zulcool...my crappy usb camera just works12:50
=== dilinger [dilinger@mouth.voxel.net] has left #ubuntu-kernel ["Leaving"]
=== viper12 [~casey@c-67-176-43-76.client.comcast.net] has joined #ubuntu-kernel
=== jbailey [~jbailey@CPE000ded9d787c-CM014260028338.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
fabbionedilinger: testing the patch now :-)09:04
fabbioneah he is not here09:07
=== smurfix [~smurf@] has joined #ubuntu-kernel
=== Seveas [dennis@seveas.demon.nl] has joined #ubuntu-kernel
=== ..[topic/#ubuntu-kernel:fabbione] : Ubuntu kernel development discussion | http://www.ubuntulinux.org/wiki/KernelTeam | kernel-team@ubuntu.com--2005/kernel-debian--pre28--2.6.10 | check the new abichecker code and GET familiar with it.
=== jbailey [~jbailey@CPE000ded9d787c-CM014260028338.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
=== Seveas [dennis@seveas.demon.nl] has joined #ubuntu-kernel
=== zul [~chuck@] has joined #ubuntu-kernel
fabbionehey zul02:38
zulhey fabbione how is it going?02:38
zuloh how i hate windows media player02:39
jbaileyYou don't have access to a better media player02:41
zulyeah but its windows that im on at work though as my main pc02:41
zulif i run winamp the windows media player just starts at the same time02:42
=== zul kicks puter
zul*sigh* i dont think you need to put your full address in your wiki page ;)02:43
jbaileyAh, I was thinking winamp, but couldn't remember what it was called. =)02:46
zulT-Bone disapeared?02:51
zuli havent seen him all weekend02:52
jbaileyHe may have for a bit.  He's really pissed.03:00
jbaileyHe's been working his brains out to try and get ia64 into a release state for Hoary because he was told that if he met certain requirements it would be let in.03:00
jbaileyBasically without any further discussion during the dev meeting, he was told otherwise.03:00
zuloh right03:00
jbaileyClearly and publiclly.03:00
fabbionezul: i am ok thanks. please see /t03:00
zulfabbione: ok will do03:01
fabbionejbailey: clearly and public the point is that if it wasn't for me pushing him to start testing ia6403:01
fabbionenobody was going to do so03:01
fabbionei don't have an ia64 myself and even tho i did take care of producing an ia64 kernel03:01
fabbioneso clearly he did ask for the buildd & co03:02
fabbioneand he expected us to do the rest03:02
fabbionewell that was not the original agreement03:02
fabbioneso he has nothing to be pissed03:02
jbaileyfabbione: As if emotions are clearly ruled by logic. =)03:03
jbaileyHe simply had his heart set on it, and beleived that if he met a couple requirements, it would be accepted.  It didn't happen that way and he's angry about it.03:04
fabbionewell he didn't meet the requirement03:04
fabbioneso there is noone to blame03:04
fabbionei was hoping to get sparc for hoary, but i can see it myself that it won't make03:05
fabbionebut that doesn't mean that i am pissed about it03:05
fabbionei know that i have been doing a lot of work03:05
fabbioneand that it will all flow in bendy03:05
fabbione(and note that i don't even have the buildds at the dc)03:05
fabbione(so it took me ages more to get stuff in place)03:06
jbaileySure.  But we don't all react the same way to things.03:06
fabbioneto be hounest i think he is a bit too impulsive03:06
fabbionebut that's just my opinion03:06
fabbionejbailey: but yes. i fully agree with you. we don't react all the same way03:07
jbaileyWe need impulsive people in the world to suggest the things that us stodgy people would never otherwise consider. =)03:07
jbaileyBut it probably sucks to be impulsive and to run against brick walls all the time. =)03:07
fabbionejbailey: STFU! ARE YOU GOING TO CONSIDER THIS? :P03:07
zulnow now :)03:08
fabbionejust kidding ;)03:08
fabbionezul: jb and i know each other03:08
fabbionefrom DC3 and DC403:08
zuli know i was be kidding as well03:08
