09 Jun, 2006

1 commit

  • There's a race between shutting down one io scheduler and firing up the
    next, in which a new io could enter and cause the io scheduler to be
    invoked with bad or NULL data.

    To fix this, we need to maintain the queue lock for a bit longer.
    Unfortunately we cannot do that, since the elevator init requires to be
    run without the lock held. This isn't easily fixable, without also
    changing the mempool API. So split the initialization into two parts,
    and alloc-init operation and an attach operation. Then we can
    preallocate the io scheduler and related structures, and run the attach
    inside the lock after we detach the old one.

    This patch has survived 30 minutes of 1 second io scheduler switching
    with a very busy io load.

    Signed-off-by: Jens Axboe
    Signed-off-by: Linus Torvalds

    Jens Axboe
     

18 Apr, 2006

1 commit

  • On rmmod path, cfq/as waits to make sure all io-contexts was
    freed. However, it's using complete(), not wait_for_completion().

    I think barrier() is not enough in here. To avoid the following case,
    this patch replaces barrier() with smb_wmb().

    cpu0 visibility cpu1
    [ioc_gnone=NULL,ioc_count=1]

    ioc_gnone = &all_gone NULL,ioc_count=1
    atomic_read(&ioc_count) NULL,ioc_count=1
    wait_for_completion() NULL,ioc_count=0 atomic_sub_and_test()
    NULL,ioc_count=0 if ( && ioc_gone)
    [ioc_gone==NULL,
    so doesn't call complete()]
    &all_gone,ioc_count=0

    Signed-off-by: OGAWA Hirofumi
    Signed-off-by: Jens Axboe

    OGAWA Hirofumi
     

19 Mar, 2006

4 commits


06 Jan, 2006

1 commit

  • Unlike other ioscheds, as-iosched handles alias by chaing them using
    rq->queuelist. As aliased requests are very rare in the first place,
    this complicates merge/dispatch handling without meaningful
    performance improvement. This patch updates as-iosched to dump
    aliased requests into dispatch queue as other ioscheds do.

    Signed-off-by: Tejun Heo
    Signed-off-by: Jens Axboe

    Tejun Heo
     

22 Nov, 2005

1 commit

  • Kill the arq->state poison statement in as_add_request(), it can trigger
    for perfectly valid code that just reuses a request after io completion
    instead of freeing it and allocating a new one. We probably should
    introduce a blk_init_request() to start from scratch, but for now just
    kill it as we will be removing the as specific poisoning soon.

    Signed-off-by: Jens Axboe
    Signed-off-by: Linus Torvalds

    Jens Axboe
     

19 Nov, 2005

1 commit


08 Nov, 2005

1 commit


04 Nov, 2005

1 commit