[12:02] not only do they not get dma at bootup, but they can never be enabled afterwards either, the kernel disallows it [12:04] https://bugzilla.ubuntu.com/show_bug.cgi?id=2773 [12:05] fixing one box to install should not result in making optical drives useless for eg burning disks for 100% of users [12:06] debian has it disabled as well [12:07] and it has been brought up before [12:08] disabled or enabled [12:08] in ubuntu CONFIG_IDEDMA_ONLYDISK=y [12:08] hm [12:08] which breaks optical drives for 100% of users instead of the few that don't work with dma [12:08] since without dma the only thing that works reliably is copying files [12:08] i haven't seen IDEDMA_ONLYDISK cause problems w/ cdroms yet [12:09] calc take it up with fabbione he is going to tell you the same thing [12:09] https://bugzilla.ubuntu.com/show_bug.cgi?id=4809 [12:09] https://bugzilla.ubuntu.com/show_bug.cgi?id=4681 [12:09] it claims in here to run hdparm -d1 /dev/foo [12:09] but with that option enabled in the kernel it won't let you run that [12:10] root@calc-amd64:~ # hdparm -d1 /dev/hda [12:10] /dev/hda: [12:10] setting using_dma to 1 (on) [12:10] HDIO_SET_DMA failed: Operation not permitted [12:10] using_dma = 0 (off) [12:10] calc: as a matter of fact, it's enabled (dma, that is) on my cdrw/dvd drive on my laptop, and works fine [12:10] so it certainly doesn't break optical drives for 100% of users [12:11] well it definitely doesn't work on ppc according to bug 4809 [12:11] and doesn't work on amd64 either [12:11] i could try downloading a i386 iso and see if works on that i suppose [12:12] note above says operation not permitted instead of failed [12:12] like when you pass invalid stuff to optical drives like -m16 [12:12] sorry, i misread your initial thing === calc will double check with a knoppix disk [12:13] but it appears to be caused by that option in the kernel [12:13] you're saying IDEDMA_ONLYDISK should be disabled; i'd agree w/ that [12:13] yea [12:13] it was enabled due to one persons hardware not working with dma [12:14] according to what i see in bugzilla [12:14] and it was enabled on all archs [12:15] i don't use my optical drive very often which is why i haven't noticed this earlier [12:15] hrm === calc bbl, testing the drive [12:15] the original bug reporter doesn't mention which ide chipset driver he's using [12:17] bbl [12:19] the correct way to have fixed #2773, imo, would've been to either add the guy's cdrom to the list of blacklisted cdroms (drive_blacklist in ide-dma.c), or make sure there's not something screwy w/ his ide chipset driver (and make sure he's not using the generic driver) === calc [~ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel [12:22] talk to fabbione [12:23] back [12:23] he'll be doing the upload [12:23] ok [12:23] fabbione: whats up? [12:23] dont think he is here though [12:23] ok [12:23] i verified it works fine under knoppix with 2.6.9 i386 without ONLYDISK set [12:24] calc: you probably missed that [12:24] the original bug reporter doesn't mention which ide chipset driver he's using [12:24] dilinger: yea i left too early [12:25] the correct way to have fixed #2773, imo, would've been to either add the guy's cdrom to the list of blacklisted cdroms (drive_blacklist in ide-dma.c), or make sure there's not something screwy w/ his ide chipset driver (and make sure he's not using the generic driver) [12:25] dilinger: yea [12:25] instead of blacklisting all ATAPI devices completely ;) [12:26] fabbione: thanks for fixing it :) [12:26] i have to go for now, but i'll be back later tonight (10-12hr probably) === T-Bone gets back from anime watching, reads backlog, agrees that IDEDMA_ONLYDISK is a bad thing === T-Bone [varenet@T-Bone.developer.debian] has