jbaileyzul: Don't worry.  There's nothing in the SC keeping Fabio and I from getting drunk and slugging it out in a bar at UDU. ;)03:08
fabbione(tho he still needs to sign my gpg key)03:08
zuljbailey: i would pay to see that ;)03:08
fabbionejbailey: at UDU it will be the 3rd time we meet :-)03:09
zul*sniff* im the outsider :)03:09
fabbionejbailey: it's time for you to consider to sign my key ;)03:09
jbaileyfabbione: Yeah, I know.  I have to find that folder. =(03:09
fabbionenah don't worry03:09
jbaileyfabbione: I left it too long, and have moved everything.  I'm not sure where it is.03:09
fabbionelet's do it at UDU03:09
fabbionejbailey: also because since DC4 i have a new key too03:09
fabbionefor ubuntu/canonical03:09
jbaileyat UDU I'll probably have met people enough times that it'll be worth just playing along with the keysigning and dropping the occasional person out.03:10
jbaileyI think I'll actually know most people there.03:10
fabbionejbailey: yeah03:10
fabbionethat's the easiest03:10
jbaileyI've been trying to decide whether to do that.03:10
jbailey(a new key for Ubuntu/Canonical)03:10
fabbionei did.03:10
fabbionefor one simple reason03:10
jbaileyI don't like the idea of putting work email addresses on my home keys.03:10
fabbionei don't like to mix private/debian/work stuff03:10
jbaileyThat's why I use the imap server instead of having stuff forwarded to my home account, too.03:11
fabbionewe all remember when lamont had to revoke his hp uid03:11
zulwell not all 03:11
fabbionei use an imap server at home for speed reasons, but separate folders/accounts03:11
fabbionezul: there a huge shake in WOT03:11
fabbionethat was tracked by all the graphers around03:12
zulah ok03:12
fabbionei think he lost something like 20 positions in the WOt in one shot03:12
fabbionethat is quite a lot03:12
jbaileyfabbione: Yeah, I just figured out the local synchronisation bits in evolution.03:13
jbaileyfabbione: I did it while working at the web caf last week so I could read my email at home.03:13
fabbionetbh i would still be using pine but i wanted to try a GUI application and i landed with thunderbird03:14
fabbionebut i can switch anytime03:14
fabbionesince all the filtering is done on the server03:14
fabbionei just have to resub to the different mailboxes03:14
jbaileyI started using evo when my wife switched from mutt (she had been using elm at her school)03:15
jbaileyShe wanted a graphical interface, and I wanted to be able to answer her questions.03:15
zulisnt that sweet 03:15
jbaileyzul: It's marriage-preserving ;)03:16
zulmine doesnt work like that03:16
fabbionehumpf i have almost finished the sync with debian 2.6.10-503:17
fabbioneand i still miss another release to check03:17
fabbionebut now we have the 31337 abi checker to help us03:18
zuldebian's hotplug is suppose to fix a bunch of firmware issues apparently03:20
fabbionei am pretty sure it would need merge to be ported03:21
zuloooh...are linux sys admin upgraded to hoary03:33
zulsorry our linux sys admin here at work upgraded to hoary03:34
zulstill too early in the morning03:34
fabbioneyou should have told them to wait a sec03:34
zultoo late03:34
fabbionetomorrow we are going out with this big fat kernel03:35
zulwith all the security updates03:35
fabbioneand a few tons of bug fixes03:35
fabbionezul: btw.. your patches?03:35
zulyep getting them ready now03:35
fabbionei can't wait forever to get them03:35
=== fabbione checks the abi
zulfabbione: http://zulinux.homelinux.net/ubuntu/kernel/2.6.10-28/03:45
fabbione6492 - no support for dell's version of sound blaster em101k03:46
fabbionehow bad is this one?03:46
fabbione55K of patch....03:46
fabbioneare you sure it works?03:46
fabbionei remember the bk commit to head03:46
fabbioneand it was pretty intrusive on the emu10k1 too03:46
zulhavent tested it but it was pulled from bitkeeper and applies cleanly and compile cleanly03:46
=== fabbione sighs...
