=== zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel [12:49] evening [12:50] lamont_r whenever you are ready [12:51] zul: can we do it in about 2 hours? [12:51] or is that toolate? [12:51] sure no probelm [12:51] problem even [12:51] ok. need to run into town with the daughter for a bit === lamont_r translocates to home again === T-Bone contemplates this conversation, notes that quoted out of context, it could sound rather strange :) [12:52] lol [12:52] perv === lamont_r whacks t-bone [12:52] even with context, it does look strange :) [12:52] T-Bone: if you're nice, I'll put the Packages files on p.u.c... :-) [12:52] LOL [12:52] T-Bone: baz primer [12:52] ok ok ok, I won't quote :) [12:52] s/quote/fortune/ :) [12:53] anyway, gotta run home, pick up kid, run to town. back in a couple hours. [12:53] see ya tomorrow then! [12:53] oh that sucks...debian dropping all but 4 arches? [12:53] yeah [12:53] they're nuts [12:53] i take people are pissed off right now [12:54] that'd be a fair bet [12:54] maybe i should go read -devel :) [12:55] i haven't even bothered, but would certainly welcome a short summary :) [12:58] bah...i dont have time...executive summary people are pissed [01:01] oooh...we are down to 18 major bugs for kernel stuff [01:11] zul: they aren't dropping them [01:11] zul: they are going to change the way that they are released [01:12] yes, people are pissed off, but mostly because they are misinterpreting the announcement (as you seemed to ;-) ) [01:14] mdz: i only saw it on slashdot i havent dealt into it [01:14] oh god [01:14] just what Debian needs, slashdot misinformation [01:14] heh [01:15] slashdot refuses to post our release announcements, and they publish this crap :-P [01:15] how come they refuse? [01:15] I have no idea [01:15] thats retarded [01:17] someone should go beat some sense into them === T-Bone is now known as T-None [02:34] la deh dah [02:40] ah, cool.. I still have 15 minutes. :-) === lamont goes to feed horses quickly [02:43] heh [02:52] ok zul [02:52] ready> [02:52] >? [02:53] stupid > key [02:54] yep [02:54] ok... things to decide. lets assume that you're going to have the archive on the local machine, and then mirror it to the http archive [02:55] ok [02:55] in my case, I chose to put the archives in /var/lib/arch - you'll want to pick a directory somewhere... [02:55] baz make-archive /var/lib/arch/zulcss@gentoo.com--2005 (or whatever) [02:56] that will create the zulcss@gentoo.com--2005 archive. [02:56] baz register-archive http://people.ubuntu.com/~lamont/Archives/kernel-team@ubuntu.com--2005 [02:56] registers the upstream beast [02:56] k hold [02:56] on [02:57] ok...made the arvhie [02:57] ok..lets go [02:58] baz branch kernel--team@ubuntu.com--2005/kernel-debian--pre28--2.6.10 zulcss@gentoo.com--2005/kernel-debian--pre28--2.6.10 [02:59] creates a branch for you to work in. [03:00] then, baz make-archive -m zulcss@gentoo.com--2005 sftp:///~zul/public_html/Archives/zulcss@gentoo.com--2005 [03:00] baz branch kernel-team@ubuntu.com--2005/kernel-debian--pre28--2.6.10 zulcss@gmail.com--2005/kernel-debian--pret28-2.6.10 [03:00] branch: invalid version name -- zulcss@gmail.com--2005/kernel-debian--pret28-2.6.10 [03:00] (I think...) [03:00] pre28--2.6.10 [03:00] not pre28-2.6.10 [03:00] oops === lamont isn't sure about the mirror - let me check.. [03:01] its running now [03:02] make-archive -m syntax works.! woot. [03:03] ok got that [03:04] zulinux.homelinux.net/Archive [03:04] baz library-config --greedy --sparse /var/cache/arch/revisions # create the revision library [03:05] baz get kernel-team@ubuntu.com--2005/kernel-debian--pre28--2.6.10 debian [03:06] will grab create the debian directory [03:07] then (inside that directory), baz archive-mirror will push things to zulinux.homelinux.net/Archive [03:07] uh, doh. [03:07] tla: indicated library is not on library path [03:08] gotta go check something [03:08] dpkg -l bazaar? [03:08] 1.2-2 [03:09] ok [03:09] im using ~/work/ubuntu/Archives as my arch repository [03:09] we forgot to tell baz to keep listings current, so you get a second step until someone in #arch tells us how to automate it... :( [03:10] doh [03:10] or we could be abusive... [03:10] abusive :) [03:10] on zulinux, rm -rf Archive/zulcss@gmail.com--2005 [03:10] done [03:11] rm ~/.arch-params/=locations/zulcss@gmail.com--2005 [03:11] on your home machine [03:11] done [03:11] ok. give me a minute to check on things... [03:11] k [03:13] baz make-archive --listing -m .... [03:13] and then baz archive-mirror [03:14] k gimme a sec [03:14] says archive already registered [03:15] which machine did you rm ~/.arch-params on? [03:15] ah, on your work machine: [03:15] local machine i did the sftp bit [03:15] rm ~/.arch-params/\=locations/zulcss@gmail.com--2005-MIRROR [03:16] uh, that 2nd rm needed to be on zulinux... [03:16] so... [03:16] same machine [03:16] ah, ok. [03:16] that helps some... [03:16] only not.. [03:17] if both are really the same machine, the best thing to do is just create the archive where http can get to it... [03:17] since you can't have the archive and the mirror have the same name on the same machine, etc... [03:17] yeah that would be easier [03:18] so best way is to just nuke the archive storage completely, and toss it from ~/.arch-params/=locations [03:18] done [03:18] then either a symlink and the needed config, or store it under the web tree somewhere.. [03:18] so... [03:18] baz make-archive --listing .../Archive/zulcss@gmail.com--2005 [03:19] you already have the other registered (unless you over-nuked...) [03:19] yeah over nuked [03:19] so baz register-archive from above [03:19] baz branch from above [03:19] doh.. [03:19] baz library-config from above [03:21] then baz get from above [03:22] then you're going to get into a love/hate relation ship with "baz lint" as you do things... [03:22] baz add files that you add, etc. [03:22] do you have a change to walk through? [03:24] gimme a sec...still going through the steps [03:24] my brain is slow today...more so than usual [03:25] still getting the same error on revision [03:25] on revision? [03:25] revisions [03:26] chuck@homer:~$ baz library-config --greedy --sparse /var/cache/arch/revisions [03:26] tla: indicated library is not on library path [03:26] dir /var/cache/arch/revisions [03:27] ah, ok. [03:28] what does baz my-revision-library say? [03:29] chuck@homer:~$ baz my-revision-library [03:29] my-revision-library: no revision library path set [03:29] there isnt a /var/cache/arch directory [03:32] yeah - I dumped my revision lib in /var/lib/arch/revisions... [03:32] pick a directory [03:32] baz my-revision-library [03:32] it'll grow over time until you prune things out of it... [03:33] ok done [03:33] then library-config [03:33] adjusted as needed [03:34] im just doing the baz get [03:36] ah, ok [03:36] so i did the baz get [03:37] and i have the two gz files in my Archive directory [03:38] so i should be done right? [03:41] thats cool [03:41] yep [03:42] actually, the branch created those files, I expect [03:42] uh, did you say --listing with your make-archive? [03:42] so when pre29 comes out i have to do a baz branch kernel-team ya deh dah [03:43] ~/arch [03:43] touch ..../Archive/zulcss@gmail.com--2005/=meta-info/http-blows [03:43] url is http://zulinux.homelinux.net/Archive/zulcss@gmail.com--2005 ?? [03:43] yep [03:44] ok. is there a http-blows file there? [03:44] no [03:44] touch it [03:44] then say baz archive-fixup zulcss@gmail.com--2005 [03:46] chuck@homer:~$ baz archive-fixup zulcss@gmail.com--2005 [03:46] usage: baz archive-fixup [options] [03:46] s/fixup/fixup -A/ [03:46] i c [03:46] done [03:50] http://localhost/Archive/zulcss@gmail.com--2005/kernel-debian--pre28--2.6.10/patch-1 [03:53] so when pre29 comes out all i have to do is a baz branch? [03:57] right im falling asleep thanks for your help lamont [04:01] yes [04:02] baz abrowse zulcss@gmail.com--2005 [04:02] zulcss@gmail.com--2005 [04:02] kernel-debian [04:02] kernel-debian--pre28 [04:02] kernel-debian--pre28--2.6.10 [04:02] base-0 .. patch-1 [04:02] woot [06:21] morning [06:21] fabbione: getting kernel-bugs created... you want me to subscribe you? [06:21] gonna move the bz target to there, instead of kernel-team, so we can have our mailing list back, [06:22] fabbione: baz register-archive http://zulinux.homelinux.net/Archive/zulcss@gmail.com--2005 [06:22] btw, [06:22] lamont: yes please [06:23] s/,// [06:23] even if all bugs are assigned to me and i get them automatically 3 times [06:23] heh. I assume you want moderator access? [06:23] mdz: yes i am cherrypicking bug fixes from Debian right now [06:23] mdz: and syncing up to their 2.6.10-6 [06:24] fabbione: ok [06:24] what will be the list id for kernel-bugs? [06:24] fabbione: we'll make the default bz target be kernel-bugs [06:24] kernel-bugs@lists.ubuntu.com [06:24] mdz: please baz get from lamont archive on chinstrap. [06:24] mdz: so you can follow yourself [06:24] fabbione: not right now; I am trying for <12 hours working today [06:25] mdz: sure.. [06:25] let me put up a changelog for you [06:25] mdz: there are only bug fixes atm [06:25] no new features [06:25] http://people.ubuntu.com/~fabbione/changelog [06:27] there are more patches to come security, memory corruption and stack handler related [06:27] we might also fix a wishlist in this upload since the change is not introsive at all [06:35] fabbione: is done. [06:35] the ml? [06:35] cool [06:35] mdz: could you whack bz and reassign all the kernel bugs from kernel-team to kernel-bugs, and change the default target? [06:36] fabbione: you're in the --pre28 tree, yes? [06:37] lamont: do a search, click "change several bugs at once" in the search results [06:37] lamont: yes [06:38] mdz: ok [06:39] but the default assignment is still you, right mdz? [06:39] lamont: default assignee for linux has been fabbione for ages [06:41] yeah [06:42] mdz: do you think it will be possible to get Kamion to create one sparc hoary CD after (and only AFTER) release? [06:42] ah, even better [06:42] well if we can get kernel-bugs to get them assigned [06:42] i wouldn't mind [06:42] fabbione: I got pretty far down the path of building my own on hppa... although it really needs some Kamion-love to be happy [06:43] fabbione: probably, but it's entirely possible that it will require changes to packages in the archive to work (so it wouldn't be hoary anymore) [06:44] in that case i agree [06:44] but i think we have been careful enough not to need that [06:44] well let see [06:45] fabbione: you'll almost certainly need d-i changes [06:45] will sparc be caught up on building packages by release? ;-) [06:45] mdz: it's almost there for main [06:45] too bad it was dead for 15 days [06:45] that was really unlucky [06:45] and i have only one or two FTBFS to handle up till now [06:45] all the rest is going in [06:45] but clearly it will never make it for hoary [06:46] i am aware of that [06:47] grumble. kernel-bugs@lists.ubuntu.com did not match anything [06:47] do I need to create a login for him? [06:47] nmostlikely [06:47] s/^n// [06:48] yeha === lamont waits for his password [06:55] crimsun: i got the patch from zul for the sis oops [06:56] i am going to apply it in a few minutes [06:56] fabbione: great :-) [06:56] do you have the bug num handy? [06:56] mdz: will new kernel bugs get kernel-bugs@lists.ubuntu.com as their qa contact then? [06:57] fabbione: from bugzilla? I don't think there is one (rather, I haven't found one), but a friend has experienced it. [06:58] fabbione: of course, mass changing 79 bugs when we get 2 copies per (well, you 3)... sorry..... [06:58] lamont: only 79? [06:58] i have approx 130 assigned for kernel [06:58] ok i am not going to touch bz [06:59] to avoid mid-air collisions [06:59] 74 bugs had kernel-team as their qa contact [06:59] probably we didn't catch all of them with qa contact [06:59] i will see after