Tuesday 3 December 2013

3.12-ck2, BFS 0.444

Here is an updated BFS patch, version 0.444:

3.12-sched-bfs-444.patch

And an updated ck tagged 3.12-ck2 patch:

3.12-ck2

The changes in this release, compared to version 0.443 and ck1 are the 2 extra patches I posted in my last announce which were designed to address various suspend to ram/disk and resume problems as discussed in previous posts. Thanks to the various people who posted bug reports and tested experimental patches along the way.

Being an even number, this is clearly a more stable patch than the last one ;)

Enjoy!
お楽しみください

32 comments:

  1. I'll give this a try on my laptop. thx

    ReplyDelete
  2. May I request that the BFS patch allows the timer frequecny to be changed to 10000Hz, like what you have done before?

    ReplyDelete
  3. Graysky one's again i got a problem with the newest kernel (linux-ck-ivybridge from repo-ck) sensors and nvidia:
    I tried to run optirun and it crashed:

    [ 81.840893] [ERROR]Cannot access secondary GPU - error: Could not load GPU driver

    [ 81.840952] [ERROR]Aborting because fallback start is disabled.

    So do you have to Re-Up the nvidia thing?

    ReplyDelete
    Replies
    1. Yes, a rebuild against 3.12.2-2-ck was needed and is now complete and online. In the future, please don't use ck's blog as I don't read it all the time. There is an an official repo-ck discussion thread where you should post. I get an email notifying me that someone posted there.

      Delete
  4. Thank you for the even number!!! ;-) Now works much better. :-)

    Besides this, does someone of you know whether the mutex-locking patch from, IIRC, Alfred Chen is still useful or maybe even needed for some configurations?

    Thank you and best regards, Manuel Krause

    ReplyDelete
    Replies
    1. I can confirm both:
      - the Chen unlock-mutex-lock.patch patches Bfs-444 patched Linux-3.12.3 with some offsets
      - the patch applied runs without problems
      Chen claimed his patch is not about performance though.

      The Chen patch I tested I put here:
      http://paste.opensuse.org/94340096

      Greetings from snowy and cold Hamburg in Germany
      Ralph Ulrich

      Delete
    2. Uups, that is odd, it seems paste.opensuse.org deletes after some hours :(
      I hope this stays longer:
      http://paste.ubuntu.com/6534536/
      Ralph Ulrich

      Delete
    3. not sure if it's Chen's patch or the switch from ck1 to ck2

      but the system really feels more responsive - also under load

      will give this more testing in the following days and see

      but I like this development =)

      Delete
    4. I really can't confirm, that this patch is relevant, neither for performance nor for interactivity with 3.12.3+BFS+BFQ.
      But I also don't see any negative impact when it's applied! :-?

      Unfortunately the search function in this blog's overall text is *eeks* unusable?!
      I haven't found in this blog, when Alfred Chen firstly introduced his patch. The first problem report I've found, so far, is
      http://ck-hack.blogspot.de/2012/08/35-ck1-bfs-424-for-linux-35.html?showComment=1345430636481#c8142340554192327796
      -- that was for kernel 3.5 / BFS 424! But when did he offer the patch? Or, did someone else make the patch (whom I haven't searched for?)?
      The only useful thing it should be good for, is to fix the circular locking issue HE had ad bootup. Just my understanding as of now.

      Another bad thing: I've made bad experiences in the last days, with my policy to always update my MESA/Xorg/graphics drivers to the newest, when they come into the openSUSE repos. These updates may (and do) affect performance much more than one is able to estimate. Especially when updating the kernel, too, and both at once. (Never change more than one parameter when testing, yes, I know...) This is sth. for kernelOfTruth perhaps. ;-)

      Best regards, Manuel Krause

      Delete
  5. Con, is there any possibility to get urw patch for BFS? There's only urw patch iself in -ck folder on server, but no patch that allows BFS to use urw.

    ReplyDelete
    Replies
    1. Add that to my time machine facilitated projects :s

      Delete
  6. O.T.: Does there today exist ONE linux GUI based download-manager that allows pattern based downloads? For instance one directory, and I only want the "*z*.tx*" files without "reindeer" in the name.
    I really drop some tears, as D4X, IIRC, did that in times long ago.

    Manuel Krause

    ReplyDelete
  7. Con, it seems that -ck breaks kernel compiling for uni-processor machines with SMP disabled with the following error:

    kernel/sched/stats.c:38:3: error: implicit declaration of function "cpu_rq" [-Werror=implicit-function-declaration]

    I've prepared experimental patch (https://gist.github.com/307dac0e59bf5e0df451), but the guy reported the issue to me hasn't tested it yet, and I have some doubts that I do things right.

    Could you please give it a look?

    ReplyDelete
    Replies
    1. Tested with no luck. Need your help.

      Delete
    2. Prepared another patch:

      https://gist.github.com/9ef1d65b05bba92aacc9

      Seems to work.

      Delete
    3. I ran across this same error on kernel 3.13.6 using the 3.13-ck1 patch, so it's still an issue.

      Delete
    4. Just encountered this error on kernel 3.12.17 on a Raspberry Pi.

      Delete
    5. Thanks to post- factum, I am able to pass the error with the link he provided. I also encountered this error on kernel 3.12 on rpi with LFS base.

      Delete
  8. @ck, your family & all other participants on here:

    Merry Christmas & a Happy New Year !!!

    Let's keep up good work in 2014.

    Best wishes, Manuel Krause

    ReplyDelete
  9. Replies
    1. 3.13 would be much appreciated indeed. Thank you for you tireless (and so often thankless!) job, Con.

      Delete
    2. Usual story, unfortunately I'm currently swamped but will get to it soon, sorry.

      Delete
    3. There are much changes in 3.13, after I ported .444 bfs to 3.13, there is huge degression, tasks take huge cpu time, and takes longer to compile kernels, suspend is also broken. Have to wait for Con's new release for 3.13.

      Delete
    4. 3.13 and 445 are... odd numbers!
      Con can get swamped until more favourable omens...

      Delete
    5. Any progress on 3.13 and the ck patchset? Alfred's indicated he's had problems already.

      Delete
    6. Seems, like aCOSwt did find the exact false words to motivate Con... ^^ ;-)

      Delete
    7. Although, it also needs to be said, that kernel 3.12.12 works very well with the still current CK/BFS and the newer BFQ v7r2.
      Best regards and wishes, Manuel Krause

      Delete
  10. post-factum already knows ;-) ...
    just to inform all you others: There's a new release out for the BFQ I/O scheduler:
    __ v7 __ (for various actual and older kernels)

    Read the announcement and preliminary release notes here: https://groups.google.com/forum/?fromgroups=#!topic/bfq-iosched/n4A5602SETA

    and find the patches here: http://algo.ing.unimo.it/people/paolo/disk_sched/patches/?C=M;O=D

    Enjoy!
    Greetings, Manuel Krause

    ReplyDelete
    Replies
    1. The BFQ team has uploaded a BUGfix release _ v7r1 _ as of this night.
      Announcement: https://groups.google.com/forum/?fromgroups=#!topic/bfq-iosched/F1Rh2sCLA8U
      You can find the most recent patches with the URL written in my previous posting.

      Sorry for writing this OT stuff related to BFQ (not BFS) on here, but I know some folks on here also believe in this different I/O scheduler, too.

      Best regards to all, also for Con's effort,
      Manuel Krause

      Delete
  11. 3.14 for the sake of pi pls con :)

    ReplyDelete