left #ubuntu-kernel ["Leaving"] === T-Bone [varenet@T-Bone.developer.debian] has joined #ubuntu-kernel [02:33] console-data is fux0red on ia64, how sweet :P [02:50] T-Bone: sigh [02:50] lamont: heh [02:51] lamont: "nobody cares" [02:51] :P [02:51] not until april 9th [02:53] i doubt it'll change then anyway :P === smurfix [~smurf@run.smurf.noris.de] has joined #ubuntu-kernel === lamont ponders d-i/hppa/package-list, feels it may be a bit lacking... [04:35] fabbione: you around yet? === lamont would like someone to review the changes in [http://people.ubuntu.com/~lamont/Archives/] lamont@ubuntu.com--2005/kernel-debian--pre34--2.6.10 [06:06] before I commit them to the kernel-team branch, that is. [06:18] morning [06:18] lamont: sure i can look at it [06:20] fabbione: the ext3 thing would be RC if we supported hppa. and the fix is localized to hppa. and we didn't bump the abi believe it or not. [06:21] lamont: i still need to checkout :) [06:21] yeah [06:22] should be able to just baz diff lamont@ubuntu.com--2005/kernel-debian--pre34--2.6.10--patch-2 [06:23] baz register-archive http://people.ubuntu.com/~lamont/Archives/ [06:23] unable to access URL: /~lamont/Archives/.listing [06:23] webdav error: 404 Not Found [06:24] fabbione: baz register-archive http://people.ubuntu.com/~lamont/Archives/lamont@ubuntu.com--2005 [06:24] that was a prefix, not the complete path... [06:25] right [06:29] lamont: the patch looks fine for me [06:29] just go ahead and merge it in pre34 [06:29] i only need to check one thing on .lnk files [06:29] because from the patch they look empty [06:29] but they should be [06:30] ok [06:30] and the DPATCH header ;) [06:30] heh [06:30] yeah - I'll fix that [06:31] ## DP: Description: Speed up some atomic ops for ext3 performance [06:31] ## DP: Patch author: Grant Grundler [06:31] ## DP: Upstream status: unknown [06:35] committted [06:35] updating now :) [06:36] A patches/parisc-ext3-perf.dpatch [06:36] i guess this is the famous I/O fix? [06:38] that's the make ext3 not suck molasses [06:38] paqtch [06:38] so, yes. [06:38] eheh [06:42] let's merge pre34 into experimental [06:42] just because i need to wake up [06:42] modules/hppa/ide-core-modules is a broken link [06:42] oops [06:43] I could just move ide-core-modules back out of existance... [06:43] wasn't sure what all that was about... [06:43] (and that was part of why I didn't want to merge it quite yet... [06:47] duh. [06:47] lamont: just commit the fix :) [06:47] a .lnk file uses what is provided by kernel-wedge [06:47] while a normal file picks up modules one by one [06:47] right [06:48] just commit the fixes when you have them [06:48] i still didn't hear anything from pitti about the 3 race conditions [06:48] so there is time for -34 [06:54] lamont: mind if i do a cosmetic change to the changelog? [06:55] go for it./ === lamont creams [06:55] screams, even [06:55] ls -la modules/hppa/ide-core-modules [06:55] lrwxr-xr-x 1 lamont lamont 49 Apr 2 21:54 modules/hppa/ide-core-modules -> /usr/share/kernel-wedge/modules/ide-core-modules [06:55] lamont@smallone:~/kernel/linux-source-2.6.10-2.6.10$ more modules/hppa/ide-core-modules [06:55] modules/hppa/ide-core-modules: No such file or directory [06:55] lamont@smallone:~/kernel/linux-source-2.6.10-2.6.10$ more /usr/share/kernel-wedge/modules/ide-core-modules [06:55] drivers/ide/ide-core.o [06:55] what am I not seeing? [06:56] that the file cannot be a symlink [06:56] that's what kernel-wedge created [06:56] but why won't more follow the symlink?" [06:56] s/more/open(2)/ === lamont considers just putting ide-core-modules back into ide-modules and forgetting about it [06:57] did you just changed the file and than rerun ./debian/rules