• Status of d80211 inclusion (in addition to bcm43xx status).
    • Kyle will work on syncing newer d80211 stack and testing.
  • Status of Unionfs. Discuss whether to include Unionfs 2.0 in Feisty kernel.
  • Phillip to work on isolating fix to backport to our current code.
  • Discussion of bug triaging methods and policy.
    • Ben will work on wiki howto to detail our current policy.
  • Assessment of current bug list.
    • It's huge. Work will begin to lower the unconfirmed and unassigned list.
  • Open Discussion.


05:00   BenC    Just waiting on rtg
05:01   pochu   BenC: is there an agenda?
05:01   BenC    https://wiki.ubuntu.com/KernelTeam/Meeting
05:01   BenC    Agenda is there
05:02   BenC    Ok, time to start
05:02   BenC    First, I want to thank anyone from the community who is joining us
05:03   BenC    This is the first of our team meetings, and hopefully it will turn into a means of us organizing with the community better, and bring in new people to help out
05:03   BenC    In case anyone missed it, the agenda is at https://wiki.ubuntu.com/KernelTeam/Meeting
05:04   BenC    I was hoping mjg59 would be here for the first item, so we'll skip to the second in case he shows up later on
05:04   BenC    pkl_: You added the second item: Status of Unionfs. Discuss whether to include Unionfs 2.0 in Feisty kernel.
05:04   pkl_    Yeah...
05:05   pkl_    A bug has been raised against the current version of Unionfs (just getting it).
05:06   BenC    pkl_: Does the bug affect just feisty, or does this affect edgy and prior as well?
05:06   pkl_    #85145...
05:06   BenC    bug 85145
Ubugtu  Malone bug 85145 in linux-source-2.6.20 "run-init crashes when root is unionfs on nfs" [Undecided,In progress]  https://launchpad.net/bugs/85145
05:06   zul     oh this is the initramfs thingy?
05:07   pkl_    It was raised against initamfs, but it isn't anything to do with it.
05:07   BenC    So dapper is ok, it's just feisty
05:07   pkl_    It is a pure Unionfs on NFS branch bug.  Recent changes (post 2.6.18) means NFS oopses on a NULL pointer.
05:08   BenC    I wonder if this has anything to do with some of the fs changes I had to do during .19/.20
05:08   zul     couldnt you just backport the fix rather than updating to 2.0?
05:08   BenC    so is the bug NFS or unionfs? Is updating to unionfs just working around the real bug?
05:08   pkl_    This has been fixed, but, only in Unionfs 2.0.  A new call lookup_one_len_nd has been added.
05:09   pkl_    Very difficult question to answer BenC:  Both the Unionfs and NFS people has engaged in a lot of finger pointing...
05:10   BenC    Ah, upstream politics
05:10   pkl_    But the major issue appears to be recent NFSv4 changes.
05:10   BenC    pkl_: Unionfs is used quite a bit in our livecd's, and I would prefer this late in the game not to bump major versions on it
05:10   BenC    pkl_: Is the fix easily isolated?
05:11   pkl_    I'll see if I can back port the essential fix (which is to add and use lookup_one_len_nd() which passes the necessary pointer)  to 1.2 Unionfs.
05:12   pkl_    This essentially should be all that's required.
05:12   BenC    pkl_: Ok, sounds like a good plan
05:12   BenC    Looks like mjg59 wont be able to make it, so we'll cover d80211 sans his expertise
05:13   BenC    _kyle: You have some newer crack than I have in the repo...how's that coming?
05:13   _kyle   good, i'm going to push it into a branch for you to look at in about 5 seconds.
05:14   BenC    flash-kyle
05:14   BenC    _kyle: Do you have any hw to test the drivers using it?
05:14   _kyle   er, unless i accidently nuked that tree... err...
05:14   _kyle   BenC: bcm43xx
05:14   Nafallo _kyle: <3 backup
05:14   BenC    I was going to do a 2.6.20-9 upload today, but if you think you can get d80211 working, I can hold off
05:14   _kyle   i have an atheros card too, but it's minipci, and gutting my thinkpad is always a pain.
05:15   _kyle   BenC: when do you intend to upload next after this?
05:15   BenC    hopefully over the weekend
05:15   BenC    uploads have been getting too far apart, I need to make them more frequent
05:16   _kyle   ok. i have some hppa crack to push for jeff so we can have a buildable.
05:16   BenC    _kyle: FYI, the bcm43xx-dscpape driver lacks a MODULE_DEV_TABLE, so if you want to use it, just manually load it (after unloading the bcm43xx driver)
05:17   _kyle   yeah, i saw those commits.
