[02:20] hey guys im havin a little trouble learning make files http://codepad.org/Uqn2MCyY....my driver source is in /home/name/Desktop/misc-drivers [02:20] http://codepad.org/Uqn2MCyY sorry (...) [02:58] now i keep getting invalid format for hello.ko === pace_t_zulu_ is now known as pace_t_zulu [08:46] hello everyone [08:49] i've just tried this one successfully - after quite some time http://www.jan-steinhoff.de/linux/synaptics-usb.html what about bringing that into the kernel -the author obviously at it [09:35] how would I want to go about interpreting the battery consumption part of the checkbox suspend tests? I suppose I should really only use it to compare against results from different kernels on the same machine? [10:00] Ng, yeah its a very subjective measure [10:01] my feeling is it should be considered only as an indication of maximum suspend time, and if its less than a day or two its fail [10:01] before some fixes i had a like 8 hour suspend time, which was definatly wrong [10:03] bah, that means I need to put jaunty back on the machine to get a baseline to support my gut feeling that suspend time has nosedived ;( [10:03] I also need to file a bug somewhere, some of the automated resumes didn't... resume. The LEDs changed, but I had to poke the power button [13:14] * popey filed bug 387272 but isn't sure it's a kernel bug or a desktop bug, feedback welcome [13:14] Malone bug 387272 in linux "[karmic] long boot time on eee 900" [Undecided,New] https://launchpad.net/bugs/387272 [13:22] cjwatson: do you have a cross compile environment for karmic ARM? Your patch series for udeb firmware isn't quite right for ARM (and probably other CPU arches without firmware) [13:22] rtg: nope [13:22] rtg: what's wrong with it? [13:23] cjwatson: it tries to copy a firmware directory that doesn't exist. I'll see if I can fix it later today [13:23] is it a bug in linux/debian/rules* or in kernel-wedge/commands/copy-firmware? [13:24] cp debian/d-i/firmware/* $(builddir)/firmware/$(arch)/ [13:24] cjwatson: linux/debian/rules [13:24] oh, is it that? [13:24] cjwatson: I think so, it was last Friday that I was messing with it [13:24] no, that can't be right, that directory does exist [13:24] if you can show me the error message I'll be happy to fix it [13:25] cjwatson: I'll get to it later today. [13:25] ok [15:27] cjwatson, could there be no files in there though? [15:27] then it could try and copy a file called '*' [15:28] maybe some incantation of cp -r debian/b-i/firmware/ is more appropriate [15:28] apw: thats exactly what is happening. I can repro the problem quite easily, so I ought to be able to fix it justr as soon as I clear up soem other issues. [15:30] rtg ack [15:43] rtg: did you apply all my patches? one of them creates a file in that very directory [15:43] so I don't see how it would be possible for that to be empty [15:43] it's not dynamic or anything, it's just a source directory [15:44] i.e. in my branch, there exists a file debian/d-i/firmware/nic-modules [15:45] cjwatson: yes, I applied all of the patches. However, I prefer patches that are bisectable so I may collapse the last 2 (once I figure out why they don't work for ARM) [15:51] fine by me, perhaps I should have committed them in the opposite order [16:02] Keybuk, hey ... did you get to any testing on union mounty stuff [16:09] hi apw ,do you want to add headers to make external dvb modules compile [16:09] which ones are missing [16:10] cp -a drivers/ieee1394/*.h $(indep_hdrdir)/drivers/ieee1394 [16:11] and for fglrx it would not hurt to add [16:11] cp -a drivers/acpi/acpica/*.h $(indep_hdrdir)/drivers/acpi/acpica [16:12] apw: not yet, was off on Friday [16:12] Keybuk, heh no worrie [16:12] apw: firedtv will not compile without [16:13] i have got a test script for dkms [16:13] http://kanotix.com/files/fix/dkms/dvb-s2api-liplianin-dkms.sh [16:13] that compiles all dvb drivers + some additional dvb-s2 drivers [16:14] the same script can be used to update gspca drivers too, just a small change [16:17] Kano, that first one seems like a reasonable request, so get that into a bug... will think about the other [16:17] let me know the number so [16:17] for the other you need an extra export too [16:17] http://kanotix.com/files/kernel/unused-patches/2.6.30-export-flush_tlb_page.patch [16:18] with both things my fglrx script works with 2.6.30 [16:18] with a relatively small patch [16:18] when you dont copy the headers the patch has to contain em, thats not really useful [16:23] btw. i have have got a verified patch for avm draft-n usb sticks, will be most likely in 2.6.31 too [16:27] http://kanotix.com/files/kernel/unused-patches/2.6.30-ar9170-support-1-stage-firmare+avm-fritz.patch [16:27] you only need the firmware === cjwatson_ is now known as cjwatson [19:02] apw: why dont you set CONFIG_HAVE_KERNEL_LZMA=y [19:03] that compresses the vmlinuz with lzma [19:05] you mean CONFIG_KERNEL_LZMA ... and that is not guarenteed to be a good thing [19:05] the kernel is smaller, but slower to extract from the compressed form [19:05] it is 100% a good thing [19:05] there is a trade off there