you change them === lamont grumbles at cupsys [07:00] fabbione: my change is done, just percolating, etc. [07:00] you didn't change to who the bugs are assigned, did you? [07:01] no just qa contact [07:01] unless, of course, I'm a loser... :-( [07:01] ok. should we reassign them to? [07:01] we could, but some are assigned to you and some to zul, etc. [07:02] lamont: package linux [07:02] we can manually reassign the ones that this check will skip [07:02] choices are to assign them to one person in charge of triage, or assign them to the list and take them as they're grabbed. [07:02] it makes it easier to have one and only one contact === lamont does the search again [07:03] ok let's talk about it with zul later [07:03] because i really have heaps more than 79 [07:03] zul: are you awake? === lamont finds 166 more bugs, many of them resolved... [07:04] don't think I'll change those [07:04] no skip the resolved... [07:04] 74+98 == 172 [07:04] just do a search with fabbione@ubuntu.com as assignee [07:05] and package linux === eruin [~eruin@eruin.user] has joined #ubuntu-kernel [07:05] copypaste myself? [07:05] yes please [07:05] did a search for package=linux, qa != kernel-bugs, status == unconfirmed thru pendingupload [07:06] I've had to compile my own kernel to set dma on for my standard ide hd/cd drives, and while doing so I noticed the ubuntu config had "CONFIG_IDEDMA_ONLYDISK=y" [07:06] get all the status [07:06] many are NEW and NEEDINFO [07:06] that was there [07:06] eruin: yes. that is correct. enabling DMA on cdrom by default causes more troubles than the ones that it solves [07:06] unconfirmed thru pendingupload includes new,assigned,reopened,needinfo,upstream [07:07] eruin: it is enough to use hdparm to enable dma on any device you want [07:07] there is no need to recompile [07:07] but not resolved,verified,closed [07:07] okay, then that's probably not what solved my issues [07:07] the problem I had was that hdparm didn't allow me to set dma on [07:08] eruin: if it doesn't, perhpas your cdrom or chipset is blacklisted from DMA [07:08] because known to be broken or something [07:08] I still haven't figured out how to get support for enabling dma on my via sata drives, but that's another thing [07:08] any idea on where I can find info on that? [07:08] there is a blacklist somewhere.... [07:08] in the source [07:08] i just can't remember where exactly [07:09] I'll have a poke [07:10] drivers/ide/ide-dma.c I bet [07:10] i think there is a blacklist.c or something [07:10] i really can't remember === lamont falls bedwards [07:10] night lamont [07:10] cya later [07:11] static const struct drive_list_entry drive_blacklist [] = { [07:11] looks like it [07:11] but that's the one for specific cdrom [07:11] can't find any of my stuff in it though [07:11] iirc [07:11] there is another one for chipset too or at least.... [07:13] bah [07:13] that blacklist blacklists known-to-work devices [07:13] if you know that they work, you should report them upstream [07:13] they are pretty responsive on that [07:14] I wonder how they get on the list in the first place if they work [07:14] eruin: sometimes developer do not have the hardware [07:14] and they hack around with the specifications [07:14] that makes the driver work [07:14] but they can't test extra features [07:14] ah [07:14] DMA can be really bad if set on broken hw [07:15] like data corruption [07:15] IDE hanging due to unneeded resets [07:15] so it is a much safer path to first blacklist it [07:15] and wait for reports that it works [07:15] than the other way around [07:15] i think we can all agree that "data corruption" = "really really bad" [07:16] compared to a minimum gain in performance [07:16] yeah ;) [07:16] well.. minimum [07:16] minimum is subjective :-) [07:16] my computer pretty much hangs when I do major operations on my sata drive [07:16] because I'm no longer able to use dma on it [07:16] only on cdrom [07:17] gathering from what I've seen around