zuldont have to include it if you want but it works with 2.6.11 apparenly03:47
fabbionedid you pull only the emu10k1x patch or did you grab it from 2.6.11?03:47
fabbionesory let me rephrase03:47
zulonly the emu101kx from bitkeeper03:47
fabbionedid you take that patch from the first commit or from 2.6.11 or from bk (like today)?03:48
fabbionei need to check on that03:48
fabbionei am not happy to add new drivers now03:48
fabbionespecially because none of us can test them03:48
zulthats cool03:48
zulbut the other patches should be ok?03:50
fabbionedid you check if there were any updates from the first commit?03:50
zulhavent been updated in 13weeks03:50
fabbioneso it has been stable since first commit...03:50
zulafaik yes03:51
fabbionecan you please triple check?03:51
fabbionei don't mind to add it03:51
fabbionei know it compiles03:51
zulyes i can do that when i get home tonight03:51
fabbionewe saw that in 2.6.11-pre03:51
fabbionewell i guess i can just check with bk here03:51
lamontfabbione: from 13 to > 100 is not 20 positions...03:53
lamontmorning all03:53
zulhey lamont03:53
fabbionehey lamont 03:54
zulbut the other patches should be fine03:54
fabbionelamont: well i couldn't really remember03:54
fabbionekeyboard-i8042 has a possible ABI change03:54
fabbioneanyway we need to bump the abi03:54
zulbleah..when you bump the abi put the inotify 0.20 stuff back in03:55
lamontfabbione: that was also about the time that they actually started paying attention to the revokations in the WoT graphs - I wasn't solely responsible for the dip03:55
lamontzul: was it just inotify that was bad?03:55
zullamont: yep looks like it03:55
lamontinotify sure seems to be shooting for most cursed status03:56
zulby the time everything got compiled on saturday fabbione already uploaded a new kernel03:56
fabbionethis is the patch that breaks the ABI03:57
lamontah, ok03:57
fabbionetell me if it is worth the troubles according to your opinion03:57
zulfo inotify03:57
fabbionezul: did you read the patch name or the entire changelog?03:57
zuli was talking about inotify03:58
fabbioneah ok03:58
fabbioneplease guys look at that patch and give me your opinion03:58
fabbionea yes here means changing the ABI03:58
fabbionethere might more patches that can do that03:59
zulwell how often does ext3 fail?03:59
fabbionei am not sure yet03:59
fabbionei only synced up to 2.6.10-5 in debian03:59
fabbionezul: ENOCLUE03:59
fabbionei really have no idea how often it can happen03:59
mjg59fabbione: Did you get my b44 patch03:59
fabbionemjg59: i read it in the mail yes..04:00
fabbionemjg59: haven't check it 100% yet04:00
fabbionemjg59: i have a pile of security updates under check right now.. sorry04:00
fabbioneit might be for -2904:00
mjg59fabbione: No problem04:00
fabbionemjg59: what is your opinion on the above ext3 patch?04:00
fabbionezul: ++ to what?04:01
zulfabbione: i say go for it04:01
zulbelt and suspenders04:01
zulbumping the abi anyways so go for it04:01
fabbionewhy anyway?04:02
fabbionethis is the only patch atm that breaks the ABI04:02
fabbione#   This is easily reproduced with a sample ext3 fs image containing an inode04:02
fabbione#   whose direct and indirect blocks refer to a block bitmap block.  Allocating04:02
fabbione#   new blocks and then deleting that inode will BUG() with:04:02
fabbione#   04:02
fabbione#   Assertion failure in journal_forget() at fs/jbd/transaction.c:1228:04:02
fabbione#   "!jh->b_committed_data"04:02
fabbionewe can skip it04:02
fabbioneeven if it is an annoying problem04:02
fabbioneit is not necessary to have04:02
fabbionebut let see lamont and mjg59 04:02
lamontseems pretty innocuous04:06
mjg59I think it's generally preferable to avoid corrupt filesystems breaking the kernel...04:07
mjg59It's better to bump the ABI now than to find we need to again later on04:07
Mithrandirmjg59: did you talk to Naima last night?04:08