dependant on disk and cpu performance characteristics [19:05] decompress is always fast [19:05] its 2x the CPU time to decompress over the default compression [19:06] as reported by a number of sources [19:06] which means in seconds [19:06] .5s longer or what [19:06] which mean double the decompress time [19:06] which with a boot budget of 10s could make a difference [19:06] t [19:07] the point is that its not a slam dunk to change, as its a trade offf which needs measuring [19:07] the load time to load the kernel is shorter [19:07] so you win [19:07] I'd be amazed if the difference in size makes up for the difference in compression [19:08] the laod time is shorter yes, but on a netbook where the cpu is poor and the disk is ssd [19:08] for live images you gain 5 mb when you compress kernel+initrd with lzma [19:08] is it going to be faster [19:08] most of that presumably comes from compressing the initrd with it [19:08] i dont have got a netbook [19:08] Kano, and how does that affect my decision? [19:08] apw: well when you dont change it, i change it [19:09] then you are happy, and i am happy. win [19:09] it is just a patch i want to get rid of [19:10] Kano: FWIW our live images only contain the kernel and initrd once, not twice as I saw you saying the other day [19:10] cjwatson, it does beg the question if we could use lzma for the initrd on the CD images as they are one time build compress time wise [19:10] cjwatson: you forget the compressed filesystem! [19:10] if the kernel understands both of course [19:11] if it works, I expect we could; speed is not so critical there [19:11] Kano: no, I don't. [19:12] the only way to get rid of storing it twice would be using for example grub2 as bootloader with a squashfs support module that would fetch the file from the squashfs [19:12] Kano: false [19:12] Kano: by demonstration, in the last several releases of Ubuntu :-) [19:13] you have got no /boot/vmlinuz-... in squashfs? [19:13] Kano: that's correct [19:13] Kano: the installer knows to copy it from the ISO9660 filesystem rather than from the squashfs [19:15] then you are forced to boot with iso-scan to isntall [19:15] no you aren't, we use ubiquity [19:15] well when you boot from network [19:16] you can do that with casper/ubiquity too [19:16] (might need some fixes since it's not tried very much, but it's basically possible) [19:16] and I don't see why iso-scan is a problem - you ought to have /cdrom set up properly during installation from local media anyway, several bits of code assume that and may operate a bit weirdly if it isn't there [19:17] what does happen when you used live-media-path override [19:17] does your installer support it [19:18] casper does but if you use that ubiquity will probably fail to understand what's going on. I invite anyone who has a problem with this to submit patches :-) shouldn't be hard to fix [19:18] it'd just need to read casper.conf [19:18] (casper does> as of karmic anywya) [19:18] anyway === bjf_ is now known as bjf_afk [19:43] Kano: ... indeed, that's simple enough to do that I've just committed support for it [19:44] do you really need the preseed option in order that your installer works correctly [19:45] huh? [19:45] what preseed option? [19:47] file=/cdrom/preseed/kubuntu.seed for example [19:47] it has sensible defaults for Ubuntu [19:48] the preseed options arrange for the right packages to be installed appropriate to the flavour you're installing, so that we don't have to ship a separate installer for every flavour [19:49] (we actually do ship a preseed file for Ubuntu too; the main relevant thing it does is to arrange to install the desktop automatically. if you don't use that preseed file then (a) you'll get asked the question on alternate/server CDs (b) it makes no difference on desktop CDs) [19:50] so bascially it is useless on the desktop ones? [19:54] when this is the case, why are those files on it then [20:01] Kano: I didn't say that, I said that the bit that arranges to install the desktop automatically makes no difference on desktop CDs [20:02] Kano: though right now it's true that none of the rest of the preseed file is relevant to desktop CDs, only DVDs; but who cares? it's easier to just tell the image building infrastructure to always include it [20:02] premature optimisation => silly [20:03] well it is at least good to know [20:03] if i want to test u live i use currently something like [20:03] http://kanotix.com/files/fix/grub2/config/70_ubuntu_iso [20:04] did you ever see such a code [20:48] -9 is not a happy kernel for me [20:51] Keybuk: do you use a dvb-c card [20:54] no [20:56] whats your problem with it [20:57] hard locks after the screen is switched off [21:03] intel gfx? [21:14] yes [21:52] sconklin: you have time for a build question? [23:08] Hi everyone. I'm having trouble recompiling the stock 2.6.28-11-generic kernel in 9.04. Everything compiles fine except for the Ubuntu supplied third-party drivers. I'm rebuilding from the tar.bz2 I retrieved through aptitude. Anyone run into this before and/or have any knowledge in this area? === sconklin is now known as sconklin-gone