the net, the via drivers only work well if compiled-in [07:17] could that be true? [07:18] no [07:18] there is basically no difference in compiling in or having it as module [07:18] hmm.. I've done so and now I can set dma on my ide hd [07:19] did you check the module parameters? [07:19] nope [07:19] some of them can accept stuff like forcel33tdma=on [07:19] :P [07:19] without the need of recompiling [07:19] curses [07:33] lamont: no, I'll need to change the default QA contact manually; please email me [07:37] mdz: unionfs... hmmmm [07:37] i think we should find something better than external projects [07:38] Al Viro is the right person to ping definetely === gabaug [~gabe@216-43-99-149.dsl.mcleodusa.net] has joined #ubuntu-kernel [07:43] why don't modules that worked in 10-3 work in 10-4 (with the error "Invalid module format")? [07:44] sorry, hoary's linux-image-2.6.10-3-686 vs 10-4 that is [07:44] -3 and -4 identify an ABI change [07:44] that is perfectly normal and wanted [07:44] ok...is it possible to compile my modules for -4? [07:44] yes [07:44] the same way you did for -3 [07:45] because I built one on a freshly installed hoary-preview-install with the -4 kernel running and headers installed and it gave me that error [07:45] probably you didn't install the correct kernel headers [07:45] also note that after preview [07:45] the kernel ABI was broken [07:46] so you might have built with a broken kernel [07:46] I have the kernel headers that the install cd provided...I have /usr/src/linux-headers-2.6.10-4[-386] [07:47] gabaug: please update to the most recent kernels and rebuild the module [07:47] okay [07:47] be sure to have 2.6.10-27 [07:48] woah...I'm confused, did we jump from -4 to -27 or did I just miss a lot of kernel updates? [07:48] no [07:48] -4 is the ABI in the package name [07:48] I've found something interesting... http://lists.debian.org/debian-amd64/2004/10/msg00099.html [07:48] -27 is the debian version [07:48] or ubuntu version [07:48] fabbione: okay, thanks [08:06] mdz: can i extend the kernel bof list for UDU (since the page is "please do not modify") [08:06] ? [12:09] fabbione: let's move here. [12:11] fabbione: the powerpc-mv643xx-enet.dpatch patch just enables to build the mv643xx_eth module on MULTIPLATFORM_PPC, and removes all mips-embedded arch specific part of it, no ABI breakage visible there. [12:12] svenl: we have a (semi) automatic abi checker [12:12] so that one is checked at build time [12:12] fabbione: the other patch detects if a marvell NB is present, and sets a couple of info (membase and irq to be used) for the driver to use. [12:12] yes i read the patches [12:12] so i don't think there should be any ABI breakage involved. [12:12] but i still need to verify them and revert an old patch we have around [12:14] svenl: i promise.. as soon as i get -28 out of the way, your patches will be the first to go in and be tested. [12:14] http://people.ubuntu.com/~fabbione/changelog [12:15] svenl: ^ [12:15] this is what i have in the boilerplate for today [12:15] fabbione: you have an old marvell patch in ? The one from debian 2.6.10 ? [12:15] svenl: i think it is from 2.6.9 that i portforwarded [12:16] but there is a patch there [12:16] fabbione: ok. we can get ride of it, and replace it with this new one. [12:16] yes i know .. i did check that too [12:16] :-) [12:16] more to this later, do your install and i will build a package locally with my proposed fix, ok ? [12:17] svenl: i will upload my package around 14:00 UTC [12:17] after that i can do directly tests for you [12:17] so don't worry too much [12:17] i can even give you the images to test [12:17] that's not an issue :-) [12:17] but before 14:00 UTC with kernel up [12:17] it's no-no [12:18] ok. [12:18] i was just asking, not knowing the schedule. tomorrow is better for everyone involved. [12:20] yup.. no problem [12:31] build orgy started :-) === fabbione -> shower -> food [12:31] later [01:42] woo woo woo [01:43]