05:17   BenC    Ok, let's move on to what I know is everyone's favorite subject...bugs
05:17   zul     yippe skippe
05:18   BenC    I didn't have time to write up a "how to triage" bugs, but there's plenty of them on the ubuntu wiki for general bug work, and the kernel isn't much different
05:19   BenC    Basically what we need to aim for is to not let the untriaged bug list get so big
05:19   zul     what is the list at now?
05:19   BenC    it's safe to assume we wont be able to fix all the bugs, because we either don't have hardware where the bug is occurring, or it's the type of bug that happened once, and will never happen again
05:19   BenC    zul: ~250 when I checked last night (2.6.20)
05:20   pkl_    that's all bugs, right?
05:20   BenC    But what I want to make sure is that we at least respond to all incoming bugs
05:20   BenC    pkl_: All open bugs on 2.6.20 kernel
05:21   BenC    For those that aren't aware, I've been working with launchpad folks to try and get some features in that will make bug triaging a lot easier
05:21   pkl_    A response along the lines of "go away we're not going to fix this" is better than no response?
05:22   BenC    pkl_: A little better wording, but pretty much :)
05:23   BenC    The main thing for a new bug right now is getting all the normal information (dmesg, screen photo of it's a solid lock with traceback, lspci, etc)
05:23   zul     please no automated bug scripts as well
05:23   BenC    9 times out of 10, the bugs just don't have enough info
05:24   BenC    Yeah, make sure you ask for information pertinent to the bug
05:24   BenC    we don't need dmidecode when someone's ext3 partition gets corrupt
05:25   BenC    The second step after getting the info is to classify the severity, there's a wiki with severity ratings (someone will paste it in a few minutes I'm sure)
05:25   BenC    Once all the info is collected, set the status (confirmed, most times)
05:25   BenC    During this process, once you take on triaging a bug, assign it to yourself
05:26   BenC    Once you start working on it, set it to In Progress, etc...Once it's in git, set it to fix committed, and once we upload it, set it to fix released
05:27   BenC    Any bug that is marked Confirmed and Importance > Medium should be fixed asap
05:27   Mithrandir      and if it's important to get fixed before release, please assign either the next milestone to it or 7.04 if it's not possible to get done before the next milestone.
05:27   BenC    either by yourself, or bringing it up to the rest of the team
05:27   BenC    Correct
05:27   Mithrandir      (7.10 for feisty+1, etc, naturally)
05:28   zul     how long per day should we spend on bug hunting (kernel team vs community)?
05:28   BenC    zul: Community is welcome to work on it as long as they want :)
05:29   BenC    I would expect core kernel team to spend some time on it every day, probably set aside 1-2 hours for just triaging, and more as needed for actual debug
05:30   fdoving looks like the bcm43xx item is done already, but for whom it may concern bug 85404 is reported by Michael Buesch concerning the state of bcm43xx in the recent feisty kernel.
05:30   Ubugtu  Malone bug 85404 in linux-source-2.6.20 "bcm43xx completely broken in feisty" [Undecided,Confirmed]  https://launchpad.net/bugs/85404
05:31   BenC    fdoving: Yes, we're aware of it, and it is fixed in git
05:31   BenC    pkl_, _kyle, rtg: Any questions?
05:31   fdoving good, thanks :)
05:31   _kyle   nyet.
05:31   rtg     Nope.
05:31   pkl_    no
05:31   BenC    Ok, let's get down to the nitty gritty of the 2.6.20 bug list :)
05:32   BenC    263 bugs, 122 unconfirmed, 233 unassigned
05:33   BenC    How many does everyone have assigned to themselves?
05:34   zul     no idea :)
05:34   rtg     I've been working on the various suspend problems, but have not assigned any to myself yet.
05:34   BenC    Only 30 bugs assigned :)
05:35   pkl_    I've working through a number, but have not assigned any to myself.
05:35   BenC    Assigning it to yourself doesn't mean you are committing to fix it
05:35   BenC    So let's work on getting that 122 and 233 down to near nothing by the end of next week
05:36   pkl_    OK.  What does assigning it signify?
05:36   BenC    the 122 is where community can help the most (confirming bugs by getting all relevant info to reproduce and/or debug it)
05:36   zul     sure..
05:37   BenC    pkl_: Means you are giving the submitter someone to talk to about the bug
05:37   pkl_    Ah good.  Terminology strikes again.  In many previous roles. assigning it basically takes an 'exclusive lock' on the bug.