Mithrandirmjg59: she had some "interesting" experiences where she did a suspend, boot normally, use the box for a little, then boot with resume.04:08
fabbioneok so we will go for it and break the ABI04:08
lamontMithrandir: that would be, um, interesting. to be sure04:09
lamontfabbione: btw, you ROCK.04:09
mjg59Mithrandir: Yeah, I talked to her about that.04:09
Mithrandirlamont: yeah; think it should then say "lalala, you've booted in the meantime, I won't resume, lalala" or something.04:10
mjg59The swapon script should check whether the swap partition has a suspend signature, and if so mkswap it04:10
fabbionelamont: what i did NOT do this time to rock? ;)04:10
Mithrandirmjg59: I think it should refuse to resume if you've booted in the meantime.04:10
mjg59Mithrandir: How could it know?04:10
mjg59We can't touch the filesystem at that point04:10
Mithrandiryou can touch the filesystem before suspending04:10
Mithrandirand remove the file when booting normally04:11
mjg59But we can't read from the filesystem04:11
Mithrandiror something like that.04:11
Mithrandirhm, that sucks.04:11
mjg59If we mount the filesystem, the journal gets replayed04:11
Mithrandireven r/o?04:11
Mithrandirthat sucks.04:11
mjg59mkswapping over the resume partition on boot fixes all but the most pathological cases04:12
Mithrandirtrue enough04:12
lamontfabbione: I finally got to my email and read your replies to pitti04:12
fabbionelamont: ah eheheh04:12
fabbionelamont: wait to do a baz update :-)04:12
mjg59fabbione: The sooner the b44 patch can go in the better from the dealing with HP point of view, but feel free to leave it to -29 if it's a problem04:13
lamontfabbione: you mean wait until I do? or that I should wait before I do?04:13
fabbionemjg59: i think i can make it for -28 given that we will break the ABI04:15
fabbionelamont: just wait a couple of more minutes04:15
mjg59fabbione: Ok, thanks04:15
fabbionemjg59: does it work for you?04:17
mjg59fabbione: It stops the machine from freezing when HAL starts, yes04:19
mjg59It means that mii ioctls won't do anything until the interface has been brought up04:20
fabbioneyes i see04:20
mjg59I'm not sure if that's the best approach, but it works04:20
fabbioneit looks sane to me04:20
fabbioneif the iface is down .. no ioctl04:20
mjg59It may mean that we don't get auto-interface up on cable plug in that chipset, but I don't think we support that anyway yet04:20
fabbioneno we don't04:21
fabbionemjg59: queued04:22
fabbioneguys you can baz update now ;)04:22
fabbioneno ABI bump done at packaging level YET04:22
mjg59fabbione: Rocking. Thanks.04:22
fabbioneno problem04:22
fabbioneupload will be tomorrow after 14:00 UTC04:23
zulnone of my stuff made it in?04:24
fabbionezul: hell gimme a break04:25
fabbionei am working on it04:25
fabbionethis is the work of the last 3 hours of merging04:25
zulfabbione: heh sorry04:25
zulsorry mybad04:26
fabbioneat least now we know 100% that the abi checker is working properly04:26
fabbionewho feel like preparing l-r-m for tomorrow?04:28
fabbioneactually we need to tell Kamion too04:29
zuli need to get my baz archive setup eventually04:29
lamontzul: any home machines have apache running on them?04:30
lamontbecause that's the simplest way to publish04:30
zulits the same machine where i put my patches for you guys04:31
lamontit can either be your actual repository (and you use an sftp:// url while we use http://), or it could be a mirror of your actual repository04:32
zulok...i did a little reading about it this weekend04:33
zulit will probably be something like http://zulinux.homelinux.net/Archive or somthing im not sure yet04:33
fabbionezul: anything is fine for us04:34
fabbionesftp would require us to have access to your box04:34
fabbionewhile with http we can just merge from you04:34
zulits going to be http i dont trust you fabbione ;)04:35