binary-udebs? [06:57] in that case the change needs to be propagated in 2 places instead of one [06:57] I changed the file and did fakeroot debian/rules clean [06:58] oh [06:58] hmm [06:58] or rather, I tried to debuild -S and that failed in the clean target. [06:59] just a sec.. i am looking [06:59] cat ide-core-modules.lnk [06:59] drivers/ide/ide-core.ko [06:59] ext3-modules and cdrom-core-modules are both valid.. [06:59] this is wrong [06:59] yeah - I fixed that [06:59] that's what I was trying to test [06:59] lrwxr-xr-x 1 lamont lamont 50 Apr 2 21:54 cdrom-core-modules -> /usr/share/kernel-wedge/modules/cdrom-core-modules [06:59] lrwxr-xr-x 1 lamont lamont 44 Apr 2 21:54 ext3-modules -> /usr/share/kernel-wedge/modules/ext3-modules [06:59] lrwxr-xr-x 1 lamont lamont 49 Apr 2 21:54 ide-core-modules -> /usr/share/kernel-wedge/modules/ide-core-modules [07:00] the first 2 work, the last one doesn't [07:00] but I can use the target of the symlink [07:00] that's just wacko [07:00] let me try something [07:01] it's weird === lamont kicks himself [07:08] so, count the characters in that link... [07:08] note that you only see 48 characters === lamont removes the trailing space in the file [07:09] doh [07:09] that's a kernel-wedge bug [07:09] hmm somebody would say that it is a feature :) [07:09] uh, yeah === lamont will verify that things build at least part way before he commits this fix [07:12] let it go all the way... [07:12] there is no need to get it fixed within 10 minutes :) [07:13] yeah [07:18] dilinger: ping? [07:34] pong [07:34] dilinger: did you review the 3 race conditions i did forward to you, by any chance? [07:35] no, i've been fighting w/ other stuff [07:35] no problem [07:35] (i'll get 2.6.11 out someday. probably after i kill herbert for writing this crap) [07:35] ehehhe [07:36] you will have the opportunity to talk with him at UDU [07:40] fabbione: herbert will be at UDU - cool. [07:41] i think he will.. that's what i heard at least [07:42] Bug #269: "Wrong merging on directories" added [07:42] ops [07:57] ACPI: AC Adapter [ACAD] (on-line) [07:57] ACPI: Battery Slot [BAT1] (battery absent) [07:57] ACPI: Battery Slot [BAT2] (battery absent) [07:57] very confused ACP [07:57] I === lamont wants support for his AC97 modem [08:39] fabbione: I'm going to commit the known fix, and go to bed.. it's building modules on the 3rd of 4 kernels right now [09:32] lamont: fine for me [09:33] fabbione: is building the 4th kernel now [09:34] lamont: rocking [10:15] Built successfully [10:15] lamont: which ac97 modem do you have? [10:16] 0000:00:07.6 Communication controller: VIA Technologies, Inc. Intel 537 [AC97 Modem] (rev 30) [10:17] there actually is support [10:17] snd-via82xx-modem [10:17] due to the alsa-related changes, it'll be cardX where X>0 [10:18] no snd-via82xx-modem* files in modules... [10:18] which means we probably don't build it [10:18] ah, I see, I'm looking at the Debian stuff [10:20] yeah, you'll have to compile that yourself using 'alsa-source' [10:24] wow. periods of 7K/sec over a dialup line === fabbione attempts to use a xen 2.6.11 patch on top of our kernel [10:29] actually it doesn't look that nice at all [10:29] apparently xen is very sensible to extra patches on top of vanilla [10:30] that doesn't really help us [10:36] sensible? [10:36] or sensitive? [10:37] (the latter) [10:38] bummer === Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-kernel [10:41] yeah [10:42] hmmmm [10:49] i have the feeling i am doing something wrong at build time [10:49] let's try another approach [10:55] crimsun: so what device file name am I looking for for the ac97 modem? === lamont heads to bed before he gets key-prints on his face. [10:57] lamont: cat /proc/asound/devices [10:57] lamont: more than likely, it'll be hw:1 [10:57] lamont: good night :) [10:57] so you'd use plughw:1,0 [10:58] nope.... [10:58] it's one of our patch :/ [10:58] :/ [10:59] let me try the xen patch from bk [10:59] at the end our kernel is more 2.6.11 than 2.6.10 [11:02] heheh [11:09] fabbione: heh, i'm trying to resist the temptation to turn 2.6.8 into 2.6.10 ;) [11:11] dilinger: eheheh [11:11] my main point is that if xen is so sensitiv [11:11] it might be the hell to maintain in time [11:16] what's the goal of getting it built with linux-source? [11:20] dilinger: one single source [11:20] meaning that one security upload is propagated to all images at once [11:20] the .11 patch doesn't build becuase of all the 4 level memory shit [11:22] extern pmd_t *FASTCALL(__pmd_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)); [11:22] i remember that pgd has been introduced with the 4Layer [11:22] do you remember what was before? [11:23] or pmd was introduced? [11:36] i thought pmd was always there? [11:36] part of the 3 level pagetables [11:36] pgd, pmd, pte [11:37] yea, pud was the thingy that was added [11:37] include/asm-xen/hypervisor.h:43:39: asm-generic/pgtable-nopmd.h: No such file or directory [11:38] bah [11:38] um.. [11:39] oh, right [11:39] hmm intersting [11:39] just adding the 2 includes it compiles [11:40] or at least it starts compiling [11:40] but heck.. we can't keep it this way [11:41] include/asm-i386/pgalloc.h has the older empty pmd_free stuff [11:42] yeah i did try to add that one and it failed for other reasons [11:48] pointless to back port.. too much to fix around [11:55] you can't use an older xen patch? [11:56] the older xen patch builds fine, but it doesn't boot === dilinger yawns. 6am, i'm gonna go to the grocery store [11:56] and now i just noticed that not even doogie did xen-images for 2.0.5 [11:56] fabbione: oh yea, that booting thing is kind of important, huh? [11:56] dilinger: kinda :) [11:56] there are only old 1.2 images [11:57] i think he is having the same problems as i do [11:57] well i need to get some stuff done [11:57] and be ready for F1 :) === smurfix [~smurf@smurfix.developer.debian] has joined #ubuntu-kernel === zenwhen [zenwhen@h-67-102-63-103.phlapafg.covad.net] has joined #ubuntu-kernel [07:45] Hey [07:46] Is anyone working on getting monitor mode orinoco drivers into the Ubuntu kernel? [07:46] not right now since there is only a couple of days to release [07:47] Oh [07:47] Well yeah I can see that. [07:47] I just hope it happens at some point. :) [07:47] perhaps...its in the list of todo === zenwhen [zenwhen@h-67-102-63-103.phlapafg.covad.net] has left #ubuntu-kernel ["555"] === dholbach [~daniel@td9091b2e.pool.terralink.de] has joined #ubuntu-kernel [09:11] hey [09:15] i just want a final confirmation: we will only need the most recent kernel-packages from debian for universe, right? [09:15] (for our architectures) === makx [~max@baikonur.stro.at] has left #ubuntu-kernel ["bella] [10:00] yep [10:01] ok === Mithrandir [~tfheen@vawad.err.no] has joined #ubuntu-kernel [10:50] hm [10:51] why the heck does mkinitrd try to find the LVM (!) device of my EVMS swap device? [10:51] Finding all volume groups [10:51] Finding volume group "vojei-sys0" [10:51] /usr/sbin/mkinitrd: /dev/evms/swap: Cannot find LVM device [10:51] Failed to create initrd image. [10:55] ouch [10:59] good night === dholbach [~daniel@td9091b2e.pool.terralink.de] has left #ubuntu-kernel ["Verlassend"] [11:16] any idea why mkinitrd cares about my swap devices? === T-Bone is clueless about initrds. you wanna ask jbailey when he's around [11:44] yeah, I filed an RC bug. [11:44] I can provide root-level access to the system just fine, though.