05:37   BenC    you get the info you need to debug it, and keep them up-to-date on the status via the bug status. milestone and comments
05:39   BenC    I would say if the bug gets to the point where you know it's confirmed, and you cannot fix it, then assign it to the ubuntu-kernel-team in general
05:40   BenC    we'll consider those bugs for discussion at meetings, or when time permits during normal hours
05:41   BenC    Community can do that as well, if the bug has all relevant info, we can pull from the ubuntu-kernel-team assigned list to assign to ourselves
05:41   BenC    I'm going to action item me to detail all this on the KernelTeam wiki
05:42   BenC    Any questions, comments?
05:42   zul     nope
05:43   pkl_    No, I'm happy.
05:43   BenC    Ok, we have about 18 minutes left if anyone has anything to bring up
05:43   BenC    This is also a good time for community to ask the kernel-team any questions
05:43   zul     i just want to confirm the kernel for feisty+1
05:43   cbx33   can I ask a question?
05:44   BenC    zul: We'll make that decision in Seville
05:44   zul     .22?
05:44   BenC    zul: It could be .23, depends on time frame
05:44   BenC    cbx33: Sure
05:44   cbx33   I'm still running edgy, but i cannot suspend it gnome-network-manager is running with ndiswrapper
05:44   cbx33   is this common?
05:45   cbx33   it just locks with gnome-screensaver instead
05:45   BenC    cbx33: This is a better question for #ubuntu-kernel, but yes, ndiswrapper is notoriously sucky...it is running win32 code afterall
05:45   cbx33   ok
05:45   cbx33   that's fine
05:45   cbx33   another question
05:46   cbx33   I work in a school and am a contributor to Edubuntu
05:46   abogani BenC: Only one question (yes or no type): Do you think that my work on my spec (https://blueprints.launchpad.net/ubuntu/+spec/realtime) will be useful in the future? Thank you
05:46   cbx33   I want to get some interactive whiteboards working in edubuntu, how hard is it to write a device driver?
05:47   BenC    abogani: Absolutely...I have to say that the stock Ubuntu kernel cannot take on the RT patchset until it is merged with Linus' tree though. However a separate kernel like what is done for xen is entirely possible.
05:47   cbx33   abogani, don't the ubuntu-studio guys have a low latency kernel?
05:48   BenC    cbx33: It all depends on if the specs are available for the device. There's the Linux Device Drivers manual available freely on the internet
05:48   cbx33   ok
05:48   pochu   I have a little question (don't know if it's stupid): will we support wpa out-of-the-box? :-)
05:48   cbx33   thanks BenC
05:48   crimsun there is a -lowlatency* set (universe & multiverse), yes
05:48   Mithrandir      the archive team is not going to accept random kernels nilly-willy and as a start, they have to be done as patch sets against linux-source-$version
05:48   BenC    cbx33: See the KernelTeam/FAQ wiki about patches like -rt :)
05:48   crimsun pochu: (already do via NM for some chipsets.)
05:48   BenC    it details what Mith said aswell
05:48   zul     Mithrandir: that will happen for feisty+1 with xen fyi
05:49   Mithrandir      zul: good to hear.
05:49   BenC    xen, the butt of all jokes
05:49   zul     thanks..
05:50   crimsun 'unconfirmed' audio bugs against linux-source-2.6.20 are now triaged.
05:51   BenC    crimsun: BTW, you rock...thanks for keeping up with audio bugs and patches
05:51   BenC    Any more random questions?
05:52   BenC    pkl_, _kyle, rtg: Anything to add?
05:52   _kyle   nope.
05:52   pkl_    no.
05:52   rtg     Lots 'o bugs to look at.
05:53   dholbach        Was that the first Kernel Team meeting?
05:53   pochu   dholbach: yes
05:53   BenC    dholbach: The first public one
05:53   dholbach        Congratulations! I hope you'll attract a lot of community people to it!
05:53   BenC    we've been doing secret-hand-shake meetings for a few weeks already :)
05:53   BenC    dholbach: Thanks
05:53   dholbach        Coooool :)
05:54   BenC    Ok, thanks to everyone for coming, especially to the community members
05:54   zul     no probs
05:54   BenC    adios all
05:54   abogani BenC: thanks again
05:55   cbx33   it was great to see BenC
05:55   cbx33   really good idea ;)
05:55   cbx33   and I like the community bit at the end

MeetingLogs/Kernel/20070221 (last edited 2008-08-06 16:15:47 by localhost)