fabbionethat's a good point to start from04:35
fabbioneconsidering i have kernel privileges on your machine :-)04:35
zulyeah i want to sleep at night :)04:35
Mithrandirfabbione: you're assuming he runs _your_ kernel on his box.. he might be... running NetBSD! :P04:37
zuland it might be chrooted04:37
fabbioneMithrandir: i still have commit access to XORG CVS :-) 04:37
fabbionezul: other than the emu10kx.. are the other patches from HEAD or did you grab them around?04:38
Mithrandirzul: chroots won't help you if he has your kernel04:38
zulfabbione: grabbed them around. the keyboard is from 2.6.10-ac9 04:38
fabbionebut i would have access to the main host even from a chroot that has proc mounted04:38
zulthe nforce one is from linux-usb ml and the via-oops is from head04:39
zulfine ill just turn my box OFF then 04:39
zulbwaha...its alive! its alive!! sorry04:46
fabbionezul: baz update :-()04:53
fabbioneyour bits are merged04:53
fabbioneother than the emu10k1x04:53
fabbionei want to look at it again04:53
fabbionetime to merge mjg59 04:53
fabbione  Changes by Matthew Garrett:04:56
fabbione  * b44 should not accept ioctls if the interface is down:04:56
fabbione    - Add patch fix-b44.dpatch.04:56
fabbione    (Closes: #7490)04:56
fabbionemjg59: ok for you? or do you want me to change something else?04:57
zulyou can ask the guy who requested the em101kx to test it04:57
fabbionezul: there won't be time for that04:57
fabbionei can't release binary image before tomorrow04:57
zuloh yes...time 04:57
fabbionenot even for testing04:57
mjg59fabbione: No, that's fine04:57
fabbionemjg59: ok04:58
lamontfabbione: one more thing for -28...04:58
lamont        kernel-wedge find-dups 2.6.10-4-hppa6404:58
lamontfind: lib: No such file or directory04:58
lamontnfs-modules-2.6.10-4-hppa32-di will be empty04:58
fabbioneuh weird...04:59
lamonthrm.. wonder how many more there were before you removed the || true04:59
=== fabbione checks
fabbioneoh that's easy05:00
fabbionenfs is compiled in on hppa05:00
fabbionethere are no modules05:00
lamontnfs-modules... is the only one that says it'll be empty05:00
fabbioneand it is empty05:01
fabbioneit's enough to remove nfs-modules.lnk05:01
fabbionein debian/d-i/hppa/modules/hppa/05:01
fabbionethis is thanks to our super alligned configs05:01
fabbionethat 2 persons (one of which in this chan right now) should have analized for hoary05:02
fabbionewe don't make names05:03
fabbioneonly surnamed05:03
fabbioneeh Short?05:03
fabbionedoesn't it ring a bell?05:03
zulsomeone called my name?05:03
lamontfabbione: maybe he didn't analyze the bastard stepchildren05:03
lamontthe find: lib: No such file or directory is kind of interesting as well.. or is that normal?05:04
fabbioneit is normal05:04
fabbionesince debian/nfs-modules-<whatever>/lib isn't there at all05:04
fabbioneno modules -> no need to create the dir05:05
fabbioneapplying patch fix-b44 to ./ ... failed.05:07
fabbioneah yeah05:08
fabbionemjg59: note for the next time: send me a dpatch :-)05:09
fabbionebaz commit -s'Merge mjg59' -- changelog patches/fix-b44.dpatch patches/00list-2805:35
fabbioneThese files would be source but lack inventory ids (`baz add' perhaps?):05:35
fabbioneM  changelog05:35
fabbionemake-changeset-files: file missing from ORIG tree (patches/fix-b44.dpatch)05:35
fabbionewhat the hell is wrong with baz?05:35
=== ddaa [~ddaa@deep-thought.xlii.org] has joined #ubuntu-kernel
fabbionei am getting a really strange error with baz05:38
fabbionebaz commit -s'Merge mjg59' -- changelog patches/fix-b44.dpatch patches/00list-2805:38
fabbionemake-changeset-files: file missing from ORIG tree (patches/fix-b44.dpatch)05:39
fabbionebut the file has been added05:39
ddaayou cannot do partial commit (yet) when the inventory was changed05:40
ddaahu... not right05:40
fabbionei did partial commits before and it was working fine05:40
