has anyone seen this before? http://paste.ubuntu.com/456338/ [12:30] ogra -^ [12:31] lag, nope [12:31] but i havent tried the kernele from lucid-proposed [12:31] (which you seem to use) [12:32] https://edge.launchpad.net/ubuntu/+source/linux-ti-omap/2.6.33-502.8 doesnt really look like there was something added to it that could cause it though [12:34] I'm using the kernel from git://kernel.ubuntu.com/ubuntu/ubuntu-lucid.git [12:34] Is that proposed? [12:35] well, 502 is in proposed [12:35] http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-lucid.git;a=shortlog;h=refs/heads/ti-omap [12:36] 501 was a security update i think [12:36] 500 was the release kernel for lucid [12:36] your paste has 2.6.33-502-omap [12:38] Yes, that's what it says [12:39] all i can say is that the 500 kernwel worked fine for me [12:39] no idea about the security or proposed versions [12:39] Okay, I'll try and find someone who has built the latest kernel [12:39] Thanks ogra_cmpc [12:40] to be honest i dont really look at lucid atm [12:42] * ogra_cmpc sighs about evolution-data-server being out of sync and breaking the images [13:13] hi [13:17] using rootstock I created an ubuntu armel rootfs. most stuff works ootb. also networkmanager sets up my usb ethernet device correctly with dhcp. but there is one problem [13:17] only root can use networking [13:17] I setup my 'user' ubuntu with all privilges seleectable in the gui but it didn't fix it [13:18] what could be the cause of the problem ? [13:24] ogra_cmpc: Can you email me your Panda kernel which you have HDMI working on please? [13:27] lag, its the plain archive kernel [13:27] As is mine I think [13:28] But I'd like to double-check [13:28] I'd like to check all avenues before I say "it's an LG monitor issue" [13:29] how close is qemu-arm-static to the hardware platform ubuntu is compiled for, especially the thumb2 stuff? [13:44] lag, http://people.canonical.com/~ogra/panda/ [13:45] Thanks ogra [14:01] cooloney: I tested your image: it boots on my 2 boards. However, the smsc95xx.macaddr parameter does not have any effect, and I can't understand why... (this is the last patch that you pulled) [14:05] ogra: That doesn't work either. It must be a problem with LG monitors [14:05] yeah, i suspected that after ndec's comment last week [14:05] seems some EDID's cant be read properly by the driver [14:06] That makes sense [14:06] It must only be a parsing issue though? No biggy? [14:10] http://people.canonical.com/~hrw/ubuntu-maverick-armel-cross-compilers/ updated to recent compilers [14:12] rcn-ee, <3 [14:13] lag, probably [15:41] bleh === bjf[afk] is now known as bjf === ian_brasil_ is now known as ian_brasil [17:03] ogra, do you know anything about x-loader and company? [17:04] cwillu_at_work: anything in particular you would like to know? [17:05] prpplague, background: I'm trying to rig up an sd card to update nand and poke at a couple odds and ends on a zippy board if and only if the user button is held down [17:05] so, I need an x-load that both initializes i2c correctly so that the zippy will work, and which will pull u-boot off nand rather than mmc [17:06] apparently the place to set that is in include/configs/omap3530beagle.h [17:06] CFG_CMD_MMC 0 [17:06] except, that doesn't make any difference [17:07] it's reading the x-loader off nand (verified with the datestamp), but the option just has no effect [17:07] if I boot without the sd card in, it loads u-boot off nand correctly, so it's not that I don't actually have a usable u-boot on there [17:07] i.e., I could just kill off MLO from the sd card, and things would work [17:08] but I'd lose the ability to use the user button to trigger updating x-load and u-boot to nand [17:08] * prpplague reads through and tries to understand your targer [17:08] target [17:09] I don't get what I'm doing wrong though; as far as I can tell from reading this, setting CONFIG_MMC to 0 should completely remove the code to even attempt to load u-boot from mmc [17:09] but yet it _still_ does it [17:11] cwillu_at_work: hmm, interesting situation [17:12] cwillu_at_work: i'd have to look at the sysboot configuration on the beagle(its been awhile) [17:12] * prpplague pulls the docs [17:13] and I've checked the voltages on the user button, it's not broken :p [17:13] 1.8v on one side, 0v on the other [17:13] and pulled down to 0v when the button is pushed [17:13] * cwillu_at_work huggles prpplague [17:14] the odd thing is that there's a version of x-load that works properly re: nand, but it doesn't initialize i2c properly, so u-boot with zippy2 support just hangs [17:16] fatload mmc 0 80200000 x-load.bin; nandecc hw; nand erase 0 80000; nand write 80200000 0 20000; nand write 80200000 20000 20000; nand write 80200000 40000 20000; nand write 80200000 60000 20000 [17:16] is what I'm using to write it [17:17] (I've named the file x-load.bin instead of .ift after signing it, as the extra extension triggers the bad behaviour re: requiring MLO to be written first on fat [17:21] cwillu_at_work: hmm, from what i am seeing, if you have a valid x-load in the nand flash, and the user button is not pressed, it should load the x-load from nand flash, not the sd card [17:21] cwillu_at_work: is that what you are seeing? [17:22] yes [17:22] Texas Instruments X-Loader 1.4.4ss (Apr 13 2010 - 22:36:28) with user [17:22] Texas Instruments X-Loader 1.4.4ss (Jun 28 2010 - 10:11:02) without [17:23] ok [17:23] both give Loading u-boot.bin from mmc [17:23] if I remove the sd card, I get Loading u-boot.bin from nand [17:23] ahh, ok [17:23] and then the u-boot prompt [17:24] cwillu_at_work: you need to look in the x-load code, there is a order of preference for loading the u-boot.bin file [17:24] cwillu_at_work: normally it is the sd card file, nand second [17:24] prpplague, mmc support isn't even compiled in as far as I can tell [17:25] include/configs/omap3530beagle.h is documented as the place to change the ordering [17:25] but the setting doesn't affect it [17:25] that is in the x-loader code? [17:25] yes [17:25] line 48 should be 1 to boot from mmc, and 0 to boot from nand [17:26] I've also tried disabling CONFIG_MMC entirely [17:26] * ogra_cmpc thought its a matter of how long you hold down the user button === hrw is now known as hrw|gone [17:27] at least it seems to work that way wiht the ubuntu x-loader here [17:27] ogra_cmpc, if I'm not holding the user button down, it shouldn't be loading from mmc [17:27] ogra, it's pulling x-loader from nand, but that xloader is just going to mmc [17:27] right, but if you hold it down after x-loader is up it should load u-boot from mmc [17:28] ogra_cmpc, ... [17:28] it's going from mmc even _without_ the user button [17:28] right, i got that [17:28] the only way it pulls u-boot from nand right now is if I don't put the sd card in at all [17:28] weird [17:29] yes :) [17:30] lib/board.c:91 is where it tries to pull it [17:30] and as far as I can tell, that shouldn't even be compiled in (CONFIG_MMC = 0) [17:31] cwillu_at_work: something doesn't sound right for your configuration and such [17:31] cwillu_at_work: i just tested on my beagle and it works fine [17:31] and you had an otherwise bootable sd card in the beagle? [17:32] yea [17:33] L:/ [17:33] this is from koen's golden git repo :./ [17:33] cwillu_at_work: i think it might be best if you have the same x-load built from scratch on both the sd and flash [17:33] no offense, but how is that relevant? [17:34] if the mmc is affecting this, then something is broken [17:34] it shouldn't be touching it, and so whatever is on it should be irrelevant [17:35] * ogra_cmpc agrees [17:36] cwillu_at_work: you need to start from a known position, if you have the same code built together at one time, you can use it as a basis for your tests [17:36] we're using 1.4.3 from the sarkoman tree in ubuntu btw [17:38] prpplague, I'm not going to lie, that's a maddeningly frustrating thing for you to say [17:39] I justed deleted MLO and x-load off the sd card completely. [17:39] same behaviour [17:39] cwillu_at_work: interesting [17:40] could you send me