Re: dcache shrink list corruption?

[Linus Torvalds]

On Wed, Apr 30, 2014 at 4:43 PM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> OK, done and force-pushed.  Should propagate in a few...

That made it more obvious how the DCACHE_MAY_FREE case ends up
working. And in particular, mind rewriting this:

    if (dentry->d_flags & DCACHE_MAY_FREE) {
        spin_unlock(&dentry->d_lock);
        dentry_free(dentry);
    } else {
        spin_unlock(&dentry->d_lock);
    }
    return parent;

as just

    bool free = dentry->d_flags & DCACHE_MAY_FREE;
    spin_unlock(&dentry->d_lock);
    if (free)
        dentry_free(dentry);
    return parent;

instead? In fact, I get the feeling that the other case later on
really fits the same model:

    spin_lock(&dentry->d_lock);
    if (dentry->d_flags & DCACHE_SHRINK_LIST) {
        dentry->d_flags |= DCACHE_MAY_FREE;
        spin_unlock(&dentry->d_lock);
    } else {
        spin_unlock(&dentry->d_lock);
        dentry_free(dentry);
  }

ends up really being better as

    spin_lock(&dentry->d_lock);
    free = 1;
    if (dentry->d_flags & DCACHE_SHRINK_LIST) {
        dentry->d_flags |= DCACHE_MAY_FREE;
        free = 0;
    }
    spin_unlock(&dentry->d_lock);
    if (free)
        dentry_free(dentry);
    return parent;

and then suddenly it looks like we have a common exit sequence from
that dentry_kill() function, no?

(The earlier "unlock_on_failure" exit case is altogether a different case).

I dunno. Maybe not a big deal, but one reason I prefer doing that
"free" flag is because I really tend to prefer the simple case of
lock-unlock pairing cleanly at the same level. NOT the pattern where
you have one lock at one indentation level, paired with multiple
unlocks for all the different cases.

                  Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/