ddaayou cannot do partial commit for new/removed/renamed files05:40
ddaaThat can be fixed.05:40
ddaaThe issue is that currently partial commit does not do inventory, so it cannot tell if the file is really new or just renamed.05:40
ddaaprobably it will be time to fix it when there will be a "baz edit" to make life livable with kernel-sized trees.05:41
ddaaBut i suggest you file a bug for the obscure error message.05:42
fabbioneddaa: malone or bugzilla?05:42
ddaaIIRC we are officially in malone again05:42
ddaamark's fault05:42
fabbionei don't want to know :-)05:43
fabbionethanks ddaa.. that was it :-)05:44
ddaabe my guest05:44
fabbionelamont: hppa ftbfs is fixed05:46
fabbioneor it should be at least05:46
lamontfabbione: thansk05:48
fabbioneno problem05:48
fabbionelamont: do you still have -27 on your buildd?05:48
fabbioneor did you trash it?05:48
fabbionewell actually.. no05:49
fabbioneit's pointless05:49
lamonthead build-hoary/chroot-hoary/build/buildd/linux-source-2.6.10-2.6.10/debian/changelog 05:49
lamontlinux-source-2.6.10 (2.6.10-27) hoary; urgency=low05:49
lamontuh, yeah. :-)05:50
fabbionei was going to ask you for the abi files, but we already know they will change05:50
fabbioneso it's useless :-)05:50
lamontah, ok05:53
fabbionezul: the emu10k1x driver has been updated regularlly. and the last commit is from 4 days ago.05:56
fabbionezul: just FYI.05:56
fabbionei am going to see if it can be built05:56
lamontfire call06:06
svenlfabbione: hi you.06:19
fabbionei was checking the code...06:21
fabbionei don't really understand why you use arch_initcall() in the beginning06:21
fabbionebut i need to look trough all the code06:21
fabbionejust gimme a few secs.. i need .11 orig tar.gz06:22
fabbioneoh there is no orig in the archive...06:22
fabbione`linux-2.6.11.tar.gz' at 1826600 (3%) 76.9K/s eta:10m [Receiving data]   06:24
fabbionesvenl: getting there.. slowly :-)06:31
svenlfabbione: well, code is not from me, at first it was to initialize the arch-specific part, which should not be in the drivers/net/mv643xx_eth.c driver (see other marvell patch)06:32
svenlfabbione: i had some code doing OF device-tree probing in place which worked, but hch suggested doing pci_find_device instead.06:33
fabbionesvenl: ok.. let's see one thing at a time06:36
fabbionethis is a network driver right?06:36
svenlfabbione: sure, a mips-and-ppc driver.06:38
fabbioneso in the first place i would move it to drivers/net/<where_appropriate> and do a proper Kconfig entry for it06:38
fabbionebecause a network card is initialized much later than the arch specific code06:38
svenlfabbione: Dale Fornsworth from montavista isolated all the arch independent code and moved them into arch/ppc|mips.06:39
svenlfabbione: the only thing i am really doing is setting a couple of arch-specific info, like where the NB chip is mapped, and what interrupt to use, in the ad-hoc structure that the arch-indep driver then uses.06:39
fabbionei think it is a bad idea because it will spread the code all over the place06:40
fabbionebut anyway06:40
svenlfabbione: so my only problem is to detect the northbridge through its pci id, and then set the stuff with platform_add_devices.06:41
=== ddaa [~ddaa@deep-thought.xlii.org] has left #ubuntu-kernel []
svenlarch_initcall is called before the pci tree is filled, but there should be a way to have mv643xx_eth_add_pds executed later, after the pci tree has been filled.06:42
fabbionearch_initcall is a synonimous for module_init() btw06:42
fabbione#define arch_initcall(fn)               module_init(fn)06:42
fabbionefrom include/linux/init.h06:43
fabbionethe problem is that pci code is initizialized as driver later in the boot process06:43
svenlyep, guessed such as pci_devices was empty :/06:44
zulfabbione: cool moe d06:46
