lkml.org 
[lkml]   [2006]   [Oct]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH take2 3/5] dio: formalize bio counters as a dio reference count

    >> +static int wait_for_more_bios(struct dio *dio)
    >> +{
    >> + assert_spin_locked(&dio->bio_lock);
    >> +
    >> + return (atomic_read(&dio->refcount) > 1) && (dio->bio_list == NULL);
    >> +}
    >
    > This function isn't well-named.

    Sure, I'll try harder.

    >
    >> @@ -1103,7 +1088,11 @@ direct_io_worker(int rw, struct kiocb *i
    >> }
    >> if (ret == 0)
    >> ret = dio->result;
    >> - finished_one_bio(dio); /* This can free the dio */
    >> +
    >> + /* this can free the dio */
    >> + if (atomic_dec_and_test(&dio->refcount))
    >> + dio_complete_aio(dio);
    >
    > So... how come it's legitimate to touch *dio if it can be freed by now?

    Indeed! This point in the patchset is retaining the existing behaviour
    where the 'should_wait' direct_io_worker() path and the
    finished_one_bio() path very carefully keep their tests in sync so that
    only one of them ends up freeing the dio.

    This particular patch doesn't change this part of the behaviour, it's
    just bringing the dec and complete that was previously hidden in
    finished_one_bio() up into its caller. So I kept the comment.

    This mess is replaced with a proper

    if (dec_and_test(dio->refcount))
    kfree(dio)

    construct in the final patch. I hope that means we don't need to patch
    in some comments that we later remove.

    - z
    -
    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: 2006-10-04 01:25    [W:2.451 / U:0.008 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site