lkml.org 
[lkml]   [2009]   [Oct]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    SubjectRe: [BUGFIX -v7] x86, MCE: Fix bugs and issues of MCE log ring buffer
    From
    2009/10/5 Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>:
    > Hi Huang,
    >
    > Huang Ying wrote:
    >> Current MCE log ring buffer has following bugs and issues:
    >>
    >> - On larger systems the 32 size buffer easily overflow, losing events.
    >>
    >> - We had some reports of events getting corrupted which were also
    >>   blamed on the ring buffer.
    >>
    >> - There's a known livelock, now hit by more people, under high error
    >>   rate.
    >>
    >> We fix these bugs and issues via making MCE log ring buffer as
    >> lock-less per-CPU ring buffer.
    >
    > Now I have a real problem on the small MCE log buffer on my new large
    > system with Nehalem which has many cpus/banks in one socket...
    > So I'd like to solve the problem asap.  I think this problem might block
    > some distros to support new processor.
    >
    > Last week I reviewed your patch again and noticed that it is doing a lot
    > of changes at once.  I suppose that this method must be one of reasons
    > why your patch seems to be so hard to review, and why it is taking long
    > time to be accepted by x86 maintainers.
    >
    > Fortunately I had some spare time so I carefully broke your patch into
    > some purpose-designed pieces.  It would be the most significant change
    > that now there are 2 steps to convert the buffer structure - 1) to make
    > it per-CPU and 2) to make it ring buffer.
    >
    > Also I fixed some problem in your patch, found on the way to make this
    > patch set.  I'll explain about my changes later using diff from your
    > change.  Comments are welcomed.
    >
    > Thanks,
    > H.Seto
    >


    Looks like the conversion of MCE log into a TRACE_EVENT is still in
    discussion whereas the
    current issues are urgent.

    So the need is to have a more stable ring buffer. But this one is an ad-hoc one.
    We already have a general purpose per-cpu/lockless ring buffer implementation in
    kernel/trace/ring_buffer.c
    And it's not only used by tracing, it's generally available.

    I think it would be nicer to use it to avoid a proliferation of
    unstable ring buffers
    inside the kernel.
    --
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2009-10-05 10:59    [W:0.023 / U:117.692 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site