fabbionei think you will also hit another problem.. PCI IRQ are assigned after PCI is initialized..06:46
fabbionethis is my best guess06:46
svenlfabbione: this is no problem, as we only tell the arch-indep code which pci irq to use, and it is dicted by what the open-firmware initializes at a lower level though.06:47
svenlfabbione: so, i understand you don't know either, and i should better wait for hch and ask him ? :)06:47
fabbionesvenl: well i am trying to help.. not knowing the arch doesn't make it easier06:48
svenlfabbione: it was no critic.06:50
svenlfabbione: actually, my only question is if there is a equivalent to arch_initcall that will happen after the pci initialization step.06:51
svenlfabbione: but i will go ask other people later if i don't find it by myself.06:51
fabbionesvenl: i think you cannot call pci_find_device at all there06:53
fabbioneif do a couple grep in platform/06:54
fabbioneyou will notice that the only driver that does it06:54
fabbioneis a mobo driver06:54
svenlfabbione: yep.06:54
fabbionein a special call only that is used to fix a pci parameter if a cache is set in a certain way06:55
fabbioneso it basically happens after the mobo and pci are initialized06:55
svenlfabbione: don't worry, i am going to ask hch as soon as he is back,06:55
fabbionei think you really can't call it there06:55
fabbionenah it's ok.. it's not like killing my time :-)06:55
svenlfabbione: i am not sure, since the code is only linked in, and its only link to the call is rge arch_initcall.06:56
svenlfabbione: so the code could be anywhere.06:56
fabbioneyes but i am more thinking in terms of code execution order06:57
svenlfabbione: the other places the code is initialized is in arch/ppc/platforms/mv64360.c, but it is a mobo driver.06:57
fabbioneyes i saw that06:57
fabbionei think it would thousand times easier to just move it drivers/net06:58
svenland same for mips, but those don't probe, they are configured at toplevel Kconfig.06:58
fabbionewhen the pci bus is initialized06:58
svenlfabbione: after Dale just removed all arch calls from there and it just got accepted upstream ? 06:58
fabbionewell he is not God you know :-)06:59
fabbioneif something cannot be done in platform/06:59
fabbione... you can guess what i mean06:59
mdzfabbione: wow, how was the kernel build time reduced for powerpc?06:59
svenlfabbione: yep, but he, hch, benh and Mark Greer told me that it was not a good idea. I guess benh is God where ppc/linux is concerned.06:59
mdzfabbione: CONCURRENCY_LEVEL?07:00
svenlfabbione: and to add to this the marvell is no pci device, it is not present in the pci tree.07:00
fabbionesvenl: ah hold on!07:00
fabbionesvenl: check the call to pci_:find_device in chestnut.c07:01
fabbioneit wants a dev structure as last parameter07:01
fabbionenot a NULL07:01
fabbionemdz: yeps :-)07:01
fabbionemdz: i set CONCURRENCY_LEVEL as Num of cpu * 207:01
fabbionemdz: i could go further since the code is ccached07:01
fabbionepro = code ccached = tons of times faster07:02
fabbionecons = no ccache = kill buildd for a little while07:02
mdzI usually do num_cpu + 107:02
mdzdid you do some tests?07:02
svenlfabbione: nope.         n = from ? from->global_list.next : pci_devices.next;07:02
fabbionemdz: yes. locally.07:03
svenlit it is null, it will searcg at the root.07:03
fabbionemdz: num_cpu * 2 is fine07:03
fabbionesvenl: yes. i didn't say it must be a filled struct..07:04
fabbionebut it might expect one there as parameter07:04
fabbionetho i am not 100% sure..07:04
fabbionebut i would still try for the sake of it07:04
svenlfabbione: ok, thanks.07:04
fabbionesorry that i can't help more :(07:05
mdzfabbione: do you use getconf to query the number of CPUs?07:05
svenlfabbione: i think that i need to investigate with Dale and/or hch about this issue, and what their intention is.07:05
svenlfabbione: no problem.07:05
fabbionemdz: no. cat /proc/cpuinfo |grep ^processor 07:06