your copy? about the only thing left it could be is my build environment and my beagle [17:40] cwillu_at_work: i'm currently at work and am limited to what i can send you [17:40] cwillu_at_work: i could post something on the wiki this evening [17:41] cwillu_at_work, just grab the ubuntu deb and unpack it if you need a working bainry [17:41] *binary [17:41] http://www.elinux.org/BeagleBoard_Zippy2#Copy_files_onto_the_BOOT_partition [17:41] ogra_cmpc: have your tried the MLO binary that i already have posted? [17:41] nope, i'm not near my beagle [17:42] oops sorry that was for cwillu_at_work [17:42] cwillu_at_work: http://www.elinux.org/BeagleBoard_Zippy2#Copy_files_onto_the_BOOT_partition [17:42] prpplague, was that MLO built for nand or for MMC though? [17:44] ... and all this to avoid teaching a technician how to write the firmware by hand :p [17:46] just to check, when you said that you had it working: you had MLO and u-boot on MMC, and x-loader and u-boot in NAND, and when you booted without the user button held down, you got the message "Loading u-boot.bin from nand"? [17:46] cwillu_at_work: correct [17:49] ... you know what would kinda tick me off? [17:49] if the damn signing program was silently not doing anything because of an existing file with the same name, which wasn't getting cleaned by make distclean [17:51] cwillu_at_work: yea you have to be careful about that [17:51] and you have to make sure that you copy the x-load.bin.ift to the file name MLO [17:51] orga: ping [17:51] no, not in this case [17:52] MLO is what you call it if you're booting off the sd card [17:52] which I'm not (unless the user button is pressed, but that's not where my grief is)\ [17:52] right, with respect to the SD card [17:53] nope, that didn't change anything [17:53] Texas Instruments X-Loader 1.4.4ss (Jun 28 2010 - 10:50:17) [17:53] Loading u-boot.bin from mmc [17:54] prpplague, which config options did you have to change in include/configs/omap3530beagle.h? [17:54] and then I'll stop bugging you and give up on this for a while :p [17:55] ogra_cmpc: ping [17:55] cwillu_at_work: sorry, i don';t have the source here, i'm currently working onsite for a customer === sbambrough is now known as scottb-lunch === gsnedder1 is now known as gsnedders === asac_ is now known as asac === scottb-lunch is now known as sbambrough [20:01] sigh. i create an image to run qemu-system-arm, and i'm left with a blinking cursor and no cpu activity. this isn't fun. === JaMa|Wrk is now known as JaMa === JaMa is now known as JaMa|AFK === JaMa|AFK is now known as JaMa [20:53] ogra_cmpc: ping [20:53] ogra: ping [20:58] any ubuntu userland folks awake? [21:00] define "awake". :P [21:00] what's up? [21:01] GrueMaster: hey, i'm mainly a kernel person, i wonder if you know about running ubuntu userland with multiple framebuffers? [21:01] Hmm. Not really. I would think X treats them as separate screens. [21:02] GrueMaster: that is my thoughts, i was just curious if there were some good examples [21:02] But the only system I have that remotely fits that description is my desktop w/ nVidia graphics. [21:02] GrueMaster: ahh [21:02] X should detect the outputs. [21:03] I would think it would be similar to a laptop with external video output ports (vga, dvi, s-video). [21:03] GrueMaster: that was my guess, but some reading online seems to indicate some people have problems with that [21:03] Depends on the card & drivers. [21:04] * prpplague will find out soon [21:04] good luck. post your results. [21:04] hehe, won't be able to for a while === XorA is now known as XorA|gone [21:06] anybody able to comment on my non-root network usage problem ? === JaMa is now known as JaMa|ZZz [23:30] is it possble to debug qemu during rootstock process ? [23:31] it seems like it gets stuck on "Switching to Virtual Machine for second stage processing" [23:31] qemu process eating 100% cpu [23:31] and mem usage doesn't change [23:31] last time let it run for 20 minutes [23:32] without any output. is this high duration without any output expected ?