mdzfabbione: getconf should be more portable across architectures07:06
fabbionemdz: so is /proc/cpuinfo for our arches :-)07:06
mdzand centralizes the code07:06
fabbionei will check it tomorrow07:07
fabbioneit's 13 hours that i am here07:07
fabbioneand i am kinda tired07:07
fabbionemdz: did you ever used getconf ?07:10
fabbionei can't see anything that tells me how many processors there are in the system.. well i must be very tired07:11
mdzfabbione: apt has used it forever07:17
mdzfor the same purpose07:17
mdzgetconf _NPROCESSORS_ONLN07:17
fabbioneusr/bin/getconf                                             base/libc607:19
fabbionefabbione@concordia:~ $ getconf _NPROCESSORS_ONLN07:20
fabbione-bash: getconf: command not found07:20
fabbionewow :-)07:20
fabbionemdz: what does it returns on hiperthreaded proc?07:20
fabbione1 proc07:20
fabbioneor 2 (if the hyper is 2)07:20
=== T-Bone [varenet@freebz.slashdirt.org] has joined #ubuntu-kernel
lamontmorning T-Bone 07:31
T-Bonehey ladude!07:31
fabbionehey T-Bone 07:31
T-Boneola fabbione07:32
fabbioneT-Bone: i did some syncing with Debian, including ia64 patches (up to 2.6.10-5)07:33
fabbionei am going to parse 2.6.10-6 tomorrow morning07:33
fabbionelet me know if you need more stuff that what i took07:33
T-Bonedon't think so07:33
fabbionewe are going to bump the ABI07:33
fabbionei didn't take all the debian patches07:33
fabbionethat's why i am asking07:33
fabbioneif you have time give it a check07:34
T-Boneia64 is mostly -ENOTMYPROBLEM lately07:34
fabbionei did try to level to important fixes07:34
fabbionedude.. i understand that it won't make it for hoary07:34
fabbionebut that's not a good reason to give up07:34
T-Bonei have reasons07:35
fabbionehoary kernel will be the base for bendy07:35
T-Boneit's not only a problem of making for hoary or not07:35
fabbionewell it's not up to me to convince you or anything07:35
crimsunfabbione: are we backporting from
fabbionecrimsun: we are porting whatever fix we can :-)07:36
crimsunfabbione: or will that just end up being too much of a hassle?07:36
crimsunfabbione: ok07:36
fabbioneit depends what..07:36
zulhey T-Bone 07:36
fabbionei am not going to backport a new VM right now :-)07:36
fabbioneif that's what you are asking for07:36
fabbionedinner is ready07:36
crimsunfabbione: I'm looking at critical ones, like "sis900 kernel oops fix"07:36
fabbionecrimsun: go ahead and gimme a patch before tomorrow 12:00 UTC07:37
fabbionecrimsun: send it via email or send me the mail with a url07:37
fabbioneand i am off for the evening07:37
crimsunfabbione: great, have a good one.07:37
fabbionei am just too tired even to concentrate07:37
=== fabbione &
T-Bonefabbione: it's more a question of "not having reasons to continue"07:37
fabbioneT-Bone: bendy :-)07:37
T-Bonehave a nice evening anyway :)07:38
fabbioneit's the same here for sparc...07:38
fabbionei have hard time to keep it up07:38
fabbionebut i still do it becuase i love sparc07:38
fabbioneit will make it for hoary+1 :-)07:38
zulcrimsun: heh i forgot to send fabio that patch07:38
T-Bonetrouble is i don't like ia6407:38
fabbioneeven if i will have to buy a cluster myseld07:38
svenlfabbione: the response to my problem was late_initcall, seems the various initcall are called at various moments ot the init.07:40
zulc ya fabbione 07:40
mdzfabbione: please review the new kernel changes with me prior to upload07:44
mdzfabbione: are you applying the release criteria to them?  currently we are only fixing high-impact bugs and simple, safe fixes07:44
zulheh down to 19 majors for the kernel09:21
=== T-Bone is now known as T-Gone
=== lamont [~lamont@mix.mmjgroup.com] has joined #ubuntu-kernel
=== lamont_r [~lamont@phantom.acmeps.com] has joined #ubuntu-kernel
=== T-Gone is now known as T-Bone

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!